# WASTE BIOREFINERIES: FUTURE ENERGY, GREEN PRODUCTS AND WASTE TREATMENT

EDITED BY : Mohammad Rehan, Abdul-Sattar Nizami, Umer Rashid and Muhammad Raza Naqvi PUBLISHED IN : Frontiers in Energy Research and Frontiers in Bioengineering and Biotechnology

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-993-3 DOI 10.3389/978-2-88945-993-3

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# WASTE BIOREFINERIES: FUTURE ENERGY, GREEN PRODUCTS AND WASTE TREATMENT

Topic Editors:

Mohammad Rehan, King Abdulaziz University, Saudi Arabia Abdul-Sattar Nizami, King Abdulaziz University, Saudi Arabia Umer Rashid, Universiti Putra Malaysia, Malaysia Muhammad Raza Naqvi, Karlstad University, Sweden

"Waste Biorefinery Concept" by Mohammad Rehan is licensed under CC-BY

Energy recovery from waste resources holds a significant role in the sustainable waste management hierarchy to support the concept of circular economies and to mitigate the challenges of waste originated problems of sanitation, environment, and public health. Today, waste disposal to landfills is the most widely used methodology, particularly in developing countries, because of limited budgets and lack of efficient infrastructure and facilities to maintain efficient and practical global standards. As a consequence, the dump-sites or non-sanitary landfills have become the significant sources of greenhouse gases emissions, soil and water contamination, unpleasant odors, leachate, and disease spreading vectors, flies, and rodents. However, waste can be utilized to produce a range of potential products such as energy, fuels and value-added products under waste biorefineries.

A holistic and quantitative view, such as waste biorefinery, on waste management must be linked to the actual country, taking into account its socio-economic situation, local waste sources, and composition, as well as the available markets for the recovered energy and products. Therefore, it is critical to understand that solutions cannot be just copied from one region to the others. In fact, all waste handling, transportation, and treatment can represent a burden to the cities' environment and macro and micro economics, except for the benefits obtained from recovered materials and energy. Equally significant is a clear and quantitative understanding of the industrial, and public potential of utilizing recovered materials and energy in the markets as these can be reached without exacerbating the environmental issues using excessive transport.

The book explores new advancements and discoveries on the development of emerging waste-to-energy technologies, practical implementation, and lessons learned from sustainable wastemanagement practices under waste biorefinery concept, which will accelerate the growth of circular economies in the world. The articles presented in this book have been written by expert researchers and academics working in institutions at different countries across the world including Germany, Greece, Japan, South Korea, China, Saudi Arabia, Pakistan, Indonesia, Malaysia, Iran, and India. The research articles have been arranged into three main subject categories; 1) Resource recovery from waste, 2) Waste to energy technologies and 3) Waste biorefineries. This book will serve as an important resource for research students, academics, industry, policy makers, and government agencies working in the field of integrated waste management, energy and resource recovery, waste to energy technologies, waste biorefineries etc. The editorial team of this book is very grateful to all the authors for their excellent contributions and making the book successful.

Citation: Rehan, M., Nizami, A.-S., Rashid, U., Naqvi, M. R., eds. (2019). Waste Biorefineries: Future Energy, Green Products and Waste Treatment. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-993-3

# Table of Contents

## 1. INTRODUCTION

*06 Editorial: Waste Biorefineries: Future Energy, Green Products and Waste Treatment*

Mohammad Rehan, Abdul-Sattar Nizami, Umer Rashid and Muhammad Raza Naqvi

#### 2. RESOURCE RECOVERY FROM WASTE

#### *09 Unlocking the Potential of Biomass Energy in Pakistan*

Muhammad Saghir, Shagufta Zafar, Amiza Tahir, Miloud Ouadi, Beenish Siddique and Andreas Hornung


## 3. WASTE TO ENERGY TECHNOLOGIES


Farida Hanum, Lee Chang Yuan, Hirotsugu Kamahara, Hamidi Abdul Aziz, Yoichi Atsuta, Takeshi Yamada and Hiroyuki Daimon

*78 Thermogravimetric Characteristics and Non-Isothermal Kinetics of Macro-Algae With an Emphasis on the Possible Partial Gasification at Higher Temperatures*

Imtiaz Ali and Ali Bahadar

### 4. WASTE BIOREFINERIES

*92 Catalytic Pyrolysis of Plastic Waste: Moving Toward Pyrolysis Based Biorefineries*

Rashid Miandad, Mohammad Rehan, Mohammad A. Barakat, Asad S. Aburiazaiza, Hizbullah Khan, Iqbal M. I. Ismail, Jeya Dhavamani, Jabbar Gardy, Ali Hassanpour and Abdul-Sattar Nizami

*109 Biorefining Waste Sludge From Water and Sewage Treatment Plants Into Eco-Construction Material*

Wai Nam Cheng, Haakrho Yi, Chun-fai Yu, Ho Fai Wong, Guoxiang Wang, Eilhann E. Kwon and Yiu Fai Tsang


# Editorial: Waste Biorefineries: Future Energy, Green Products and Waste Treatment

#### Mohammad Rehan<sup>1</sup> \*, Abdul-Sattar Nizami <sup>1</sup> , Umer Rashid<sup>2</sup> and Muhammad Raza Naqvi <sup>3</sup>

<sup>1</sup> Centre of Excellence in Environmental Studies (CEES), King Abdulaziz University, Jeddah, Saudi Arabia, <sup>2</sup> Institute of Advanced Technology, Universiti Putra Malaysia, Seri Kembangan, Malaysia, <sup>3</sup> Department of Engineering and Chemical Sciences, Karlstad University, Karlstad, Sweden

Keywords: waste-to-energy, waste biorefinery, green products, biofuels, bioenergy

#### **Editorial on the Research Topic**

#### **Waste Biorefineries: Future Energy, Green Products and Waste Treatment**

Energy recovery from waste resources holds a significant role in the sustainable waste management hierarchy to support the concept of circular economies and to mitigate the challenges of waste originated problems of sanitation, environment, and public health. Today, waste disposal to landfills is the most widely used methodology, particularly in developing countries, because of limited budgets and lack of efficient infrastructure and facilities to maintain efficient and practical global standards. As a consequence, the dump-sites or non-sanitary landfills have become the significant sources of greenhouse gases emissions, soil and water contamination, unpleasant odors, leachate, and disease spreading vectors, flies, and rodents. However, waste can be a potential source of energy, fuels, and value-added products, if appropriately and wisely managed.

This Frontiers Research Topic was designed to collect new advancements and discoveries on the development of emerging waste-to-energy technologies, practical implementation, and lessons learned from sustainable waste management practices under waste biorefinery concept (**Figure 1**), which will accelerate the growth of circular economies in the world. This Frontiers Research Topic has attracted and compiled 13 top quality research and review articles. The articles have been written by researchers and academics working in institutions at different countries across the world including Germany, Netherlands, Greece, South Korea, Japan, Hong Kong, Saudi Arabia, Pakistan, Indonesia, Malaysia, Iran, and India. The editorial team of this research topic is very grateful to all the authors for their excellent contributions and making the research topic successful.

## RESOURCE RECOVERY FROM WASTE

A wide range of resources, such as energy, fuels, chemicals, and other by-products, can be recovered from waste using different technologies. The article Saghir et al. in this research topic demonstrates how energy deficient countries like Pakistan can generate significant amounts of renewable energy from biomass waste. The authors further discussed the two major classes of the techniques for the processing of biomass into bioenergy, such as (1) thermo-chemical conversion techniques including direct combustion, liquefaction, transesterification, gasification and pyrolysis; and (2) biochemical decomposition techniques including anaerobic decomposition and fermentation.

Biofuels are emerging as alternative clean fuels to replace the fossil fuels that contributed to huge environmental damage and climate change. An interesting study, Akhtar et al. showed the production of biodiesel from non-edible sterculia foetida (Bottle Tree) using CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> nanocatalysts using transesterification technology. Organic residual streams, such as food waste,

#### Edited and reviewed by:

Uwe Schröder, Technische Universitat Braunschweig, Germany

\*Correspondence: Mohammad Rehan dr.mohammad\_rehan@yahoo.co.uk

#### Specialty section:

This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research

Received: 04 May 2019 Accepted: 23 May 2019 Published: 12 June 2019

#### Citation:

Rehan M, Nizami A-S, Rashid U and Naqvi MR (2019) Editorial: Waste Biorefineries: Future Energy, Green Products and Waste Treatment. Front. Energy Res. 7:55. doi: 10.3389/fenrg.2019.00055

**6**

also have great potential as an alternative resource to produce chemicals and fuels, since they are renewable and do not compete with the human food chain. For example, medium chain fatty acids (MCFAs), such as n-caproate, are potential valuable platform chemicals that can be produced from lowgrade organic residues. The study Roghair et al. focused on the production of MCFAs from food waste by anaerobic reactor microbiomes through two subsequent biological processes: hydrolysis combined with acidogenesis and chain elongation. The authors have claimed to set the record for producing the highest concentration of n-caproate (25.7 g/L) observed in a chain elongation process to date. Moreover, the study notably demonstrates that such high concentrations can be obtained from food waste under practical circumstances in a continuous process.

### WASTE TO ENERGY TECHNOLOGIES

Energy can be recovered from waste in the form of fuels, heat, and electricity using a wide range of available waste to energy technologies. However, more research and development work is needed to optimize these technologies for achieving maximum economic, environmental and social benefits. A study Prajapati et al. published in this research topic proposed a new method for enhanced methane production from agriculture waste. The same idea can be tested and applied for other wastes such as food waste to produce cleaner biogas for energy applications.

Another study Hanum et al. highlighted the challenges and possible solutions of treating sewage sludge via anaerobic digestion. The authors presented the case study of Malaysia and how the energy can be recovered from huge waste in this country. The deep understanding of chemistry, thermodynamics, and kinetics of reactions involved in any process is crucial for process control and optimization. A study, Ali and Bahadar presented detailed kinetics of pyrolysis of macro-algae using a simple orderbased model compared with the versatile Šesták-Berggren (SB) model considering combined and multiple reactions.

### WASTE BIOREFINERIES

Waste biorefineries are attracting significant interest worldwide as sustainable waste management solutions as well as facilities to produce fuels, power, heat, and value-added products. This research topic attracted many articles on recent developments and challenges associated with waste biorefineries based on a range of waste. For example, authors have presented detailed experimental data of pyrolysis of plastic waste and way forward to develop pyrolysis based biorefineries using plastic, biomass, and other waste in Miandad et al.. The authors have also discussed the many technical, operational, and socio-economic challenges, and the potential of biorefineries to achieve circular economies.

Other interesting studies published on possibilities of waste biorefineries using waste sludge from water treatment plants (Cheng et al.), used cooking oil (Rahimzadeh et al.), co-digestion of plant biomass and poultry waste (Shah et al.), and co-digestion of catering and agro-industrial waste (Anjum et al.). These studies not only illustrate the possibilities of obtaining different products from a wide range of wastes, but some also showcase studies of corresponding countries.

There could be many ideas and prospects for optimizing waste based biorefineries. For example, the study Wambugu et al. showed the positive role of biochar in food waste based biorefineries. The biochar enhanced the performance of the anaerobic digestion system. The sustainable waste biorefinery facilities represent multifactorial systems that necessitate the organization, cooperation and the acceptance of different social stakeholders. A comprehensive study Kokkinos et al. was published in the research topic in which a novel Fuzzy Cognitive Map (FCM) modeling approach was proposed in order to analyze the socio-economic implications and to overcome multiple uncertainties occurring in sustainable WBF development and implementation.

## OUTLOOK

Waste management in a sustainable and environment-friendly manner is a crucial issue to protect the environment and human health. This research topic highlights the potential and recent developments in waste biorefineries as a sustainable solution to treat waste. Multiple resources such as energy, fuels, chemicals, and other value-added products can be recovered from waste through waste biorefineries. The research topic also highlights the challenges and potential solutions in developing waste based biorefineries from treating different types of waste such as food waste, plastic waste, organic waste, biomass, and agriculture waste. One of the most essential area to work on is to figure out smart ways of integrating different technologies under one platform of biorefinery for achieving maximum socioeconomic and environmental benefits.

## AUTHOR CONTRIBUTIONS

MR wrote the initial editorial draft. A-SN critically reviewed and corrected the initial draft. UR and MN read and approved the final draft. A-SN was responsible for editing 5 chapters, MR: 4 chapters, and MN: 1 chapter, in this research topic.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Rehan, Nizami, Rashid and Naqvi. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Unlocking the Potential of Biomass Energy in Pakistan

Muhammad Saghir <sup>1</sup> \*, Shagufta Zafar <sup>2</sup> , Amiza Tahir <sup>2</sup> , Miloud Ouadi <sup>1</sup> , Beenish Siddique<sup>1</sup> and Andreas Hornung1,3

*<sup>1</sup> Department of Chemical Engineering, University of Birmingham, Birmingham, United Kingdom, <sup>2</sup> Department of Chemistry, Government Sadiq College Women University, Bahawalpur, Pakistan, <sup>3</sup> Fraunhofer Institute for Environmental, Safety, and Energy Technology UMSICHT, Sulzbach Rosenberg, Germany*

Being a developing economy, Pakistan is facing a severe energy crisis that limits its economic development. Pakistan relies heavily on energy resources like natural gas, oil, hydropower, nuclear, coal and liquefied petroleum gas (LPG) which contribute as 48.3, 32.1, 11.3, 7.6, and 0.6% of the primary energy supply, respectively. Because of the rapidly growing population and economy, Pakistan's energy needs are huge; to accomplish these energy necessities, Pakistan is continually spending \$7–9 billion on the import of fossil fuels. To resolve the serious issues of energy availability, the Alternative Energy Development Board (AEDB) of Pakistan is currently exploring the development of renewable energy technologies in Pakistan that will be beneficial for the developing economy so that Pakistan might be able to minimize the growing energy crisis. Out of all the renewable energy resources, biomass is considered the best and most easily accessible source of energy with its unique environmentally friendly nature, constant supply, wider availability, and ease of integration into existing infrastructure. Despite the presence of an abundance of biomass energy resources, there is still a need for work on the use of these sources to produce energy. This literature review explores the availability of biomass resources in Pakistan and their potential for addressing rapidly growing energy demand in the country, which can assist in the stabilization of a Pakistan's energy demand for challenged economic development.

#### *Lucian Toma, Politehnica University of Bucharest,*

Keywords: biomass, bioenergy, pyrolysis, gasificaiton, energy, biochar

## INTRODUCTION

The current population of Pakistan is 198,139,348 (198 million) with an annual growth rate of 1.97% based on the most recent United Nations approximations.<sup>1</sup> Within various sectors of, such as domestic consumption, commercial consumption, agriculture, public lighting and bulk supply, the expenditure of energy is 45.7, 7.5, 28.1, 11.8, 0.5, and 6.4%, respectively (Aziz, 2013). Agriculture is the second largest sector of Pakistan's economy, comprising five subsectors with major crops, minor crops, livestock, fisheries, and forestry having a collective share of 21% of the total Gross Domestic Product (GDP) (Raza et al., 2012; Chandio et al., 2016). Major crops, minor crops, livestock, and forestry contribute 5.3, 2.3, 11.8, and 0.4%, respectively, to the total GDP (Chandio et al., 2016). Agriculture is the backbone of Pakistan's economy as most of the exports from Pakistan are based on agricultural products including rice, cotton, wheat, sugarcane, and many other major and minor

#### Citation:

Edited by:

Reviewed by: *Mohammad Zain Khan, Aligarh Muslim University, India*

*Romania*

\*Correspondence: *Muhammad Saghir m.saghir@bham.ac.uk*

Specialty section: *This article was submitted to Bioenergy and Biofuels,*

*Muhammad Raza Naqvi, Karlstad University, Sweden*

*Saghir M, Zafar S, Tahir A, Ouadi M, Siddique B and Hornung A (2019) Unlocking the Potential of Biomass Energy in Pakistan. Front. Energy Res. 7:24. doi: 10.3389/fenrg.2019.00024*

*a section of the journal Frontiers in Energy Research* Received: *12 November 2018* Accepted: *14 February 2019* Published: *22 March 2019*

<sup>1</sup>Worldometer: Population. http://www.worldometers.info/world-population/pakistan-population/ 2017

products (Faridi, 2012). Presently, Pakistan is facing a severe and unmanageable energy crisis that is hindering socio-economic development (Mirza et al., 2008a; Aziz, 2013).

#### The Energy Situation in Pakistan

The industrial and agricultural sectors are the backbones of the economy, and the progress of the economy is directly linked to the constant availability of energy resources (Abbas, 2015). The main hindrance to the economic development of Pakistan is its poor division of risky energy resources. The power sector of Pakistan remains unsuccessful in overcoming the energy-related issues despite being rich in coal resources (Arif, 2011). Michael Kugelman (Javed et al., 2016) acknowledged in his remarks on Pakistan's energy situation in the National Bureau of Asian Research of America, "Pakistan is mired in an acute energy crisis, one with immense implications for both the nation's floundering economy and its volatile security situation." At periods of high demand in Pakistan, the electricity supply gap is ∼6,000 MW, which accounts for 33–35% of the total expenditures (Abbas, 2015; Saeed et al., 2015). Pakistan has more than 30% of its population (55 million) with no access to electricity (Harijan et al., 2008; Saeed et al., 2015). In urban and rural areas the duration of power outages is 6–12 and 12–18 h per day, respectively (Saeed et al., 2015).<sup>2</sup> Pakistan relies heavily on energy resources like natural gas, oil, hydro, nuclear, coal, and liquefied petroleum gas (LPG), each of which contributes 48.3, 32.1, 11.3, 7.6, and 0.6%, respectively, to the primary energy supply (Rehman et al., 2013). According to recent statistics by AEDB, the share of various resources of energy, along with a contribution of 2% from renewable energy, resources involve natural gas (21%), hydropower (31%), Oil (39%), LNG (4%), and nuclear power (3%).<sup>2</sup> Moreover, relevant figures for the energy mix have also been supported by the Hydrocarbon Development Institute of Pakistan (HDIP). During 2013, dependence on various sources made up an 80.8% share of natural resources [oil (30.8%), gas (49.5%), coal (87.4%), and liquefied petroleum gas (0.5%)], with a 12.5% share coming from the remaining sources [hydro-power (10.5%), nuclear electricity (1.9%) and imported electricity (0.1%)] (Shaikh et al., 2015). Widening power supply and demand gaps from 1980 to 2012 are shown in **Table 1** (Mahmood and Ayaz, 2018).<sup>3</sup> Kessides (2013) explored the entire energy scenario of Pakistan, developing a projection of the future stating that electricity outages are expected to increase to 13,000 MW by 2020 (**Figure 1**).

#### Effects of Energy Crisis

The economic development of any country relies on energy availability as energy is the engine of the economy (Lee and Chang, 2008; Sadorsky, 2010; Kakar et al., 2011). Economical productivity exhibits an increase concomitant with that of the increase in available energy resources. Likewise, energy crisis leads to serious impacts on the economy of a country through its adverse impacts on employment, marketing, and poverty



(Cleveland et al., 1984; Kessides, 2013). Because of the heavy shortfall in energy production vs. demand, not only domestic but also various dimensions of the industrial sectors are also badly affected (Khan and Ahmad, 2008; Aziz, 2013; Saeed et al., 2015). Sixty to seventy percent (60–70%) of industrial investors have shifted their businesses to other countries, such as China, India, and Bangladesh, which is attributed to the increase in the energy shortfall, which reaches 40% (Asif, 2009; Saeed et al., 2015). In this period of industrialization and globalization, increasing energy demand cannot be accompanied by only depending on natural resources. Since the mid-1970s, natural energy resources have been running low and considerable efforts have been made to produce energy through renewable energy technologies (Mohan et al., 2006). Not only is the industrial sector affected, but agricultural productivity is also adversely affected by the rapid decrease in affordable energy availability, as the proper functioning of all the tools for agriculture—i.e., for transportation, pesticides and irrigation—are based on the application of energy (Pimentel et al., 1973). Pakistan is not only facing this severe electricity crisis, but a shortage of petroleum products and water scarcity have also become serious issues (Khan et al., 2012). Kessides (2013) in his recent publication described the severe energy crisis in Pakistan. He related severe energy issues with the lack of proper forecasting and planning together with the increasing domestic necessity. Severe blackouts in Pakistan are pushing the country toward concerning levels of poverty and create a negative impact on employment, global competitiveness and exports. The effects of energy disruption and electrical failures are so intensive that they declined the growth rate of the economy by 2% of the original figure of 6.5% per annum (GOP, 2013; Komal and Abbas, 2015). Kugelman (2013) stated that the unavailability of indispensable energy to the various economical sectors has resulted in an annual decline of up to 4% in the total GDP. Siddiqui et al. (2008) reported a survey describing the impacts of power crisis on the employment, production cost, and supply orders. Employment demand for labor in the industrial sector of Pakistan is adversely impacted. The resulting unemployment and labor loss per day varies from industry to industry and the results for different industries are shown in **Figure 2**.

#### Energy or Electricity Theft in Pakistan

Of all the electricity generated every year in Pakistan, a large segment is either stolen or lost in the distribution process. Besides the transmission losses and electricity theft, a significant fraction of the electricity consumed remains unpaid for. Islamabad Electricity Supply Company (IESCO), Lahore Electricity Supply

<sup>2</sup>Alternative-Energy-Development-Board, Ministry of Water and Power, Government of Pakistan [http://www.aedb.org/]. <sup>3</sup>https://www.eia.gov/beta/international

Company (LESCO), Gujranwala Electric Power Company (GEPCO), Faisalabad Electricity Supply Company (FESCO), Multan Electric Power Company (MEPCO), Peshawar Electricity Supply Company (PESCO), Quetta Electricity Supply Company (QESCO), Hyderabad Electricity Supply Company (HESCO), and Karachi Electric Supply Company (KESC) are the major power distribution companies in Pakistan (Jamil and Ahmad, 2014). The collective transmission loss of all the distributing companies was reported to be 20.4% by 2010, with the average loss exceeding 20–25% per year. Jamil and Ahmad (2014) submitted a detailed analysis of electricity theft and transmission losses throughout the country. During 201–13, the recorded transmission losses for the IESCO, PESCO, and KESC were 9.5, 36, and 40% respectively. They discovered that 17.8 million consumers were supplied with 92,480 GWh, and ∼18,919 GWh remained unpaid for in 2010. The illegal consumption and stealing of electrical power can be prevented by the implementation of strict laws of accountability, strategies and action plans alongside the establishment of good political governance.

#### Effects of Gas Shortage

In the previous decade, Pakistan was self-reliant in natural gas. However, currently, widespread utilization of natural gas in industry, domestic affairs, and transportation has led to a gas shortage in Pakistan. During 2001–2008, a significant elevation of 10% yearly in gas use was reported. Natural gas provides 31.9% of the overall national energy expenditure and 29% of electricity production (Awan and Rashid, 2012). The Ministry of Petroleum and Natural Resources reported that the growth rate of production of natural gas increased by 2.6%, which occurred alongside a growth rate of 4.4% in the consumption of gas. The consumption of natural gas in various sectors of Pakistan is shown in **Figure 3** (Dawood et al., 2013). The textile industry of Pakistan is one of the biggest industries of the country, covering about 60% of total exports and having a net worth of \$9 billion. This textile industry is also adversely affected by the gas crisis (Khan and Khan, 2010).

#### Effects of Oil-Shortage

The International Energy Agency has reported that the rise of oil consumption, on average, will be from 85 million barrels to 113 million barrels by 2030 (Howden, 2007). In Pakistan, the energy yearbook describes a 35.5% increase in the share of oil consumption, which surely can be a reason for oil depletion (**Figure 4**).<sup>4</sup> In Pakistan, oil utilization is equivalent to 490 kilograms, with the commercial share of oil in the production of electricity being 19% (Husain, 2010; Javaid et al., 2011). There is a huge gap between oil production and consumption, with production equaling 64,000 barrels per day and consumption surpassing 351,400 barrels per day (Sahir and Qureshi, 2008; Zuberi et al., 2013). The consumption of oil and gas has grown so fast that the remaining reserves are reported to last for 19 and 10 years respectively (Muneer et al., 2006; Asif, 2009). The depletion of natural gas and oil reserves not only affects agriculture and transport, but a reduction in industrial productivity has also been noticed (Asif, 2009). The majority of the natural oil reserves of Pakistan have been consumed, and due to this reason Pakistan is spent \$6.7 billion on oil imports during 2005–06 with an increase in growth rate of 1% per year (Ashraf Chaudhry et al., 2009; Awan and Rashid, 2012).<sup>5</sup> Currently, studied sources of renewable energy in Pakistan are solar, wind, hydro and biomass. A great number of industries in Pakistan are at present dependent on liquid fuels to fulfill their electricity and heating requirements (Saeed et al., 2015). Unfortunately, accessing universal energy is also a large concern in the country. More than 20,920 villages in the country are so far to be electrified.<sup>5</sup> This energy dilemma is further exacerbated because there is no connection to the national electricity grid for some of these remote villages. Thus, decentralized electricity provision is the only cost-effective solution for them.

### Reasons for Critical Power Outages

Continuous power interruptions negatively affect economic activities. The obvious reduction in Pakistan's economic productivity is due to power cuts that exceeded 8,000 MW until 2012. Experts analyzed various aspects of the critical power situation in Pakistan and provided reasons for these drastic shortfalls; one of these is that the relevant organizations and institutes have failed to enhance electricity generation and capacity according to supply demands, which leads to drastic energy deficits. Other major reasons include the decline in natural energy reserves, severe water shortfalls, rapidly increasing demand, excessive dependency on imported petroleum products, circular debt and other significant political and governance issues (Malik, 2012; Munir and Khalid, 2012; Nawaz et al., 2013; Rashid Amjad, 2015). Mismanagement and poor administration are also significant factors behind energy issues due to nepotism, corruption, incompetent generation or limited capacity addition, inefficient and outdated transmission systems, financial malefaction, non-optimal tariffs and reliance on expensive fuels (Malik, 2012; Nawaz et al., 2013; Ullah, 2013; Ahmed et al., 2015; Rashid Amjad, 2015). In general, the unskilled and undereducated political leaders and policymakers, low investments by the government, lacking proper vision and accountability, and instability of governance have all led to the power crisis in the country. Delays in payments by the government to the relevant power generation agencies is another huge issue that has lead to the growth of this crisis (Malik, 2012; Munir and Khalid, 2012; Kugelman, 2013; Rashid Amjad, 2015).

## The Policy for Energy Production

Efficient policies and proper planning can prevent severe power outages and expanding crisis. Pakistan needs good and efficient governance for broad-based planning to overcome the energy shortfalls. In the designing of reliable policies, many factors are taken into consideration, such as valid and reliable forecast for future demand on the basis of past trends, appropriate development and assessment of technologies, energy conservation, improved policies of long-term and short-term planning to compensate for the energy demand and effective project management (Mathur, 2001; Chikkatur et al., 2009; Østergaard and Sperling, 2014). There are numerous concerns for policy-makers in the formation and management of long-term reliable energy policies, such as energy security, climate change and natural fuel depletion (Turton and Barreto, 2006; Østergaard and Sperling, 2014; Overland, 2016). Energy security refers to the availability of affordable energy without interruptions, which is crucial for economic development (Turton and Barreto, 2006; Hogan et al., 2007).

Reliable energy policy and planning take into account the following important factors, such as the following (Mathur, 2001; Qureshi, 2009; Balat, 2010; Valasai et al., 2017):


Various past governments of Pakistan have also put forward a number of different policies which were reshaped occasionally, including the national energy policies of 1994, 1995, 1998, 2002, 2005, 2006, 2010–2012, and 2013 (**Table 2**) (Munir and Khalid, 2012; Dawood et al., 2013; Abbas, 2015; Valasai et al., 2017).

For bridging the supply and demand gap, the Government of Pakistan (GOP) has recognized alternative and renewable energy (ARE) sources as one of the best preference (Aziz, 2013; Saeed et al., 2015). Pakistan had made a strategy to add a minimum

<sup>4</sup>Pakistan Energy Year Book 2015 published by Hydrocarbon Development Institute of Pakistan (HDIP) MoPNR, Government of Pakistan.

<sup>5</sup>An overview of Fossil Fuels in Pakistan PEY, 2004 by Ministry of Petroleum and Natural Resources, Government of Pakistan.

of 9.7 GW of renewable energy by 2030 in the national energy mix. The industrial sector contributes about 40% of the country's GDP. But owing to the deficiency in energy, a large number of these industrial units are dependent on expensive and polluting diesel generation units to meet their energy demand.<sup>5</sup> Various industries, such as rice mills, wood processing mills, brick kilns, cement kilns, and metal processing are heavily dependent on heat energy. They frequently burn fossil fuels and some biomass to sustain their energy needs.<sup>5</sup> So now, Pakistan is in dire need of advanced conversion technologies, such as biomass combustion, gasification and pyrolysis. Such technologies have been effectively installed in various countries around the globe (Mirza et al., 2008a; Saeed et al., 2015).<sup>5</sup> The global daily expenditure of oil currently stands at 85 million barrels, which will be increased to 113 million barrels by 2030 (Howden, 2007). Energy resources that are used at present will not be able to cope with future energy necessities because Pakistan's fossil fuel reserves are diminishing rapidly and the demand of energy is expected to exceed 66,000 MW or 66 GW by 2030 (Howden, 2007; Zuberi et al., 2013). Modern society and economic development would not be realistic without energy; thus, a continuous decline in fossil fuels has forced the relevant agencies to plan alternative ways to produce sustainable energy (Qureshi, 2009; Valasai et al., 2017). The energy produced by renewable sources or



inexhaustible natural resources is usually referred to as renewable or sustainable energy (Howden, 2007; Ashraf Chaudhry et al., 2009).<sup>2</sup> Renewable energy sources including solar, wind, biomass and others are prevalent in Pakistan (Raja and Abro, 1994; Farooq and Kumar, 2013).<sup>2</sup> As stated in a recent survey, the extent of power generation by renewable energy sources in Pakistan is accounted for by wind (340,000 MW), solar (2,900,000 MW), hydro (small/mini, 3,000 MW), and bagasse cogeneration (2,000 MW) (Syed et al., 2014).<sup>2</sup>

### RENEWABLE OR SUSTAINABLE ENERGY RESOURCES IN PAKISTAN

Pakistan can resolve energy-related concerns by utilizing renewable energy resources rather than relying on conventional fossil energy resources (non-renewable). Sheikh (2010) described in his study that Pakistan pays 60% of the foreign exchange on the import of expensive fuels, which can be saved by diverting to renewable energy resources. The renewable energy resources of Pakistan are solar, wind, hydro, geothermal and biomass (Sheikh, 2009, 2010; Rafique and Rehman, 2017).

#### Wind-Power

The coastal area of Pakistan is spread over 1,050 km (250 km in Sindh and 800 km in Balochistan) from the Indian border in the east to the Iranian border in the west (Bhutto et al., 2013).<sup>6</sup> According to an estimation of AEDB, Pakistan has the potential to produce 360 GW, but currently, Pakistan ranks 44th in generating wind power, with a total installed capacity of 106 MW (Bhutto et al., 2013; Khahro et al., 2014a). The estimated wind speed on the coastlines of Pakistan ranges between 5 and 7 m/s which leads to wind power generation at the coastal areas of Sindh and Balochistan (Karachi, Ormara, Jivani, Pasni, Baburband, Kati Bandar, Gharo, and many others), and wind power plants have been installed (Ahmed et al., 2006; Sheikh, 2009; Ullah and Chipperfield, 2010; Khahro et al., 2014a,b; Siddique and Wazir, 2016). Wind power plants with energy capacities of 4 KW and 20 KW have been installed at various locations on the coastline of Sindh and Balochistan (Ahmed et al., 2006). Ahmed et al. (2006) reported that the generators of 4 kW capacities work efficiently throughout the year, but there are only a few locations, such as Pasni and Jivani where the 20 KW plants can work because they need high wind speed for operation. The Pakistan Meteorological Department has reported the wind power potential of 45 different Pakistani locations on the coastline. The resulting measures are shown in **Table 3** (Harijan, 2008; Harijan et al., 2009, 2011). Despite having considerable wind power generation capacity, regrettably the wind power share has no significant contribution to the total energy mix, but wind energy can be employed in the development of various rural and remote areas in Pakistan (Ahmed et al., 2006). Wind power generation can be enhanced by the formation and proper implementation of long-term productive policies, significant investment and proper vision (Mirza et al., 2007).

### Solar-Power

Pakistan possesses exceptional solar potential, which is suitable for decentralized commercial utilization of energy. Therefore, solar energy could be the best option to balance the increasing supply and demand energy gap (Khalil and Zaidi, 2014). Solar rays occur on 95% of the surface of Pakistan, which is about 1,500–3,000 h per annum with 5–7 kWh/m<sup>2</sup> /day of annual average solar radiation, accounting for 200–250 watt/m<sup>2</sup> or 6,840–8,280 MJ/m<sup>2</sup> /day (Mirza et al., 2003; Ashraf Chaudhry et al., 2009; Solangi et al., 2011; Harijan et al., 2015). According to the solar map of Pakistan issued by NREL and USAID, the province of Balochistan shares more than 5–7 kWh/m<sup>2</sup> /day of annual average global insolation, with an energy potential of 18–25 MJ/m<sup>2</sup> /day as the sun warms the surface of Balochistan for 6–8 h daily (Sheikh, 2009). Adnan et al. (2012) presented a report stating that solar irradiation intensity was observed to be varied, as the irradiation intensity exceeded 200 W/m from February–October in province of Sindh, March–October in Balochistan, April–September in the various regions of KPK, Gilgit-Baltistan, Azad Kashmir, and March–October in the areas of Punjab. Solar energy resources can be exploited in rural and urban areas for an assortment of applications in two major fields, including photovoltaic (PV) and solar

<sup>6</sup>Wildlife of Pakistan. Introduction to Pakistan: Coastline WoP: Available online at: http://www.wildlifeofpakistan.com/IntroductiontoPakistan/coastlineofPakistan. htm



thermal (for instance solar water heating, steam generation, solar cookers, solar dryers and solar desalination) (Mirza et al., 2003; Ashraf Chaudhry et al., 2009). Solangi et al. (2011) reviewed the complete profile of solar power units' installment in Pakistan. They stated that since the 1990s, Pakistan had established 18 photovoltaic projects with 440 kW capacities that afterward diminished due to the inadequacy of technical expertise and mismanagement. Other solar power units of photovoltaic and solar thermal were installed by the relevant companies, with <1,000 kW in photovoltaic and 10,000 kW solar thermal power, which is considerably less than the estimated potential of solar energy generation in Pakistan. There are many reasons behind the current regrettable solar energy scenario in Pakistan, for instance lack of managing and operating proficiency, appropriate policies, awareness in remote areas and lack of sufficient investment together with the extravagant PV technology (Sheikh, 2010; Solangi et al., 2011); (Khan and Pervaiz, 2013).

#### Hydropower

Hydropower, as the name suggests, is the form of energy that is interconnected with the force of flowing water (potential energy), which can be converted into electricity for diversified applications (Wagner and Mathur, 2011). Globally, hydro-power shares a great deal of about 20% contribution in the total electricity production (Boyle, 2004; Asif, 2009). Total water resources (such as rivers, glaciers, snow melt, etc.) of Pakistan have been estimated to be 400,000 Km<sup>2</sup> with an anticipated potential of about 42 GW (Ashraf Chaudhry et al., 2009; Sheikh, 2009). Regardless of being rich in natural water resources, the contribution of hydropower to the total energy mix of the country is declining day after day due to many reasons (such as climate change, short-term policies, proper execution of project, land clearance, resettlement site development, long construction periods and many other financial and political issues). Until now, only 15% of the total hydropower potential, accounting for 6.5 GW, has been installed (Asif, 2009; Bhutto et al., 2012). The generation and exploitation of hydropower have reduced since the 1970s, from 70 to 33% in 2006. Various major hydropower projects are operational, for instance Tarbela, Mangla, Warsak, Chashma, Ghazi, and Barotha contributing about 6,355 MW (Mirza et al., 2008b). The major hydropower projects including Tarbela, Mangla, and Chashma have recently been reported as displaying declines in their storage capacity by 20%, yet alternate undertakings, for instance Ghazi and Barotha (introduced in 2004), are noteworthy contributors toward the complete hydropower capability of Pakistan (Asif, 2009; Bhutto et al., 2012). Hydropower development in Pakistan is shown in **Figure 5**. 7 In the Indus basin of Pakistan, 800 potential sites have been recognized, with a hydroelectric power of 60 GW. Only the hydropower plants at 134 sites are operational, amounting to 11% (6,720 MW) of the total, but the remaining 89% of this project are under implementation and completely undeveloped (Siddiqi et al., 2012). Pakistan can boost the hydroelectric share of the nation's overall energy output by effectively working on the following major projects: Kalabagh, Bhasha, Bunji, Dasu, Kohala, Patan, Neelam-Jhelum, Thakot, Munda, and Ahori, with energy generation capacities of about 3,600–3,800, 4,500–4,600, 5,400, 3,800, 1,100, 2,800, 950, 2,800, 750, and 600 MW, respectively (Asif, 2009; Malik and Sukhera, 2012). Besides these major projects, there are many mini- and micro-hydropower sites in Pakistan, with an identified potential of <5%, which can be employed to harness <1,000 MW of hydroelectric power (Sheikh, 2009, 2010).

#### Geothermal

Geothermal energy is another form of clean, renewable energy that is obtained from the heat energy available on the earth. There are various sources of geothermal energy, such as hot springs, volcanoes, fumaroles, and geysers. Pakistan also has great potential for electricity production from geothermal energy. There are numerous hot springs with temperatures of 30–170◦C in the regions of Balochistan, Sindh, Karachi, Azad Kashmir, and KPK that can be exploited for the generation of geothermal energy. Despite having plentiful geothermal resources, there is no substantial work or planning to produce usable energy from these resources in Pakistan (Sheikh, 2009, 2010; Awan and Rashid, 2012; Malik and Sukhera, 2012).

<sup>7</sup>Hydropower Development in Pakistan [http://www.wapda.gov.pk/index.php/ projects/hydropower-development-in-pakistan].

## ENERGY FROM BIOMASS

Biomass energy has the potential to play a substantial role in combating the expanding energy crisis in Pakistan owing to the massive and diversified biomass potential of the country. Biomass is counted as a clean energy source since it recycles carbon dioxide through photosynthesis during biomass growth (Mirza et al., 2008a). Biomass can be used to produce electricity, thermal energy, and various chemicals. Zuberi et al. (2013) depicted the diverse advantageous aspects of the biomass resources in Pakistan including economic, environmental and employment benefits. Understanding the different properties of the biomass resources is crucial for the success of a biomass-to-energy project. The properties include calorific value, moisture content, ash content, carbon content, hydrogen content, cellulose, hemicellulose and lignin (McKendry, 2002a).

### Indigenous Biomass Resources in Pakistan

Pakistan is gifted with biomass resources; however, its overall energy potential is yet to be scientifically determined. Biomass is not only limited to habitat and animal waste but also includes feedstocks, such as agricultural stalk, straw and trash, agro-industrial bagasse, paddy husks and shells, forestry and woodchips, barks and trims, and riverside greens (Farooq and Kumar, 2013; Naqvi et al., 2018).<sup>2</sup> Agricultural waste, agroindustrial waste or lignocellulosic waste and wood-based residues are presently estimated to be around 20,494, 25,271, and 1,121 million tons, respectively (Anwar et al., 2014).<sup>2</sup> The total estimated biomass potential of Pakistan is 50,000 GW h/year (Farooqui, 2014), which contributes up to 36% of the total nation's energy scenario (Asif, 2009). The various types of biomass feedstock/sources available for power generation are shown in the following **Figure 6** (Asif, 2009; Naqvi et al., 2018).<sup>2</sup>

## Potential of Agricultural Residues

Owing to the huge agricultural sector, Pakistan produces a large number of agricultural residues including wheat husks, rice husks, cotton sticks and sugar cane residues. In addition to nonwoody agricultural residues, the woody portion also contributes a great share to the production of energy (Mirza et al., 2008a; Aziz, 2013; Zuberi et al., 2013). The total land structure of Pakistan is presented in **Figure 7** (Book, 2017).

Cotton is being planted on a large scale with an average crop production of 2,00,000 tons, 2.67 million hectares (11% of the agricultural land) during 2011–12. From 2011 to 2012, the cotton stalks obtained were about 5,898,771 tons, containing an energy potential of 614 GWh. Pakistan is positioned as the fourth largest sugarcane producing country. According to a survey from 2011 to 2012, Pakistan's sugar cane crop cultivation was measured to be 63,920,000 tons, with 5,752,800 tons of residue possessing a power potential of 9,475 GWh (**Table 4**) (Naqvi et al., 2018). Sugar mills contribute ∼2,000 MW in the total national energy (Naqvi et al., 2018).

Regarding the cultivation of wheat, Pakistan is listed as the world's third largest country in the production of wheat, with its total share of the agricultural sector reaching 10.1%, and it has been identified as a potential source to produce clean energy, such as bio-oil (Iqbal et al., 2018). Together with these major crops, other crop residues (maize, rice, gram, etc.) also encompass a significant share of the total of agricultural residues that are available for energy conversion (Farooq and Kumar, 2013). The total residue production from the agriculture sector was estimated to be 62 million tons, with 2% growth rate per

annum since 2000, which became 81 million tons per annum in 2012–13 (Farooq and Kumar, 2013; Ghafoor et al., 2016). Woody residues are utilized extensively in the household and cottage industries. Forest covers 5.2% (4.224 million hectares) of the available land and provides about 80% of the total generated bio-energy (Sheikh, 2009; Naqvi et al., 2018). Banana production is also considered as a source of renewable energy in Pakistan. According to one estimation, there are 173,000 banana trees per km<sup>2</sup> countrywide, with an average caloric value of 17.8 MJ/k. Cutting of the banana trees is usually carried out three times per annum, resulting 7 kg of heavy residues per tree (Saeed et al., 2015). Saeed et al. (2015) presented a comprehensive study on the estimated energy production from banana cultivation in Pakistan (**Table 5**). They stated that the share of banana tree remains in the annual consumption of electricity is 2.4% and, by efficiently processing agricultural waste materials, 56% of the total electricity requirement of the nation can be achieved.

#### Potential of Animal Manure

The bovine population of Pakistan is estimated to be 67,294,000 (**Table 6**) with an annual growth rate of 4% (Amjid et al., 2011; Aziz, 2013; Sakib Sherani, 2009-10). The resulting animal manure measures at 368,434,650 kg, which is deployed in the production of biogas with an energy potential of 23,654 GWh (Aziz, 2013; Naqvi et al., 2018). Livestock animals (such as cows, buffaloes, sheep, cattle, etc.) account for 55.3% of the total agricultural sector, which was elevated to 55.4% a year later, with total contributions of 11.9 and 12.1% to the GDP, respectively (Uddin et al., 2016).

### Potential of Municipal Solid Waste (MSW)

Being a populous country, urban areas of Pakistan produce more than 64,000 tons of municipal solid waste daily, which comprises

TABLE 4 | Major crop contributions to Pakistan's total energy content.



#### TABLE 5 | Energy production from banana crop per year (Saeed et al., 2015).

TABLE 6 | Livestock population (in millions) in Pakistan from 2010 to 2013 (Uddin et al., 2016).


both organic and inorganic components and can be efficiently processed into various energy forms with an average calorific value of 6.89 MJ/kg (Harijan et al., 2008; Aziz, 2013; Valasai et al., 2017). In the major cities of Pakistan (such as Karachi, Islamabad, Lahore, Multan, Faisalabad, Peshawar, Gujranwala, Rawalpindi, and Quetta), the estimated MSW production is about 7,121,626 tons (Aziz, 2013). MSW can contribute 13,900 GWh of energy production (Naqvi et al., 2018). Pakistan is confronting issues of waste (damped papers, plastics, metals, rubbers, glass, textile effluents, cardboards, bones, leaves, grass, straws, wood, fodder, and so on) mismanagement as the effluent is not disposed of properly. The typical composition of the MSW is shown in **Figure 8** (Zuberi and Ali, 2015). According to an approximation, 9.8 million tons of MSW were produced in 2005, 57% of which was collected for utilization in the production of electricity (Farooq and Kumar, 2013). Muhammad Khalid Farooq and S. Kumar (Farooq and Kumar, 2013) reported the complete scenario of MSW in Pakistan shown in **Table 7**.

## METHODOLOGIES FOR BIOMASS-TO-ENERGY CONVERSION

There are two major classes of the techniques for processing of biomass into bioenergy, such as (i) thermo-chemical conversion techniques including direct combustion, liquefaction, transesterification, gasification and pyrolysis; and (ii) biochemical decomposition techniques including anaerobic decomposition and fermentation, as shown in **Figure 9** (McKendry, 2002a,b; Naqvi et al., 2018). Another unique set of techniques receiving attention is bioelectrochemical systems (BES). In bioelectrochemical techniques, microbes are utilized to yield valuable inorganic and organic byproducts along with power generation by oxidization of biological substratum (Pant et al., 2012; Kelly and He, 2014; Wang et al., 2015; Bajracharya et al., 2016; Khan et al., 2017). BES is a potential platform for producing sustainable bio-energy by minimum power consumption and it can also be applied as a robust tool for remediation of contaminants, such as carbon dioxide, nitrates, phosphates, and micropollutants in wastewater, etc. (Jin and Fallgren, 2014; Kelly and He, 2014; Sultana et al., 2015; Wang et al., 2015). Amongst various thermo-chemical processes, the most preferable are gasification and pyrolysis. Biomass pyrolysis has drawn much consideration recently owing to the production of bio-oil, biochar, and hydrogen-rich fuel gas (Ferdous et al., 2001; Li et al., 2004; Goyal et al., 2008). Pyrolysis of biomass takes place at elevated temperatures above 400◦C in the absence of oxygen or air, and the nature of the resulting products is determined by the chemical composition of the biomass and reaction conditions of the pyrolysis (Li et al., 2004; Naqvi et al., 2018). The efficiency of the pyrolysis process mainly depends upon various factors, such as particle size, temperature, heating rate, residence time, biopolymer composition, and type of catalyst (Li et al., 2004). The process of pyrolysis has certain limitations including the production of unstable crude bio-oil, which is comprised of high contents of water and oxygen and needs to be further upgraded for direct use in combustible engines. Moreover, an air decontamination installation is essential as the pyrolysis process generates a high concentration of detrimental gases, such as carbon dioxide, carbon monoxide, etc. As a result of pyrolysis, the delivered ashes are reported to have high heavy metal content that is regarded as hazardous waste (Zaror and Pyle, 1982; Serio et al., 2001; Yaman, 2004; Jahirul et al., 2012). Sometimes, biochemical conversion methodology is applied to wet biomass (anaerobic digestion, ethanol fermentation, lignocellulosic conversions) for biomass energy conversions (Goyal et al., 2008; Naqvi et al., 2018).<sup>2</sup> The anaerobic digestion process is well-recognized in the generation of biogas from biomass including MSW, remnants of fruits, vegetables, leaves, grasses, woods, weeds, as well as marine and freshwater biomass. As an outcome of this process, the biogas obtained is rich in methane (60%) together with hydrogen and carbon dioxide, which can be utilized for numerous applications, such as transportation fuel, heating fuel, as a substitute to natural gas and so on (Jones and Ogden, 1984; Chynoweth, 1987; Gunaseelan, 1997; Chynoweth et al., 2001). Almost all types of dry biomass can be effectively utilized as fuel in combustion-based processes with the limited end-use of products. For gasification, woody

biomass—for example husks and wood chips—are good fuels in dry form. Gasification of biomass is carried out in the presence of oxygen, air and/or steam at temperatures exceeding 700◦C with or without a catalyst, which subsequently generates a syngas consisting of hydrogen, methane, carbon dioxide, carbon monoxide, nitrogen and water vapor (Nunes et al., 2016). Various reactor configurations are utilized, such as fixed bed (updraft or downdraft) and fluidized bed (bubbling or circulation) reactors. Moreover, with the application of gas turbines or diesel generators, the generated gases can be converted into electricity (Chynoweth et al., 2001). The complete process of biomass conversion to syngas consists of different steps (**Figure 10**) (Kumar et al., 2009). Some of the semi-woody biomass, such as corncobs and corn stalks, can also be used for gasification. Leafy biomass, such as straw and stalks need to be densified (pellet) for gasification (Demirbas et al., 2009; Kumar et al., 2009; Farooq and Kumar, 2013).<sup>2</sup> An alternative sustainable solution of solid waste management including agricultural waste materials and animal manure is composting, which is frequently practiced in various underdeveloped countries including Pakistan. In the process of composting, the organic matter is subjected to aerobic or anaerobic decomposition by microbes, which act as an effective fertilizer or soil conditioner for the agricultural land. Moreover, composting results in the production of harmful greenhouse gases—for instance, methane, ammonia, nitrous oxide, and carbon dioxide—which, together with deteriorating the atmosphere, emits an unpleasant odor and causes contamination of the underground water sources (Inbar et al., 1993; Smet et al., 1999; Alfano et al., 2008; Ngoc and Schnitzer, 2009; Shen et al., 2011).

### Biomass Composition

The woody biomass is categorized into three types softwood, hardwood, and eucalypt. Lignocellulosic biomass primarily consists of three natural polymers (lignin,

TABLE 7 | The potential of Municipal Solid Waste (MSW) in Pakistan (Farooq and Kumar, 2013; Valasai et al., 2017).


cellulose, and hemicelluloses). The chemical composition of different biopolymers of wood is given in **Table 8** (Papari and Hawboldt, 2015).

#### Biomass Products

The final product of pyrolysis of biomass depends on feedstock that can be sourced from agricultural waste, such as rice, wheat, sugar cane, and forestry residues, such as sawdust, bark, wood chips, shavings, and algae. Three main products obtained from biomass are bio-oil, biochar, and non-condensable fuel gas, in addition to solid pellets, chemicals and biofuels (Papari and Hawboldt, 2015).

## Bio-Oils

During pyrolysis, the breakage of bonds between the biopolymers (lignin, cellulose, and hemi-cellulose) leads to volatile vapors being produced. These vapors are condensed to form a liquid called bio-oil. Fast pyrolysis leads to higher bio-oil yields and slow pyrolysis to lower yields. Bridgwater (2012) and Duman et al. (2011) describe the bio-oil as a complex mixture of different organic compounds, which requires blending with conventional fuels or hydro-deoxygenation to make it usable as a liquid fuel. Bio-oil is a complex mixture of around 300 different compounds as stated by Zhang (Zhang et al., 2007). Various researchers (Oasmaa and Kuoppala, 2003; Oasmaa and Meier, 2005; Mullen et al., 2010) have described the weighted

percentage of compounds present in bio-oil to include water (10–30%), aldehydes (1–17%), acids (3–10%), carbohydrates (3– 34%), phenolics (2–15%), alcohols (<4%), ketones (2–11%), and other unclassified compounds (5–58%). Calorific value, pH and viscosity of bio-oil are shown by Zhang (Zhang et al., 2007) to vary between 16 and 30 MJ/kg, 2.5–3.4 and 40–100 (cP), which makes it an inferior fuel compared to diesel.

Pyrolysis of biomass generates bio-oil which is a CO2-neutral fuel with limited contents of sulfur oxide gases and can be readily stored and transported (Mohan et al., 2006). Bio-oils have been catalytically improved to high-quality hydrocarbon fuels and can be used as an alternative to petroleum fuels, which are of exceptionally high cost (Czernik and Bridgwater, 2004; Mohan et al., 2006). Iqbal et al. (2018) presented a study based on the production of bio-oil from the wheat residues by using the fast pyrolysis methodology. They studied the energy products of the fast pyrolysis of wheat stalks under a temperature range of 300–650 degree Celsius, which resulted in the phenolics, linear ketones, anhydrosugars, ketones, furan, anhydrosugar and acids, with percentage shares of 27.32, 8.53, 7.66, 8.53, 10.74, and 11.12%, respectively (Iqbal et al., 2018).

TABLE 8 | Chemical composition of different biopolymers of wood (Papari and Hawboldt, 2015).


#### Biofuels

Biofuels are harnessed from food crops that require, for instance, bio-ethanol from corn, wheat or sugar beet, and biodiesel from oil seeds (such as rapeseed, soybean, palm, sunflower, and so forth) (Demirbas, 2011; Naqvi et al., 2018). Some of the biofuels are highly controversial as they use land that is used for growing food. Forestry and agricultural remains along with the MSW can also be exploited in the production of biofuels (Balat, 2007). Another classification of bio-fuels presented based on the conversion technology including first generation biofuels,

second-generation biofuels, third generation biofuels, and fourth generation biofuels as shown in **Figure 11**. Variant products obtained by using first-generation biofuels are bio-alcohols, vegetable oil, biodiesel, bio-syngas, and biogas. Bio-alcohols, bio-oil, bioDMF, biohydrogen, and bio-Fischer–Tropsch diesel are generated by employing the second-generation biofuel technology. Third-generation biofuel technology results in the production of vegetable oil and biodiesel, and the fourth-generation technology gives bio-gasoline (Demirbas, 2011).

Vegetable oils and animal fats are also frequently deployed nowadays in the production of biodiesels (a biofuel) through trans-esterification, which is used in compression–ignition engines or diesel engines. Low sulfur content, flash point, aromatic content, and biodegradability make biodiesel superior to diesel fuel (Mirza et al., 2008a). Currently, Pakistan possesses the largest stocks of the methanol and ethanol, which react with vegetable oil to form bio-diesels. Pakistan is estimated to have the potential to produce 56 million tons of bio-diesels, which is far greater than the required amount (8.5 million tons). It has been reported in the literature that Pakistan has the potential to produce ∼930,000 tons of edible oils, which is expected to be elevated in the future (Khan and el Dessouky, 2009).

## Biogas

Biogas is generated by using the microbially-controlled anaerobic digestion process from waste byproducts including animal dung, urban waste, and crop residues with high moisture content. On average, biogas contains methane (50 to 70%), carbon dioxide (30 to 50%) and traces of other gases, such as hydrogen, hydrogen sulfides, moisture, and siloxanes. Biogas plants of varying scales are deployed in domestic biogas production for cooking and in industrial installations to generate electricity. In the literature, the reported calorific value of biogas is 21–24 MJ/m<sup>3</sup> (Sheikh, 2010; Amjid et al., 2011; Jiang et al., 2011; Saleh, 2012). Rural areas of Pakistan are replete with a variety of biomass resources (animal dung waste, MSW, agri-industrial, and agricultural waste), which can be used to produce about 12 million cubic meters of biogas per day, which is enough to fulfill the energy requirements of 28 million rural people. Currently, Pakistan possesses ∼5,357 operational biogas units, with a varying production capacity of 3–15 m<sup>3</sup> /day; however, the targeted total biogas potential is 12–16 million m<sup>3</sup> /day (Mirza et al., 2008a; Sheikh, 2010; Javed et al., 2016; Naqvi et al., 2018). In Pakistan, the installment of biogas plants was started in 1974, and from then 4,137 plants were installed until 1987, but the growth rate of this technology became extremely slow, resulting in an increase of the figure to only 6,000 installations by 2006. Nevertheless, new policies are being developed by the government to install 3,00,000 biogas units throughout Pakistan within 10 years. This program was initiated in 2009, but only 5,360 plants have been successfully installed as of 2014 (Saleh, 2012).8,9 Biogas is also used to produce electricity in Pakistan on a small scale, as 92 biogas plants are working in this regard resulting in the generation of 790 KW electricity (Uddin et al., 2016). There is significant potential that remains to be explored to install biogas plants for electricity production.

#### Biochar

Biochar is a charcoal-like substance (Chynoweth et al., 2001) produced by the thermochemical conversion of organic matter, such as crop residues, various grasses, and agricultural plant residues. Char can be produced either by heating biomass with or without air or oxygen (pyrolysis) by processes known as gasification and pyrolysis (run in the regime that leaves charcoal residue) (Jones and Ogden, 1984; Chynoweth et al., 2001).

Carbon is found in all living organisms. It is the major building block for life on Earth. Carbon exists in many forms, predominately as plant biomass, soil organic matter and as the gas carbon dioxide (CO2) in the atmosphere and dissolved in seawater (Demirbas et al., 2009). Carbon dioxide (CO2) is a potent greenhouse gas along with other gases, such as methane (CH4) and nitrous oxide (NOx) (Jones and Ogden,

<sup>8</sup>EKN-RSPN Pakistan Domestic Biogas Programme (PDBP): [http://www.rspn. org/index.php/projects/completed/ekn-pdbp/].

<sup>9</sup> (PDBP) E-RPDBP: [http://www.rspn.org/wp-content/uploads/2014/09/ Newsletter-Jan-Mar-2012.pdf].



1984; Demirbas et al., 2009; Kumar et al., 2009; Ngoc and Schnitzer, 2009; Nunes et al., 2016). CO<sup>2</sup> is released through natural processes; for example via respiration and volcano eruptions or by human activities, such as the burning of fossil and biomass fuels. It has been reported that in the last 150 years, the amount of carbon in the atmosphere has increased by 30% (Alfano et al., 2008). According to the World Meteorological Organization (WMO) "concentrations of carbon dioxide in the atmosphere surged at a record-breaking speed in 2016 to the highest level in 800,000 years." This high level of CO<sup>2</sup> in the atmosphere affects the global climate. The scientists believe that the rising global temperature has a direct relationship with the increased level of CO<sup>2</sup> (Alfano et al., 2008). It is estimated that global temperatures will rise for decades to come, largely due to greenhouse gases emitted by human activities. At present, Pakistan contributes <1% of the total global greenhouse gas emission on a per-capita basis. However, due to the geographical location of Pakistan, average temperatures are predicted to rise faster than elsewhere, increasing 7.2◦F (4◦C) by the year 2100 (2012 World Wildlife Fund report) (Inbar et al., 1993). Pakistan's per capita energy consumption and cumulative CO<sup>2</sup> emissions are extremely low (Inbar et al., 1993). However, the situation will not remain so as Pakistan is projected to add nearly 100 million people by 2050, causing great strains on its resources. According to projections, the country will increase carbon emissions by 300 percent in the next 15 years due to more cars which clog roads and an increased demand for electricity (Inbar et al., 1993). Hence, it is high time for Pakistan to focus on using renewable sources for energy production that will lead the country to a low-carbon pathway. In this direction, biochar production via pyrolysis of biomass is a carbon-negative system that leads to the production of energy products and biochar for carbon sequestration. This technology is carbon-negative as it removes net carbon dioxide from the atmosphere and can store CO<sup>2</sup> in the soil via a process known as carbon sequestration. Electrochemical reduction of the carbon dioxide is also actualized by deploying modified microbial electrochemical technologies (METs) including microbial fuel cells (MFCs), which aim to convert of biomass into electrical power using micro-organisms (Rabaey and Verstraete, 2005; Logan et al., 2006; Oh and Logan, 2007; Wang et al., 2010). In low-cost microbial electrochemical technologies (METs) or bioelectrochemical systems (BES), such as microbial electrolysis cells (MECs), the biological substrate is oxidized at the surface of the anode by utilizing bacteria that act as exoelectrogens, leading to electron generation and also CO<sup>2</sup> (Aelterman et al., 2006; Logan, 2009; Qiao et al., 2010; Villano et al., 2010; Zhao et al., 2012). Then, these electrons are captured by the cathode. On the cathode surface, the electron and diffused protons cause a reduction process of CO<sup>2</sup> and produce valuable products; for example, hydrogen gas or water and methane [CO<sup>2</sup> + 8H<sup>+</sup> + 9e−<sup>1</sup> → CH<sup>4</sup> + 2H2O] (Call and Logan, 2008; Sun et al., 2008; Cheng et al., 2009; Villano et al., 2010; Wang et al., 2010; Kim and Logan, 2011; Logan and Rabaey, 2012; Zhao et al., 2012; Khan et al., 2017). It is estimated that biochar has the potential to sequester almost 400 billion tons of carbon by 2100 and to lower atmospheric CO<sup>2</sup> concentrations by 37 parts per million (Ngoc and Schnitzer, 2009). Overall, it seems reasonable to conclude that biochar production from biomass has the potential to offset the entire annual CO<sup>2</sup> increase in the atmosphere (4.8 vs. 4.1 GtC/yr), as shown in **Table 9** (Jones and Ogden, 1984).

#### CONCLUSION

Being a developing country, Pakistan is facing a severe energy crisis that limits its economy. Because of the rapidly growing population and economy, Pakistan's energy needs become potentially huge; to resolve these serious issues, AEDB is currently functioning on the development of recent renewable energy technologies in Pakistan, which will be beneficial for the developing economy of Pakistan to minimize the growing energy crisis. The sources of renewable energy are hydel, solar, wind and biomass, which has significant potential to compete with growing energy requirements. Thus far, Pakistan envisions setting up plants to generate 10,000 MW through renewable resources by 2030. These plants will be installed with a 50% share of the beneficiary. Out of all the renewable energy resources, biomass is considered the best and most easily accessible source of energy with its unique environmentally friendly nature. Despite the bountiful presence of biomass energy resources, there is still a need to work on the use of these sources to produce energy.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### REFERENCES


Muneer, T., Maubleu, S., and Asif, M. (2006). Prospects of solar water heating for textile industry in Pakistan. Renew. Sustain. Energy Rev. 10, 1–23. doi: 10.1016/j.rser.2004.07.003

Munir, K. A., and Khalid, S. (2012). Pakistan's Power Crisis: How Did We Get Here?


utilization in Pakistan. Renew. Sustain. Energy Rev. 12, 290–298. doi: 10.1016/j.rser.2006.07.002


extracellular electron transfer by a hydrogenophilic methanogenic culture. Bioresour. Technol. 101, 3085–3090. doi: 10.1016/j.biortech.2009.12.077


self-assembly carbon nanotube/cobalt phthalocyanine modified electrode. Environ. Sci. Technol. 46, 5198–5204. doi: 10.1021/es300186f


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Saghir, Zafar, Tahir, Ouadi, Siddique and Hornung. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Comparative Study of Liquid Biodiesel From *Sterculia foetida* (Bottle Tree) Using CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> Nano Catalysts

Maryam Tanveer Akhtar <sup>1</sup> , Mushtaq Ahmad<sup>2</sup> \*, Anjuman Shaheen<sup>1</sup> , Muhammad Zafar <sup>2</sup> , Riaz Ullah<sup>3</sup> , Maliha Asma<sup>1</sup> , Shazia Sultana<sup>2</sup> , Mamoona Munir <sup>2</sup> , Neelam Rashid<sup>2</sup> , Khafsa Malik <sup>2</sup> , Muhammad Saeed<sup>4</sup> and Amir Waseem<sup>4</sup>

*<sup>1</sup> Department of Environmental Science, International Islamic University, Islamabad, Pakistan, <sup>2</sup> Biodiesel Lab, Department of Plant Sciences, Quaid-i-Azam University, Islamabad, Pakistan, <sup>3</sup> Medicinal Aromatic and Poisonous Plants Research Center, College of Pharmacy, King Saud University, Riyadh, Saudi Arabia, <sup>4</sup> Analytical Lab, Department of Chemistry, Quaid-i-Azam University, Islamabad, Pakistan*

#### *Edited by:*

*Umakanta Jena, New Mexico State University, United States*

#### *Reviewed by:*

*Mohammad Zain Khan, Aligarh Muslim University, India JunLi Ren, South China University of Technology, China Jabbar Gardy, University of Leeds, United Kingdom*

> *\*Correspondence: Mushtaq Ahmad mushtaqahmadqau@gmail.com*

#### *Specialty section:*

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

*Received: 29 October 2018 Accepted: 17 January 2019 Published: 01 March 2019*

#### *Citation:*

*Akhtar MT, Ahmad M, Shaheen A, Zafar M, Ullah R, Asma M, Sultana S, Munir M, Rashid N, Malik K, Saeed M and Waseem A (2019) Comparative Study of Liquid Biodiesel From Sterculia foetida (Bottle Tree) Using CuO-CeO2 and Fe2O3 Nano Catalysts. Front. Energy Res. 7:4. doi: 10.3389/fenrg.2019.00004* This study examined the potential of nanocatalyst CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> for efficient conversion of *Sterculia foetida* seed Oil into biodiesel. *S. foetida* contains 40% oil content and low free fatty acid value (0.18 mg KOH/g). The synthesized nanocatalyst was characterized using X-Ray Diffraction (XRD), Fourier Transform Infrared Spectroscopy (FT-IR), and Scanning Electron Microscopy (SEM) techniques. The maximum conversion was achieved (92% yield) using CuO-CeO<sup>2</sup> at 0.25% catalyst loading. The optimized reaction was carried out by experimental variables included molar ratio (1:9), temperature (70◦ C), reaction time (3 h) and stirring rate (600 rpm) using reflux transesterification method. The XRD results showed the size of crystals with order 54.4 nm for CuO-CeO<sup>2</sup> and 31.3 nm for Fe2O3. The SEM images of CuO-CeO<sup>2</sup> showed spherical structure having an average particle size of 32.3 nm. SEM images of Fe2O<sup>3</sup> showed the size ranges from 46.27 to 28.76 nm having regular morphology, including spherical shape. The FT-IR analysis of this nanocatalyst also reinforced the results of this study. Gas Chromatography Mass Spectroscopy (GC-MS) and Fourier Transform Infrared Spectroscopy (FT-IR) confirmed the efficient conversion of *S. foetida* seed oil into biodiesel using prepared nanocatalysts. The prepared nanocatalysts are cheaper, readily available and can be used for industrial scale biofuel production assembly, making it economically feasible and more cost effective.

Keywords: *Sterculia foetida*, biodiesel, nanocatalyst, conversion efficiency, XRD, SEM, FT-IR, GC-MS

## INTRODUCTION

Sterculia foetida, commonly known as Bottle tree, Java Olive, Bastard Poon Tree, Jangli Badam, Hazel Sterculia, and Wild Almond tree, belongs to the genus Sterculiae. The family Sterculiaceae is considered one of the non-drying and non-edible oil yielding species. It is considered as a potential non-edible feed stock for biodiesel production amongst the previously studied framework of nonedible seed oil. This family represents 10 genera and 20 species in Pakistan. Most of them are introduced as a decorative plant (Rao et al., 2015). S. foetida is planted in many parts of the world, such as Ethiopia, Australia, Bangladesh, India, Indonesia, Malaysia, Oman, Uganda, Thailand, Somalia, Sri Lanka as well as Pakistan. It is a wild plant that is modified by tropical and sub-tropical areas (30 North Latitude to 35 South Latitude). The plant has an average life span of more than 100 years (Heyne, 1987; Munarso, 2010). S. feotida is a large, straight and deciduous tree which can grow up to 40 m in height and 3 m in width. It is commonly grown across roadsides and in parks for shadow, variegating flashy flowers. The fruit is 10 m long, woody red, nearly smooth in appearance. The single large seed is enclosed by a shell and a thin 1–2 mm layer of pulp. Each fruit contains 10–15 seeds, giving about 250–300 kg seeds annually (Munarso, 2010). Seed collection from the fruit is somewhat problematic due to the presence of harsh curls in it. The main compositions of S. foetida seeds are fats and protein which are 51.78 and 21.61%, respectively (Heyne, 1987; Munarso, 2010).

Alternative fuel has acquired more attention over the past twenty-five years due to fossil fuel depletion, climate change and human health hazards (Gardy et al., 2018). Biodiesel, being eco-friendly, non-toxicity, and biodegradability are gaining popularity. Worldwide, 90% of biodiesel is produced by a homogeneous catalyzed transesterification process, using edible feedstock (Rehan et al., 2018). Substantial amount of fresh water and arable land are required for its cultivation, which could lead to food security problems in developing countries (Salam et al., 2016). The above mentioned issues can be overwhelmed by using non-edible feedstock, such as waste cooking oil, Jatropha oil, Neem, Pongamia, Karanja, Rocket seed oil, Chinese wild tallow tree, Castor beans, Hemp, Milo, etc. (Rashid et al., 2015). This non-edible feedstock contain large amount of Free Fatty Acids (FFA) resulting in soap formation. Therefore, the nanocatalyst is currently used for synthesis of biodiesel to overcome such problems. These materials have significant permeability, high contact area, high thermal stability, and good chemical properties, culminating in a more effective participation in the chemical reactions to produce biodiesel.

To the best of our knowledge based on detail literature reviewed, qualitative, and quantitative analysis of fatty acid methyl esters (FAMESs) profile of non-edible seed oils has been the focus of many advanced publications. This study focused, for the first time, on a detailed investigation on seed oil source of S. feotida as a new addition in the list of non-edible feedstock for biodiesel synthesis using application of nano-catalysts. Biodiesel can be defined as fuel consisting of mono alkyl esters of long chain fatty acids derived from renewable sources. Biodiesel have various reigns over conventional diesel, such as biodegradability, non-toxicity, renewability, environment friendly, high flash point, and lubricity (Kumar et al., 2011; Okitsu et al., 2014). There are several problems related to biodiesel production, i.e., expensive raw material, shortage of food supply and high energy consumption. Recently, an increase in energy consumption causes a steady depletion of conventional petrol-diesel reserves. According to this theory, the higher demand for oil will require its exceeded the supplies, (Fusco, 2006; Ashnani et al., 2014; Datta and Mandal, 2014; Ullah et al., 2015). To fill this gap, it is significant to investigate the potential of appropriate feedstock like S. feotida as non-edible seed oil production which must be increased significantly if biodiesel possesses the potential impact to contribute in solving energy crises.

Different approaches have been used to reduce the viscosity of oil such as dilution, micro emulsions with short chain alcohols, pyrolysis, catalytic cracking, and transesterification (Farobie and Matsumura, 2015). Transesterification is one of the viable processes used for biodiesel synthesis at industrial scale (Edward and Peggy, 2013) because it gives high yield with low temperature and pressure at short reaction time (Shikha and Chauhan, 2012).

Transesterification method is used for biodiesel production. In this process triglycerides first convert into diglycerides followed by the conversion of diglycerides into monoglycerides in presence of methanol and suitable catalyst (Farobie and Matsumura, 2015). Homogeneous catalysts are typically used at large scale for biodiesel production. Although, they show good catalytic activity with several inadequacies, such as high production cost, difficulty in product isolation and requirement of large quantity of water. A new drift in biodiesel production is the use of heterogeneous alkali catalysts. Heterogeneous catalytic methods are time consuming, low catalytic stability with significant escape of catalyst components to the deactivation of the catalyst, incompetent, and robust to mass transfer. In contrast, nanocatalysts are gaining much more attention in biodiesel production due to its latent role to overcome the disadvantages associated with the use of homogeneous and heterogeneous catalyst. Nanocatalysts show several advantages such as large surface to volume ratio, high catalytic efficiency due to its Nano dimension and morphological structure, good rigidity, low separation cost, and reusability, resistance to saponification, simple operational procedures, and decrease in pollution regeneration.

Among nano-catalysts, transition metals catalysts are more effective than alkali or acid catalyst for biodiesel production (Hu et al., 2011). In the development of nanoparticles, transition metals are intensively perused because of their prominence in the field of applications in science and technology (Aparna et al., 2012). The selected metal oxide nanocatalysts CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> have high catalytic efficiency, higher stability, and lower cost as compared to noble metals. Due to turbulent flow and shock waves, metal particles can move toward each other at high-speed and may also melt at collision point. The suspension solution occurs because of very quick Inter-particle collisions; as a result, the mass of the particle is formed. Collisions can cause a crushing blow between particles, and so, increased specific surface, and finally to achieve high reactivity and good conversion (Karimi et al., 2013).

In recent years, several mixed oxide catalysts have been practiced for synthesizing biodiesel owing to their highest catalytic activity during transesterification reaction. Since mixed oxides possess strong basic sites that increases catalytic surface area which increases catalytic stability and activity during the process. This also helps in the reutilization of catalysts after the completion of its reaction. For instance, CaCeO<sup>3</sup> gives 80% biodiesel yield during transesterification reaction with the reusability of about 5 to 7 times (Okoronkwo et al., 2012). CuO-CeO<sup>2</sup> gives 92.59% biodiesel via transesterification of waste cooking oil and proves to be stable during the process (Tajammul Hussain et al., 2013). In the methonolysis of Pistaciachinensis seed oil via CuO-CeO<sup>2</sup> gives 91% biodiesel yield and can be used with five times consecutively, while Fe-Zn double metal cyanide gives 92% biodiesel yield with the reusability of five times(Chouhan and Sarma, 2011). On the basis of reported work, transition metal mixed oxide Nano catalysts (CuO-CeO<sup>2</sup> and Fe2O3) are therefore, selected for the present work on the basis of its stability, reusability, and highest catalytic performance under mild reaction conditions.

Previously, several mixed oxide catalysts have been practiced for synthesizing biodiesel owing to their highest catalytic activity during transesterification reaction. Since mixed oxides possess strong basic sites that increase the catalytic surface area which increases catalytic stability and activity during the process. This also helps in the reutilization of catalysts after the completion of its reaction. For instance CaCeO<sup>3</sup> gives 80% biodiesel yield during transesterification reaction with the reusability of about 5–7 times (Okoronkwo et al., 2012). CuO-CeO<sup>2</sup> gives 92.59% biodiesel via transesterification of waste cooking oil and proves to be stable during the process (Tajammul Hussain et al., 2013). In the methonolysis of Pistacia chinensis seed oil via CuO-CeO<sup>2</sup> gives 91% biodiesel yield and can be used with five times consecutively, while Fe-Zn double metal cyanide gives 92% biodiesel yield with the reusability of five times (Chouhan and Sarma, 2011). On the basis of reported work transition metal mixed oxide Nanocatalysts (CuO-CeO<sup>2</sup> and Fe2O3) is therefore, selected for the present work on the basis of its stability, reusability, and highest catalytic performance under mild reaction conditions.

Several studies focused on the Biodiesel production using nano-catalysts. On the other hand, comparative studies are very limited. In the present study, comparative evaluation has been made to deliberate the effects of two different nanocatalysts (CuO-CeO2, Fe2O3) in terms of their characterization i.e., X-ray powder diffraction (XRD), Scanning Electron Microscopy (SEM), Fourier transfer Infrared spectroscopy (FT-IR), on FAMEs synthesis using non-edible seed oil of S. feotida. In addition to this, synthesized FAMEs have been quantitatively analyzed using analytical techniques such as Gas Chromatography-Mass Spectrometry (GC-MS), FT-IR.

#### METHODS

The experimental work was conducted at Wet Lab, Department of Environmental Science, International Islamic University Islamabad and Nano Science and Technology research lab of the National Centre for Physics, Quad-i-Azam University Islamabad, Pakistan. Shells of S. feotida plant was collected through several field trips across the country. The collected shells were placed in sun light to remove insects and moisture, then processes for seed separation. Seeds were washed with soft warm distilled water to eradicate the dirt then dried in an oven at 55◦C. The Soxhlet apparatus was used to determine seed oil contents. Crude oil was extracted using an electric oil expeller (KEK P0015-10127) from Germany.

#### Equipment and Chemicals

Digital Weighing Balance (GF-3000), Teflon Magnetic Stirrer (AM4, VELP, SCIENTIFICA), Thermometer, pH Paper, Conical Flasks, Beakers (100 and 500 ml), Filter Paper (Wattman 42), Aluminum Foil, Pipette (10 ml), Iron Stand, Electric Oil Expeller (KEK P0015, 10127 Germany), Burette, Soxhlet Assembly (Behr Labor – Technik), Magnetic Stirrers, Pestle mortar, Drying Oven (DHG-9053A), Calcination Furnance (Neycraft), Sodium hydroxide (NaOH), Methanol 99.9% purity, Cerium (IV) oxide, Iron Sulfate, Copper Oxide, Sulfuric acid. All the reagents used were of analytical grade and purchased from Merck (Germany), Scharlau (Spain), and Sigma Aldrich (USA).

#### Oil Extraction

Soxhlet apparatus was used for solvent extraction. It is a sequence by which chemical constituents are removed from sample by using organic solvent. Dried seeds were wrinkled into fine powder using pestle and mortar. Round bottom flask was filled with 250 ml of n-hexane. Thimble was filled with 5 g of seeds powder and heated up to 60◦C for 5 h. In this process, solvent is continually recovered and reused. Oil droplets were also visible in the round bottom flask. The resulting wet sample was again oven dried at 60◦C to evaporate the solvent and weigh the sample. The reduction in the sample weight was determined to calculate the amount of oil extracted.

Percentage of oil content = **W4** = [(**W3** − **W1**)/**W2**] × **100** (1)

Where **W1** = Weight of Empty Flask **W2** = Weight of Empty Flask + powdered Sample **W3** = Weight of Sample Used **W4** = Weight of Extracted Oil

The collected oil was pre-treated by filtration and titrated using acid base titration method to determine its FFAs contents. The resulting crude oil contained 3.7 wt% FFA.

## Synthesis of Nanocatalyst (CuO-CeO2) and Fe2O<sup>3</sup>

CuO-CeO<sup>2</sup> composite was synthesized by top down approach through the ball mill method. Equivalent amount of CuO (2 g) and CeO<sup>2</sup> (2 g) were weighed through analytical balance and then placed in ball mill apparatus. It was made airtight with the help of a rubber piece. At room temperature the revolution per minute was 200 for 8 h. The powder was subjected to high energy collision from the balls. Then, these processed nanoparticles were calcined for 3 h at 500◦C in the muffle furnace. After calcinations, the composite of CuO and CeO2was ready for characterization.

Fe2O<sup>3</sup> was prepared by co-precipitation route. 0.1 M solution of FeSO<sup>4</sup> was prepared. Then, 0.1 M solution of NaOH was prepared and put into the burette. Take 100 ml of Iron Sulfate into beaker and titrate it against 0.1 M NaOH, to get its basic pH i.e., 10–12. NaOH solution dropped very slowly to produce fine nanoparticles. The color of solution turns to reddish brown Iron (II) oxide, due to oxidation. The solution was frequently stirred at 600 rpm till its pH reached up to 10. When the solution pH reached 10, titration was stopped. The solution was left over night for phase separation then washed with distilled water to remove impurities of NaOH. After washing, the sample was oven dried at 90◦C for 10–12 h to evaporate extra water then calcined in muffle furnace for 3 h at 500◦C. Following calcination, the FeSO<sup>4</sup> nanoparticles were ready for characterization.

#### Characterization of Nanocatalyst X-Ray Diffraction

The Nanocatalyst was characterized using XRD Model No. D8 Advance Bruker, to confirm the formation of crystallite size. The Scherer Debye equation was used to calculate crystallite size. XRD measurement was performed at 2<sup>θ</sup> value between 10 and 80◦C.

#### Scanning Electron Microscopy (SEM)

SEM was performed using scanning electron microscopy Model JEOL JSM-5910 to understand the morphology of nanocatalyst. Field emission electron microscopy with 5 kV accelerating voltage was used to attain SEM image. This allows the qualitative characterization of catalyst and support, to understand the phenomena occurring in calcination and pretreatment.

#### Fourier Transform Infrared (FT-IR) Analysis

FT-IR spectroscopic technique was used to identify and measure organic functional groups and inorganic ions of S. feotida biodiesel using MODEL BRUKER-TENSOR 27 in the range from 3,500 to 400 cm−<sup>1</sup> . The resolution was 1 cm−<sup>1</sup> and 15 scans for biodiesel analysis. FT-IR of seed oil, biodiesel and catalysts were analyzed.

#### Biodiesel Synthesis Using Reflux Transesterification

Reflux transesterification method was used for biodiesel production from seed oil. Materials used for transesterification were Nanocatalyst, seed oil, hot plate, magnetic stirrer, thermometer, reflux condenser. The reaction was carried out in a 250 ml round bottom flask equipped with thermometer, reflux condenser and magnetic stirrer. 0.1 g Nanocatalyst was refluxed with 1:3 oil to methanol ratio for 1 h at 70◦C. After cooling, preheated oil was added into flask and refluxed again at 70◦C for 2 h. Then mixture was allowed to cool down and transferred into separating funnel for phase separation. The upper liquid phase was crude ester and the lower liquid phase was the glycerol. Biodiesel yield was calculated using the equation

$$\% \text{ Yield} = \frac{Grams \text{ of } biodiesel \text{ produced}}{Grams \text{ of } oil \text{ used}} \times 100 \tag{2}$$

## Characterization of Synthesized Fatty Acid Methyl Esters/Biodiesel

#### FT-IR Analysis

FT-IR spectroscopic technique was used to determine structural composition of Sterculia feotida biodiesel using FT-IR MODEL BRUKER-TENSOR 27 in the range of 4,000 to 600 cm−<sup>1</sup> . The resolution was 1 cm−<sup>1</sup> and 15 scans for biodiesel analysis. FT-IR of seed oil, biodiesel, and catalysts were analyzed.

#### GC-MS Analysis

It was used to determine the chemical composition of fatty acids in S. foetida oil biodiesel. The synthesized biodiesel was subjected to GC-MS model QP 210 ultra (Shimadzu, Japan). A sample volume of 1 ml was dissolved with 5 ml chloroform and stirred for 3 min. One microliter of this sample was injected into GC-MS model using a split mode with split ratio of 1:3. Helium was used as a vector with a flow rate of 1.44 mL/min. The column temperature was set from 50 to 300◦C at the rate of 80◦C/min. The temperature of detector and injector was set at 250 and 120◦C.

#### Physico-Chemical Properties of Biodiesel

Fuel properties of synthesized biodiesel from Sterculia Oil were determined quantitatively from Pakistan State Oil (PSO) and Rawalpindi, and were compared to ASTM ;biodiesel standards. Flash point (◦C) was recorded at ASTM D-93,Pour point at ASTM D-97, Cloud Point ◦C at ASTM D-2500, Density @ 15◦C kg/L at D-1298, Kinematic Viscosity @ 40◦C at ASTM D-445, Sulfur % ASTM D-4294 and Total Acid Number (mg KOH/mg) at ASTM D-974 (Atabani et al., 2013) was recorded and compared to the ASTM.

#### RESULTS AND DISCUSSION

#### Biodiesel Synthesis From *S. foetida* Seed Oil

S. foetida seed oil was converted into biodiesel, and oil content and FFA contents were determined. Based on dry biomass, the extracted oil percentage was 40% which is higher as compared to other edible and non-edible oil sources studied in previous literature, such as Pongamia oil (37%) (Baskar et al., 2016), Melia Azedarch fruit oil (39%) (Stavarache et al., 2008), Rocket seed (35%) (Chakrabarti et al., 2011), Soybean (18–22%) (Williams, 2005; Moser, 2010). Biodiesel quality and yield highly depend upon the FFA contents. As reported in literature, FFA content of crude oil up to 2.5% is suitable for biodiesel synthesis. Efficiency of crude oil gradually decreases outside the prescribed limit, and soap formation occurs which ultimately causes difficulty in separation process (Naik et al., 2010; Shikha and Chauhan, 2012).

Two nano-catalysts CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> in different concentrations were used for efficient conversion of S. foetida seed oil (SFSO) into biodiesel. Data presented in **Table 1** depicts that conversion rate of oil is mainly dependent on type and amount of catalyst. At 0.25% of said catalysts concentration, the conversion rate is relatively high, up to 92%. While, by increasing the catalyst amount shows declining trends in biodiesel yield, keeping all the variables constant. Loss in biodiesel yield is proportional to the catalyst amount as shown in **Table 1**. Higher amount of catalyst emulsification which hampers the glycerin separation that results low ester yield (Vicente et al., 1998; Gurunathan and Ravi, 2015). As reported in literature, different nanocatalysts were used for biodiesel production from non-edible feedstock. Hashmi et al. (2016) investigated the use of CaO-Al2O<sup>3</sup> nano-catalyst for biodiesel production from Jatropha oil. The highest 82.3 %


TABLE 1 | Effect of nanocatalyst on biodiesel production of *S. foetida* oil using molar ratio (1:6), temperature (70◦C), reaction time (3 h), and stirring speed (600 rpm).

biodiesel yield was obtained. Baskar et al. (2017) demonstrated the biodiesel production from Pongame oil using magnetic composite of Zinc oxide nano-catalyst. Transesterification of Pongame oil yielded 93% of biodiesel. In this study, CuO-CeO<sup>2</sup> showed the highest conversion efficacy of crude oil into FAMEs that is 92%.

## Characterization of CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> Nano-Catalysts

#### XRD of CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup>

X-Ray Diffraction pattern of CuO-CeO<sup>2</sup> nano-catalyst (**Figure 1**) shows strong diffraction peaks at 2θ angles i.e., 28.6, 33.1, 47.4, and 56.3 with diffraction pattern (1 1 1), (2 0 0), (2 2 0), and (3 1 1), respectively. The peak appears at 2θ angles 28.6, 33.1, 47.4, and 56.3 confirms the cubic fluorite structure of CeO2. The peak observed at 2θ angles 35.7, 38.7 with diffraction pattern (1 1 1), and (0 2 2), respectively shows the hexagonal structure of CuO. The average crystalline size of CuO.CeO<sup>2</sup> 54.4 nm was determined by Debye-Scherer equation (D=Kλ/βcosθ) (Mustafa et al., 2013). Where D is the crystallite size in nm, λ stands for X-ray wavelength = 0.154095 nm, θ is the half diffraction angle, K stands for the factor shape which takes a value of 0.89 and β stands for full width at half maximum (Pathaka et al., 2017). crystallite size of CuO-CeO<sup>2</sup> nano-particles in literature was 40 nm (Amaniampong et al., 2018). XRD pattern of Fe2O<sup>3</sup> nano catalyst (**Figure 2**) shows strong diffraction peaks at 2θ angles 33.1, 35.6, 40.6, 49.5, 54.0, 62.5, and 64.16 with diffraction pattern of (1 0 4), (1 1 0), (1 1 3), (0 2 4), 1 1 6), (2 1 0), and (3 0 0), respectively. These peaks indicate the rhombohedral structure of Fe2O<sup>3</sup> which is confirmed from (JCPDS File No. 24-0072) with particle size 32 nm.

#### Scanning Electron Microscopy (SEM)

SEM was used to see the morphology and particle size of synthesized nano-catalysts. The SEM images have been shown in different magnifications. The SEM results showed that CuO-CeO<sup>2</sup> nanoparticles have average crystalline size 32.3 nm. The results showed that the nanoparticles have spherical shape with no agglomeration (**Figure 2**).

The SEM study of Fe2O<sup>3</sup> nano-catalyst demonstrated the average size and morphology. The average size ranged from 46.27 to 28.76 nm. The shape of Iron oxide nanoparticles appeared to be spherical with no agglomeration. The larger iron

oxide aggregated particles remained visible may be due to the aggregation of the smaller particles (**Figure 3**).

#### FTIR of CuO-CeO<sup>2</sup>

FT-IR spectroscopic analysis is used to determine characteristic peaks in functional group (4,000–1,400 cm−<sup>1</sup> ) and finger print (400–1,400 cm−<sup>1</sup> ) region. FT-IR spectrum of CuO-CeO<sup>2</sup> shows characteristic absorption peaks in functional group region at 3648.13, 3565.93, 1738.54, 1558.63, 1540.72, 1507.07, 1456.98, 1365.10 cm−<sup>1</sup> and in finger print region at 1229.86, 1217.11, 715.05, 690.37, 681.98, 668.78 cm−<sup>1</sup> which is shown as (**Figure 5**). The peak observed at 3648.13 and 3565.93 cm−<sup>1</sup> due to –OH stretch of water molecule present in sample as a moisture. The peak appears in spectra at 1738.51 and 1558 cm−<sup>1</sup> is due to bending vibrations of H-O-H molecule present in sample. The C-H stretching and bending vibration appears at 2,969 and 1456, 1365 cm−<sup>1</sup> , respectively. The peak appears at 690 and 681 cm−<sup>1</sup> is due to of Cu-O particles, but an additional peak is observed at 1,229 cm−<sup>1</sup> , which is due to stretching mode of Cu+<sup>2</sup> and O−<sup>2</sup> . The peak appearing at 668 cm−<sup>1</sup> is due to CeO<sup>2</sup> vibration. These peaks confirm the formation of CuO-CeO<sup>2</sup> (Karimi et al., 2013) (**Figure 4**).

FIGURE 3 | (A–D) 1µm. SEM image of Fe2O3 nano-catalyst at different resolution.

#### FT-IR of Fe2O<sup>3</sup>

The peak appearing in the spectra of Fe2O<sup>3</sup> at 3,375 cm−<sup>1</sup> is due to –OH stretching vibration of water molecule present in sample as moisture. The peak observed at 1,635 cm−<sup>1</sup> is due to bending vibration of H-O-H. The characteristic peak appears at 674 cm−<sup>1</sup> is due to Fe-O stretching vibration mode. An additional stretching mode of Fe-O is observed at 911 cm−<sup>1</sup> in spectra.The previous study also supports the results. Farahmandjou and Soflaee (2015) Synthesize and Characterize Fe2O<sup>3</sup> nanoparticles by simple co-precipitation method (Farahmandjou and Soflaee, 2015). These peaks confirm the formation of Fe2O<sup>3</sup> (**Figure 5**).

#### Characterization of Synthesized FAMEs GC-MS Analysis of S. foetida FAMEs

GC-MS analysis is an extensively used analytical technique to quantify chemical composition, structure and type of FAMEs present in biodiesel. The given hierarchy shows the efficiency of



synthesized Nanocatalyst i.e., CuO-CeO<sup>2</sup> >Fe2O<sup>3</sup> on biodiesel synthesis, the percentage of synthesized methyl esters was in the order of 92 > 84%. The catalytic efficiency of CuO-CeO<sup>2</sup> was highest due to its specific characteristics. GC spectrum of CuO-CeO<sup>2</sup> catalyzed biodiesel showed 9 different peaks (**Figure 6**). Each peak corresponds to Methyl Ester and identified from the library match software (NO. NIST II). The major FAMES were found to be tetradecanoic acid, Hexadecanoic acid, 6- Nonezal, 9-Octadecaenoic Acid, Methyl Ester, 7,10-Octadecadienoic Acid, Heptadecanoic Acid, 1-Azuleneethanol, Acetate, Sterculic acid 8-(-2-Octacyclopropen-1-yl) octanoic acid. The retention time of these methyl esters is illustrated in **Tables 2A,B**. FAMEs were identified by arranging the retention time data; moreover, verification was made by GC internal standards (**Figures 7**, **8**).

#### FT-IR of S. foetida Seed Oil

FT-IR spectroscopic analysis is used to determine characteristic peaks in functional group (4,000–1,400 cm−<sup>1</sup> ) and finger print (400–1,400 cm−<sup>1</sup> ) region. The peak appearing at 3007.82 cm−<sup>1</sup> confirms the presence of aromatic compounds. Strong peaks at 2923.91, 2852.93 cm−<sup>1</sup> confirm the (sp<sup>3</sup> and sp2C-H) stretch. An intense peak in FT-IR spectrum is observed at 1743.67, 1710.28 cm−<sup>1</sup> due to C=O stretching vibrations that confirmed the presence of saturated aliphatic Esters. A medium peak at 1458.49 and 1377.20 cm−<sup>1</sup> appears in spectra due to bending vibrations of CH<sup>2</sup> and CH3. A peak appeared at 1236.01 cm−<sup>1</sup> shows C-H deformation vibration, whereas, at 1162.02 cm−<sup>1</sup> stretching vibration of C-C bond was observed. Peak observed at 1099.06 cm−<sup>1</sup> confirm the presence of cyclic compound shows C-O stretching vibration as shown in **Figure 9**.

#### BIODIESEL OPTIMIZATION

The conversion rate of Sterculia oil to biodiesel depends on various parameters, such as molar ratio, reaction temperature, reaction time, catalyst loading. A series of experiments were performed to achieve the maximum conversion of crude oil into biodiesel. Five variables were selected to check out their effect on biodiesel yield (**Figure 10**).

#### Molar Ratio

It is one of the most important variables in biodiesel production. Stoichiometrically 3 mole of methanol is required for 1 mole of triglyceride but practically, a large amount of methanol is required to shift the equilibrium favorably(Khan and El Dessouky, 2009) when other variables were kept constant methanol to all ratio shows positive influence on the biodiesel yield as it increases from 1:3 to 1:9. But as the molar ratio increase from 1:9 to 1:15, the incremental decrease in the yield was observed (Atapour and Kariminia, 2013). The maximum biodiesel yield was obtained at 1:9 oil to methanol ratio. Biodiesel yield was reduced at 1:15 oil to methanol ratio, due to decrease in density difference between the two phases obtained after transesterification reaction. High amount of methanol was used to minimize the contact of triglyceride on the catalyst active sites, ultimately decreasing the catalyst activity.

#### Reaction Temperature

The reaction temperature is one of the main parameters which affects the biodiesel yield. At certain range, elevated temperatures will speed up the reaction rate. The transesterification proceeded slowly at 45◦C because at lower temperatures, only small numbers of molecules were able to get over the required energy barrier. Based off of the literature, the maximum conversion results acquired at the range of 60–70◦C were studied (Chelladurai and Rajamanickam, 2014). In this study, the reaction temperature was kept at 60, 65, 70, 75, and 80◦C by keeping other variables constant. Maximum conversion of esters was obtained at 70◦C. At lower temperature, the energy is not enough to promote the collisions among reactant molecules. Whereas at higher temperature the possibility of collision among reactant molecules becomes greater and the activation energy is easily reached (Sivakumar et al., 2012).

#### Reaction Time

Reaction time has great influence on the rate of transesterification reaction. Oil must be stirred at constant rate to convert it into biodiesel. The conversion increased progressively with increasing

reaction time and then reached a plateau value that represents the equilibrium (Pathaka et al., 2017). By keeping all other variables constant, the reaction time was kept from 1 to 3 h at the optimal reaction conditions. The results show the maximum conversion of biodiesel was obtained at 3 h. After that, further increase in time, up to 5 h, showed the decrease in yield of biodiesel. This decrease in yield is due to the backward reaction. Results show that increasing reaction time has influence on FAME yield production (Feyzi et al., 2013) (**Figure 11**).

### Catalyst Loading

The concentration of fatty acid methyl ester decreases with the increase in catalyst concentration (Gupta and Agrawal, 2015). The catalyst concentration used in transesterification reaction was 0.1, 0.15, 0.2, 0.25, 0.3, and 0.35 gm by keeping other variables such as reaction time, reaction temperature, oil to methanol ratio and catalyst type constant. The results showed a decrease in biodiesel yield with increase in catalyst concentration up to 0.25 gm because emulsification process hampers the glycerine separation and decreases the ester yields. The maximum biodiesel yield is obtained by using low catalyst concentration (Leung et al., 2010).

## Physico-Chemical Properties of *Sterculia* Oil Biodiesel (Sob)

Physico-chemical properties of SOB were quantitatively determined and compared with the American Society for Testing and Materials (ASTM) standards for biodiesel testing (Atadashi et al., 2010). Fuel properties of pure biodiesel was tested where its density, cloud point, viscosity, density, sulfur content, Flash point, and total acid number were determined. The obtained values were then compared with the research and with the standard values of ASTM. The results for B-100 are mentioned in **Table 3**.

## Color

The color of Sterculia biodiesel is on visual 2 according to ASTM standards.

## Flash Point◦C

The temperature at which fuel can ignite, is known as a flash point. It is the measure of affinity of a material to form a flammable mixture in the presence of air. Fuel having high flash point is safe during handling, transportation and storage (Candeia et al., 2009). Karmee and Chadha (2005) reported the flash point value of Pongamia pinnata i.e., 150◦C using ASTM D-93. In this sample, the flash point was recorded as 73◦C within the range of ASTM D-93 standards, which makes it safer fuel. High Speed Diesel (HSD) have flash point in the range of 60–80◦C.

## Density @ 15◦C Kg/L

Fuel density is defined as the mass per unit of volume that measures in a vacuum. It plays a vital role in the determination of biodiesel quality. Fuel density has direct effect on fuel efficiency. Cetane number and viscosity are strongly linked with density. Small mass and speed of less viscous fuel make very accurate spray (Candeia et al., 2009). The density of Sterculia Oil Biodiesel was analyzed at ASTM D-1298 and the result was 0.0832 kg/L (**Table 3**), while High Speed Diesel has a density of 0.834. It confirms that SOB has less density than that of High Speed Diesel (HSD).

## Kinematic Viscosity @ 40◦C

Kinematic viscosity provides the indication of the ability of a material to flow. Engine efficiency becomes reduced due to the deposition of higher viscosity fuel (Gunstone and Hamilton, 2001; Sarin et al., 2007). Esterification process helped to reduce the viscosity (Sarin et al., 2007). Less viscous oil is easier to pump and to achieve fine droplets of fuel (Goodrum, 2002). (Morshed et al., 2011)studied that rubber oil biodiesel have kinematic viscosity of 4.5 @ 40◦C which was greater as compared to HSD viscosity i.e., 4.223. The kinematic viscosity of SOB was analyzed at ASTM D-445 and result was 4.15 @ 40◦C which lie within the

range of ASTM standards. The value is also higher than that of HSD value i.e., 4.22.

## Pour Point◦C and Cloud Point◦C

The lowest temperature at which the fuel can flow in chilled conditions is known as Pour point. While the cold point is a temperature at which paraffin becomes crystallize or begin to separate from solution. When the fuel is ice cold under suggested conditions, (Rashid et al., 2013) discovered that the Pour point and Cloud Point value of Rocket seed oil was −15 and −3 ◦C, respectively. The Pour point and Cloud point of SOB was analyzed at ASTM D-97 and ASTM D-2500. Pour point value was −8 ◦C which were according to ASTM standards whereas its cloud point value is −13◦C which is higher than HSD.

## Sulfur Contents % (wt)

Fuel with decreasing number of sulfur contents is ideal for most polluted areas. Low sulfur is considered more effective for the environment and for the engine's life (Candeia et al., 2009). Rashid et al. (2011) found that Muskmelon seed oil biodiesel contains 0.02% sulfated ash. Ahmad et al. (2009)reported that the 0.01% of sulfated ash present in Sesame oil biodiesel. The sulfur contents in SOB were found to be 0.0002% by weight, which was very low by comparing with ASTM D-4294 standard. It is stated that biodiesel is superior to HSD due to its decreasing number of sulfur contents.

#### Total Acid Number mg KOH/gm

Total acid number or Acid Value is termed as the number of free fatty acids present in fuel samples. It is expressed in mg KOH/gm for neutralizing 1 gm of FAMEs (Raj and Sahayaraj, 2010). Biodiesel is not acidic in nature but presence of FFAs in the diesel can make it acidic. High number of acid value is not good for engine efficiency. It results in severe corrosion in fuel supply system and internal engine combustion (Atabani et al., 2013). Rashid et al. (2011) studied that the acid value of Cotton Seed oil is 0.45 mg KOH/gm. According to ASTM D-974 standard, the acid value of SOB was 0.173 mg/KOH.


#### Kinetic Study of Biodiesel

Rate of Chemical reaction is the decrease in reactants concentration and increase in product concentration and was calculated experimentally. The order of reaction for the transesterification process to produce biodiesel was calculated by: second order (1/Conc. vs. time), first order (-(ln(conc.) vs. time), zero order (conc. vs. time) (Shu et al., 2011). The R 2 -values calculated from the graph is given in the table below. The zero order reaction rate of the chemical reaction was calculated from Equation (1);

$$\frac{d[P]}{dt} = \frac{Kpdx}{dt}$$

Where [P] is the product concentration, Kp is reaction rate constant, x is the fraction of biodiesel.

The pseudo first order was calculated from Equation (2);

$$r = \frac{-dp}{dt} = k[P]$$

Where K is the rate constant of pseudo first order of reaction and was calculated from Equation (3);

$$\kappa = \frac{\ln\left(1 - x\right)}{time}$$

Where x is the methanol content and the Reaction rate constant of various kinetic study at different temperatures for zero and pseudo first order reaction, are given in **Table 4**. The activation energy was calculated from Arrhenius equation (Aransiola et al., 2013) and value of R 2 calculated from graph of (conc. vs. time) and (-ln (1-x) vs. time) and is given in **Table 5**. By comparing R 2 values of different models via Arrhenius equation, pseudo first order reaction appeared to be well fit with experimental yield because it has greater value of (R <sup>2</sup> = 0.9417) as compared to zero order reaction rate.

TABLE 4 | Rate constant values at different temperature.


TABLE 5 | Activation energy for the reaction.


The reusability of transition metal mixed oxide Nano catalysts (CuO-CeO<sup>2</sup> and Fe2O3) was checked by collecting catalyst after each round of reaction through reflux system. It was noted that CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> can be used four times without decrease in their catalytic activity. This fact was supported by the literature values, where CuO-CeO<sup>2</sup> and Fe-Zn double metal cyanide can be used five times (Chouhan and Sarma, 2011). Therefore, CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> are stable and active catalysts for large scale biodiesel synthesis.

Long term use and feasibility of catalysts can be checked by examining the reusability of transition metal mixed oxide Nano catalysts (CuO-CeO<sup>2</sup> and Fe2O3) on biodiesel yield at optimum operating conditions. **Figure 12** shows the reusability of (CuO-CeO<sup>2</sup> and Fe2O3) under optimum conditions of 0.25% catalyst loading with oil to methanol ratio of 1:9 at 70◦C for 3 h. After each cycle of catalyst used, it was washed with absolute ethanol followed by oven drying (100◦C) for 4 h. The catalysts were found to be active over four times without decrease in their catalytic activity. However, after the fifth round of catalyst testing, their activity significantly decreased. This fact was supported by the literature values where CuO-CeO<sup>2</sup> and Fe-Zn double metal cyanide can be used five times (Chouhan and Sarma, 2011). Therefore, CuO-CeO<sup>2</sup> and Fe2O<sup>3</sup> are stable and active catalysts. The decrease in biodiesel yield might be attributed to the deactivation of active sites and leaching of catalyst in reaction media (**Figure 12**).

## CONCLUSION

This study focused on the comparative analysis of nano-catalysts on biodiesel production from S. foetida. Nanocatalysts (CuO-CeO<sup>2</sup> and Fe2O3) were synthesized chemo-mechanically, then characterized and applied for biodiesel production from SSO to SOB. Among these two catalysts, CuO-CeO<sup>2</sup> showed the auspicious results; highest conversion efficiency was achieved 92%, using CuO-CeO<sup>2</sup> followed by Fe2O<sup>3</sup> at 0.25% catalyst loading. The best optimized variables were: alcohol to oil molar ratio (1:6), reaction time (3 h), temperature (70◦C) and stirring rate (600 rpm) using reflux transesterification method. Moreover, pseudo first order reaction appeared fit with experimental yield with greater R <sup>2</sup> = 0.9417 as compared to zero order reaction rate. The synthesized nanocatalyst exhibit good stability and reusability during the reaction. The encouraging results of this study strongly recommended S. foetida an economically feasible potential feedstock for biodiesel production as a renewable energy on industrial level.

## AUTHOR CONTRIBUTIONS

MA, RU, MZ, and MTA designed the whole experiment and methodology. AS collected the seeds. MTA, MAs, and SS performed the whole experiment. MM, MS, and AW revised the whole manuscript carefully to its present form, and assisted with checking the consistency of the data. NR and KM drafted manuscript with contribution from all co-authors. All authors read the final manuscript and agreed to its submission.

## ACKNOWLEDGMENTS

The authors extend their appreciation to Deanship of Scientific Research at King Saud University for funding their group through research Group No (RG 1440–100). Authors are also thankful to Chengdu Institute of Biology for providing lab facilities.

#### REFERENCES


Space Technology Services and Climate Change (RSTSCC) (Chennai: IEEE), 80–86.


oil and its process augmentation for biodiesel production. Indus. Eng. Chem. Res. 51, 8992–8998. doi: 10.1021/ie300882t


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Akhtar, Ahmad, Shaheen, Zafar, Ullah, Asma, Sultana, Munir, Rashid, Malik, Saeed and Waseem. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Development of an Effective Chain Elongation Process From Acidified Food Waste and Ethanol Into n-Caproate

Mark Roghair <sup>1</sup> , Yuchen Liu<sup>1</sup> , David P. B. T. B. Strik <sup>1</sup> \*, Ruud A. Weusthuis <sup>2</sup> , Marieke E. Bruins <sup>3</sup> and Cees J. N. Buisman<sup>1</sup>

*<sup>1</sup> Sub-department of Environmental Technology, Wageningen University and Research, Wageningen, Netherlands, <sup>2</sup> Bioprocess Engineering, Wageningen University and Research, Wageningen, Netherlands, <sup>3</sup> Wageningen Food and Biobased Research, Wageningen University and Research, Wageningen, Netherlands*

#### Edited by:

*Abdul-Sattar Nizami, King Abdulaziz University, Saudi Arabia*

#### Reviewed by:

*Qaisar Mahmood, COMSATS Institute of Information Technology, Pakistan Baskar Gurunathan, St. Joseph's College of Engineering, India Weichao Wang, Purdue University, United States*

> \*Correspondence: *David P. B. T. B. Strik david.strik@wur.nl*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Bioengineering and Biotechnology* Received: *10 March 2018* Accepted: *13 April 2018*

Published: *27 April 2018*

#### Citation:

*Roghair M, Liu Y, Strik DPBTB, Weusthuis RA, Bruins ME and Buisman CJN (2018) Development of an Effective Chain Elongation Process From Acidified Food Waste and Ethanol Into n-Caproate. Front. Bioeng. Biotechnol. 6:50. doi: 10.3389/fbioe.2018.00050* Introduction: Medium chain fatty acids (MCFAs), such as n-caproate, are potential valuable platform chemicals. MCFAs can be produced from low-grade organic residues by anaerobic reactor microbiomes through two subsequent biological processes: hydrolysis combined with acidogenesis and chain elongation. Continuous chain elongation with organic residues becomes effective when the targeted MCFA(s) are produced at high concentrations and rates, while excessive ethanol oxidation and base consumption are limited. The objective of this study was to develop an effective continuous chain elongation process with hydrolyzed and acidified food waste and additional ethanol.

Results: We fed acidified food waste (AFW) and ethanol to an anaerobic reactor while operating the reactor at long (4 d) and at short (1 d) hydraulic retention time (HRT). At long HRT, n-caproate was continuously produced (5.5 g/L/d) at an average concentration of 23.4 g/L. The highest n-caproate concentration was 25.7 g/L which is the highest reported n-caproate concentration in a chain elongation process to date. Compared to short HRT (7.1 g/L n-caproate at 5.6 g/L/d), long HRT resulted in 6.2 times less excessive ethanol oxidation. This led to a two times lower ethanol consumption and a two times lower base consumption per produced MCFA at long HRT compared to short HRT.

Conclusions: Chain elongation from AFW and ethanol is more effective at long HRT than at short HRT not only because it results in a higher concentration of MCFAs but also because it leads to a more efficient use of ethanol and base. The HRT did not influence the n-caproate production rate. The obtained n-caproate concentration is more than twice as high as the maximum solubility of n-caproic acid in water which is beneficial for its separation from the fermentation broth. This study does not only set the record on the highest n-caproate concentration observed in a chain elongation process to date, it notably demonstrates that such high concentrations can be obtained from AFW under practical circumstances in a continuous process.

Keywords: food waste, chain elongation, caproate, HRT, sludge, ethanol, sodium hydroxide

## INTRODUCTION

Organic residual streams, like food waste, have great potential as alternative resource for production of fuels and chemicals because they are renewable and because they do not compete with the human food chain (Coma et al., 2017). The challenge is to convert these mixed residues into the desired products and purify them in an energy-efficient and economically viable process. An emerging technology that facilitates conversion of (derivatives of) organic residues into (precursors of) fuels and chemicals is chain elongation. Chain elongation is an anaerobic openculture biotechnological process that converts volatile fatty acids (VFAs) and an electron donor into more valuable medium chain fatty acids (MCFAs) (Angenent et al., 2016). The conversion of VFAs into MCFAs with ethanol as electron donor is done by chain elongating micro-organisms (e.g., Clostridium kluyveri) that use the reverse β-oxidation pathway. In this pathway, 1 additional mole of ethanol is oxidized into acetate for every 5 chain elongation reactions (Equation 1) (Seedorf et al., 2008).

$$\begin{aligned} \text{Reverse } \beta &\quad - \text{ oxidation pathway} \quad 5C\_{\text{x}}H\_{2\text{x}-1}O\_{2}^{-} + 6C\_{2}H\_{6}O\\ &\quad \rightarrow \ 5C\_{\text{x}+2}H\_{2\text{x}+3}O\_{2}^{-} + \ ^{\cdot}C\_{2}H\_{3}O\_{2}^{-} + 4H\_{2}O + \ H^{+} + 2H\_{2} \end{aligned} \tag{1}$$

VFAs can be obtained through hydrolysis and acidogenesis of organic residues. Electron donors that are suitable for chain elongation processes, such as ethanol (Steinbusch et al., 2011), hydrogen (Steinbusch et al., 2011), methanol (Chen et al., 2016b), and lactic acid (Zhu et al., 2015), can also be produced from organic residues (e.g., lignocellulosic bioethanol). Particularly, MCFAs can be used for production of aviation fuels (Harvey and Meylemans, 2014; Khor et al., 2017) and for other endproducts such as solvents, lubricants, feed additives for poultry (Evans et al., 2017) and piglets (Hanczakowska et al., 2013), plastics and dyes (Angenent et al., 2016). The main advantage of chain elongation is that it is catalyzed by an anaerobic open-culture reactor microbiome (i.e., sludge). Open-culture microbiomes can tolerate mixtures of residual streams while they convert the residues under mild and non-sterile conditions. Chain elongation, therefore, does not need a chemical catalyst and proceeds under mild and non-sterile conditions. Although inhibition of competing processes is important, it is not necessary to do this by adding bioactive chemicals such as antibiotics or methanogenic inhibitors such as 2-bromoethanesulfonate (e.g., Roghair et al., 2018). As such, solid residual streams from the chain elongation process itself could be used as soil fertilizer upon composting.

MCFA production from organic residues through biomass hydrolysis, acidogenesis and chain elongation can be executed in a single-stage system (Agler et al., 2012) as well as in a two-stage system (Grootscholten et al., 2014). In a two-stage system, hydrolysis and acidogenesis are done in one stage and chain elongation in a subsequent stage. The advantage of a twostage system over a single-stage system is that both stages can be optimized independently. Grootscholten et al. (2014) concluded that MCFA production from the organic fraction of municipal solid waste and additional ethanol in a two-stage system resulted in higher MCFA production rates and concentrations compared to a single-stage system. Another advantage of a two-stage system is that it allows easier control of the hydrogen partial pressure (pH2) in the chain elongation stage by e.g., manipulating CO<sup>2</sup> loading rate (Grootscholten et al., 2014; Roghair et al., 2018). The pH<sup>2</sup> is important because it can thermodynamically inhibit competing processes such as anaerobic oxidation of MCFAs and anaerobic oxidation of ethanol, also known as excessive ethanol oxidation (EEO; Equation 2).


Suppression of EEO is essential to make efficient use of ethanol because ethanol is a major cost factor. EEO is considered to be performed by ethanol-oxidizing microorganisms which do not perform chain elongation (Roghair et al., 2018). Earlier work demonstrated that ethanol is oxidized due to hydrogenotrophic methanogenesis (Equation 3) (Agler et al., 2014) and that the overall reaction can be referred to as syntrophic ethanol oxidation (Roghair et al., 2018) (Equation 4). By limiting the CO<sup>2</sup> loading rate to a chain elongation process, EEO was reduced from 28.8 to 15.9% of total ethanol consumption (Roghair et al., 2018). No CO<sup>2</sup> addition resulted in low and decreasing MCFA production rates. When working with organic residues, however, CO<sup>2</sup> loading rate may be more difficult to control because CO<sup>2</sup> is also a product of acidogenesis. Even though acidogenesis and chain elongation can be separated, dissolved CO<sup>2</sup> could still complicate the control over the actual CO<sup>2</sup> supply to the chain elongation process. In such a case, alternative strategies than CO<sup>2</sup> loading rate to suppress EEO are needed. Although EEO can be beneficial for ethanol upgrading processes to n-caproate (in situ ethanol oxidation into acetate and subsequent chain elongation into even-numbered fatty acids), one can consider ethanol upgrading as an inefficient use of ethanol per produced MCFA (Roghair et al., 2018). Furthermore, EEO acidifies the fermentation broth and this requires extra base addition for pH correction. Because the use of both ethanol and base cause major environmental impact over the life cycle of chain elongation processes (Chen et al., 2017), their consumption should be reduced in the development of this technology.

A high MCFA concentration in chain elongation processes would be beneficial because this improves its separation from the fermentation broth (López-Garzón and Straathof, 2014). Grootscholten et al. (2014) achieved a maximum n-caproate concentration of 12.6 g/L in a two-stage MCFA production

**Abbreviations:** VFAs, Volatile Fatty Acids; MCFAs, Medium Chain Fatty Acids; HRT, Hydraulic Retention Time; AFW, Acidified Food Waste; FW, Food Waste; NaOH, Sodium Hydroxide; EEO, Excessive Ethanol Oxidation; VS, Volatile Solids; VSS, Volatile Suspended Solids; TS, Total Solids; COD, Chemical Oxygen Demand; UASB, Upflow Anaerobic Sludge Blanket.

system from organic waste. Recently, considerably higher concentrations of n-caproate (>20 g/L) have been reported from chain elongation processes which were conducted in batch at near-neutral pH (Zhu et al., 2015; Liu et al., 2017). These studies suggest that such high concentrations can also be reached in a continuous chain elongation process as long as the hydraulic retention time (HRT) is long enough to allow product accumulation. The HRT, however, was also shown to influence the volumetric MCFA productivity as the highest reported MCFA production rate was achieved at a short HRT of 4 h (57 g/L/d) (Grootscholten et al., 2013a). A high MCFA productivity is desired to make effective use of the bioreactor; though MCFA production by chain elongation (62.4 g COD/L/d, Grootscholten et al., 2014) already exceeds the rate of methane production by anaerobic digestion (6.7–11.2 g COD/L/d<sup>1</sup> , Syngellakis, 2015). Chain elongation studies that operated at near neutral pH usually maintained an HRT shorter than 1 d (Grootscholten et al., 2013b,c, 2014; Roghair et al., 2016, 2018). To date, however, there are no studies that report the effect of a long HRT in combination with a near-neutral pH in a chain elongation process from acidified organic waste and ethanol.

The objective of this study was to develop an effective continuous biological chain elongation process from acidified food waste (AFW) and ethanol to produce n-caproate at a high concentration while EEO is limited. The effect of 2 HRTs (1 and 4 d) was compared and evaluated based on an extensive set of performance indicators including MCFA production rates, MCFA concentrations, MCFA production efficiency, substrate consumption efficiency, rate of EEO and base consumption. Finally, an outlook is given on the potential of the developed bioprocess.

## MATERIALS AND METHODS

### Preparation of Food Waste

Food waste was collected from Rotie, a recycling company in Lijnden, the Netherlands. The waste consisted of outdated food scraps and had a total solids (TS) content of 15.5 ± 0.2% (w/w), a volatile solids (VS) content of 13.8 ± 0.4% (w/w) and a sodium content of 2.7 ± 0.01 g/L. The waste was stored at −20◦C until further use. Before use as fermentation feed, the waste was thawed at 4◦C, diluted until ∼4.0% VS (w/w) with tap water and the pH was adjusted to 5.5 with 5 M NaOH.

## First Stage: Hydrolysis and Acidogenesis of Food Waste

Hydrolysis and acidogenesis of food waste was executed in a batch reactor with a working volume of 20 L as described by Chen et al. (2016a). After loading 20 L diluted food waste into the reactor, the reactor content was sparged with N<sup>2</sup> for 10 min to remove oxygen. Thereafter, 200 mL inoculum (derived from a previous hydrolysis and acidogenesis run that used the same substrate and reactor configuration) and 5 mL Antifoam B Emulsion (Sigma-Aldrich, the Netherlands) were added. The reactor was then operated at 35◦C, 1 atm, stirred at 44 rpm while the pH was maintained at 5.5 using a pH sensor (model QP-635-E275-S8, ProSense BV - QiS, Oosterhout, The Netherlands) and 5M NaOH. The slightly acidic pH was selected to inhibit methanogenesis (Agler et al., 2012; Ge et al., 2015). After 18 days of operation, reactor content (acidified food waste; AFW) was centrifuged (15,000 rpm for 15 min) and decanted to remove solids and sieved (1 mm) to remove floating particles (e.g., lipids). This was performed for in total four 20 L batches to generate sufficient AFW as feedstock for the chain elongation stage. The resulting centrifuged and sieved AFW from the four batches was pooled together and stored at 4◦C until further use. The following compounds in the pooled AFW were measured (concentration in g/L): inorganic carbon (0.011), sodium (4.8), ethanol (0.1), butanol (0.2), acetate (8.1), propionate (1.7), isobutyrate (0.6), n-butyrate (9.3), isovalerate (0.4), valerate (0.4) and n-caproate (1.4). The mentioned organic compounds account for a chemical oxygen demand (COD) of 35 gO2/L. AFW had a soluble COD of 37.1 gO2/L (LCK 014 COD, Hach Lange GMBH, Germany).

Average VS consumption in the hydrolysis and acidogenesis stage was determined based on the mean VS content at the beginning of two batches (n = 6) and on the mean VS content at the end of these two batches (n = 6). Average NaOH consumption in the hydrolysis and acidogenesis stage was determined based on the difference between the mean sodium content of diluted food waste (n = 3) and the mean sodium content of AFW (n = 3).

## Seconds Stage: Chain Elongation of Acidified Food Waste and Ethanol

Chain elongation of AFW and ethanol was performed in one single process using a continuously stirred anaerobic reactor as described by Roghair et al. (2016). In short, a continuous reactor with a working volume of 1 L was operated at 30◦C, 1 atm, stirred at 100 rpm while the pH was maintained at 6.8 using a pH sensor (Applisens model Z001023551, Schiedam, The Netherlands) and 2M NaOH. Gaseous CO<sup>2</sup> was supplied with a mass-flow controller (Brooks Instruments 5850S, the Netherlands) at 1 LCO2/L/d. The reactor was started with a synthetic medium that contained 13.8 g/L propionic acid (≥99.5%, Sigma-Aldrich) and 32.2 g/L ethanol (Absolute, VWR). These starting conditions have formerly shown to induce formation of granular sludge (Roghair et al., 2016) and can also be used to distinguish the carbon flux of ethanol upgrading from VFA upgrading (Roghair et al., 2018). The composition of other components (salts, yeast extract, vitamins and trace elements) were as previously described (Roghair et al., 2016). The reactor was inoculated in batch mode on day 1 with 50 mL chain elongation sludge from a previous run; the inoculum contained chain elongating micro-organisms, ethanol oxidizers and hydrogenotrophic methanogens (Roghair et al., 2018). On day 9, the reactor operation mode was set from batch to continuous with an HRT of 4 d. From day 19 onwards, the reactor was fed with AFW to which 32.2 g/L ethanol was added. On day 58, the HRT was set from 4 to 1 d. On day 103, the HRT was set back from 1 to 4 d.

Liquid samples were taken from the reactor content and from the influent tank 1-3 times per week. Gas samples were taken

<sup>1</sup>a Based on 5–6 m<sup>3</sup> /m<sup>3</sup> /d with a methane content of 50–70% Syngellakis, 2015

from the headspace 1-3 times per week. The reactor was assumed to be in steady state when n-caproate production rates were similar (with a maximum relative standard deviation of 16%) over a period of at least 7 HRTs. Average concentrations and rates and their corresponding standard deviations were based on at least nine measurements within a steady state. Average NaOH consumption in the chain elongation stage was determined based on the difference between average sodium concentration in the effluent (n = 3 different samples during a steady state) and on the average sodium concentration in the influent tank (n = 3 different samples during a steady state).

#### Analytical Procedures

Alcohols (C2-C6) and fatty acids (C2-C8) were analyzed by gas chromatography using an Agilent 7890B (USA), equipped with HP-FFAP capillary column (l = 25 m, ID = 0.32 mm, film = 0.5µm). 1 µl from a diluted sample was injected into a liner with glass wool at 250◦C. Vaporized compounds entered the column, along with helium as carrier gas, with a flow rate of 1.25 mL/min (first 3 min) and 2 mL/min (until the end of the run). The oven temperature program was as follows: 60◦C for 3 min; 21◦C/min up to 140◦C; 8◦C/min up to 150◦C and constant for 1.5 min; 120◦C/min up to 200◦C and constant for 1.25 min; 120◦C/min up to 240◦C and constant for 3 min. Compounds were detected with a flame ionization detector at 240◦C, fed with 30 mL/min hydrogen and 400 mL/min air.

Gaseous compounds (N2, H2, CO2, CH4, and O2) were analyzed by gas chromatography using an Agilent Varian CP4900 µGC (USA) equipped with a thermal conductivity detector and two parallel columns: a Mol Sieve 5A PLOT column (l = 10 m, ID = 0.32 mm, film = 0.15µm) and a PoraPlot U column (l = 10 m). The oven temperature was 80◦C for the Mol Sieve 5A PLOT column and 65◦C for the PoraPLOT U column. The carrier gas was argon and had a flow rate of 1.47 mL/min.

Sodium was measured by ion chromatography using a Metrohm Compact IC Flex 930 (Schiedam, the Netherlands) equipped with a pre-column (Metrohm Metrosep RP 2 Guard/3.6), a cation column (Metrosep C 4-150/4.0) and a conductivity detector. The mobile phase was 3mM nitric acid.

TS, VS, and VSS were determined following Standard Methods (APHA, 1998). The filter for VSS measurements (Whatman GF/F 0.7µm) was preheated at 450◦C prior to filtration. Inorganic carbon was measured using a total organic carbon analyser (Shimadzu TOC-VCPH, Japan).

#### Mathematical Expressions

The volumetric production or consumption rate of aqueous compounds is based on the difference between effluent concentration and influent concentration divided by the HRT:

Rate [g/L/d] = (effluent concentration [g/L] – influent concentration [g/L]) / HRT [d]

Excessive ethanol oxidation is the difference between total ethanol consumption and ethanol consumption through the reverse β-oxidation pathway:

Excessive ethanol oxidation (EEO) [g/L/d] = rate total ethanol consumption [g/L/d] – rate ethanol consumption through the reverse β-oxidation pathway [g/L/d] (Roghair et al., 2018)

Ethanol consumption through the reverse β-oxidation pathway (g/L/d) = ethanol use for elongation of fatty acids through the reverse β-oxidation pathway (g/L/d) + ethanol oxidation into acetate through the reverse β-oxidation pathway (g/L/d)

Ethanol use for elongation of fatty acids through the reverse βoxidation pathway (g/L/d) = (rate n-butyrate [mmol/L/d] + rate valerate [mmol/L/d] + 2· rate n-caproate [mmol/L/d] + 2· rate n-heptanoate [mmol/L/d] + 3 · rate n-caprylate [mmol/L/d]) · 46.05 / 1000

Ethanol oxidation into acetate through the reverse βoxidation pathway (g/L/d) = Ethanol use for elongation of fatty acids through the reverse β-oxidation pathway (g/L/d) · 0.2

Selectivity is defined as product produced relative to substrates consumed on an electron basis (Grootscholten et al., 2014):

Selectivity [mol e %] = product formation rate [mol e/L/d] / total substrate consumption rate (mol e / L/d) · 100

Selectivity values that are based on a carbon basis are reported in the supplementary material but are not presented and discussed in the results and discussion section.

Substrate consumption efficiency is defined as substrate consumed relative to the organic loading rate on an electron basis:

Substrate consumption efficiency [mol e %] = (|substrate consumption rate [mol e/L/d]| / organic loading rate [mol e/L/d] · 100

Product production efficiency is defined as product produced relative to the organic loading rate on an electron basis

Product production efficiency [mol e %] = product formation rate [mol e/L/d] / organic loading rate [mol e/L/d] · 100

### RESULTS

## Higher MCFA Concentrations and Selectivities at Long HRT Than at Short HRT

Acidified food waste (AFW) and ethanol were fed to a continuous biological chain elongation process, resulting in production of MCFAs (n-caproate, isocaproate, n-heptanoate and n-caprylate). n-Caproate, the dominant MCFA, was produced (5.5 ± 0.4 g/L/d) at a high steady state concentration of 23.4 ± 1.0 g/L. This was observed at long HRT (4 d) from day 28 through day 58 (**Figure 1**). After the HRT was decreased from 4 to 1 d (short HRT), another steady state was observed from day 79 through day 103. Here, n-caproate was produced at a similar rate (5.6 ± 0.9 g/L/d) but at a lower concentration (7.1 ± 0.9 g/L). On day 103, the HRT was increased from 1 to 4 d. Again, n-caproate was continuously produced at a high steady state concentration (23.2 ± 1.9 g/L). This was observed from day 114 through day 147. The maximum n-caproate concentration was 25.7 g/L on day 120. Reactor performance of the first steady state at long HRT was similar as the reactor performance of the second steady state at long HRT. This shows that the effect of HRT on reactor performance is reversible. From here on, however, "long HRT"

FIGURE 1 | Graphical summary of experimental results with concentrations over time (A) and net production and consumption rates over time (B). T <sup>=</sup> <sup>30</sup>◦C, pH <sup>=</sup> 6.8, V <sup>=</sup> 1 L, CO<sup>2</sup> loading rate <sup>=</sup> 1.0 LCO2/L/d.

refers to the first steady state (day 28–58). Mean steady state rates, concentrations and selectivities of all identified substrates and products at long and at short HRT are reported in Tables S1 and S2 respectively. A summary of reactor performance indicators and properties of the steady states at both long and short HRT are reported in **Table 1** for comparison.

At long HRT, n-heptanoate and n-caprylate were both produced at a low rate (∼0.15 g/L/d) and concentration (∼0.6 g/L) compared to n-caproate. At short HRT, however, these MCFAs were produced at insignificant amounts (<0.1 g/L). Isocaproate was produced in trace amounts at both long and short HRT. The selectivity of MCFAs was 81.6 mol e % at long HRT and 46.3 mol e % at short HRT. The remaining consumed carbon ended up as VSS (biomass), VFAs, methane, propanol, and unidentified products as given in Tables S1 and S2. The MCFA productivity at both long and short HRT (∼12.5 g COD/L/d) was somewhat higher compared to the typical methane productivity in an anaerobic digester (6.7–11.2 g COD/L/d<sup>1</sup> , Syngellakis, 2015).

## Excessive Ethanol Oxidation, Methane Production and NaOH Consumption

Long HRT did not only result in higher MCFA concentrations and selectivities compared to short HRT, it also resulted in less EEO and in less sodium hydroxide (NaOH) consumption per MCFA produced. EEO occurred at a 6.3 times lower rate at long HRT (0.9 ± 0.4 g/L/d) than at short HRT (5.6 ± 1.4 g/L/d). This process was also relative to the total ethanol consumption rate lower at long HRT (14.7 ± 5.5%) than at short HRT (45.0 ± 9.7%). Methane production was 3.4 times lower at long HRT (12.8 ± 2.6 mmol/L/d) than at short HRT (43.8 ± 2.5 mmol/L/d). A previous study showed that hydrogenotrophic methanogenesis and EEO are coupled and that the overall reaction can be referred to as syntrophic ethanol oxidation (Roghair et al., 2018). In syntrophic ethanol oxidation, the stoichiometric ratio between methane production and ethanol oxidation is 0.5 mol/mol (Equation 4). The present study shows a similar ratio at long HRT (0.7 ± 0.3) and at short HRT (0.4 ± 0.1) which implies that EEO, like in the previous study, was a result of syntrophic ethanol oxidation. Less EEO also led to fewer NaOH consumption for pH correction because EEO is an acidifying process that releases a proton (Equation 2). NaOH consumption per produced MCFA was two times lower at long HRT (0.92 ± 0.04 molNaOH/molMCFA) than at short HRT (1.93 ± 0.31 molNaOH/molMCFA). These results not only show that EEO can be limited at long HRT but also that consumption of NaOH is hereby reduced.

### Consumption of VFAs and Ethanol

Acetate, propionate and ethanol were consumed at both long and short HRT. However, whereas butyrate was consumed at long HRT (−0.9 ± 0.1 g/L/d), it was produced at short HRT (1.6 ± 1.1 g/L/d). This resulted in a lower n-butyrate concentration in the reactor at long HRT (5.6 ± 0.6 g/L) than at short HRT (10.8 ± 1.0 g/L). A net consumption of n-butyrate instead of production indicates that caproate production is more efficient in ethanoluse. This is because n-butyrate consumption (i.e., elongation) requires less ethanol than acetate elongation to caproate or ethanol upgrading. Whereas ethanol upgrading requires 3 moles of ethanol to produce 1 mole of n-caproate, VFA upgrading requires only 1.2 moles of ethanol from n-butyrate or 2.4 moles of ethanol from acetate (Roghair et al., 2018). Indeed, the chain elongation process at long HRT was more efficient in ethanoluse than at short HRT. Approximately two times less ethanol was consumed per produced MCFA at long HRT (0.87 ± 0.07 mol C/mol C) than at short HRT (1.83 ± 0.31 mol C/mol C). The VFA consumption per produced MCFA at both long (0.29 ± 0.04 mol C/mol C) and short HRT (0.20 ± 0.07 mol C/mol C) were found to be similar.

The concentration of ethanol in the reactor (and thus also in the effluent) was much lower at long HRT (2.8 ± 1.1 g/L) than at short HRT (20.1 ± 1.6 g/L). The ethanol consumption efficiency, therefore, defined as consumed ethanol relative to supplied ethanol, was higher at long HRT (98.6 ± 5.4 mol e %) than at short HRT (40.3 ± 3.5 mol e %). A high ethanol consumption efficiency (or a low ethanol concentration in the effluent) is desired because any unconsumed ethanol requires an additional recovery or treatment step after the chain elongation stage which makes the overall process more expensive. The VFA consumption efficiency was also higher at long HRT (45.8 ± 6.9 mol e %) than at short HRT (7.2 ± 2.3 mol e %). Although VFAs are not as costly as ethanol, it is evident that a higher VFA consumption efficiency is preferred because the remaining VFAs (e.g., after selective extraction of the MCFAs) also have to be recovered or treated with a waste water treatment system.

### VSS

The mean VSS concentration in the reactor at long HRT (0.34 ± 0.23 g/L) was similar compared to the mean VSS concentration TABLE 1 | Performance indicators and properties of the chain elongation process at long and at short HRT.


in the reactor at short HRT (0.33 ± 0.03 g/L). These reactor concentrations were in the same order of magnitude as the VSS concentrations in the effluent (0.43 ± 0.32 g/L at long HRT and 0.35 ± 0.20 g/L at short HRT), implying that the reactor was ideally stirred with no biomass retention (i.e., CSTR). Formation of granular sludge, however, was observed like in the earlier experiment with the same set-up while using a synthetic medium (Roghair et al., 2016). The earliest observation of granules (by eye visible) was on day 82, at short HRT. Granules disappeared within a few days after the HRT was increased on day 103. Because the formation of granular sludge coincided with highrate syntrophic ethanol oxidation (at short HRT) it is likely that this syntrophic process attributed to sludge granulation. Syntrophic processes may benefit from granulation because granules could facilitate a more efficient interspecies hydrogen transfer due to the decreased intermicrobial distances (Kouzuma et al., 2015).

The mean VSS concentration in the influent was 0.25 ± 0.06 g/L. From the mentioned values the VSS production rate and VSS specific growth rate were calculated (**Table 1**). Although one could expect a four times lower growth rate at long HRT (0.07 ± 0.16 g/g/d) than at short HRT (0.54 ± 1.0 g/g/d), there was no significant difference due to the large standard deviations.

## DISCUSSION

## Continuous n-Caproate Production at a High Concentration From Acidified Food Waste

In this study, an effective two-stage MCFA production process from food waste and ethanol was developed. The microbiome in the second stage (chain elongation) was able to continuously produce n-caproate at 23.4 ± 1.0 g/L while EEO was limited to



*N.A., Not applicable.*

*N.D., Not determined.*

14.7 ± 5.5% of total ethanol consumption. This was achieved at long HRT (4 d) and at near-neutral pH (6.8) but without in-line product extraction. Thus, a long HRT was shown to be effective for this specific waste stream. The n-caproate production rate was similar for both long and short HRT (∼5.5 g/L/d) and was much lower compared to the highest reported n-caproate production rate to date (55.8 g/L/d) (Grootscholten et al., 2013a). This high production rate, however, occurred at a substantially lower ncaproate concentration (9.3 g/L) than obtained in the present study. An overview of comparable studies that reported high n-caproate concentrations and/or rates using open cultures is shown in **Table 2**.

A high n-caproate concentration is beneficial for its separation from the fermentation broth. The obtained n-caproate concentration in the present study is more than twice as high as the maximum solubility of the undissociated form of n-caproate, n-caproic acid, in water (∼10.8 g/L, Yalkowsky et al., 2016). Thus, in theory, the high n-caproate concentration allows it to be separated from the effluent by phase separation after lowering the pH to 4.9 or lower. A recent study, executed by Zhu et al. (2015), also reported a high n-caproate concentration (23.4 g/L) using an anaerobic reactor microbiome. The n-caproate was produced from lactate as the sole carbon source in a batch process using an inoculum that was derived from mature pit mud, a microbiome used for the production of Chinese strong-flavored liquor. Their process reached a higher n-caproate selectivity (81.4 mol e %) than the process in the present study at long HRT (76.5 mol e %). However, the process in the present study achieved a slightly higher MCFA selectivity (81.6 mol e %) because it also produced other MCFAs (isocaproate, n-heptanoate and caprylate).

Liu et al. (2017) also reported a high n-caproate concentration (21.2 g/L) from ethanol and acetate using an anaerobic reactor microbiome. They were able to produce this in a batch process upon addition of biochar and 2-bromoethanesulfonate (Liu et al., 2017). The maximum n-caproate selectivity was lower (65.0 mol e %) than the maximum n-caproate selectivity in the present study (76.5 mol e %).

The studies by Zhu et al. (2015), and by Liu et al. (2017) already showed that high n-caproate concentrations (>20 g/L) can be reached using anaerobic reactor microbiomes. However, they were using synthetic media and batch systems. As such, the present study does not only show that such high concentrations can be obtained from organic residues, it also shows that this can be obtained in a continuous process and without the use of bioactive compounds such as 2-bromoethanesulfonate. The organic residue that was used, food waste, is a suitable substrate for MCFA production not only because such conversion was recently subjected to a life cycle assessment (Chen et al., 2017) but also because it is currently being developed to a demonstration factory, processing ∼40 ton/day, by ChainCraft in Amsterdam, the Netherlands.

#### Why Was Reactor Performance so Much Better at Long HRT Than at Short HRT?

The performance of the chain elongation process was far better at long HRT than at short HRT. A long HRT did not only result in a higher concentration of MCFAs, it also led to a lower rate of syntrophic ethanol oxidation, a net n-butyrate consumption instead of production, and to less NaOH consumption for pH correction. Why was reactor performance so much better at long HRT?

The difference in MCFA concentration can be explained as follows: at long HRT, the microbiome had sufficient time to accumulate MCFAs while these products were washed out (as effluent) at a relatively low rate. This resulted in a high caproate concentration (23.4 ± 1.1 g/L). Vice versa, at short HRT, the microbiome had little time to accumulate MCFAs while these products were washed out at a relatively high rate. This resulted in a considerably lower caproate concentration (7.1 ± 0.9 g/L). Because production of n-caproate occurred at the same volumetric production rate at both long and short HRT (∼5.5 g/L/d), it is a logical consequence that n-caproate reached a higher concentration at long HRT than at short HRT. The high n-caproate concentration could also be achieved because the process was not limited by availability of substrates as sufficient ethanol (2.8 ± 1.1 g/L), acetate (4.2 ± 0.8 g/L) and n-butyrate (5.6 ± 0.6 g/L) was observed in the reactor. At short HRT, however, substantially more ethanol (20.1 ± 1.6 g/L), acetate (7.3 ± 0.5 g/L) and n-butyrate (10.8 ± 1.0 g/L) was observed but no higher MCFA production rates. This seems to be limited by the biomass concentration or by the high (i.e., inhibitory) ethanol concentration (Lonkar et al., 2016).

Syntrophic ethanol oxidation was more limited at long HRT than at short HRT. Two explanations can be given: firstly, the low ethanol concentration at long HRT may have resulted in a low rate of EEO. Vice versa, the high ethanol concentration at short HRT may have resulted in a high rate of EEO. However, a previous chain elongation study showed that ethanol loading rate and ethanol concentration does not substantially influence reactor performance as long as ethanol is not depleted (Roghair et al., 2018). Secondly, the high MCFA concentration may have caused an inhibitory effect on (one of) the involved competing syntrophs (i.e., ethanol oxidizers and hydrogenotrophic methanogens). It is known that undissociated MCFAs are toxic to microorganisms because they can damage the cell membrane (Royce et al., 2013). The average undissociated MCFA concentration at long HRT (at pH 6.8) was 0.27 g/L. Ge et al. (2015) determined that chain elongation proceeds until a toxic limit of 0.87 g/L undissociated n-caproic acid is reached (Ge et al., 2015) although recent work demonstrated chain elongation activity up to 1.46 g/L undissociated n-caproic acid (Andersen et al., 2017). Because chain elongating microorganisms were producing MCFAs at ∼5.8 g/L/d in the present study, evidently these organisms were not rigorously inhibited by the undissociated MCFAs. It is well possible, however, that these undissociated MCFAs (at 0.27 g/L) were selectively inhibitory to either ethanol oxidizers or hydrogenotrophic methanogens. This would explain the limited rate of syntrophic ethanol oxidation at long HRT compared to a short HRT while chain elongation could proceed at a similar rate at both HRTs. It is also possible that the dissociated form of n-caproate was selectively inhibitory to one of the syntrophs. This means that dissociated n-caproate (i.e., the conjugate base) becomes toxic to ethanol oxidizers or hydrogenotrophic methanogens at a concentration around 20 g/L.

The data is not consistent to point out whether hydrogenotrophic methanogens or ethanol oxidizers were more inhibited at long HRT: whereas the first steady state at long HRT (day 28 to day 58) had a pH<sup>2</sup> below the detection limit of the gas chromatograph (<0.1%), the second steady state at long HRT (day 138–147) had a pH<sup>2</sup> of up to 30%. This means that the data from the first steady state suggests that ethanol oxidizers were more inhibited whereas data from the second steady state suggests that hydrogenotrophic methanogens were more inhibited. To what extent n-caproic acid and n-caproate is toxic to hydrogenotrophic methanogens and ethanol oxidizers could be elucidated in further studies. In any way, irrespective how syntrophic ethanol oxidation was limited, it is clear from the results that longer HRTs should be applied in chain elongation processes at near-neutral pH to allow n-caproate accumulation and to limit the rate of syntrophic ethanol oxidation. This limited rate of syntrophic ethanol oxidation results in a more efficient use of ethanol, more VFA consumption and in less base addition for pH control and as such, in a more effective chain elongation process.

## Methanogenic UASB Sludge Can Acclimate Into a Reactor Microbiome That Is Able to Produce n-Caproate at High Concentrations

This study can be compared with a previous study (Grootscholten et al., 2014). Both studies focused on twostage MCFA production from organic residues and ethanol using anaerobic reactor microbiomes. In the previous study, a continuous chain elongation process was operated at similar pH (6.5–7.0) but at shorter HRT than in the present study (11 h instead of 1 and 4 d). This resulted in a lower maximum n-caproate concentration (12.6 g/L) and selectivity (72.0 mol e %). Besides, n-butyrate was produced instead of consumed, indicating that the process was not efficient in ethanol use. A hypothesis is that the process in the previous study could have performed better, including a high n-caproate concentration, if both the HRT and the ethanol concentration in the influent were increased. Of course, the origin of the inoculum may also have an effect on reactor performance because this determines which pathways or microorganisms are introduced and to what extent the initial microbiome is acclimated. The inoculum used by Grootscholten et al. (2014) and also the inoculum used in the present study were both eventually derived from a study executed by Steinbusch et al. (2011), who used granular sludge from an upflow anaerobic sludge blanket (UASB) reactor treating brewery wastewater (Steinbusch et al., 2011). This shows that high n-caproate concentrations can be obtained using various types of inocula and not only with mature pit mud (e.g., Zhu et al., 2015) or with mesophilic sludge from an anaerobic reactor treating paper mill wastewater (e.g., Liu et al., 2017). Under the right circumstances (e.g., as described in this study), UASB sludge from a methanogenic reactor will likely acclimate within a matter of weeks to a reactor microbiome that is able to produce n-caproate at high concentrations (>20 g/L).

## Consumption of Food Waste, Ethanol and Base in the Overall Two-Stage System

In this study, MCFAs were produced from food waste and ethanol using a two-stage system. In the hydrolysis and acidogenesis stage, part of the food waste was converted into VFAs while some MCFAs were also produced. In the chain elongation stage, a part of the VFAs from the hydrolysis and acidogenesis stage were converted with additional ethanol into MCFAs while some ethanol was also used for ethanol upgrading. Both stages required NaOH addition to keep the pH constant in the reactors. Based

on the results of this study it was possible to calculate how much food waste (expressed as g VS), ethanol and NaOH would be consumed to yield 1,000 g MCFAs. This was done for two scenarios; at long and at short HRT in the chain elongation stage. Equations are shown in Table S3. Parameters that were used as input for the equations are shown in Table S4 and were based on:


A graphical representation on how much food waste (in g VS), ethanol and NaOH would be consumed to yield 1,000 g MCFAs at long and short HRT in the chain elongation stage is shown in **Figure 2**.

To produce 1,000 g MCFAs, the two-stage system with a long HRT in the chain elongation stage would consume 53% less ethanol, 41% less NaOH and 42% more VS compared to a two-stage system with a short HRT in the chain elongation stage. This does not only show that a long HRT in the chain elongation stage result in a more efficient consumption of ethanol and base in the chain elongation stage itself, it also shows that base consumption is hereby reduced in the overall twostage system. Unfortunately, such quantification on waste-use and chemical-use is not common in chain elongation studies to date.

Addition of ethanol could be minimized or even completely avoided by steering the hydrolysis and acidogenesis stage to MCFAs and/or lactate production. In this study, for example, 1.6 g/L n-caproate was produced in the hydrolysis and acidogenesis stage without addition of an external electron donor. Lim et al. (2008) also demonstrated production of n-caproate (up to 5 g/L) from food waste without an external electron donor (Lim et al., 2008). Xu et al. (2017) demonstrated conversion of acid whey waste into MCFAs via lactate (Xu et al., 2017) using a two-stage system; also without an external electron donor. This shows that the effectiveness of MCFA production processes from diverse organic waste streams can be further improved by optimizing the hydrolysis and acidogenesis stage. Studies should also report on how much base or acid or electricity was used for pH control for a better comparison in terms of effectiveness.

Chemical base consumption could be fully eliminated through membrane electrolysis using electricity and separation of fatty acids, as was demonstrated by Andersen et al. (2015). They fermented thin stillage into VFAs and MCFAs using a membrane electrolysis system and no chemical pH control. Such system, however, consumes a substantial amount of energy. Based on their experiments, they estimated a power input of 2 kWh per produced kg CODfatty acids. The developed twostage system in the present study, at long HRT, consumed 0.22 kg NaOH per produced kg CODfatty acids. Assuming an electricity consumption of 3 kWh/kg NaOH via the chloralkali process (Euro Chlor, 2010), the two-stage system would require less energy for pH control per produced kg COD (0.67 kWh/kg CODfatty acids) compared to the in-site membrane electrolysis system. However, whereas the membrane electrolysis system already separated fatty acids from the fermentation broth, the two-stage system would require an additional product separation step to be comparable in electricity consumption.

Effective MCFA production from organic waste can be further developed by reducing the need for chemicals and/or electricity. Possibly, also water-use can be reduced too since the two-stage system used a substantial amount of water to dilute the waste before use as fermentation feed. The presented scenario in the present study, as well as the mentioned alternative scenarios (e.g., Lim et al., 2008; Andersen et al., 2015; Xu et al., 2017) can be optimized and assessed with a case-specific life cycle assessment to make a complete justified discussion on the total environmental impact.

#### Future Outlook

In a previous study, it was shown that EEO in a chain elongation process can be limited to 15.9% of total ethanol consumption by reducing the CO<sup>2</sup> loading rate to 0.5 LCO2/L/d (Roghair et al., 2018). In the present study, an alternative strategy to suppress EEO is provided: by applying a long HRT, EEO was limited to a similar extent (14.7%). A major advantage of this strategy is that the n-caproate concentration can become much higher. By further increasing the HRT possibly higher n-caproate concentrations can be reached. This could potentially lead to even more limited rate of EEO and thus base consumption in the chain elongation process. Of course, this is only feasible when the process is not limited in substrates (ethanol and VFAs) and CO2.

Based on this study and the availability of 88 million ton wet food waste per year in the European Union (Stenmarck et al., 2017), the developed process has the prospects to produce 29 million ton MCFAs per year. This was calculated using the ratios in **Figure 2** (at long HRT) and by assuming that wet food waste has the same VS content as in this study; calculations are shown in the supplementary material. The required ethanol (∼311 million barrels) would be 37% of total annual global ethanol production (∼844 million barrels, Renewable Fuels Association, 2017). The required electrical power for NaOH production (38.4 tWh) would be 13% of total annual wind energy production in Europe (305.8 tWh, International Energy Agency, 2015).

After selective extraction of the MCFAs, they can be further processed into a fuel (i.e., a mixture of hydrocarbons) via Kolbe electrolysis (Khor et al., 2017; Urban et al., 2017). Assuming no losses during extraction and a theoretical efficiency of 0.61 ghydrocarbons/gMCFA in the Kolbe electrolysis process and a fuel density of 0.73 g/cm<sup>3</sup> , it is possible to produce ∼202 million barrels of fuel per year. This is approximately 7% of the

#### REFERENCES


nowadays annual aviation fuel consumption (∼2,710 million barrels, International Air Transport Association, 2017).

#### AUTHOR CONTRIBUTIONS

MR planned and performed the experiments, analyzed the results, and wrote the manuscript. YL assisted in experimental work and analytical work. DS assisted in the design of the study and revised the manuscript. MB, RW, and CB revised the manuscript. All authors read and approved the final manuscript.

#### FUNDING

This work has been carried out with a grant from the BE-BASIC program FS 01.006 (www.be-basic.org).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbioe. 2018.00050/full#supplementary-material

Biomass Bioenergy 93, 201–208. doi: 10.1016/j.biombioe.2016. 07.008


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Roghair, Liu, Strik, Weusthuis, Bruins and Buisman. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Pretreatment and Multi-Feed Anaerobic Co-digestion of Agro-Industrial Residual Biomass for Improved Biomethanation and Kinetic Analysis

#### Kishan Kumar Prajapati <sup>1</sup> , Nidhi Pareek <sup>2</sup> and Vivekanand Vivekanand<sup>1</sup> \*

*<sup>1</sup> Centre for Energy and Environment, Malaviya National Institute of Technology, Jaipur, India, <sup>2</sup> Department of Microbiology, School of Life Sciences, Central University of Rajasthan Bandarsindri, Kishangarh, India*

Batch biochemical methane potential (BMP) test of agro-industrial, agricultural, and municipal solid waste (MSW) individuals like groundnut straw, rice bran, and guar husk, mung bean husk (MBH), wheat straw (WS), and organic fraction of municipal solid waste (OFMSW), under mesophilic conditions, were performed to evaluate the biogas potential. BMP test for multi-feed anaerobic co-digestion (AcoD) of WS, OFMSW, and MBH at five mixing ratio was performed to evaluate the synergistic effect of multi feed-stocks of blending the feedstocks. Mixture ratio having OFMSW:WS:MBH of 25:5:70% composition resulted into 37, 20, and 4% higher methane yield up to 280 ml/g Volatile solid (VS) in comparison with mono-digestion of OFMSW, WS and MBH. Ultrasonic pretreatment was also performed and the experimental results showed that varying the sonication time have a significant improvement on substrate's biodegradation and solubilization augmenting the methane yield from OFMSW, WS, and MBH by 71, 75, and 46%, respectively at sonication period of 60 min. Effect of pretreatment on the substrate studied through Scanning electron microscopy (SEM), XRD and FT-IR analysis. Cone and Exponential models showed an average coefficient of determination *R* 2 = 0.9855 and *R* <sup>2</sup> = 0.9893, respectively, and RMSE values for cone model was lower than that of Exponential model showing Cone model was more précised.

Keywords: agricultural waste, agro-industrial waste, organic fraction of municipal solid waste, anaerobic digestion, ultrasonic pretreatment, kinetics

## INTRODUCTION

Developing country India facing energy security problem and striving to fulfill its energy demand by natural resources like coal, crude oil, and natural gas. Renewable energy sources like wind, solar, hydro, biomass, and geothermal have huge potential to wipe out this problem of energy demand with sustainability and an excellent alternative to declining fossil fuel reserves. Biomass is considered as carbon neutral for energy production through biogas, ethanol, and pellets production, still, its direct burning practiced over the country, its conversion to renewable bioenergy production is a sustainable way. As 686 MT (million tons) biomass is generated in the country annually (Hiloidhari et al., 2014; Wang et al., 2017; Kumar et al., 2018). Agricultural crops residues, organic fraction of municipal solid waste (OFMSW), food wastes, agro-industrial wastes, aquatic plants, and algae, animal dung etc. all are considered as biomass.

#### Edited by:

*Mohammad Rehan, King Abdulaziz University, Saudi Arabia*

#### Reviewed by:

*Qaisar Mahmood, COMSATS University Islamabad, Pakistan Samir Jaiwantrao Deshmukh, Prof.Ram Meghe Institute of Technology & Research, India*

\*Correspondence:

*Vivekanand Vivekanand vivekanand.cee@mnit.ac.in*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *30 June 2018* Accepted: *02 October 2018* Published: *23 October 2018*

#### Citation:

*Prajapati KK, Pareek N and Vivekanand V (2018) Pretreatment and Multi-Feed Anaerobic Co-digestion of Agro-Industrial Residual Biomass for Improved Biomethanation and Kinetic Analysis. Front. Energy Res. 6:111. doi: 10.3389/fenrg.2018.00111*

Agriculture, MSW, and agro-industrial waste materials embody potential sources of renewable energy production and making a contribution toward energy production independent from fossil fuels.

## Organic Waste Generation

Agriculture biomass or waste are alternative feedstock to energy and chemical production. Leftover straw, stalk and husk of a crop plant are the crop residues after its harvesting (Devi et al., 2017). Agricultural wastes are agriculture crop residues i.e. residues from group of cereals, oilseeds, pulses, horticulture, and other crops. Most of the agricultural residues used for animal feeding, fuel for domestic application or gasification (Hiloidhari et al., 2014; Report RRECL, 2017).

Around 25% of biomass from crop residues is burnt on a global basis (Devi et al., 2017). Its on-farm burning, categorized as underutilization of resources and leading to environmental impacts (Hiloidhari et al., 2014; Kumar et al., 2018). Agricultural crop residues can be utilized into biogas production, textile making process, fuel and manure, etc. to mitigate the problem of on farm in-situ burning of surplus crop residues which is a source of GHG, smoke, aerosols, and particulate matters etc. (Nguyen et al., 2016; Devi et al., 2017). Animal dung from cattle, goats, ships, and poultry are highly loaded from organic matter content and having huge pollution potential of methane gas. It is utilized as potential substrate or inoculum in co-digestion in the biogas plants (Bundhoo et al., 2016). Also, Shah et al. (2015) has reported the anaerobic co-digestion of blends of maize residues, water hyacinth, giant reed, and poultry litter for biogas production at community scale.

India generates crop residues 686 MT annually, out of which 34% of crop residue is estimated as surplus (Hiloidhari et al., 2014). For Rajasthan state, agricultural activity generates 52.6 MT/year residues out of which 9.25% remain surplus (Report RRECL, 2017). The surplus fraction of residue availability from cereals is 29%, from oilseeds is 18%, and 23% from pulses crop in Rajasthan (Hiloidhari et al., 2014). **Wheat** (Triticum) crops cultivation is 2nd highest among cereals in the country, and 5th highest in the Rajasthan state. Wheat straw [Crop Residue Ratio (CRR) = 1.5] production is 15.8 MT/year having 30% share to total biomass generation in the state (Annual Report MA FW, 2017; Report RRECL, 2017). Wheat straw has composition of 30.1–39.2% cellulose, 22.2–34.0% hemicellulose, and 6.5–22.1% lignin (Chandra et al., 2012a,b; Bolado-Rodríguez et al., 2016; Yadav et al., 2018).

**MSW** generation is increasing in almost all cities with urbanization and population growth, accompanied by its economy. MSW handling is one of the chief issue facing Indian cities. Its poor and incautious management entails enormous impacts on public health, environment, and climate change (Shekdar, 2009; Yadav and Samadder, 2017; Gollapalli and Kota, 2018). Currently, India generates 62 MT MSW per year where collection and treatment efficiency is around 90 and 27%, respectively, which means 79% MSW directly goes to unsanitary landfilling (Planning Commission Report, 2014; MSW Report, 2017; Yadav and Samadder, 2017). Jaipur, one of the metro cities of India, generates 1,000 tons per day (TPD) MSW, 31% share alone in total state MSW generation with ∼3.04 million population (Census, 2011; Municipal solid waste, 2017). Waste to energy potential in India with organic and inorganic portions are open for conversion to biofuel and for power generation. Presently, India has 2,554 MW potential as renewable energy potential from waste generated i.e., MSW and waste water (Kalyani and Pandey, 2014; MNRE Annual Report, 2017).

Improved income and population in a city led to a changed lifestyle of urban dwellers, and also causing an escalation in waste quantity and MSW composition altered. Knowledge of the composition of MSW is important for selecting the suitable waste processing and disposal practices since MSW volume and its composition differs considerably with the places having changes in food habits, cultural traditions, lifestyles, socio-economic conditions, and climate (Yay, 2015; Mboowa et al., 2017). OFMSW has composition around 17.5% cellulose, 10.7% hemicellulose, 9.6% lignin, 37.8% raw fiber, 7.7% protein, and 9.6% fat or oil (Rao and Singh, 2004). However, its composition varies with their nutrient content due to its heterogeneous nature with respect to place to place even in the same city. Through AD process, OFMSW is capable of generating 300–400 m<sup>3</sup> biogas/ ton of VS of OFMSW (Dhar et al., 2017).

The **agro-industrial** sector represents one of the important sectors of the economy, responsible for processing or production of different foodstuffs and meanwhile also generating large quantity of waste. Major agro-industrial waste comes from food processing, juice and wine industry, edible oil refineries, slaughterhouse, and dairy processing. Many food processing wastes are sent to animal feeding, and chemical production and rest goes to landfilling (Bundhoo et al., 2016; Pellera and Gidarakos, 2017). Although glycerine as ideal co-substrate being pure and 100% degradability, other products like cheese whey, olive mill waste, sugar by-products, have been used for codigestion with positive results (Mata-Alvarez et al., 2014). Here, Mung bean husk, groundnut shell, guar husk and rice bran have been studied as agro-industrial waste for biogas potential.

**Mung bean** (Vigna radiata) production is 4,77,245 tons sharing 20% of pulse production as 2nd highest production after gram pulse which has 44% share in pulses production in the state (Rajasthan Agri-Statistics Report, 2017). **Groundnut** (Arachis hypogea) crop production is 8,08,143 tons production in Rajasthan with 15.56% share to country (Rajasthan Agri-Statistics Report, 2017). Its shells (CRR = 0.3) are discarded waste, which is a carbohydrate rich biomass having composition 25.57% carbohydrate, crude protein 4.43% and lipid 0.50%, crude fiber 59%, ash content 2.5%, and moisture 8% (Abdulrazak et al., 2014; Hiloidhari et al., 2014). **Guar** or cluster bean, with the botanical name Cyamopsis tetragonoloba drought tolerant legume grown in the Rajasthan state of India. Guar meal is a protein and fiber rich nutritional food for livestock containing germ and hull part of it (Janampet et al., 2016). Rajasthan produces Guar crop 2,204,931 tons annually in 2014–15. Its residues are stalks and husk. **Rice** (Oryza sativa) crop is most widely consumed food grain in the world. This cereal has 103.73 MT cultivation in India being 2nd highest in India after sugarcane crop. Its production is 2,84,131 tons annually in Rajasthan state. Residues from rice milling industry are rice husk and bran. (Rajasthan Agri-Statistics Report, 2017). Rice bran has composition around 4.6% cellulose, 8.4% hemicellulose, 2.8% lignin, 28.5% starch, and 13.9% protein (Favaro et al., 2017).

## Anaerobic Digestion

The AD is biological process categorized in the effective waste management as well as waste to energy technologies. As it avoids the natural emission of greenhouse gases like CH<sup>4</sup> and CO<sup>2</sup> from self-degradation of organic waste. Main products of AD process are CH<sup>4</sup> for renewable energy production and a digestate as bio-fertilizer for soil amendment (Vivekanand et al., 2013; Bundhoo et al., 2016). As AD owing several benefits concerning environment, economy and maintenance, also have been used for processing agricultural waste, MSW, sewage sludge, livestock effluents, agro-industrial waste, therefore AD process being dealt in this study.

The AD consists of four consecutive processes i.e., Hydrolysis, acidogenesis, acetogenesis, and methanation as shown in **Figure 1**. In first phase, large polymers breakdown into monomers, carbohydrates into simple sugar, fat into fatty acid, and glycerol, protein into amino acid and peptide. In second phase, these monomers break down into carbonic acid, alcohols, H2, CO2, NH3. In this phase, O<sup>2</sup> and CO<sup>2</sup> consumed by facultative bacteria to create the anaerobic condition. In third phase i.e., acetogenesis, previous phase products are converted into acetic acid, H<sup>2</sup> and CO2. This acetic acid and hydrogen are converted into methane by different methanogens i.e. archea in the last phase of AD i.e., methanogenesis (Kumar et al., 2018; Negi et al., 2018).

Nutritional imbalance and operating factors like biodegradability and chemical composition of single substrate make its direct utilization tough. Therefore, different nutrition composition substrates are digested together, also biogas yield gets improved. Anaerobic co-digestion provides benefits like process stabilization, pH and buffer capacity, avoid inhibitory or toxic compounds (formation of Volatile fatty acids (VFA)/ammonia), microorganism's good synergistic effect (Mata-Alvarez et al., 2014; Hagos et al., 2017; Pellera and Gidarakos, 2017). AcoD is a technique for simultaneous treatment of different organic residues proving better synergism through diverse co-substrates and microbial activity by improving methane yield in process. Suitable co-substrate is that which compensate with excess metabolites produced by another substrate which inhibit the methanogens in AD (Zhang et al., 2016). In simple words, AcoD is performed to balance the C/N ratio of the feedstock.

BMP test is a batch experiment of AD determining the biomethane potential of a substrate. This test is performed in laboratory for examining the biodegradability and methane conversion efficiency of organic substrates, assessing different combination of co-substrates, feasibility of different biomasses for AD, also helps in achieving optimal condition for AD process (Cho et al., 2013; Vivekanand et al., 2013; Hagos et al., 2017). A feasibility study of organic waste energy potential in college campus shows mixed food and green waste has biogas potential of 400–660 ml/g VS of the mixed substrate (Paritosh et al., 2018).

Kacprzak et al. (2010), experimented combination of agriculture waste (corn silage, carrot residues), agro-industrial waste (beet pulp silage and cheese whey) with industrial waste: glycerine (waste from biodiesel production), resulting highest methane yield with mixture of corn silage, cheese whey and glycerine (Kacprzak et al., 2010; Mata-Alvarez et al., 2014). Blending of few agro-industrial waste i.e. brewery spent grain, carbonated soft drink sludge, with either powdered rice husk or soya bean cake resulted into enhanced biogas production than individuals (Uzodinma et al., 2007). Vivekanand et al. (2018) co-digested manure, fish ensilage and whey as multi feed-stocks and produced 84% higher methane production than individual substrate digestion. Food waste co-digestion with straw (containing straw of wheat, maize, and sorgos) showed 39.5 and 149.7% augmented methane yield at 5:1 mixing ratio as compared to mono-digestion of food waste and straw, respectively (Yong et al., 2015). Co-digestion of OFMSW and rice straw in the ratio of 2:1 yielded 57% higher methane (403 ml/g VS) than the other combinations (Negi et al., 2018). Batch test of co-digestion of OFMSW with pretreated sludge and rice straw experimented at different blending ratio, and ratio 3:0.5:0.5 yields maximum biogas 558 ml/g VS (Abudi et al., 2016). Shekdar (2009) has reported the different treatment methods for improving the biogas production from OFMSW. MSW, bovine slaughterhouse, manure and crop residues were mixed and evaluated for methane production in mesophilic

and thermophilic condition and mesophilic condition results showed 57% lower methane in comparison with the thermophilic condition (Pagés-Díaz et al., 2013). Methane yield from corn straw and manure get increased up to 22.4% by means of mixing another substrate i.e., fruit and vegetable waste (Wang et al., 2018).

The organic fraction of municipal waste and agro-industrial waste are mostly investigated and used for co-substrates with manure (Mata-Alvarez et al., 2014; Abudi et al., 2016). Selection of suitable co-substrates and the ratio of mixing are key factors aiming synergism to enhanced methane production. Municipal, agro-industrial and agricultural waste have been individually


TABLE 1 | Characterization of feed-stocks.

\**O estimated by deducting other constituents from 100%, NA-Not Analyzed.*

utilized as a source of energy through this technique. So far little work performed regarding synergistic effect from these three different sector waste or biomass (Sharholy et al., 2008; Vivekanand et al., 2018).

#### Pretreatment Process

Utilizing bioenergy from biomass with higher efficiency is also a major concern, for this different pretreatment techniques are applied. Lignocellulosic biomass composed of cellulose, hemicellulose, and lignin, where lignin has the amorphous 3D structure of phenylpropanoid units covering cellulose fibers hindering access from enzymes. For deconstructing lignin structure, biological, physical or chemical, or physiochemical pretreatment techniques are available (Bussemaker and Zhang, 2013; Kumar et al., 2018; Yadav et al., 2018).

#### Ultrasonic Pretreatment

Sound waves or energy having frequency beyond 20 kHz, inaudible to human beings, called as ultrasonic waves. These waves are responsible for producing vibration in the suspension leading the formation of cavitation bubbles of vapor. The formation of microbubbles and its growth and collapse called cavitation effect, it occurs only due to more ultrasonic energy (e.g., 1 W/cm<sup>3</sup> for water) than molecular attractive forces. Microbubbles collapse during wave compression generating high temperature and pressure spot on the biomass disrupting the outer structure. This cavitation delivers physical effects on the substrate due to shear forces to the surface of the substrate leading its structure lysis (Nakashima et al., 2016; Rodriguez et al., 2017; Wang et al., 2017).

Depending on the ultrasonication condition, effective physical disintegration can be achieved augmenting AD yield. It was also proven to be effective and versatile in comparison with other pretreatment such as acid, base thermal and bacterial for fat-rich solid substrates especially waste coming from meat processing (Cesaro and Belgiorno, 2014). Pretreatment parameters should require to be near at room temperature and pressure, mild severity with significant delignification. Ultrasonication exclusiveness is due to its parameters ranges near to atmospheric conditions i.e., room temperature, atmospheric pressure, no chemicals required (Nakashima et al., 2016; Xiong et al., 2017). Factors influencing cavitation effect are vibration frequency, ultrasound power (or amplitude), temperature and viscosity of liquid (solvent), surface tension, and time duration (Clark and Nujjoo, 2000; Bussemaker and Zhang, 2013; Rasapoor et al., 2016).

Deepanraj et al. (2017) found that the ultrasonication (20 kHz, 130 W, 30 min) as more effective pretreatment than autoclave and microwave for biogas production from food waste. Castrillón et al. (2011) analyzed ultrasound (20 kHz, 100 W, 4 min) pretreatment and co-digestion of cattle manure with 4% glycerin both showing 121 and 400% increment, respectively on biogas production in mesophilic and thermophilic condition. Cesaro et al. (2012) sonicated mixture of organic solid waste and sewage sludge for 30 and 60 min duration at different energy densities (0.1, 0.2 and 0.4 W/ml). Twenty-four percent enhanced biogas from the sonicated mixture due to enhanced solubilization and enhanced biodegradability of organic matter enhanced during the AD. Cho et al. (2013) conducted sonication for microalgae at different power amplitude for 15 and 30 min duration without temperature control on microalgae Scenedesmus, and methane increment was around 14–75%. Effect of ultrasonic power density and sonication time was analyzed by performing ultrasonic pretreatment of OFMSW on biogas production and proving low power density and higher sonication time give better results than high power density and low sonication time due to simultaneously temperature increment with respect to time making more effective. Effect of sonication on municipal sludge has widely investigated, however, there is an inadequate study on different organic residues (Cesaro et al., 2014; Rasapoor et al., 2016; Zeynali et al., 2017).

As such this study investigates the feasibility and potential of energy production from AcoD of multi feed-stocks from different waste sectors i.e., agricultural, agro-industrial, and municipal solid waste, through BMP tests to alleviate the energy as well as waste management problem. Also exploring the ultrasonic pretreatment for organic residues from above-mentioned sectors at controlled temperature condition and to validate experimental

pretreatment results with two classical kinetic models which can be used to describe and evaluate the batch BMP test for AD process.

## MATERIALS AND METHODS

#### Feed Stocks Collection and Preparation Inoculum

Microbial inoculum used in this study was collected from the continuous Durgapura biogas plant feeding cow dung as substrate. The inoculum was incubated anaerobically at 37◦C for a week to reduce endogenous biogas production. Inoculum has 7.5% TS and VS content as 59.3% of TS. The inoculum was diluted to a TS content of 1.2% with water. Furthermore, divided into 400 mL aliquots in 610 mL batch serum bottles. This diluted inoculum has pH 7.4 and conductivity 2.3 mS.

#### Feed-Stocks or Substrates

The organic fraction of MSW was collected from MNIT campus and two residence of Malviya Nagar, which is a source segregated waste i.e. without any contamination of non-biodegradable waste**.** This waste is mainly composed of leftover cooked food, vegetable peelings, dry green waste of fallen leaves, grass trimmings. With the help of household blender organic wastes were comminuted and mixed homogenously to have better solubility shown in **Figure 2A**, then stored at 4◦C till use (Izumi et al., 2010; Negi et al., 2018).

Wheat straw as an agricultural waste collected from Jaisinghpura village, Jaipur, the collected WS sample shown in **Figure 2B**. Agro-industrial waste: (i) Mung bean husk sample shown in **Figure 2C** is procured from Sunny Daal Mill industry, Sitapura, Jaipur, MBH residue from mill contains almost equal amount of broken small beans and husk, (ii) Groundnut shell waste sample (**Figure 2D**) collected from groundnut processing industry (Rajasthan Agro Product, Sikar Road, Jaipur), which is latter being burnt or disposed, (iii) Guar husk and (iv) Rice bran samples shown in **Figures 2E, F,** respectively which were


BMP test of these mixtures was performed for an incubation period of 60 days at 37◦C with shaking of 90 rpm.

collected from Unique Organic Ltd, Sitapura Industrial Area, Jaipur.

#### Biochemical Methane Potential Test

A BMP test for biogas production was performed in triplicates of each 6 different substrates, with control bottles i.e., negative control having inoculum alone, and positive control having cellulose as substrates. Subsequently, total 24 bottles flushed with nitrogen and closed with a rubber cap, and transferred to the shaker (REMI CIS 24, India) for incubation (37 ◦C, 90 rpm, 30 days) to figure out best agro-industrial substrate for codigestion. 1.55 g substrate VS was added to diluted inoculum bottles (610 ml) having a working volume of 400 ml (Vivekanand et al., 2013; Paritosh et al., 2018).

#### Biogas Volume and Methane Composition

Pressure measurement in head space of each batch reactors was made periodically through Digital pressure meter (TESTO model 512, Germany, shown in **Figure 3A**) for biogas volume measurement. After this, biogas was released, reducing head space pressure to atmospheric pressure. The ideal gas law was used for calculating the biogas volume in the headspace volume of the batch reactors. Biogas production from inoculum control bottles is subtracted from biogas production by each substrate (Donoso-Bravo et al., 2010; Vivekanand et al., 2018).

Ten millimeters gas-tight glass syringe used for sampling of biogas for methane composition analysis. Methane composition was analyzed by Gas chromatography (Thermo SCIENTIFIC Trace 1110) equipped with Porapak column and Thermal Conductivity Detector (TCD) where helium gas was used as carrier gas (Zhen et al., 2016).

#### BMP Test of Mixtures

Five ratios were selected based on nutritional content (C/N ratio) of individual substrates and mixed on VS basis (Mata-Alvarez et al., 2014; Pellera and Gidarakos, 2017; Vivekanand et al., 2018).

## Ultrasonic Pretreatment

Each selected substrate suspension (10 g in 200 mL of water) in a beaker (500 mL) was sonicated (Wang et al., 2017; Xiong et al., 2017). Ultrasonic pretreatment was carried out in a probe-type sonicator (Ultrasonic Processor-sonicator EI-250UP, Electrosonic Industries, Mumbai, India) equipped with a probe of 15 mm diameter (shown in **Figure 3B**) operating at frequency 22 kHz and power of 250 W.

Ultrasonication pretreatment was performed by immersing probe to a half depth of suspension. Here, only pretreatment time is accounted as pretreatment variable i.e., for 30, 60, and 90 min (Nakashima et al., 2016). The temperature of suspension was maintained constant at 32 ± 1 ◦C by recirculating cold water in the provided water bath and manual stirring is performed to have uniform pretreatment effect on substrates. Here, the project aims to investigate the influence of ultrasonic pretreatment duration without temperature effect on methane production.

Afterward, BMP test was performed to evaluate the effect of ultrasonic pretreatment on methane production from different substrates.

#### Analytical Studies

Proximate analysis i.e. Moisture content, Total solids (TS) and VS of the inoculum and substrates were determined by standard methods from American Public Health Association (APHA, 1999). Ultimate analysis i.e. carbon, hydrogen and nitrogen content were determined by CHNS analyser (Thermo Finnigan, FLASH EA 1112 series, Italy). pH of batch bottles was measured by a pH meter (LMPH 10, Labman Scientific Instruments Pvt. Ltd. India) in the beginning and end of the BMP test. The proximate and ultimate results of selected substrates are shown in **Table 1**.

#### Surface Morphology

Surface morphology of different pretreatment level of the respective substrate was observed by morphological change through scanning electron microscope SEM (Nova NanoSEM 450, Netherland) observation.

#### Crystallinity

Crystallinity index of untreated and pretreated substrates was examined through X-ray diffraction test using a XPERT-PRO Xray diffractometer (XRD), and the samples were scanned in 2θ ranged from 5–50◦ with a step size of 0.02◦ (Liu et al., 2015; Wang et al., 2017). The decrease in CrI ascribed to disruption of crystalline structure, whereas an increase in CrI ascribed to removal of lignin and remained crystalline cellulose (Nakashima et al., 2016).

The crystallinity index (CrI) is calculated by following equation (1):

$$\text{CrI} = \frac{I\_{cr} - I\_{amor}}{I\_{cr}} \ast 100 \text{ \%} \tag{1}$$

Icr = Diffraction intensity from 002 plane of crystalline part (cellulose) at 2θ = 22.6◦ , and Iamor = Diffraction intensity of amorphous part (cellulose, hemicellulose, and lignin) 2θ = 18◦ (Kumar et al., 2009; Liu et al., 2015; Wang et al., 2017).

#### Chemical Structure

Effect of pretreatment on different substrates can be observed by a change in chemical compositions through Spectrum 10.4.00 FTIR spectrophotometer (PerkinElmer, USA). FT-IR can be used to characterize the intrinsic biodegradability of a biomass. FT-IR spectroscopy analysis of selected samples of untreated and pretreated FW was performed in the range of 4000–400 cm−<sup>1</sup> to characterize the effect of treatment on the functional groups and the chemical structure.

#### Kinetics Study

The different kinetics models have been used in previous studies to determine the methane yield by fitting the measured and predicted methane yield. Cone model and Exponential kinetic (or First order kinetic) model were chosen to fit the methane yield. Principal kinetic patterns (digestion mechanism of the organic substrate) of methane production from BMP test can be described through these models. Both models have an assumption of methane production in the batch test is proportional to methanogenic bacteria growth rate. These two models were used to compare and understand the kinetics of methane production from AD (El-Mashad, 2013; Zhen et al., 2016; Syaichurrozi, 2018).

Hydrolysis process as rate limiting step is the basis for using Exponential kinetic model describing the AD assuming no accumulation of intermediary elements (Veeken and Hamelers, 1999; Li K. et al., 2015). And, cone model gives better fit with methane production than other models reported by El-Mashad (2013), Li K. et al. (2015) and Paritosh et al. (2017). The equations of Cone model (Equation 2), and Exponential model (Equation 3) were shown below:

$$\text{Exponential model}: B = P \times (1 - \exp^{-k \times t}) \tag{2}$$

$$\text{Cone model}: B = \frac{P}{1 + (k \times t)^{-n}} \tag{3}$$

B = Accumulative methane yield (ml/g VS), P = Final methane yield (ml/g VS), k = hydrolysis rate constant (1/day), n = shape factor, t = digestion time (day) for t ≥0. With the help of non-linear least square fitting of accumulative methane yield, parameter "k" and "n" were estimated (Syaichurrozi, 2018).

The coefficient of determination (R 2 ) and Root mean square error (RMSE) were used as indicators for model fitness, and to compare the accuracy of both models. R 2 is an index for the goodness of fit, and RMSE gives standard deviation between measured and predicted values, the better fit model shows low RMSE (Li K. et al., 2015). RMSE can be calculated through following formula:

$$RMSE = \sqrt{\frac{\sum\_{i=1}^{n} \left(Y\_p - Y\_m\right)^2}{n}}$$

Where, Y<sup>p</sup> and Y<sup>m</sup> are predicted and measured methane yield (ml/g VS), respectively, and n is a number of measurements performed (El-Mashad, 2013).

## RESULTS AND DISCUSSION

### BMP Test of Different Agro-Industrial Waste

The results of I-BMP test of individual substrates showing biogas production are shown in **Figure 4** of different agroindustrial waste for digestion time of 30 days. Accumulated biogas production in 30 days of digestion period from raw Groundnut shell, Mung bean husk, Guar husk, and Rice bran are also defined in **Table 2**. Mung bean husk showed maximum potential of biogas among other agro-industrial waste (i.e., groundnut shell, guar husk, and rice bran) therefore it was selected as co-substrate with wheat straw and OFMSW for the multi feed-stock AD. Biogas production from rice bran was low due to its high protein content making unstable AD process (Favaro et al., 2017).

#### BMP Test of Selected Substrates and Their Mixtures

The II-BMP test of selected substrate i.e., OFMSW, WS, and MBH, and anaerobic co-digestion of five different mixing ratio shown in **Table 2** of these substrates were performed for a retention time of 60 days. Accumulative biogas production in 60 days from above-mentioned substrates and mixtures are shown in **Figure 5**. Biogas production of OFMSW, WS, and MBH in mono-digestion mode were 397, 431, and 468 ml/g VS, respectively. AcoD having mixture M3 having a composition of OFMSW, WS, and MBH at 42, 7.5, and 50.5%, respectively ratios produced maximum biogas production in comparison with other mixing ratios.

Comparison of biogas and methane production of substrates in mono-digestion and co-digestion mode with total methane fraction are shown in **Table 2**. Methane produced and its total fraction in biogas from WS, OFMSW, and MBH are 225 (52%), 203 (51%), and 267 (57%) ml/g VS. WS cellulosic enzymatic hydrolysis is the slow due presence of hemicellulose and lignin can be seen from accumulative biogas production curve in

**Figure 5** (Zheng et al., 2018). Methane production through AcoD of mixtures were 256 (M1), 226 (M2), 275 (M3), 280 (M4), and 84 (M5) ml/g VS. Mixture M4 showed maximum methane production with 280 ml/g VS with methane fraction of 59%, whereas mixture M5 has minimum methane production 84 ml/g VS, due to the nutritional imbalance i.e., high C/N ratio of 52 which led to the accumulation of VFA inhibiting the methanogens.

The enhanced methane yield from mixtures M1-M4 showed the good synergistic effect of co-digestion. For better methane yield mixture of such substrates should have a composition of OFMSW and MBH in the range of 25–42%, and 50– 70%, respectively. The synergistic effect of co-digestion can be strengthened by optimizing the above-mentioned range of specific substrate fractions. The mixture M4 methane yield showed 37, 20, and 4% higher methane yield in comparison with mono-digestion of OFMSW, WS, and MBH.

## Analytical Study of Ultrasonic Pretreatment on Different Substrates

Untreated and pretreated substrates were digested anaerobically at mesophilic conditions, for which III-BMP test was performed whose results were shown in **Table 2**. The effect of pretreatment was evaluated by analyzing following sections of respective substrates.

#### Wheat Straw (WS)

#### **Methane yield**

Ultrasonication enhance biogas and methane yield due to improved solubility of organic matters. This can be validated with enhancement of accumulative biogas yield from WS are shown in **Figure 6**. Biogas yield due to pretreatment are similar with results from previous pretreatment studies on WS (Li Y. et al., 2015). Methane yield and methane fraction in biogas from pretreated WS samples 314 (67%), 357 (65%), and 314 (61%) ml/g VS at sonication time of 30, 60, and 90 min, respectively. Highest methane yield (357 ml/g VS) was obtained at 60 min of sonication pretreatment. The increment of 75% was observed in methane yield of ultrasonic pretreated WS at sonication period of 60 min than untreated WS. After 60 min sonication, methane yield decreases in WS, it was due to an increased particle size of samples because of reflocculation phenomena so lowered the hydrolysis rate. Re-flocculation occurred due to excess radical formation in the suspension (Rasapoor et al., 2016; Zeynali et al., 2017).

#### **Surface morphology**

The morphological changes induced by ultrasonic pretreatment on Wheat straw samples with increment in sonication time were observed by SEM images which are shown in **Figures 7A–D**. The untreated sample has smooth surface whereas pretreated samples showing erosion on the surfaces which can be recognized as the action of ultrasound pretreatment caused by the cavitation

effect of sonication. As sonication time increases more surface erosion occurs on the outer surface of the WS samples which can be observed from **Figures 7A–D**. Also, images **Figures 7B–D** of pretreated WS samples reflecting some of the lignin cover on the surface are removed and broken resulting into the more accessible area of microfibrils for faster enzymatic hydrolysis, leading methane augmentation (Wang et al., 2016; Zheng et al., 2018).

#### **Crystallinity**

Crystallinity is a factor affecting hydrolysis significantly, its higher value represents better pretreatment effect if the purpose is to destroy amorphous fraction i.e. hemicellulose and lignin structure (Wang et al., 2017). Untreated and pretreated WS samples have similar XRD patterns shown in **Figure 8**, revealing significant disruption in the structure of WS samples. The crystallinity index (CrI) of pretreated WS samples calculated according to equation (1) whose values for untreated WS, WS-30, WS-60, and WS-90 are 39.9, 40.8, 38.7, and 37.4, respectively. Here, crystallinity increases from 39.9 to 40.8 after 30 min sonication showing deconstruction the lignin cover of WS samples due to cavitation effect induced by ultrasound frequency in water. Afterward, as sonication time is increases resulted in decreased crystallinity. The decrease in CrI may be ascribed as partial disruption of crystalline cellulose structure (Zheng et al., 2018). This shows pretreatment destroys concurrently lignin,

hemicellulose, and cellulose structure i.e., in a non-selective manner (Xiong et al., 2017).

#### **Chemical structure**

The FT-IR spectra of untreated and pretreated WS samples are shown in **Figure 9**. The absorbance band of 3350–3450 cm−<sup>1</sup> assigned to the –OH stretching vibrations, peak intensity is weakened as pretreatment increases showed cellulose content in pretreated samples is higher than untreated one (Zheng et al., 2018). In FT-IR spectroscopy, 1400–1700 cm−<sup>1</sup> band shows the presence of C=C of aromatic rings which are found in lignin. In this spectra, the peak near 1430 and 1640 cm−<sup>1</sup> are diminishing in pretreated samples that indicates the degradation of lignin. The peak at 1735 cm−<sup>1</sup> which represents ester bond C=O also diminishing with pretreatment indicating degradation of ester linkages between lignin and carbohydrates corresponding to the hemicellulose (Liu et al., 2009). Apart from lignin and hemicellulose, the crystalline cellulose defined by peak 1049 cm−<sup>1</sup> also degrades indicating reduction in crystalline cellulose part. The intensity of peak 2922 and 1375cm−<sup>1</sup> which is related to asymmetric methylene stretching and CH<sup>2</sup> wagging, respectively of cellulose, becomes weaker with increasing pretreatment severity (Singh et al., 2011; Raj et al., 2015). The pretreatment effect is consistent with results of XRD analysis.

#### Mung Bean Husk (MBH)

The effect of pretreatment was evaluated by analyzing methane yield, changes in surface morphology and chemical structure of pretreated MBH samples.

#### **Methane yield**

Enhancement in accumulative biogas yield from pretreated MBH substrate are shown in **Figure 10**. Methane yield and methane fraction (% of biogas) of pretreated MBH samples were 350 (68%), 382 (69%), and 379 (65%) ml/g VS at sonication time of 30, 60, and 90 min, respectively. Highest methane yield (382 ml/g VS) was obtained at 60 min of sonication. Ultrasonication pretreatment enhanced methane fraction from 57 to 69% of biogas volume. In case of MBH pretreatment, biogas volume was consistently increasing with an increment of sonication time in performed pretreatment conditions. Ultrasonic pretreated MBH has a maximum increment of methane yield up to 46% at sonication 60 min.

#### **Surface morphology**

SEM images of different pretreated MBH samples were obtained shown in **Figure 11** to prove the substrate structural change caused by ultrasonication pretreatment. Before and after pretreatment SEM images show surfaces of MBH samples have significant changes can be observed. Untreated samples exhibit compact and regular surface structure, where microfibers

can't be observed. After sonication pretreatment, MBH samples becomes coarse and scattered making microfibrils more porous compared to untreated MBH. Thus, pretreatment enhances surface defibrillation and coarseness as sonication time increases, confirming the positive effect in disrupting the structure of MBH samples.

#### **Chemical structure**

The FT-IR analysis demonstrates that untreated and pretreated samples of MBH have similar spectra pattern shown in **Figure 12**, showing the surface functional groups were significantly affected by pretreatment. The –OH bond at 3430 cm−<sup>1</sup> represents the OH groups in cellulose whose degradation can be observed with pretreatment. As can be seen from FT-IR spectra, peak 1659 and 1530 cm−<sup>1</sup> corresponds to C=C stretching of aromatic ring originated from lignin, were diminishing with pretreatment severity to MBH samples. Also, peak at 1247 cm−<sup>1</sup> of ether bonds in pretreated MBH samples were almost disappearing in comparison to untreated MBH sample. This indicates ultrasonic pretreatment could disrupt or remove ether linkages between carbohydrates and lignin (Liu et al., 2009). Reduction in the peak of 1018 cm−<sup>1</sup> of C-O stretching corresponding to hemicellulose TABLE 3 | Parameters of Cone and Exponential kinetic model obtained from untreated and pretreated samples.


showing degradation of hemicellulose. Also, the intensity at 2928 cm−<sup>1</sup> of –CH aliphatic stretching represents CH<sup>2</sup> and CH<sup>3</sup> groups corresponding to cellulose and hemicellulose which were diminishing as sonication time increases for MBH samples. Also, peak of –CH<sup>2</sup> wagging at wavenumber 1403 cm−<sup>1</sup> is also weakening than the untreated samples (Smidt et al., 2002; Chandra, 2015; Li et al., 2018). FT-IR spectra results showed pretreatment affect lignin, hemicellulose, and cellulose simultaneously.

#### Organic Fraction of Municipal Solid Waste (OFMSW)

The effect of pretreatment was evaluated by analyzing methane yield, and changes in surface morphology of pretreated OFMSW samples.

#### **Methane yield**

Accumulated biogas yield were measured for defined sonication time to OFMSW is shown in the **Figure 13**. **Table 2** also showing methane yield and methane fraction of pretreated OFMSW samples 291 (60%), 331 (66%), and 287 (58%) ml/g VS at sonication time of 30, 60, and 90 min, respectively with respect to 193 (50%) ml/g VS from untreated OFMSW. Highest methane yield (331 ml/g VS) was obtained at 60 min of sonication pretreatment. The increment of 71% in methane yield and methane fraction enhanced from 50 to 66% of biogas, were observed from ultrasonic pretreated OFMSW at sonication period of 60 min than untreated OFMSW. Methane yield resulted from ultrasound pretreatment of OFMSW were comparable with results of Cesaro and Belgiorno (2013). Methane yield from pretreated OFMSW samples increases with ultrasonication time due to the organic matter solubilization making the higher availability of substrate for digestion. After 60 min of ultrasonication, methane yield decreases in OFMSW, it was due to increased particle size of samples because of radical formation initiating the polymerization reaction during ultrasonic pretreatment (González-Fernández et al., 2012).

#### **Surface morphology**

SEM analysis of untreated and pretreated OFMSW samples helped to comprehend the ultrasonication effect. SEM images of OFMSW samples are shown in **Figures 14A–D**. Untreated OFMSW has more tight particles having composites of organic compounds having clusters of particles, whereas ultrasonicated OFMSW samples showed bigger particle cluster were rugged, and broken into smaller particles sizes, thereby augmenting surface area of pretreated samples and improving the organic matters solubility. Thus ultrasonication effect led changes in physical structure to OFMSW samples achieving better disintegration, so pretreated OFMSW samples have the more microbial accessibility of carbon for enzymatic hydrolysis than untreated one (Deepanraj et al., 2017).

#### Kinetic Study Results

The degradation kinetics of selected substrates were described by using Cone and Exponential kinetic models for III-BMP test. **Table 3** shows the experimental methane yield (ml/g VS) from BMP-III test and predicted methane yield (ml/g VS) for respective kinetic models with relevant kinetic parameters. For both models, hydrolysis rate constant "k" increasing with pretreatment duration of respective substrate samples with few exceptions. Hydrolysis rate constant "k" reflects biodegradability of substrates. High "k" refers to the faster rate of degradation of organic substrate, enhancing methane production (Veeken and Hamelers, 1999; Zhen et al., 2015; Mao et al., 2017).

The studied model results of OFMSW shows improvement in measured methane and maximum methane potential was independent of "k," that rise in "k" doesn't convey high methane production. This inference was similar to Zhen et al. (2016) and Keymer et al. (2013). In case of cone model, "k" ranges between 0.21 and 0.24 day−<sup>1</sup> for MBH, 0.064–0.083 day−<sup>1</sup> for WS, and 0.135–0.16 day−<sup>1</sup> for OFMSW, in case of exponential kinetic model "k" ranges 0.13–0.15 day−<sup>1</sup> for MBH, 0.046–0.06 day−<sup>1</sup> for WS, and 0.086–0.097 day−<sup>1</sup> for OFMSW. The "k" values are

comparatively lower for lignocellulosic biomass, therefore lignin content is higher in WS, than substrate MBH, and OFMSW, as it contains green waste.

The calculated RMSE and R 2 values for both models are shown in **Table 3**, their lower values confirming the model's good agreement with experimental results. RMSE values vary in between 4.4–12.1 for Cone model and 3.9–19.2 for the Exponential model. This reflects Cone model with a lower range

#### REFERENCES


of RMSE, is more accurate than the Exponential kinetic model. The cone model (R 2 : 0.9592–0.9929) and Exponential kinetic model (R 2 : 0.9723–0.9973) showed a good fit to predicted and measured data of different untreated and pretreated substrates in mono-digestion. **Figure 15A,B** shows relation between predicted methane yield against measured values for both Cone and Exponential kinetic model through R 2 .

#### CONCLUSION

This study shows multi-feed of wheat straw, mung bean husk, and OFMSW anaerobic co-digestion shows good synergy for enhanced methane production. Mixing of such co-substrates results into 37, 20, and 4% higher methane yield in comparison with mono-digestion of OFMSW, wheat straw and mung bean husk. The experimental results showed ultrasonic pretreatment with varying sonication time have a significant effect on methane yield from OFMSW, wheat straw, and Mung bean husk. Pretreatment significant effects on different substrates were validated through BMP test, SEM, XRD, and FTIR analysis. Ultrasonication time can be increased up to 60 min with stable temperature pretreatment constraint. Methane yield of ultrasonic pretreated wheat straw, mung bean husk and OFMSW have a maximum increment of 75, 46, and 71%, respectively at sonication period of 60 min. Two classical kinetic models i.e., Cone and Exponential were used for validating the experimental results of III-BMP test. Cone and Exponential model have an average coefficient of determination R <sup>2</sup> = 0.9855 and R <sup>2</sup> = 0.9893, respectively, but the range of RMSE for cone model results is lower than that of Exponential model showing Cone model is more précised.

## AUTHOR CONTRIBUTIONS

KP performed the experiments and VV conceived and designed the experiments. KP carried out the experiments and wrote the manuscript. VV and NP worked on and corrected the manuscript critically.

#### ACKNOWLEDGMENTS

KP thanks MNIT for assistantship and SAIF, IIT Bombay for CHN characterization. Authors are thankful to Department of Biotechnology-GoI (Grant No. BT/RLF/Re-Entry/04/2013) and Department of Science and Technology-GoI (Grant No. ECR/2016/000989) for financial support.

of TWAS and RS pretreatment and mixing ratio. Energy 107, 131–140. doi: 10.1016/j.energy.2016.03.141


pretreatments on the biochemical methane potential and kinetics of the anaerobic digestion of wheat straw and sugarcane bagasse. Bioresour. Technol. 201, 182–190. doi: 10.1016/j.biortech.2015.11.047


(Scenedesmus sp., Chlorella sp.) and food waste: kinetic modeling and synergistic impact evaluation. Chemical Eng. J. 299, 332–341. doi: 10.1016/j.cej.2016.04.118


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Prajapati, Pareek and Vivekanand. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Treatment of Sewage Sludge Using Anaerobic Digestion in Malaysia: Current State and Challenges

Farida Hanum1,2, Lee Chang Yuan<sup>1</sup> , Hirotsugu Kamahara<sup>3</sup> , Hamidi Abdul Aziz <sup>4</sup> , Yoichi Atsuta<sup>1</sup> , Takeshi Yamada<sup>1</sup> and Hiroyuki Daimon1,3 \*

*<sup>1</sup> Department of Environmental and Life Sciences, Toyohashi University of Technology, Toyohashi, Japan, <sup>2</sup> Department of Chemical Engineering, Universitas Sumatera Utara, Medan, Indonesia, <sup>3</sup> Institute for Global Network Innovation in Technology Education, Toyohashi University of Technology, Toyohashi, Japan, <sup>4</sup> School of Civil Engineering, Universiti Sains Malaysia, Nibong Tebal, Malaysia*

Anaerobic digestion is widely considered as an environmentally friendly technology for various organic waste including sewage sludge. Although the implementation of anaerobic digestion as an alternative treatment method for sewage sludge can be seen in many countries, its status in Malaysia is not clear. This study reviewed the current state of sewage sludge treatment in Malaysia and discussed the challenges to promote anaerobic digestion in sewage sludge treatment. Other than the common constraints faced, namely technical, political and economic, the characteristics of sewage sludge in Malaysia are considered to be a factor regarding feasibility. Anaerobic co-digestion is the simultaneous anaerobic digestion of two or more substrates which is a promising possible option to overcome the disadvantages of mono-digestion, and improve the economic viability due to higher methane production. There are a variety of biomasses as co-substrates in Malaysia. However, the anaerobic co-digestion of food waste and sewage sludge might be the most feasible method to overcome such constraints. Adding food waste as co-substrate is suggested as the possible approach to not only improve the process's performance but also help to handle the increasing volume of food waste in Malaysia. This study aims to highlight the potential as well as to provide a starting point for further studies regarding the treatment of sewage sludge using anaerobic digestion in Malaysia.

Keywords: sewage sludge, anaerobic digestion, co-digestion, food waste, Malaysia

## INTRODUCTION

Sewage sludge is the residue produced by a wastewater treatment process and has the largest volume amongst all the components removed during the process. In global terms, the amount of sludge is expected to increase. This is due to an increase in the percentage of households connected to central wastewater treatment plants (WWTPs), increasingly tightening regulations on effluent discharges, as well as improved technologies in achieving higher efficiency in wastewater treatment (Werther and Ogada, 1999; Appels et al., 2008). As the sludge itself may contain concentrated levels of contaminants that were initially included in the wastewater, the increasing sewage sludge has become a growing concern. Current primary disposal methods for sewage sludge are agricultural use, landfill, and incineration. Nevertheless, each technique has its particular issues which relate to public health as well as its environmental impacts (Fytili and Zabaniotou, 2008).

#### Edited by:

*Mohammad Rehan, King Abdulaziz University, Saudi Arabia*

#### Reviewed by:

*Muzammil Anjum, Pir Mehr Ali Shah Arid Agriculture University, Pakistan Muhammad Waqas, Kohat University of Science and Technology, Pakistan*

> \*Correspondence: *Hiroyuki Daimon daimon@ens.tut.ac.jp*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *02 July 2018* Accepted: *11 February 2019* Published: *05 March 2019*

#### Citation:

*Hanum F, Yuan LC, Kamahara H, Aziz HA, Atsuta Y, Yamada T and Daimon H (2019) Treatment of Sewage Sludge Using Anaerobic Digestion in Malaysia: Current State and Challenges. Front. Energy Res. 7:19. doi: 10.3389/fenrg.2019.00019*

Anaerobic digestion is generally considered to be an economical and environmentally-friendly technology for treating various organic waste, including sewage sludge (Appels et al., 2011; Nasir et al., 2012). In the absence of oxygen, the organic waste is biologically degraded and converted into a form of biogas, and other energy-rich organic compounds as end products. The biogas, which is generally composed of 48–65% methane, could be used for power generation (Ward et al., 2008). Since the introduction of both commercial and pilot plant designs during the early 1990s, anaerobic digestion has gained worldwide attention (Karagiannidis and Perkoulidis, 2009). Throughout the world, more than 1,300 anaerobic digestion systems, based on sewage sludge, are in operation or under construction (IEA Bioenergy, 2001). Although China and India lead the initiative among developing countries, the thrust in of the developed world mainly comes from Western Europe (Abbasi et al., 2012). The biogas generation from anaerobic digestion in Europe has reached 4.5–5.0% growth annually, with Germany being the largest biogas producer (Bodík et al., 2011). It is notable that a high proportion of the biogas produced in anaerobic digestion plants is from those situated in WWTPs and thus anaerobic digestion has become a major and essential part in modern WWTPs. The current trend has also seen the introduction of anaerobic co-digestion, in which two or more substrates are mixed and digested simultaneously, with successful examples reported from WWTPs in Denmark, Germany, and Switzerland dedicated to combined sewage sludge and biowaste (Braun and Wellinger, 2009).

As a developing country, Malaysia is no exception from the global trends of the volume of sewage sludge increasing annually. It is reported that about 3 million metric tons of sewage sludge is produced annually in Malaysia, and is expected to increase to 7 million metric tons by 2020 (Indah Water Konsortium Sdn Bhd, 2010). Almost 50 % of the sewerage systems in Malaysia uses mechanized plants while others still use septic tanks and oxidation ponds. Some modern mechanized plants have been upgraded with anaerobic digesters whereby the sewage sludge are treated anaerobically to produce CH<sup>4</sup> which is a valuable energy source to generate electricity (Kumaran et al., 2016).

Malaysia is the third-largest consumer of energy in Southeast Asia. Petroleum, natural gas, and coal are the primary fuel sources consumed in Malaysia. In 2014, Malaysia consumed 43% of natural gas, 37% of petroleum and other liquids, 17% of coal, 2% of other renewable, and 1% of hydroelectricity (EIA, 2017). The growing consumption of energy has resulted in becoming increasingly dependent on fossil fuels such as coal, oil, and gas. Fossil fuel depletion and environmental issues have encouraged the government and researchers to investigate renewable energy as alternative sustainable energy resources (Hosseini and Wahid, 2013). Recently, bioenergy sources like animal and agricultural wastes, municipal solid waste and wastewater effluents as the renewable energy sources, have attracted attention and applied for preparing a fraction of the global energy demand (Petinrin and Shaaban, 2015).

The main objective of this study is to assess the potential of treating sewage sludge in Malaysia using anaerobic digestion. The current state regarding sewage sludge in Malaysia was reviewed, with activities around other regions included for comparison. Challenges and tasks faced when treating sewage sludge using anaerobic digestion in Malaysia were discussed, with anaerobic co-digestion which would include food waste, suggested as a possible approach. This study aims to provide a greater understanding as well as a starting point for further study regarding the future aspect of utilizing sewage sludge in Malaysia.

## OVERVIEW OF SEWERAGE SYSTEM AND PRODUCTION OF SEWAGE SLUDGE IN MALAYSIA

Malaysia's sewerage industry has expanded over the last half of the twentieth century. At the end of the 1960s, the Malaysian government launched a series of 5-year plans to construct appropriate sanitation facilities in both urban and rural areas. At that time septic tanks were used as the primary sewerage system. In the 1970s, the government started the "National Sewerage Development Program" to develop sewerage facilities in major cities with the aim of introducing modern sewerage systems in urban areas. During the 1980s, the government introduced a policy that obliged housing developers to build sewerage systems for regions comprising more than 30 households of a 150 population equivalent (PE). This resulted in the diffusion of many small-scale wastewater treatment facilities all over the country, which were gradually connected to large-scale sewerage systems to complete the public sewerage system (Japan Sanitation Consortium, 2011). A profile of sewerage systems in Malaysia can be seen in **Table 1**. Although the large-scale sewerage systems, or public sewage treatment plants in Malaysia have increased in number, as seen from **Table 1**, the amount of operational individual septic tanks constitutes an enormous part of the overall wastewater treatment in Malaysia. This current situation could be a crucial factor in the attempt to utilize sewage sludge, as the septic tanks only provide partial treatment of the sewage that flows into it and needs to be desludged on a regular basis (Suruhanjaya Perkhidmatan Air Negara, 2017).

Under the Environmental Quality (Scheduled Waste) Regulations 2005, the Department of Environment (DOE) categorizes sludge generated from WWTPs as scheduled waste, and shall be disposed of under a prescribed premise only. Since 1994, the management of wastewater in most of the states of Peninsular Malaysia, is undertaken by a private company named Indah Water Konsortium (IWK) Sdn Bhd. According to a recent sustainable report by IWK, it has been operating and maintaining 16,328 km of sewer pipelines, 5,997 public sewerage treatment plants and 926 network pump stations. On average, IWK assumes control of 300 treatment facilities and 1,000 km of sewer network yearly (Indah Water Konsortium Sdn Bhd, 2013). It was estimated that the total cost of managing sewage sludge is US\$ 0.33 billion per year (Kadir and Mohd, 1998). Currently, the sewage sludge is commonly being disposed of either at landfill sites or being burned in incinerators (Bradley


TABLE 1 | Profile of sewerage systems in Malaysia.

*n.a., Data not available.*

and Dhanagunan, 2004). However, as mentioned before, the sewage sludge itself may contain particular contaminants, and in case of incineration, approximately 30% of the solids remain as ash. Hence, these current disposal methods remain a concern (Malerius and Werther, 2003).

IWK had begun planning to reuse the sewage by-products, yet this mainly occurs only at local plants with a PE of 100,000 (Kadir and Mohd, 1998). There are very few wastewater treatment plants where digestion system could be found operating, and in most cases, the biogas produced is not utilized for power generation but only for combustion (Liew, 2008; Indah Water Konsortium Sdn Bhd, 2013). Bunus Centralized Sewage Treatment Plant (Kuala Lumpur) generates 2,500 m<sup>3</sup> /days of biogas. From that amount of biogas, 15,000 kWh/m<sup>3</sup> of power without conversion losses could be produced. In fact, to date, all methane gas has been burned without effective use. As the average plant consumes electricity of 43,503 kWh/day (David et al., 2014), which is about 40% of the overall operations and maintenance costs (Baki et al., 2005), then the power generated through biogas from anaerobic digestion could be helpful in reducing total energy consumption, as well as serving as an alternative power generation source for public use. Recently, Malaysia has three modern WWTPs which are equipped with anaerobic digestion. There are the Pantai 2 WWTP in Greater Kuala Lumpur, the Jelutong WWTP in Pe18nang, and the Langat WWTP in Selangor. Pantai 2 and Jelutong WWTP are in operation, but at Langat is due to be completed in 2018. Pantai 2 WWTP has been designed to produce up to 9,600 m<sup>3</sup> /day of biogas. Presently, in 2018, they produce around 450–500 KW electricity (Pantai 2 regional sewage treatment plant Kuala Lumpur)<sup>1</sup> . The Jelutong WWTP is one of the largest plants in Malaysia. It operates to treat 800,000 PE (Jelutong WWTP-Malaysia).<sup>2</sup>

Overall, not many investigations or studies have been conducted in Malaysia regarding the treatment of sewage sludge using anaerobic digestion, which reflects that the focus and awareness of these issues are indeed not high amongst the general public.

## TREATMENT OF SEWAGE SLUDGE USING ANAEROBIC DIGESTION

Anaerobic digestion is a proven technology for sewage sludge treatment, in which the high water content sewage sludge can be processed without any pretreatment (Ward et al., 2008). The properties of sewage sludge are modified during anaerobic digestion, with not only biogas being produced as a result, but also several positive consequences for the sludge management that follows the process. Anaerobic digestion enhances the stabilization of sewage sludge, reduces pathogens and odor emission, and dry matter of sludge is reduced, which leads to a significant reduction in the final sludge volume. These benefits of anaerobic digestion of sewage sludge are widely recognized, and the technology is well established in many countries. The total biogas production in several countries and biogas production from WWTPs are listed in **Table 2** based on the statistics of the International Energy Agency (IEA Bioenergy Task 37, 2017) and the World Bioenergy Association (2017). Germany dominates the worldwide biogas production with 10,431 biogas plants generating 55,108 GWh/y of electricity. It has 1, 258 sewerage plants generating 3,517 GWh/y of electricity. Although India has 83,540 biogas plants, bioenergy production is still at a household mode and has been limited to small size biogas plants which are used by the people especially in rural areas. On the other hand, in Malaysia, IWK has taken over 4,281 small plants for upgrading to large plants by 2013. Overall IWK has 35 large plants (Indah Water Konsortium Sdn Bhd, 2013) which is 247 GWh/y electricity generated in 2017 (Sustainable Energy Development Authority Malaysia, 2018)<sup>3</sup> .

Meanwhile, in Japan, policies such as SPIRIT21 (Sewage Project, Integrated and Revolutionary Technology for Twenty First Century) and Sewerage Vision 2100 have been implemented over the last decade regarding the improvement of WWTPs and utilization of sewage sludge. Led to the active movements on the usage of sewage sludge, which included energy recovery by anaerobic digestion. According to the Japan Sewage Works Agency, 280 out of 2,150 WWTPs have anaerobic digestion system operating, with 0.4–0.6 m<sup>3</sup> CH4/kg-VS produced averagely (Japan Ministry of Land, Infrastructure, Transport and Tourism).<sup>4</sup> One of the most successful WWTPs has sewage sludge treated using anaerobic digestion and is located in Yamagata Prefecture. The daily biogas produced is 4,082 m<sup>3</sup> , in which 7,226 kWh/d of electricity is generated. This corresponds to 48.7% of the total energy consumption of the plant itself (Li and Kobayashi, 2010).

Hanum et al. Anaerobic Digestion Using Sewage Sludge

<sup>1</sup>English brochure. Published by IWK.

<sup>2</sup>Available online at: http://sfcu.at/portfolio\_category/asia/?lang=en

<sup>3</sup>Available online at: http://www.seda.gov.my/

<sup>4</sup>Available online at: http://www.mlit.go.jp/index\_e.html


*n.a., Data not available.*

## CHALLENGES TO UTILIZE SEWAGE SLUDGE USING ANAEROBIC DIGESTION IN MALAYSIA

#### Technical, Political and Economic Issues for Sewage Sludge Anaerobic Digestion

Several issues are surrounding the water services industry that needs the attention of policymakers and stakeholders to ensure the success of restructuring the water industry and achieving the goal of a sustainable water services industry in Malaysia. Anaerobic digestion is a complex system which involved several stages of biotransformation of organic matter, which are hydrolysis, acidogenesis, acetogenesis, and methanogenesis. Each of these steps has operational requirements, especially for methanogenesis, where methanogenic bacteria convert acetate into CH<sup>4</sup> and CO2. In term of the technical aspect, improper operation of the anaerobic digestion will lead to low biogas yield. Sun et al. (2014) showed that increasing fat content in municipal solid waste (< 60%) could increase CH<sup>4</sup> yield significantly. The study also observed a decrease in CH<sup>4</sup> yield when the fat content is higher than 65% and displayed inhibitory effect on biogas production. There is still a lack of research and development effort in this aspect. Besides, the operation and maintenance of the anaerobic digestion plant will require skillful engineers and technicians, which could be lacking in Malaysia as anaerobic digestion is still not a common practice (Ali et al., 2012).

Climate change and depletion of fossil fuel along with the rise in petroleum prices have forced the Malaysian Government to rethink strategies to be more effective including the decision to embark on renewable energy resources as a better energy source amongst the global energy mix. The Malaysian Government established policies, renewable energy programs, and incentives more effective to promote and develop the use of renewable energy. In 2000, under the eight Malaysia Plan (MP), the Fifth Fuel Diversification Policy was launched, and Renewable Energy was included to be the fifth primary energy source, following oil, gas, coal, and hydropower. Following this, there has been continuous implementation of renewable energy-promoting policies and actions, such as the Small Renewable Energy Program, National Green Policy 2009, National Renewable Energy Plan 2010, Renewable Energy Act 2011, Feed-in Tariff (FiT) mechanisms, renewable energy business fund and Green Technology Financial Schemes under different Malaysia Plans (Bong et al., 2017). On the other hand, regarding social aspect, the public is generally not aware of anaerobic digestion plant and can be reluctant to have the anaerobic digestion plant near residential area and city area.

## Characteristics of Sewage Sludge in Malaysia

Despite producing a massive amount of sewage sludge annually, the utilization of sewage sludge in Malaysia did not get as much attention or progress as seen in other regions. The current state of sewage sludge treatment sees it primarily being used for landfill. Several studies had reported using sewage sludge to produce clay bricks (Liew et al., 2004), compost (Kala et al., 2009), as well as fuels (Abbas et al., 2011). Nevertheless, there are very few focused on anaerobic digestion. One aspect that should be considered regarding anaerobic digestion is the characteristics of sewage sludge. Characterization of sewage sludge is essential for determining a suitable rate of application of sewage and for investigating pollutant risks that may be associated with the use of sewage sludge. Sewage sludge is characterized by high concentrations of solid and organic matter, with a significant presence of pathogens, nutrients, organic and inorganic pollutants. In an anaerobic digester, proper carbon to nitrogen (C/N) ratio is essential for efficient digestion (Kim et al., 2012). Characterization of sewage sludge from wastewater treatment plants in Malaysia can be seen in **Table 3** (Abbas et al., 2011). The C/N ratio of sewage sludge is 6. However, in general, the suggested optimum C/N ratio for anaerobic digestion is in the range of 20–30 (Parkin and Owen, 1986), but the C/N ratio of sewage sludge is ranged between 6 and 16 typically (Tchobanoglous et al., 1993). Similar research also reported by Rosenani et al. (2008) in which they analyzed sewage sludge samples taken from 10 WWTPs throughout Peninsular Malaysia, showed that the C/N ratio of sewage sludge ranged from 4.1 to 38.0 with a mean of 16.6. As the fermentation substrate, this low C/N in sewage sludge, which goes against the nutrition balance of microorganisms. The pH of sewage sludge was reported as acidic, ranging from 3.57 to 6.43 because no lime was added in WWTPs. While the optimum pH of anaerobic digestion is in the range 6.5–7.5 (Liu et al., 2008).

Besides, the concentration of nutrients, such as nitrogen, phosphorus, iron, nickel, etc., it is crucial as the bacteria needs them for optimum growth. Although these elements are needed in low concentrations, the lack of these nutrients interferes microbial growth and performance. Methanogenic archaea have relatively high internal concentrations of iron, nickel, and cobalt, which are 1,800, 100, and 75 mg/kg, respectively (Rajeshwari et al., 2000). Rosenani et al. (2008) determined the mean of the heavy metal content (mg/L) of 10 selected sewage sludge samples taken from different wastewater plants in Malaysia and found that elements like potassium, magnesium, iron and nickel were at significantly lower levels than the suggested concentration. Therefore, precise design and a good understanding of the anaerobic digestion process would be necessary to apply the process based on the different characteristic of each WWTPs sewage sludge.

On the other hand, Baki et al. (2005) reported that 85% of the WWTPs are serving <5,000 PE. Therefore, the volumes generated from those small size treatment plants might be too small to be considered for power generation, and even if the sludge were to be collected to one centralized location, it would be uneconomical concerning logistics and transportation costs. Under such circumstance, studies based on the smallscale WWTPs seen in Germany could be used as a reference when adopting anaerobic digestion in small-scale WWTPs in Malaysia (Dichtl et al., 2013). Therefore, the results determined may indicate that the current state of sewage sludge currently would be considered as low feasibility if anaerobic digestion is applied. To cope with the challenges mentioned above, more detailed studies as well as a different approach are needed.

### Anaerobic Co-digestion of Sewage Sludge

The complex microstructure components such as extracellular polymeric substances make sewage sludge difficult to hydrolyze and digest. These components have low volatile solid degradation (30–50%) even at a long retention time (20–30 day), caused low methane yield (Appels et al., 2008; Pereira et al., 2015). To accelerate the hydrolysis and enhance subsequent methane productivity, a variety of sewage sludge pretreatment options, such as mechanical, thermal, chemical, biological process or integration of these, have been developed at laboratory or pilot level recently (Kim et al., 2010; Houtmeyers et al., 2014; Kinnunen et al., 2015; Zhen et al., 2017). Although pretreatment can increase methane yield but leads to high energy demand and capital cost, complex operation, and maintenance, as an aspect of uneconomically feasible, caused most of them were still in the research stage and could not be utilized in real or commercial scale.

Anaerobic co-digestion is the simultaneous digestion of a homogenous mixture of two or more substrates. By combining two or more substrates or biomass resources, anaerobic codigestion could improve the economic viability of anaerobic digestion plants due to higher methane production (Alvarez et al., 2000), and hence is a well-known and feasible option to overcome the drawbacks of a conventional digestion process. Codigestion can improve the C/N ratio, enhance biogas production, overcome the nutrient imbalance of substrate, and improve the stable condition of anaerobic digestion (Kumaran et al., 2016). Anaerobic co-digestion of sewage sludge with biomasses resources which is a municipal solid waste, livestock manure, industrial waste, forestry, and agriculture waste can be used as cosubstrate in anaerobic co-digestion. Recently, different organic waste materials such as food waste with higher content of organic carbon have been mixed with sewage sludge in anaerobic codigester to improve C/N ratio which is leading to an increase in biogas production (Tanimu et al., 2014).

Anaerobic co-digestion between sewage sludge and the organic fraction of municipal solid waste (OFMSW) is the most reported co-digestion research (Alvarez et al., 2014). It was found that the cumulative biogas production from mixtures of sewage sludge and the OFMSW increased with increasing proportions of the OFMSW (Sosnowski et al., 2003). One study which evaluated the economic and environmental suitability of using OFMSW as co-substrate in two WWTPs had concluded that using OFMSW as co-substrate with sewage sludge was the most advantageous solution when compared to OFMSW composting and conventional digestion (Krupp et al., 2005).

The introduction of anaerobic co-digestion would be an advantageous approach to the current situation of OFMSW treatment in Malaysia as well. Nearly half of the Malaysian municipal waste is made up of food waste, and by 2020 the amount of municipal waste is estimated to increase to 9,820,000 tons in Peninsular Malaysia only (Tarmudi et al., 2009). Conventionally this municipal waste is disposed through landfill, but most of the landfill sites are mere open dumpsites where capacity has been exceeded. The high amount of food waste at landfill sites would lead to issues such as foul odor, toxic leachate and vermin infestation (Lee et al., 2007). There is currently no


clear strategy regarding the increasing food waste in Malaysia. Thus, the co-digestion of sewage sludge and food waste might be the most feasible breakthrough method for handling both biomass resources problems in Malaysia. In this case, WWTPs with low PE could be benefited by including food waste collected from the surrounding area. A well-organized system or structure that includes collection and separation of municipal solid waste would certainly lead to a more feasible anaerobic digestion process of sewage sludge.

#### CONCLUSIONS

This this study we reviewed the current state of sewage sludge treatment in Malaysia and discussed the challenges to promote anaerobic digestion in sewage sludge treatment. Currently, the sewage sludge produced from WWTPs is mostly disposed of through landfill and incineration. Although noteworthy movements can be seen in many regions including developing countries, where sewage sludge is utilized using anaerobic digestion for energy recovery as well as alternative to replace conventional methods, the focus on sewage sludge within Malaysia is not strong. The increasing volume of sewage sludge produced, corresponding with the growth of population, should not be treated lightly as the process itself consumes a lot of energy. Furthermore, the majority of WWTPs in Malaysia are operated on a small scale, and so the feasibility of anaerobic digestion, if applied, would be a challenge. Though various technical, political, and economic factors constrain anaerobic digestion of sewage sludge, the anaerobic co-digestion of food waste and sewage sludge might be the most feasibility method to overcome such constraints. It would not only improve the performance of anaerobic digestion when applied on sewage sludge but also simultaneously solve the problem of the increasing food waste

#### REFERENCES


in Malaysia. Nevertheless, considering the lack of information as well as the complexity of anaerobic digestion itself, more research, especially with some demonstration projects, are needed to verify the proper design, suitable conditions, and practical approach for the utilization of sewage sludge using anaerobic digestion in Malaysia.

Overall anaerobic digestion is a proven technique and at present applied in a variety of waste streams. To cope with the increasing energy demand, as well as to mitigate the environmental impacts as a result of the present situation, immediate but effective actions have to be taken, both by the government and the private sectors in Malaysia. Promotion and application of utilizing sewage sludge using anaerobic digestion could be not only one of the solutions to various environmental problems, but also a breakthrough in the field of waste recycling for Malaysia, where the accumulation of knowledge and understanding about anaerobic digestion could be shared and used for other biomass resources.

#### AUTHOR CONTRIBUTIONS

FH and LY carried out the literature review and wrote the initial draft. HA, YA, and TY revised the manuscript. HK and HD revised the manuscript and provided over-all guidance for work and editing of the text.

#### ACKNOWLEDGMENTS

We would like to acknowledge financial support from the Indonesian Endowment Fund for Education (LPDP), Ministry of Finance; and the Directorate General of Higher Education (DIKTI), Ministry of Research, Technology and Higher Education, Republic of Indonesia.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Hanum, Yuan, Kamahara, Aziz, Atsuta, Yamada and Daimon. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Thermogravimetric Characteristics and Non-isothermal Kinetics of Macro-Algae With an Emphasis on the Possible Partial Gasification at Higher Temperatures

#### Imtiaz Ali\* and Ali Bahadar

*Department of Chemical and Materials Engineering, King Abdulaziz University, Rabigh, Saudi Arabia*

Pyrolysis of *Turbinaria ornata* was realized in a thermogravimetric analyzer. The process was crudely classified into primary and secondary reaction zones. In the primary reaction zone, the thermal decompositions of the low-thermal stable components produced volatiles and biochar. The solid products obtained from the primary decomposition reactions contained inorganics with heavy metals. At mild-to-high temperatures, the catalytic effects accompanied gasification using oxygen, which was partially supplied by the oxygen carriers present in the solids and evolved gases. In order to study the pyrolytic conversion, combined, and multiple reaction schemes were employed. While the model-free methods helped to provide the accurate activation energies and the initial value of the pre-exponential factor, the non-linear regression optimized the chosen model parameters. In this study, a simple order-based model was compared with the versatile Šesták-Berggren (SB) model considering combined and multiple reactions. The application of a multiple reaction scheme to the primary and secondary reaction zones concluded that a simple order-based model suffices for the kinetic analysis. The secondary decomposition was shown to start with a high activation energy, which decreased appreciably when the conversion proceeded toward completion.

#### Edited by:

*Mohammad Rehan, King Abdulaziz University, Saudi Arabia*

#### Reviewed by:

*Meisam Tabatabaei, Agricultural Biotechnology Research Institute of Iran, Iran Vivekanand Vivekanand, Malaviya National Institute of Technology, Jaipur, India*

> \*Correspondence: *Imtiaz Ali imtiaz\_che@hotmail.com*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *19 October 2018* Accepted: *18 January 2019* Published: *13 February 2019*

#### Citation:

*Ali I and Bahadar A (2019) Thermogravimetric Characteristics and Non-isothermal Kinetics of Macro-Algae With an Emphasis on the Possible Partial Gasification at Higher Temperatures. Front. Energy Res. 7:7. doi: 10.3389/fenrg.2019.00007* Keywords: Turbinaria ornata, seaweed, pyrolysis, pseudo-components, independent parallel reactions

## INTRODUCTION

The last few decades have seen an increase in the use of biofuels and high-value biochemicals derived from biomass. Biomass is a renewable carbonaceous source, an alternative to fossil fuel, and a precursor of biochemicals. Non-food biomasses are widespread in nature and are also cultivated at ∼100 billion tons annually (Sheldon, 2014). According to an estimate, biofuels are expected to satisfy around 10% of the global energy demand by 2035, and they have the potential to replace 27% of global transportation fuel by 2050. The main reason for the growing interest in the use of biomass, besides its renewable nature, is its carbon-neutral and less polluting characteristics. The Paris climate agreement requires clean energy transformation in order to reach net zero emissions by 2060, which can only be achieved by scaling-up efficient energy conversion systems to generate bio-energy with carbon capture and storage (BECCS) (Agency, 2017).

Marine macro-algae or seaweeds are efficient photosynthetic organisms. Besides their use in the food, pharmaceutical, and cosmetic industries, they are also well-known for their use in bio-sequestration and bioremediation, thereby mitigating climate change (Maceiras et al., 2011; Oliveira, 2016). There has been increased interest in macro-algae as a renewable and sustainable energy source over the past decade among the scientific and industrial communities because of its high biomass productivity (Baghel et al., 2017). Macro-algae biomass is mainly produced from the harvesting of wild stocks, a recent report of FAO reveals that a volume of around 12 million tons is produced annually (Ferdouse et al., 2018). Its cultivation is normally carried out through ocean farming and thus does not require land or fresh water, allowing these to be used for food crops and aquaculture. Macro-algae is a cheap source of carbon with a varying chemical composition depending on the type and the environmental conditions. It can be crudely classified into four main groups depending on the stored nutrients, pigments and chemical composition, namely red seaweeds (Rhodophyceae), green seaweeds (Chlorophyceae), brown seaweeds (Phaeophyceae), and blue-green seaweeds (Cyanophyceae) (Manivannan et al., 2009).

Seaweeds, like terrestrial forms of biomass, are lignocellulosic in nature with a complicated embedding of cellulose, hemicellulose, lignin, and intracellular substances. However, compared to terrestrial biomass, they may have higher costs attached to their cultivation in conjunction with the presence and effects of heavy metal concentrations in aquatic environments, while the removal of harmful contents (sulfur and nitrogen) from the resulting fuel (Ghadiryanfar et al., 2016) is another problem. Furthermore, some studies have revealed that the transesterification of oil extracted from macro-algae is restricted due to low lipid contents, and they have suggested the conversion of carbohydrates into biofuel but with a higher gas yield (Roesijadi et al., 2010; Hong et al., 2017). Therefore, the most suited conversion technologies, such as anaerobic digestion and hydrous pyrolysis (Ross et al., 2008), are likely to be employed for macro-algal biomass containing high levels of ash and alkali metals. Macro-algae are rich in macro and micronutrients, but also display an uptake of heavy metals from the environment. The heavy metals in macro-algae may indicate the fractions of industrial effluents affecting the aquatic ecosystems, which is also a serious concern. These accumulations depend upon their biochemical composition, specifically the properties of the cell wall (Davis et al., 2003) and the constituents present in the macro-algae cell walls. The cell wall contains the fibers and matrix of polysaccharides (alginates and fucoidan). The alginates in brown algae have a high accumulation of divalent cations, while sulfated polysaccharides exhibit a tendency toward trivalent cations (Saha and Orvig, 2010; Ortiz-Calderon et al., 2017). **Figure 1** presents the various major constituents of the seaweed cell wall.

The uptake of heavy metals takes place in macro-algae through bio-sorption or ion exchange processes by the chemical constituents present in the cell wall. The Red Sea, with its diverse ecosystem, has over 500 species of seaweeds, which are spread over 1,900 km covering 438,000 km² of coral reef

stretching along its coasts and part of the Indian-Pacific Ocean (Spalding et al., 2001; Siddall et al., 2004; Bruckner et al., 2013). Red Sea seaweeds primarily consist of small green and brown algal filaments found in the northern and central parts, while the southern area of the Red Sea has large brown algal species like Sargassum and Turbinaria spp (Bruckner et al., 2013). Turbinaria ornata (T. ornata) is a well-known brown alga suitable for human consumption due to its antioxidant characteristics (Deepak et al., 2017a). Several studies have discussed its antibacterial, antioxidant, and anticancer properties (Omar et al., 2012; Ponnan et al., 2017; Stranska-Zachariasova et al., 2017; Tenorio-Rodriguez et al., 2017) as well as the biogenic synthesis of crystalline Ag/Au nanoparticles (Rajeshkumar et al., 2013; Kayalvizhi et al., 2014; Deepak et al., 2017b), but to the best of the authors' knowledge, no previous studies are available for its use as a source of biofuel. The mineral and heavy metal uptakes by different Turbinaria species are presented in **Table 1**.

Chemical looping pyrolysis gasification (CLPG) and chemical looping combustion (CLC) are proven concepts for syngas production and CO2-capturing, respectively. The use of oxygen carriers such as the oxides of Ni, Mn, Cu, Fe, and Co have previously been investigated in these processes (Adanez et al., 2012; Huang et al., 2017). In the present study, the pyrolysis of T. ornata was realized in a thermogravimetric analyzer (TGA). The unique and novel concept of pyrolysis-gasification with the oxides of heavy metals already embedded in the matrix of polysaccharides of brown macro-algae (T. ornata) has also been considered besides the condensation of polymers at higher temperatures. The understanding of the role of inherent inorganic compounds of T. ornata and the investigation into evolved gases at high temperatures will lead to process optimization, efficient use, and the removal of heavy metals in the form of biochar.

## MATERIALS AND METHODS

## Macro-Algae Biomass

T. ornata is a brown alga that has phytochemical constituents like carbohydrates, proteins, flavonoids, alkaloids, and phenols as well as sulfated fucan-polymers like polysaccharide and alginate


structures that display tremendous potential for use in various applications. Its ability to bind heavy metals is of particular interest. Its biochar can also be used as a bio-indicator for identifying the type and quantity of heavy metals present in aquatic environments.

T. ornata was collected from the Red Sea coastal area (N22◦ 48′ 32′′, E38◦ 56′ 37′′). The fresh macro-algae was thoroughly washed to remove adherent dirt and sand with deionized water several times. The cleaned biomass was air-dried for 2 days, then crushed and sieved. A fine powder with a particle size of <0.147 mm (the fraction of the material passed through a 100 mesh sieve) was kept in an air-dried container for further use. **Table 2** shows the physicochemical properties of T. ornata.

#### Thermogravimetric Analysis

Pyrolysis is a promising process to convert biomass whereby the distribution of the product depends greatly on the conditions used. It has been reported that slow pyrolysis yields more biochar whereas fast pyrolysis produces more volatiles (Mohan et al., 2006). Extensive development in the pyrolysis process has been made recently and there has been a growing interest in its use because of its simplicity, flexibility, and efficiency. Biochar, a product of pyrolysis, finds in addition to volatiles potential application in the agricultural and environmental sectors. A thermogravimetric analyzer (TGA) is a commonly employed thermo-analytic instrument for the thermal degradation study of biomass. Under well-defined conditions of slow pyrolysis, it provides precise time and temperature-series mass loss data, which are used to estimate the kinetic parameters.

In the present study, the TGA experiments were performed on a Q5000 IR electro-balance analyzer (TA Instruments, USA). Approximately 15 mg of algal biomass sample was used for each run in triplicate to ensure accuracy and precision. The TGA temperature was raised from 50 to 900◦C using different heating rates (5, 10, 15, and 30 <sup>o</sup>C·min−<sup>1</sup> ), while a flow of high purity N<sup>2</sup> gas was maintained at 70 ml min−<sup>1</sup> . Additionally, an isothermal step was introduced before the heating ramp to achieve pyrolysis stabilization. The acquisition, diagnosis, and processing of curves were performed on PlatinumTM software.

#### Kinetic Evaluation and Analysis

The devolatilization of algal biomass through pyrolysis is useful in understanding multi-component and multi-pyrolytic phases for process design, optimization, and scale-up (Ranzi et al., 2017).



*<sup>a</sup>ASTM standard methods (E1756, E1755-01, and D3174-12) were used except for fixed carbon, which was calculated from the difference.*

*<sup>b</sup>C, H, and N were measured with a CHN/O 2400 series II Perkin Elmer elemental analyzer.*

TABLE

1


Macro/micro

 mineral and heavy metal uptake by Turbinaria macroalgal

 species.

A global single-step reaction is always used in assessing the kinetic mechanism of pyrolysis. In a simplistic approach, the overall reaction process of T. ornata pyrolysis can be described with the following reaction:

$$\begin{array}{rcl} \text{Turbinaria orrata (s)} \rightarrow \text{ Volatilis (g)} + \text{Biochar (s)} & \end{array} \quad \text{(1)}$$

A non-isothermal TGA analysis technique is usually preferred over an isothermal technique for three main reasons. Firstly, the isothermal procedures are simply not possible to realize at higher temperatures because of the large non-isothermal heat-up and cool-down durations. Secondly, the estimation of the kinetic parameters at low heating rates is rather difficult due to weight loss during the heat-up time. Thirdly, there is the existence of the almost non-zero extent of conversions (Vyazovkin and Wight, 1998; Grigiante et al., 2016). For the aforementioned reasons, a non-isothermal technique was selected for this study. The process data recorded by TGA during non-isothermal pyrolysis can be expressed in terms of conversion (α) at any time (t) using the Arrhenius law [43]:

$$\frac{d\alpha}{dt} = \beta \frac{d\alpha}{dT} = A \cdot \exp\left(-\frac{E\_a}{RT}\right) \cdot f\left(\alpha\right) \tag{2}$$

where E<sup>a</sup> (J·mol−<sup>1</sup> ) is the apparent activation energy, A (s−<sup>1</sup> ) is the frequency factor, T is the absolute temperature, R is the ideal gas constant (J·mol−<sup>1</sup> ·K −1 ), β is the heating rate (K·s −1 ), and f(α) represents the algebraic form of the reaction model. The conversion (α) is defined as the fractional mass loss expressed by the following equation:

$$\alpha = \frac{m\_o - m\_i}{m\_o - m\_f} \tag{3}$$

where m0, m<sup>i</sup> , and m<sup>f</sup> are the intial, instantaneous, and final mass, respectively.

Integrating Equation (1) with respect to conversion and temperature in the ranges α ∈ [0, α] and T ∈ [To, T] gives Equation (4):

$$\log\left(\alpha\right) = \int\_0^a \frac{d\alpha}{f\left(\alpha\right)} = \frac{A}{\beta} \int\_{T\_0}^T \exp\left(-\frac{E\_a}{RT}\right) \mathrm{d}T \tag{4}$$

Equation (4) can be used to describe the conversion when the temperature lag between the sample and its environment is very small. E<sup>a</sup> is an important kinetic parameter, defined as the energy required to have a reaction, which is usually measured by model-free and model-fitting methods. Each method has its own advantages, and they are complementary and are not in competition. A model-free analysis can be performed without having specified the reaction mechanism, while the model-fitting analysis uses the approach of data-model best fitting to estimate Ea, A, and f(α) (Šesták and Berggren, 1971).

#### Model-Free Methods

Model-free kinetics methods are mostly isoconversional in nature. They provide an accurate estimation of E<sup>a</sup> from isothermal and non-isothermal measurements and are most commonly employed in the pyrolytic kinetics of biomass (Opfermann et al., 2002). These methods have the capability to solve complex processes and are based on the fact that the reaction rate at a particular heating rate (β) is a function of temperature (T) alone at a specific conversion (α), while f(α) remains constant. Model-free methods provide a detailed kinetic analysis without knowledge of the reaction mechanism while indicating multiple processes if E<sup>a</sup> varies strongly with α (Vyazovkin et al., 2011). Depending on the mathematical form of the kinetic equation, model-free methods can be differential or integral. The Friedman method (Friedman, 1964) is the most common differential while the Kissinger-Akahira-Sunose (KAS) method (Kissinger, 1957) is a widely used integral method. The Friedman method reveals the dependency of the conversion rate ln dα dt αi over 1/Tα<sup>i</sup> as a straight line:

$$\text{Friendman}: \quad \ln \left( \frac{\text{d}\alpha}{\text{d}t} \right)\_{\alpha\_i} = \text{const} - \frac{E\_{\alpha}}{R \cdot T\_{\alpha\_i}} \tag{5}$$

The apparent activation energy (Eα) at a particular conversion (α) is estimated from the slope of the straight line. Integral methods are relatively less sensitive to data noise as compared to differential isoconversional methods and offer consistent E<sup>α</sup> measurements. The KAS method is represented by the following linear Equation (6):

$$\mathbf{KAS}: \qquad \ln\left(\frac{\beta\_i}{T\_{\alpha\_i}^2}\right) = \text{const} - \frac{E\_{\alpha}}{R \cdot T\_{\alpha\_i}}\tag{6}$$

The slope of the linear fit of the ln βi T2 αi vs. 1/Tα<sup>i</sup> is used to evaluate the apparent Eα.

#### Model-Fitting Methods

The model-fitting analysis is conventionally used for the pyrolytic assessment of biomass materials on the basis of their devolatilization weight loss measurements using 1storder decomposition single-step reaction modeling or adding the weight of multiple pseudo-components like cellulose, hemicellulose, lignin, and others (Aboyade et al., 2012). The resulting characteristics from the kinetic analysis by model-fitting yield detailed simulations of biomass pyrolysis experiment data. Also, many studies have reported more precise simulations when using nth-order for specific reaction mechanisms where an optimization of the unknown constant is needed (Osman et al., 2017). Thus, in model-fitting procedures many models were applied to data collected from experiments to achieve the best fit to assess the kinetic parameters.

#### Combined Kinetics

The simple nth-order reaction has been used extensively to study the pyrolytic kinetics of biomass materials (Xu et al., 2017). The nth-order reaction mechanism for a solid state generally uses an expression like f(α) = (1 − α) n , which when used in Equation (2) becomes:

$$\frac{d\alpha}{dt} = A \cdot \exp\left(-\frac{E\_a}{RT}\right) \cdot (1 - \alpha)^n \tag{7}$$

Another function is the Šesták—Berggren model (Šesták and Berggren, 1971), which was employed for the kinetic fitting of the experimental data. This is because of the high flexibility of the function to investigate the various types of physicogeometrical mechanism of the solid-state reaction (Kissinger, 1957; Opfermann et al., 2002; Vyazovkin et al., 2011; Aboyade et al., 2012), deviating cases with non-integral or fractal reaction geometry (Šimon, 2011; Dussan et al., 2017; Osman et al., 2017), the size distribution of the reactant particles (Orfão et al., 1999), and so on (Zubia et al., 2005). The high flexibility makes it difficult to interpret the physicochemical meanings of the kinetic exponent, but the SB model with the three parameters indicates the highest performance as a fitting function.

$$\text{SB}: \qquad f\left(\alpha\right) = \alpha^m . \left(1 - \alpha\right)^n \cdot \left[ -\ln\left(1 - \alpha\right) \right]^p \tag{8}$$

The SB model with three fit parameters (m, n, and p) is considered as one of the highest performing model-fitting methods, where m is accelerating behavior, n represents the order of the reaction and p shows the nuclei growth. By incorporating the SB model, Equation (7) becomes:

$$\frac{d\alpha}{dt} = A \cdot \exp\left(-\frac{E\_a}{RT}\right) \cdot (1-\alpha)^n \cdot \alpha^m \cdot \left[-\ln\left(1-\alpha\right)\right]^p \tag{9}$$

The SB (m, n, p), for specific reaction model functions and other parameters like E<sup>a</sup> and A can be optimized by employing a nonlinear least square analysis, whereby the residual sum of squares, also known as the sum of squared errors (SSE), is minimized by fitting the data curve of <sup>d</sup><sup>α</sup> dt exp from the predicted curve of dα dt pred against time.

$$\text{SSE} = \sum \left( \left. \frac{\text{d}\alpha}{\text{d}t} \right|\_{\text{exp}} - \left. \frac{\text{d}\alpha}{\text{d}t} \right|\_{\text{pred}} \right)^2 \tag{10}$$

A non-linear least square analysis always requires the initial values for SB (m, n, p), E<sup>a</sup> and A for optimization. After fitting the data, the goodness of the fit was measured by Fit (%), as in Equation (11).

$$\text{Fit } (\%) = \left[ 1 - \frac{\sqrt{\text{SSE}}}{\left( \left. \frac{\text{d} \alpha}{\text{d}t} \right|\_{\text{exp}} \right)\_M} \right] \times 100 \tag{11}$$

where dα dt exp M is the peak-maximum conversion rate.

#### Independent-Parallel Reactions

Independent reaction models are mainly employed to identify the roles of the components in the thermal decomposition process. Biomass consists of many components and each component behaves differently. Therefore, these models have a set of discrete multiple reactions and may be extended to continuous distribution, which is largely employed to study the degradation kinetics of lignocellulosic materials. It is commonly accepted that each pseudo-constituent degrades independently at a particular temperature range. The weight loss calculations are performed by taking the sum of the mass fractions from the individual pseudo-components' decomposition rates (Orfão et al., 1999). The polysaccharides of T. ornata are mainly hemicellulose, cellulose, and others. The proximate composition of different Turbinaria species is given in **Table 3** below.

T. ornata contains many components and each component can be divided into subcomponents. During decomposition, the components which degrade during a similar temperature range may be grouped together and termed as pseudocomponent. Lignocellulosic materials are normally classified into three pseudo-components, whereby each pseudo-component is involved in an independent reaction (Orfão et al., 1999). Similarly, the pyrolysis of T. ornata can be presented in terms of three pseudo-independent components, as given in Equation (12).

$$\begin{array}{c} \text{Pseudo} - \text{Hemeicellulose} \stackrel{\frac{\text{dav}\_1}{\text{d}t}}{\rightarrow} \\ \text{Pseudo} - \text{Cellulose} \stackrel{\frac{\text{dav}\_2}{\stackrel{\text{dav}\_3}{\text{d}t}}}{\rightarrow} \\ \text{Ohers} \end{array} \begin{array}{c} \text{\(\frac{\text{dav}\_1}{\stackrel{\text{dav}\_3}{\stackrel{\text{dav}\_3}{\text{d}t}}}\)} \\ \text{Volatules1} + \text{Biochar1} \end{array} \text{\(12\)}$$

During the main stage of the pyrolysis process, the pseudocomponents are subjected to the thermal splitting of chemical


bonds for the production of primary Volatiles1, gases (e.g., CH4, CO2, and CO), and condensates and yield biochar solids at <500◦C, named as Biochar1. Biochar contains minerals and even trace elements which were present in parent biomass. The pyrolytic conversion of three pseudo-components (i.e., pseudohemicellulose, pseudo-cellulose, and others) mainly depends on the temperature, particle size, and reactor type. The decomposition of hemicellulose and cellulose usually occurs between 470 to 530 K and 510 to 620 K, respectively, while others like lignin degrade at 550–770 K (Wang et al., 2017). The decomposition of cellulosic materials is mainly induced by depolymerization reactions forming oligosaccharides an then levoglucosan, which is finally decomposed into smaller molecules (Patwardhan et al., 2011). Similarly, the hemicellulose pyrolysis is initiated by the depolymerization of polysaccharides into oligosaccharides and, subsequently, leads to the rearrangement of the molecules, producing 1,4-Anhydro-D-xylopyranose, which further disintegrates into smaller molecular weight components.

At high temperature, secondary decomposition leads to the formation of secondary volatiles (Volatiles2) and biochar (Biochar2). This may include cracking, reforming, condensation, polymerization, oxidation, and gasification (Di Blasi, 2008). During the secondary reactions, Biochar1 forms Volatiles2 and Biochar2, as presented in Equation (13).

$$\stackrel{\text{d}\omega\_4}{Biochar1} \stackrel{\text{d}\omega\_4}{\rightarrow} Volatelles2 + Biochar2\tag{13}$$

Likewise, the inorganic compounds are also greatly affected during pyrolysis. These inorganic compounds, including heavy metals, may be accumulated in the cell wall and bind ionically and covalently to polysaccharides, which can initiate autocatalytic pyrolysis (Keown et al., 2005). Here, a concept of the partial gasification of T. ornata polluted with heavy metals from the Red Sea is presented in **Figure 2**.

These inorganic compounds may include macro/micro minerals (e.g., K, Ca, Na, Mg, Fe, Zn, and Mn) and heavy metals (e.g., Ag, Al, As, Cd, Cr, Co, Hg, Mo, Ni, Pb, and Ti). The pyrolysis process has been previously employed for processing these minerals and heavy metals for valuable products like biochar and bio-oil. For example, alkali and alkaline earth metals (such as K, Ca, and Mg) have notable catalytic effects on the pyrolytic conversion. Vaporization of K has been shown to take place at lower temperatures, while Ca and Mg vaporize at higher temperatures depending on their ionic or covalent bonding with organic molecules (Wei et al., 2012). Liu et al. (2017) reviewed biomass pyrolysis in the presence of inorganic elements including heavy metals and contributed a critical analysis for optimizing the pyrolytic mechanisms with productive resource utilization and a reduction in pyrolysis-based pollutants. Liu et al. (2017) also found that the prior literature has exhibited the enrichment of heavy metals in the solid biochar phase rather than being released into the volatile (gaseous/liquid) phase, and that there is no need to focus on the release of heavy metals in later research. Also, Ni and Cu have been previously used to enhance the quality of pyrolysis products (Richardson et al., 2013; Zhao et al., 2015). Hence, there is an urgent need to investigate the mechanisms for the use of inherited available inorganics and heavy metals to improve the processes and products (Liu et al., 2017). **Table 4** presents a few studies on biomass pyrolysis containing salts of heavy metals.

The summation of the rate equations in Equation (14) gives the overall degradation rate.

$$\frac{\mathrm{d}\alpha}{\mathrm{d}t} = \sum\_{j=1}^{4} \frac{\mathrm{d}\alpha\_{j}}{\mathrm{d}t} = \sum\_{j=1}^{4} \boldsymbol{\kappa}\_{j} \cdot \boldsymbol{A}\_{j} \cdot \exp\left(-\frac{E\_{aj}}{RT}\right) \cdot \boldsymbol{f}\left(\alpha\right) \tag{14}$$

The separation of the pseudo-component peaks was carried out assuming an asymmetric Fraser–Suzuki (FS) distribution with the following function (Perejón et al., 2011):

$$\frac{\mathrm{d}\alpha\_{j}}{\mathrm{d}t} = a\_{0} \exp\left[-\ln 2 \left\{ \frac{\ln \left(1 + 2a\_{I} \frac{T - T\_{m}}{\mathrm{HW}}\right)}{a\_{I}} \right\}^{2} \right] \tag{15}$$

where a<sup>0</sup> is the amplitude, a<sup>1</sup> is the asymmetry, T<sup>m</sup> is the peak maximum temperature and HW is the half width of the peak.

Initial values of kinetic parameters were obtained from modelfree methods applied to the separated pseudo-components. The optimization of the kinetic parameters was performed by nonlinear least square fitting using Equation (10).

#### RESULTS AND DISCUSSIONS

#### Thermal Behavior and Reactivity Analysis

**Figure 3** shows the TG and DTG profiling of T. ornata against temperature at different heating rates. T. ornata biomass was pyrolyzed to 850◦C at different heating rates. The DTG curves feature two prominent zones, as illustrated in **Figure 3B**: A primary pyrolysis zone and a secondary reaction zone. In the primary pyrolysis zone, more than 90% of the total amount of mass loss was observed, resulting from reactions such as depolymerization, decarboxylation, and cracking. The maximum mass loss rates increased from 0.043 to 0.290%·s <sup>−</sup><sup>1</sup> with an incremental increase in the heating rates of 5 to 30◦C·min−<sup>1</sup> . Secondary reaction zones (Zone II), initiated at 875 K, and the remaining mass was decomposed at a slower rate. This slow rate is caused by the trapped volatile matter present in the biochar and delays the decomposition (Richardson et al., 2013). The most common characteristic of all the DTG and TG curves is the increase in the conversion rate and the shifting of the peaks toward the higher temperatures when the heating rates are increased. This increase in conversion rates may be ascribed to the improved transport phenomenon. Meanwhile, the shifting of peaks is rendered to the slowing down of the pyrolysis process owing to the difficulty in the heat transfer from the surrounding to the sample for a shorter time and larger thermal lag (Shuping et al., 2010; Han et al., 2017; Özsin and Pütün, 2017). It can also be perceived from the DTG curves that the peaks in the secondary reaction zone (900–1,100 K) may be caused by the catalytic effect and gasification of the biochar due to the presence of inorganic oxides.

#### Isoconversional Methods

The pyrolytic characteristics were measured using the two isoconversional methods Friedman and KAS. **Figure 4** establishes the Arrhenius plot using Equations (5, 6) at different conversions. The data were fitted using linear regression lines for both differential and integral isoconversional methods at the progressive increase of conversion from 10 to 90% with a step size of 5%, as depicted in **Figure 4**.

The apparent activation energy E<sup>a</sup> was measured from the slope of the straight lines from the plots. The quality of the regression models was estimated by the coefficient of determination (R 2 ), which strengthens our confidence in the estimation of E<sup>a</sup> from the observed data. The mean apparent activation energy Eα, calculated from the Friedman and KAS isoconversional models, are 164.03 and 146.41 kJ·mol−<sup>1</sup> , respectively.

### Model-Fitting Methods

Model-fitting methods have two stages, namely (a) the estimation of the initial parameters and (b) the application of iterative routines to obtain the optimized values of the kinetic parameters.

### Combined Kinetics

The average activation energies (Ea) obtained from the isoconversional methods (both differential and integral) were used to obtain the order-based and SB model parameters by optimizing Equations (7, 9) using MS Excel Solver. The differential best-fit curves of T. ornata with optimized orderbased and SB model parameters at different heating rates (5, 10, 15, and 30◦C·min−<sup>1</sup> ) are shown in **Figure 5**. It is worth noting here that this displays the fitting of the data to order-based and SB models through a single reaction. Moreover, a single heating rate may render inaccuracies; therefore, the mean values calculated at different heating rates are presented in **Table 5**.

The SB model is a powerful tool that allows the measurement of complex degradation mechanisms (Gámiz-González et al., 2017). As is evident from **Table 5** and has been previously found as well, one of the SB model parameters is numerically absent (Šesták and Berggren, 1971; Ali et al., 2018). In the present study for all heating rates, parameter m, which corresponds to the deceleration mechanism, was not involved and hence the pyrolysis of macro-algae (T. ornata) was most probably driven by random nucleation and growth (Wang et al., 2006; Li et al., 2010; Ye et al., 2010). **Table 5** shows a Fit% higher for the SB model, which indicates that the SB model can better explain thermal degradation than the order-based model. Thus, the parameters obtained from the SB model can better represent the pyrolysis mechanism. The average values of A obtained from the Friedman method were 6.4 × 10<sup>13</sup> and 1.3 × 1014, while for the KAS method these were 1.0 × 10<sup>12</sup> and 2.3 × 10<sup>12</sup> for the order-based and SB models, respectively. The values of n for the order-based and SB models were 7.47 and 8.08 for the Friedman method and 6.76 and 7.44 for the KAS method, respectively. The high value of n in the SB model with a higher Fit% (>92) highlights the roles of order-based, diffusion and geometrical contraction in T. ornata pyrolysis. The order-based mechanism can be attributed to the nucleation process in a solid state reaction where the nucleus growth is induced by nuclei collisions with each other (Poletto et al., 2012). The optimum p-values (0.40 and 0.43) signify that the degradation process is partially controlled by the random nucleation mechanism, which randomly promotes the generation of activation centers. It is highly likely that the presence of oxygen carriers (gases and inorganic salts containing heavy metals) could initiate partial gasification (Huang et al., 2013) and catalyze T. ornata pyrolysis. This would increase the

#### TABLE 4 | Influence of heavy metals present in biomass during pyrolysis.


weight loss rate and increase the gases produced (like CH4, H2, and CO2) from the pyrolysis of T. ornata and could remove lattice oxygen from the oxygen carriers. Depending upon the amount of inorganics accumulated by the marine algae T. ornata, the catalytic characteristics of heavy metals stimulating pyrolysis and partial gasification could vary.

55.13 to 59.81% for *Arundo donax* and from 44.62 to 52.12 % for *Broussonetia papyrifera.*

#### Independent-Parallel Reactions

From the DTG curves of samples in **Figure 3B**, it is indicated that the degradation process is taking place in a multiple reaction scheme since the peaks are overlapping in the close temperature range. The thermal decomposition of hemicellulose in T. ornata biomass started at about 473 K, followed by a

major mass loss between 550 and 660 K and tailing up to 900 K; then there was a slight shoulder followed by a tail in the temperature range 880–1100 K, which could have induced the partial gasification mechanism using inherited minerals. Most of the mass loss occurs in Zone (I), as shown in **Figure 3B**, which takes the shape of a shoulder with overlapping peaks, which corresponds to hemicellulose and cellulosic material, followed by the contribution of other components with an interplay of the inherited minerals present in T. ornata collected from the Red Sea. The shoulder and tailing in Zone (II) may correspond to the decomposition of inorganics and could proceed further to a partial gasification process. Hence, it is necessary to apply three parallel-reaction models for the primary reactions in Zone (I) pertaining to the decomposition of pseudo-hemicellulose, pseudo-cellulose and pseudo-others. Overall, the DTG curves at 5, 10, 15, and 30◦C·min−<sup>1</sup> were decomposed into three primary and one secondary decomposition reactions by employing asymmetric Fraser-Suzuki (FS) distributions, as presented in **Figure 6**. Recent studies show have shown that the application of the FS function suits the thermal degradation of a solid state reaction into its individual processes (Millán et al., 2017). The mass loss contributions (xj) of separated peaks obtained from **Figure 6** are presented in **Table 6**.

The average relative mass loss contribution (xj) for pseudohemicellulose, pseudo-cellulose, pseudo-others, and biochar at 5, 10, 15, and 30◦C·min−<sup>1</sup> was 40.5, 13.0, 38.5, and 8.0%, respectively. It can also be seen from **Figure 5** that the degradation rates of pseudo-components increase with



increasing heating rates. Furthermore, with increasing heating rates the Fit% increases.

The major mass loss comes from the degradation of pseudohemicellulose and others, accounting for relative mass loss contributions of 40.5 and 38.5%, respectively. The decomposition continues over wide temperature ranges. The secondary decomposition in Zone (II) contributes to an additional mass loss of 8%, which results in the evolution of more volatiles. Therefore, the remaining mass after secondary decomposition, termed as Biochar2, further decreases. The decompositions of carbohydrates, others, and Biochar1 in the presence of inorganics indicates the cumulative effects of catalysts and gasification. This could encourage flash pyrolysis or rapid heating in order to fully utilize the gasification and catalytic effects of these inorganically rich marine biomasses.

The kinetic parameters of the pseudo-component degradations were estimated using the Friedman and KAS model-free methods on separated deconvoluted peaks. The E<sup>a</sup> of the independent-parallel reactions were calculated from the regression plots of Friedman and KAS and the evolution of



TABLE 7 | Optimized kinetic parameters of pseudo-components.


activation energies with respect to conversion; these are shown in **Figure 7**.

It is clear from **Figure 7** that the primary decompositions overlap and that the interplay of the components causes the overall activation energy decrease until around 90% of the conversion. This aspect has been regarded by some researchers to the condensation of polymers at higher temperature to form char (Chorley, 1892; Basile et al., 2016; Anca-Couce and Scharler, 2017; Ciuta et al., 2018). The secondary conversion of Biochar1 at a higher temperature with a higher activation energy, which drops as the conversion proceeds toward completion. Average activation energies of the pseudo-components and initial parameter values obtained from the Friedman and KAS methods were used to optimize the kinetic parameters via non-linear regression. The SB parameters m and p were absent, hence both order-based and SB models were found to be equivalent. The obtained parameters using differential and integral methods at different heating rates are presented in **Table 7**.

A summary of the key thermal and kinetic parameters is tabulated in **Table 8**. It is evident that the peak-maximum temperature (Tm) for the decomposition of pseudo-hemicellulose is 541.2 K for the differential Friedman method and 541.5 K for the integral KAS method. The T<sup>m</sup> for pseudo-cellulose is 598.4 and 583.4 K using the Friedman and KAS methods, respectively. Pseudo-others decompose with a T<sup>m</sup> of 641.8 to 634.2 K according to the Friedman and KAS methods. The decomposition at mild-to-high temperatures may also involve condensation of polymers. Depending on the heating rate, the


secondary decomposition of Biochar1 occurs in a temperature range of 870–1,100 K with a T<sup>m</sup> of 963.9 and 964.2 K using the Friedman and KAS methods, respectively.

The average reaction order (n) of the decomposition of pseudo-hemicellulose, pseudo-cellulose, others, and biochar was found to be 2.09, 1.03, 1.03, and 1.69 for the Friedman method and 2.33, 1.27, 1.07, and 1.66 for the KAS method, respectively. Similarly, the average pre-exponential factor (A) of the decomposition of pseudo-hemicellulose, pseudo-cellulose, others, and biochar was found to be 4.7 × 1018, 2.1 × 10<sup>16</sup> , 0.06, and 9.5 × 10<sup>12</sup> s −1 , respectively, for the Friedman method and 5.9 × 1019, 1.5 × 1015, 0.07, and 1.6 × 10<sup>10</sup> for the KAS method, respectively.

The average E<sup>a</sup> of pseudo-hemicellulose obtained from the Friedman and KAS methods was 183.7 and 204.2 kJ.mol−<sup>1</sup> , respectively. Similarly, the average E<sup>a</sup> of pseudo-cellulose using the Friedman and KAS methods was 205.2 and 203.4 kJ.mol−<sup>1</sup> , respectively. Pseudo-others decomposed with a lower average E<sup>a</sup> of 25.8 and 25.1 kJ.mol−<sup>1</sup> , as estimated from the Friedman and KAS methods, respectively. The average E<sup>a</sup> of pseudo-others is very low, which may be due to the formation of polymers owing to the metal oxides from inherited inorganics taken up by T. ornata macro-algae. Jian et al. (2014) found that the E<sup>a</sup> of metal oxides was low at higher heating rates, indicating oxygen transport to be a rate limiting step at high temperatures and high heating rates. This may induce a partial gasification phenomenon at this stage. The partial gasification, according to Equation (8), could be due to the inherited metal oxides present in the cell wall of the polysaccharide matrix of T. ornata. The average E<sup>a</sup> was calculated for the degradation of Biochar1, which was 240.1 and 236.5 kJ.mol−<sup>1</sup> using the Friedman and KAS methods, respectively. It can also be deduced that the

#### REFERENCES

variation in the values of A, n, and E<sup>a</sup> was the greatest during the secondary decomposition of Biochar1 in Zone (II) among the other pseudo-components. Moreover, the sharp decrease in the E<sup>a</sup> of Biochar1 is a mere indication of the spontaneity of the process.

#### CONCLUSIONS

Marine macro-algae contains inorganics, including heavy metals, the amount of which depends on the macro-algae's ability to uptake these as well as the concentration of inorganics dissolved in the marine environment. While this help mitigates the heavy metals, their disposal needs care. Pyrolysis offers not only a means to handle such materials but also the ability to use these to our benefit in the form of valuable products such as volatiles and biochar. The Red Sea is a rich source of many minerals, which are taken up by the organisms living there. The role of these organisms in this process is important and needs to be evaluated using complex models. A four-component independent-reaction model using a simple order-based mechanism for each reaction is sufficient to analyze the pyrolytic kinetics. This shows that three primary reactions proceed simultaneously, producing a solid product which degrades through a secondary decomposition reaction at a higher temperature. Interestingly, the activation energy drops considerably for the secondary reaction, hinting at the partial gasification and the catalytic effects of the minerals inherent in the macro-algae.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

Abdallah, A. M. A., Abdallah, M. A., Beltagy, A., and Siam, E. (2006). Contents of heavy metals in marine algae from Egyptian Red Sea coast. Toxicol. Environ. Chem. 88, 9–22. doi: 10.1080/02772240500414911

Aboyade, A. O., Carrier, M., Meyer, E. L., Knoetze, J. H., and Görgens, J. F. (2012). Model fitting kinetic analysis and characterisation of the devolatilization of coal blends with corn and sugarcane residues. Thermochim. Acta 530, 95–106. doi: 10.1016/j.tca.2011. 12.007


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling Editor declared a shared affiliation at the time of review, though no other collaboration, with the authors.

Copyright © 2019 Ali and Bahadar. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Catalytic Pyrolysis of Plastic Waste: Moving Toward Pyrolysis Based Biorefineries

Rashid Miandad<sup>1</sup> , Mohammad Rehan<sup>2</sup> \*, Mohammad A. Barakat 3,4, Asad S. Aburiazaiza<sup>3</sup> , Hizbullah Khan<sup>1</sup> , Iqbal M. I. Ismail <sup>2</sup> , Jeya Dhavamani <sup>2</sup> , Jabbar Gardy <sup>5</sup> , Ali Hassanpour <sup>5</sup> and Abdul-Sattar Nizami <sup>2</sup>

*<sup>1</sup> Department of Environmental Sciences, University of Peshawar, Peshawar, Pakistan, <sup>2</sup> Centre of Excellence in Environmental Studies, King Abdulaziz University, Jeddah, Saudi Arabia, <sup>3</sup> Department of Environmental Sciences, Faculty of Meteorology, Environment and Arid Land Agriculture, King Abdulaziz University, Jeddah, Saudi Arabia, <sup>4</sup> Central Metallurgical R&D Institute, Helwan, Egypt, <sup>5</sup> School of Chemical and Process Engineering, University of Leeds, Leeds, United Kingdom*

Pyrolysis based biorefineries have great potential to convert waste such as plastic and biomass waste into energy and other valuable products, to achieve maximum economic and environmental benefits. In this study, the catalytic pyrolysis of different types of plastics wastes (PS, PE, PP, and PET) as single or mixed in different ratios, in the presence of modified natural zeolite (NZ) catalysts, in a small pilot scale pyrolysis reactor was carried out. The NZ was modified by thermal activation (TA-NZ) at 550◦C and acid activation (AA-NZ) with HNO3, to enhance its catalytic properties. The catalytic pyrolysis of PS produced a higher liquid oil (70 and 60%) than PP (40 and 54%) and PE (40 and 42%), using TA-NZ and AA-NZ catalysts, respectively. The gas chromatography-mass spectrometry (GC-MS) analysis of oil showed a mixture of aromatics, aliphatic and other hydrocarbon compounds. The TA-NZ and AA-NZ catalysts showed a different effect on the wt% of catalytic pyrolysis products and liquid oil chemical compositions, with AA-NZ showing higher catalytic activity than TA-NZ. FT-IR results showed clear peaks of aromatic compounds in all liquid oil samples with some peaks of alkanes that further confirmed the GC-MS results. The liquid oil has a high heating value (HHV) range of 41.7–44.2 MJ/kg, close to conventional diesel. Therefore, it has the potential to be used as an alternative source of energy and as transportation fuel after refining/blending with conventional fuels.

Keywords: catalytic pyrolysis, pyrolysis based biorefineries, natural zeolite, plastic waste, aromatic compounds, modified natural zeolite, catalyst

### INTRODUCTION

Plastic waste production and consumption is increasing at an alarming rate, with the increase of the human population, rapid economic growth, continuous urbanization, and changes in life style. In addition, the short life span of plastic accelerates the production of plastic waste on a daily basis. The global plastic production was estimated at around 300 million tons per year and is continuously increasing every year (Miandad et al., 2016a; Ratnasari et al., 2017). Plastics are made of petrochemical hydrocarbons with additives such as flame-retardants, stabilizer, and oxidants that make it difficult to bio-degrade (Ma et al., 2017). Plastic waste recycling is carried out in different ways, but in most developing

#### Edited by:

*Wei-Hsin Chen, National Cheng Kung University, Taiwan*

#### Reviewed by:

*Gopalakrishnan Kumar, University of Stavanger, Norway Tzu-Hsien Hsieh, CPC Corporation, Taiwan*

\*Correspondence: *Mohammad Rehan dr.mohammad\_rehan@yahoo.co.uk*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *15 November 2018* Accepted: *22 February 2019* Published: *19 March 2019*

#### Citation:

*Miandad R, Rehan M, Barakat MA, Aburiazaiza AS, Khan H, Ismail IMI, Dhavamani J, Gardy J, Hassanpour A and Nizami A-S (2019) Catalytic Pyrolysis of Plastic Waste: Moving Toward Pyrolysis Based Biorefineries. Front. Energy Res. 7:27. doi: 10.3389/fenrg.2019.00027* countries, open or landfill disposal is a common practice for plastic waste management (Gandidi et al., 2018). The disposal of plastic waste in landfills provide habitat for insects and rodents, that may cause different types of diseases (Alexandra, 2012). Furthermore the cost of transportation, labor and maintenance may increase the cost of recycling projects (Gandidi et al., 2018). In addition, due to rapid urbanization, the land available for landfills, especially in cities, is reducing. Pyrolysis is a common technique used to convert plastic waste into energy, in the form of solid, liquid and gaseous fuels.

Pyrolysis is the thermal degradation of plastic waste at different temperatures (300–900◦C), in the absence of oxygen, to produced liquid oil (Rehan et al., 2017). Different kinds of catalysts are used to improve the pyrolysis process of plastic waste overall and to enhance process efficiency. Catalysts have a very critical role in promoting process efficiency, targeting the specific reaction and reducing the process temperature and time (Serrano et al., 2012; Ratnasari et al., 2017). A wide range of catalysts have been employed in plastic pyrolysis processes, but the most extensively used catalysts are ZSM-5, zeolite, Y-zeolite, FCC, and MCM-41 (Ratnasari et al., 2017). The catalytic reaction during the pyrolysis of plastic waste on solid acid catalysts may include cracking, oligomerization, cyclization, aromatization and isomerization reactions (Serrano et al., 2012).

Several studies reported the use of microporous and mesoporous catalysts for the conversion of plastic waste into liquid oil and char. Uemichi et al. (1998) carried out catalytic pyrolysis of polyethylene (PE) with HZSM-5 catalysts. The use of HZSM-5 increased liquid oil production with the composition of aromatics and isoalkanes compounds. Gaca et al. (2008) carried out pyrolysis of plastic waste with modified MCM-41 and HZSM-5 and reported that use of HZSM-5 produced lighter hydrocarbons (C3–C4) with maximum aromatic compounds. Lin et al. (2004) used different kinds of catalysts and reported that even mixing of HZSM-5 with mesoporous SiO2-Al2O<sup>3</sup> or MCM-41 led to the maximum production of liquid oil with minimal gas production. Aguado et al. (1997) reported the production of aromatics and aliphatic compounds from the catalytic pyrolysis of PE with HZSM-5, while the use of mesoporous MCM-41 decreased the aromatic compounds produced due to its low acid catalytic activity. The use of synthetic catalysts enhanced the overall pyrolysis process and improved the quality of produced liquid oil. However, the use of synthetic catalysts increased the cost of the pyrolysis process.

The NZ catalysts can be used to overcome the economic challenges of catalytic pyrolysis which comes with the use of expensive catalysts. In recent years, NZ has gained significant attention for its potential environmental applications. Naturally, NZ is found in Japan, USA, Cuba, Indonesia, Hungary, Italy, and the Kingdom of Saudi Arabia (KSA) (Sriningsih et al., 2014; Nizami et al., 2016). The deposit of NZ in KSA mostly lies in Harrat Shama and Jabbal Shama and mainly contain minerals of mordenite with high thermal stability, making it suitable as a catalyst in plastic waste pyrolysis. Sriningsih et al. (2014) modified NZ from Sukabumi, Indonesia by depositing transitional metals such as Ni, Co, and Mo and carried out pyrolysis of low-density polyethylene (LDPE). Gandidi et al. (2018) used NZ from Lampung, Indonesia for the catalytic pyrolysis of municipal solid waste.

This is the first study to investigate the effect of modified Saudi natural zeolite, on product quality and yield from catalytic pyrolysis of plastic waste. Saudi natural zeolite catalyst was modified via novel thermal activation (TA-NZ) at 550◦C and acid activation (AA-NZ) with HNO<sup>3</sup> to enhance its catalytic properties. The catalytic pyrolysis of different types of plastics waste (PS, PE, PP, and PET) as single or mixed in different ratios, in the presence of modified natural zeolite (NZ) catalysts in a small pilot scale pyrolysis reactor, was carried out for the first time. The quality and yield of pyrolysis products such as liquid oil, gas, and char were studied. The chemical composition of the liquid oil was analyzed by GC-MS. Furthermore, the potential and challenges of pyrolysis-based biorefineries have been discussed.

## MATERIALS AND METHODS

## Feedstock Preparation and Reactor Start-Up

The plastic waste used as the feedstock in the catalytic pyrolysis process was collected from Jeddah and included grocery bags, disposable juice cups and plates, and drinking water bottles, which consist of polyethylene (PE), polypropylene (PP) polystyrene (PS), and polyethylene terephthalate (PET) plastics, respectively. The selection of these plastic materials was made based on the fact that they are the primary source of plastic waste produced in KSA. To obtain a homogenous mixture, all the waste samples were crushed into smaller pieces of around 2 cm<sup>2</sup> . The catalytic pyrolysis was carried out using an individual or mixture of these plastic wastes in different ratios (**Table 1**). 1000 g of feedstock was used, with 100 g of catalyst in each experiment. Saudi natural zeolite (NZ), collected from Harrat-Shama located in the northwest of Jeddah city, KSA (Nizami et al., 2016), was modified by thermal and acid treatment and used in these catalytic pyrolysis experiments. NZ was crushed into powder (<100 nm) in a ball-milling machine (Retsch MM 480) for 3 h using 20 Hz/sec, before modification and its usage in pyrolysis. For thermal activation (TA), NZ was heated in a muffle furnace at 550◦C for 5 h, while for acidic activation (AA) NZ was soaked in a 0.1 M solution of nitric acid (HNO3) for 48 h and continuously shaken using an IKA HS 501 digital shaker with 50 rpm. Afterward, the sample was washed with deionized water until a normal pH was obtained.

The experiments were carried out in a small pilot-scale pyrolysis reactor at 450◦C, using a heating rate of 10◦C/min and reaction time of 75 min (**Figure 1**). The obtained yield of each pyrolysis product was calculated based on weight, after the completion of each experiment. The produced liquid oil characterization was carried out to investigate the effect of feedstock composition on the quality of liquid oil produced in the presence of modified NZ. TGA was carried out on feedstock to obtain the optimal process conditions such as temperature and reaction time (75 min) under controlled conditions. In TGA, a 10 µg of each type of plastic waste was taken and heated with a rate of 10◦C from 25 to 900◦C under a continuous flow of nitrogen (50

#### TABLE 1 | Experimental scheme.


ml/min). The authors of this study have recently published work on the effect of feedstock composition and natural and synthetic zeolite catalysts without catalyst modification on different types of plastic waste (Miandad et al., 2017b; Rehan et al., 2017).

## Experimental Setup

The small pilot scale reactor has the capability to be used as both a thermal and catalytic pyrolysis, using different feedstocks such as plastic and biomass materials (**Figure 1**). In this study, modified NZ catalysts were added in the reactor with the feedstock. The pyrolysis reactor can hold up to 20 L of feedstock and the maximum working safe temperature of up to 600◦C can be achieved with the desired heating rates. Detailed parameters of the pyrolysis reactor were published earlier (Miandad et al., 2016b, 2017b). As the temperature increases above certain values, the plastic waste (organic polymers) converts into monomers that are transferred to the condenser, where these vapors are condensed into liquid oil. A continuous condensation system using a water bath and ACDelco Classic coolant was used to ensure the condensation temperature was kept below 10◦C, and to ensure the maximum condensation of vapor to liquid oil. The produced liquid oil was collected from the oil collection tank, and further characterization was carried out to uncover its chemical composition and characteristics for other potential applications.

## Analytical Methods

The pyrolysis oil was characterized using different techniques such as gas chromatography coupled with mass spectrophotometry (GC-MS), Fourier transform infrared spectroscopy (FT-IR),

Bomb Calorimeter and TGA (Mettler Toledo TGA/SDTA851) by adopting the standard ASTM methods. The functional groups in pyrolysis oil was analyzed by a FT-IR, Perkin Elmer's, UK instrument. The FT-IR analysis was conducted using a minimum of 32 scans with an average of 4 cm−<sup>1</sup> IR signals within the frequency range of 500–4,000 cm−<sup>1</sup> .

The chemical composition of oil was studied using a GC-MS (Shimadzu QP-Plus 2010) with FI detector. A capillary GC 30 m long and 0.25 mm wide column coated with a 0.25µm thick film of 5% phenyl-methylpolysiloxane (HP-5) was used. The oven was set at 50◦C for 2 min and then increased up to 290◦C using a 5◦C/min heating rate. The temperature of the ion source and transfer line were kept at 230, and 300◦C and splitless injection was applied at 290◦C. The NIST08s mass spectral data library was used to identify the chromatographic peaks, and the peak percentages were assessed for their total ion chromatogram (TIC) peak area. The high heating values (HHV) of produced liquid oil obtained from different types of plastic waste were measured following the standard ASTM D 240 method with a Bomb Calorimeter (Parr 6200 Calorimeter) instrument, while production of gas was estimated using the standard mass balance formula, considering the difference of weights of liquid oil and char.

## RESULTS AND DISCUSSION

## TGA Analysis of Feedstock

TGA was carried out for each type of plastic waste on an individual basis to determine the optimum temperature for thermal degradation. All types of plastic waste show similar degradation behavior with the rapid loss of weight of hydrocarbons within the narrow range of temperature (150– 250◦C) (**Figure 2**). The maximum degradation for each type of plastic waste was achieved within 420–490◦C. PS and PP showed single step decomposition, while PE and PET showed a two-stage decomposition under controlled conditions. The single step decomposition corresponds to the presence of a carbon-carbon bond that promotes the random scission mechanism with the increase in temperature (Kim et al., 2006). PP degradation started at a very low temperature (240◦C) compared to other feedstocks. Half of the carbon present in the chain of PP consists of tertiary carbon, which promotes the formation of carbocation during its thermal degradation process (Jung et al., 2010). This is probably the reason for achieving maximum PP degradation at a lower temperature. The PS initial degradation started at 330◦C and maximum degradation was achieved at 470◦C. PS has a cyclic structure, and its degradation under the thermal condition involves both random chain and end-chain scission, which enhances its degradation process (Demirbas, 2004; Lee, 2012).

PE and PET showed a two-stage decomposition process; the initial degradation started at lower temperatures followed by the other degradation stage at a higher temperature. PEs initial degradation started at 270◦C and propagated slowly but gradually

until the temperature reached 385◦C. After that temperature, a sharp degradation was observed, and 95% degradation was achieved with a further increase of around 100◦C. A similar two-stage degradation pattern was observed for PET plastic and the initial degradation started at 400◦C with a sharp decrease in weight loss. However, the second degradation started at a slightly higher temperature (550◦C). The initial degradation of PE and PET may be due to the presence of some volatile impurities such as the additive filler used during plastic synthesis (Dimitrov et al., 2013).

Various researchers have reported that PE and PET degradation requires higher temperatures compared to other plastics (Dimitrov et al., 2013; Rizzarelli et al., 2016). Lee (2012) reported that PE has a long chain branched structure and that its degradation occurs via random chain scission, thus requiring a higher temperature, while PET degradation follows the ester link random scission which results in the formation oligomers (Dziecioł and Trzeszczynski, 2000; Lecomte and Liggat, 2006). The initial degradation of PET was perhaps due to the presence of some volatile impurities such as diethylene glycol (Dimitrov et al., 2013). Literature reports that the presence of these volatile impurities further promotes the degradation process of polymers (McNeill and Bounekhel, 1991; Dziecioł and Trzeszczynski, 2000). The difference in TGA curves of various types of plastics could be due to their mesoporous structure (Chandrasekaran et al., 2015). In addition, Lopez et al. (2011) reported that the use of catalysts decreases the process temperature. Therefore, 450◦C could be taken as the optimum

temperature, in the presence of activated NZ, for catalytic pyrolysis of the aforementioned plastic waste.

## Effect of Feedstock and Catalysts on Pyrolysis Products Yield

The effect of thermal and acid activation of NZ on the product yield of the pyrolysis process was examined (**Figure 3**). The catalytic pyrolysis of individual PS plastic using TA-NZ and AA-NZ catalysts showed the highest liquid oil yields of 70 and 60%, respectively, compared to all other types of individual and combined plastic waste studied. The high yield of liquid oil from catalytic pyrolysis of PS was also reported in several other studies (Siddiqui and Redhwi, 2009; Lee, 2012; Rehan et al., 2017). Siddiqui and Redhwi (2009) reported that PS has a cyclic structure, which leads to the high yield of liquid oil from catalytic pyrolysis. Lee (2012) reported that PS degradation occurred via both random-chain and end chain scissions, thus leading to the production of the stable benzene ring structure, which enhances further cracking and may increase liquid oil production. Furthermore, in the presence of acid catalysts, PS degradation followed a carbenium mechanism, which further underwent hydrogenation (inter/intramolecular hydrogen transfer) and βscission (Serrano et al., 2000). In addition, PS degradation occurred at a lower temperature, compared to other plastics such as PE, due to its cyclic structure (Wu et al., 2014). On the other hand, the catalytic pyrolysis of PS produced a higher amount of char (24.6%) with AA-NZ catalyst than with TA-NZ (15.8%) catalyst. Ma et al. (2017) also reported the high production of char from the catalytic pyrolysis of PS with an acidic zeolite (Hβ) catalyst. The high char production numbers were due to the high acidity of the catalyst, which favors char production via intense secondary cross-linking reactions (Serrano et al., 2000).

The catalytic pyrolysis of PP produced higher liquid oil (54%) with the AA-NZ catalyst than the TA-NZ catalyst (40%) (**Figure 3**). On the other hand, the TA-NZ catalyst produced large amounts of gas (41.1%), which may be due to the lower catalytic activity of the TA-NZ catalyst. According to Kim et al. (2002) catalyst with low acidity and BET surface areas with microporous structures, favor the initial degradation of PP which may lead to the maximum production of gases. Obali et al. (2012) carried out pyrolysis of PP with an alumina-loaded catalyst and reported the maximum production of gas. Moreover, formation of carbocation during PP degradation, due to the presence of tertiary carbon in its carbon chain, may also favor gas production (Jung et al., 2010). Syamsiro et al. (2014) also reported that catalytic pyrolysis of PP and PS with an acid (HCL) activated natural zeolite catalyst produced more gases than the process with a thermally activated natural zeolite catalyst, due to its high acidity and BET surface area.

The catalytic pyrolysis of PE with TA-NZ and AA-NZ catalysts produced similar amounts of liquid oil (40 and 42%). However, the highest amounts of gases (50.8 and 47.0%) were produced from PE, using AA-NZ and TA-NZ respectively, compared to all other types of plastic studied. The char production was lowest in this case, 7.2 and 13.0% with AA-NZ and TA-NZ, respectively. Various studies also reported the lower production of char from the catalytic pyrolysis of PE (Xue et al., 2017). Lopez et al. (2011) reported that catalysts with high acidity enhanced the cracking of polymers during the catalytic pyrolysis. The increase in cracking, in the presence of a high acidic catalyst, promotes the production of gases (Miandad et al., 2016b, 2017a). Zeaiter (2014) carried out catalytic pyrolysis of PE with HBeta zeolite and reported 95.7% gas production due to the high acidity of the catalyst. Batool et al. (2016) also reported the maximum production of gas from catalytic pyrolysis of PE, with highly acidic ZSM-5 catalyst. According to Lee (2012) and Williams (2006), PE has a long chain carbon structure, and its degradation occurs randomly into smaller chain molecules via random chain scission, which may promote gas production. During the pyrolysis of PE, which holds the C-H and C-C bonds only, initially, macromolecule backbone breaking occurred and produced stable free-radicals. Further, the hydrogenation steps occurred, leading to the synthesis of secondary free-radicals (new stable C-H bond), which resulted into β-scission and produced an unsaturated group (Rizzarelli et al., 2016).

The catalytic pyrolysis of PP/PE (50/50% ratio) did not show any significant difference in the overall product yields when using both AA-NZ and TA-NZ. The liquid oil produced from the catalytic pyrolysis of PP/PE was 44 and 40% from TA-NZ and AA-NZ catalysts, respectively. A slight decrease in the liquid oil yield from AA-NZ could be due to its high acidity. Syamsiro et al. (2014) reported that AA-NZ with HCl has high acidity compared to TA-NZ, produced less liquid oil yield and had high production of gases. Overall catalytic pyrolysis of PP/PE produced the maximum amount of gas with low amounts of char. The high production of gas may be due to the presence of PP. The degradation of PP enhances the carbocation process due to the presence of tertiary carbon in its carbon chain (Jung et al., 2010). Furthermore, the degradation of PE in the presence of catalyst also favors the production of gas with a low yield of liquid oil. However, when PP and PE catalytic pyrolysis was carried out separately with PS, a significant difference was observed in the product yield.

There was a significant difference in the liquid oil yield of 54 and 34% for catalytic pyrolysis of PS/PP (50/50% ratio) with TA-NZ and AA-NZ catalysts, respectively. Similarly, a significant difference in the char yield of 20.3 and 35.2% was observed, whereas the high yield of gases were 25.7 and 30.8% using TA-NZ and AA-NZ catalysts, respectively. Lopez et al. (2011) and Seo et al. (2003) reported that a catalyst with high acidity promotes the cracking process and produces maximum gas production. Furthermore, the presence of PP also enhances gas production due to the carbocation process during degradation (Jung et al., 2010). Kim et al. (2002) reported that PP degradation produces maximum gas in the presence of acid catalysts.

The catalytic pyrolysis of PS with PE (50/50% ratio) in the presence of TA-NZ catalyst produced 44% liquid oil, however 52% liquid oil was obtained using the AA-NZ catalyst. Kiran et al. (2000) carried out pyrolysis of PS with PE at different ratios and reported that an increase in the concentration of PE decreased the liquid oil concentration with the increase in gas. The presence of PS with PE promotes the degradation process due to the production of an active stable benzene ring from PS (Miandad et al., 2016b). Wu et al. (2014) carried out TGA of PS with PE and observed two peaks, the first one for PS at a low temperature, followed by PE degradation at a high temperature. Moreover, PE degradation follows a free radical chain process and hydrogenation process, while PS follows a radical chain process including various steps (Kiran et al., 2000). Thus, even when considering the degradation phenomena, PS resulted in higher degradation compared to PE and produced stable benzene rings (McNeill et al., 1990).

Catalytic pyrolysis of PS/PE/PP (50/25/25% ratio) showed slightly lower liquid oil yields as compared to catalytic pyrolysis of all individual plastic types. The oil yield from both catalysts, TA-NZ and AA-NZ, in this case, is similar, 44 and 40%, respectively. The char production was higher (29.7%) with the AA-NZ catalyst than (19.0%) with the TA-NZ catalyst, which may be due to polymerization reactions (Wu and Williams, 2010). Furthermore, the addition of PET with PS, PE and PP (20/40/20/20% ratio) reduced the liquid oil yields down to 28 and 30% overall, using TA-NZ and AA-NZ catalysts, respectively, with higher fractions of char and gas. Demirbas (2004) carried out pyrolysis of PS/PE/PP and reported similar results for the product yield. Adnan et al. (2014) carried out catalytic pyrolysis of PS and PET using he Al-Al2O<sup>3</sup> catalyst with ratios of 80/20% and reported only 37% liquid oil. Moreover, Yoshioka et al. (2004) reported the maximum production of gas and char with negligible liquid oil production from catalytic pyrolysis of PET. In addition, maximum char production was also reported when PET catalytic pyrolysis was carried out with other plastics (Bhaskar et al., 2004). The higher production of char from PET pyrolysis was due to the carbonization and condensation reactions during its pyrolysis at a high temperature (Yoshioka et al., 2004). In addition, the presence of the oxygen atom also favors the high production of char from catalytic pyrolysis of PET (Xue et al., 2017). Thilakaratne et al. (2016) reported that production of benzene-free radicals, with two activated carbons, is the precursor of catalytic coke from PET degradation.

## Effect of Catalysts on the Composition of Liquid Oil

The chemical composition of liquid oil produced by the catalytic pyrolysis of different plastic waste using TA-NZ and AA-NZ catalysts were characterized by GC-MS (**Figures 4**, **5**). The produced liquid oil composition is affected by different types of feedstock and catalysts used in the pyrolysis process (Miandad et al., 2016a,b,c). The liquid oil produced from the individual plastic types such as PS, PP and PE contained a mixture of aromatics, aliphatic and other hydrocarbon compounds. The aromatic compounds found in oil, from PS and PE, were higher than PP using the TA-NZ catalyst. The aromatic compounds increased in oil from PS and PP but reduced in PE when using the AA-NZ catalyst. The mesoporous and acidic catalyst leads to the production of shorter chain hydrocarbon due to its high cracking ability (Lopez et al., 2011). However, microporous and less acidic catalysts favor the production of long chain hydrocarbons as the cracking process occurred only on the outer surface of the catalysts. Overall, in the presence of catalysts, PE and PP follow the Random-chain scission mechanism, while PS follows the unzipping or end chain scission mechanism (Cullis and Hirschler, 1981; Peterson et al., 2001). The end-chain scission results in monomer production while random chain scission produces oligomers and monomers (Peterson et al., 2001).

The liquid oil produced from the catalytic pyrolysis of PE, when using both catalysts, produced mainly Naphthalene, Phenanthrene, Naphthalene, 2-ethenyl-, 1-Pentadecene, Anthracene, 2-methyl-, Hexadecane and so on (**Figures 4A**, **5A**). These results agree with several other studies (Lee, 2012; Xue et al., 2017). The production of a benzene derivate reveals that TA-NZ enhances the process of aromatization compared to AA-NZ. Xue et al. (2017) reported that intermediate olefins produced from catalytic pyrolysis of PE, further aromatized inside the pores of catalysts. Nevertheless, the aromatization reaction further leads to the production of hydrogen atoms that may enhance the aromatization process. Lee (2012) reported that ZSM-5 produced more aromatic compounds compare to the mordenite catalyst, due to its crystalline structure.

There are two possible mechanisms which may involve the degradation of PE in the presence of a catalyst; hybrid ion abstraction due to the presence of Lewis sites or, due to the carbenium ion mechanism via the addition of a proton (Rizzarelli et al., 2016). Initially, degradation starts on the external surface of the catalysts and later proceeds with further degradation in the inner pores of the catalysts (Lee, 2012). However, microporous catalysts hinder the entrance of larger molecules and thus higher carbon chain compounds are produced from catalytic pyrolysis of PE with microporous catalysts. In addition, in the presence of acidic catalysts, due to carbenium mechanism, the formation of aromatic and olefin compound production may increase (Lee, 2012). Lin et al. (2004) reported highly reactive olefin production, as intermediate products during the catalytic pyrolysis of PE, that may favor the production of paraffin and aromatic compounds in produced liquid oil. Moreover, the presence of an acidic catalyst and free hydrogen atom may lead to alkylation of toluene and benzene, converting intermediate alkylated benzene to the production of naphthalene due to aromatization (Xue et al., 2017).

The liquid oil produced from catalytic pyrolysis of PS with TA-NZ and AA-NZ, contains different kinds of compounds. Alpha-Methylstyrene, Benzene, 1,1′ -(2-butene-1,4-diyl)bis-, Bibenzyl, Benzene, (1,3-propanediyl), Phenanthrene, 2- Phenylnaphthalene and so on were the major compounds found in the produced liquid oil (**Figures 4A**, **5A**). The liquid oil produced from catalytic pyrolysis of PS, with both activated catalysts, mainly contains aromatic hydrocarbons with some paraffins, naphthalene and olefin compounds (Rehan et al., 2017). However, in the presence of a catalyst, the maximum production of aromatic compounds was achieved (Xue et al., 2017). Ramli et al. (2011) also reported the production of olefins, naphthalene with aromatic compounds from catalytic pyrolysis of PS with Al2O3, supported with Cd and Sn catalysts. PS degradation starts with cracking on the outer surface of the catalyst and is then followed by reforming inside the pores of the catalyst (Uemichi et al., 1999). Initially, the cracking of polymer is carried out by the Lewis acid site on the surface of catalysts to produce carbocationic intermediates, which further evaporates or undergoes reforming inside the pores of the catalyst (Xue et al., 2017).

The catalytic pyrolysis of PS mainly produces styrene and its derivate as the major compounds in the produced liquid oil (Siddiqui and Redhwi, 2009; Rehan et al., 2017). Conversion of styrene into its derivate was increased in the presence of protonated catalysts due to hydrogenation (Kim et al., 2002). Shah and Jan (2015) and Ukei et al. (2000) reported that hydrogenation of styrene increased with the increase of the reaction temperature. Ogawa et al. (1982) carried out pyrolysis of PS with the alumina-silica catalyst at 300◦C and found the hydrogenation of styrene to its derivate. Ramli et al. (2011) reported the possible degradation mechanism of PS on acid catalysts that may occur due to the attack of a proton associated with Bronsted acidic sites, resulting in the carbenium ion mechanism, which further undergoes β-scission and is later followed by hydrogen transfer. Moreover, cross-linking reaction was favored by strong Bronsted acidic sites and when this reaction occurred the completing cracking may decrease to some extent and enhance the production of char (Serrano et al.,

2000). Furthermore, silica-alumina catalysts do not have strong Bronsted acidic sites, though it may not improve the crosslinking reaction but favor the hydrogenation process. Thus, it may be the reason that styrene was not found in the liquid oil, however, its derivate was detected at high quantities (Lee et al., 2001). Xue et al. (2017) also reported the dealkylation of styrene, due to the delay in evaporation inside the reactor, which may lead to an enhanced reforming process and result in the production of

a styrene derivate. TA-NZ and AA-NZ contain a high amount of alumina and silica that leads to the hydrogenation of styrene to its derivate, resulting in the production of styrene monomers instead of styrene.

The catalytic pyrolysis of PP produced a complex mixture of liquid oil containing aromatics, olefins and naphthalene compounds. Benzene, 1,1′ -(2-butene-1,4-diyl)bis-, benzene, 1,1′ - (1,3-propanediyl)bis-, anthracene, 9-methyl-, naphthalene, 2-phenyl-, 1,2,3,4-tetrahydro-1-phenyl-, naphthalene, phenanthrene etc. were the major compounds found in the liquid oil (**Figures 4A**, **5A**). These findings are in line with other studies that carried out catalytic pyrolysis of PP with various catalysts (Marcilla et al., 2004). Furthermore, degradation of PP with AA-NZ resulted in the maximum production of phenol compounds. The higher production was perhaps due to the presence of high acidic sites, as it favors phenol compound production. Moreover, the presence of a high acidic site on catalysts enhanced the oligomerization, aromatization and deoxygenation mechanism that led to the production of poly-aromatic and naphthalene compounds. Dawood and Miura (2002) also reported the high production of these compounds from the catalytic pyrolysis of PP with a high acidic modified HY-zeolite.

The composition of oil from the catalytic pyrolysis of PP with PE contains compounds found in the oil from both individual plastic type feedstocks. Miandad et al. (2016b) reported that feedstock composition also affects the quality and chemical composition of the oil. The produced liquid oil from catalytic pyrolysis of PE/PP contains aromatic, olefin, and naphthalene compounds. The major compounds found were; benzene, 1,1′ -(1,3-propanediyl)bis-, mono(2 ethylhexyl) ester, 1,2-benzenedicarboxylic acid, anthracene, pentadecane, phenanthrene, 2-phenylnaphthalene and so on (**Figures 4B**, **5B**). Jung et al. (2010) reported that the aromatic production from PP/PE catalytic pyrolysis might follow the Diels–Alder reaction mechanism and is then followed by dehydrogenation. Furthermore, catalytic pyrolysis of PP and PE carried out individually with PS, mainly produced aromatic compounds due to the presence of PS. The produced liquid oil from PS/PP contains benzene, 1,1′ -(1,3-propanediyl)bis, 1,2-benzenedicarboxylic acid, disooctyl ester, bibenzyl, phenanthrene, 2-phenylnaphthalene, benzene, (4-methyl-1 decenyl)- and so on (**Figures 4A**, **5A**). PS catalytic pyrolysis with PE mainly produced liquid oil with major compounds of azulene, naphthalene, 1-methyl-, naphthalene, 2-ethenyl, benzene, 1,1′ - (1,3-propanediyl)bis-, phenanthrene, 2-phenylnaphthalene, benzene, 1,1′ -(1-methyl-1,2-ethanediyl)bis- and some other compounds as well (**Figures 4B**, **5B**). Miskolczi et al. (2006) carried out pyrolysis of PS with PE with a ratio of 10 and 90%, respectively, and reported the maximum production of aromatics even at a very low ratio of PS. Miandad et al. (2016b) reported that thermal pyrolysis of PE with PS without a catalyst, resulted in the conversion of PE into liquid oil with a high composition of aromatics. However thermal pyrolysis of the only PE without a catalyst converted it into wax instead of liquid oil due to its strong long chain branched structure (Lee, 2012; Miandad et al., 2016b). Wu et al. (2014) carried out TGA of PS with PE and reported that the presence of PS favors the degradation of PE, due to the production of stable benzene rings.

The chemical composition of pyrolysis oil, by different functional groups, was studied using FT-IR. The obtained data revealed the presence of aromatics and aliphatic functional groups in the oil (**Figures 6**, **7**). A very strong peak at 696 cm−<sup>1</sup> was observed in most of the liquid oils obtained using both catalysts, which corresponds to the high concentration of aromatic compounds. Two more peaks, that are obvious, were visible at around 1,456 and 1,495 cm−<sup>1</sup> for C-C with single and double bonds, corresponding to aromatic compounds. Furthermore, at the end of the spectrum, strong peaks at 2,850, 2,923, and 2,958 cm−<sup>1</sup> were observed in all types of liquid oils except the PS, corresponding to the C-H stretch of alkanes compounds. Overall, the liquid oil obtained from catalytic pyrolysis of different plastic waste using the AA-NZ catalyst, showed more peaks than the samples from the TA-NZ catalysts. These extra peaks corresponded to aromatics, alkanes and alkene compounds. This indicates that, as expected, the AA-NZ had better catalytic properties than the TA-NZ. Various researchers have reported similar results, that liquid oil produced from PS was dominant with aromatics. Tekin et al. (2012) and Panda and Singh (2013) also reported the presence of aromatics with some alkanes and alkenes from catalytic pyrolysis of PP. Kunwar et al. (2016) carried out the thermal and catalytic pyrolysis of PE and reported that produced liquid oil contained alkanes and alkenes as a major functional group. Overall, the FT-IR analysis provided more insight into the chemical composition of liquid oil produced, from catalytic pyrolysis of different plastic waste, using modified NZ catalysts and further confirmed our GC-MS results.

### Potential Applications of Pyrolysis Products

The liquid oil produced from the catalytic pyrolysis of different types of plastic feedstock has a high number of aromatic, olefin, and naphthalene compounds that are found in petroleum products. Moreover, the HHV of the produced liquid oil has been found in the range of 41.7–44.2 MJ/kg (**Table 2**) which is very close to the energy value of conventional diesel. The lowest HHV of 41.7 MJ/ kg was found in liquid oil obtained from PS using the TA-NZ catalyst, whereas the highest HHV of 44.2 MJ/kg was from PS/PE/PP using the AA-NZ catalyst. Thus, the pyrolysis liquid oil produced from various plastic wastes has the potential to be used as an alternative source of energy. According to Lee et al. (2015) and Rehan et al. (2016), the production of electricity is achievable using pyrolysis liquid oil in a diesel engine. Saptoadi and Pratama (2015) successfully used pyrolytic liquid oil as an alternative in a kerosene stove. Moreover, the produced aromatic compounds can be used as raw material for polymerization in various chemical industries (Sarker and Rashid, 2013; Shah and Jan, 2015). Furthermore, various researchers utilized the produced liquid oil as transportation fuel after blending with conventional diesel at different ratios. The studies were carried out to explore the potential of produced liquid oil in the context of engine performance and vehicle exhaust emission. Nileshkumar et al. (2015) and Lee et al. (2015) reported that 20:80% blend ratio of pyrolytic liquid oil and conventional diesel, respectively, gave similar engine performance results than conventional diesel. Moreover, at the same blended ratio the exhaust emissions were also similar, however the exhaust emissions increased with the increase in the blended amount of pyrolysis oil (Frigo et al., 2014; Mukherjee and Thamotharan, 2014).

The residue (char) left after the pyrolysis process can be utilized for several environmental applications. Several researchers activated the char via steam and thermal activation (Lopez et al., 2009; Heras et al., 2014). The activation process increased the BET surface area and reduced the pore size of the char (Lopez et al., 2009). Furthermore, Bernando (2011)

upgraded the plastic char with biomaterial and carried out the adsorption (3.6–22.2 mg/g) of methylene blue dye from wastewater. Miandad et al. (2018) used the char obtained from pyrolysis of PS plastic waste to synthesize a novel carbon-metal double-layered oxides (C/MnCuAl-LDOs) nano-adsorbent for the adsorption of Congo red (CR) in wastewater. Furthermore, the char can be used as a raw material for the production of activated carbon as well.



## Limitations of GC-MS Analysis of Pyrolysis Oil

There are some limitations in conducting the accurate quantitative analysis of chemical components in pyrolysis oil using GC-MS. In this study, we used the mass percentage of different chemicals found in oil samples, calculated based on the peak areas identified by a normal phase DP5-MS column and FID. The identified peaks were matched with the NIST and mass bank spectra library. The compounds were chosen based on the similarity index (SI > 90%). Further comparison with known (CRM) standards enabled confirmation of the identified compounds. The used column and detectors were limited only with hydrocarbons. In reality however, oil from most plastic waste has a complex chemical structure and may contain other groups of unidentified chemicals such as sulfur, nitrogen, and oxygen-containing hydrocarbons. This is why a more in-depth and accurate qualitative chemical analysis is needed to fully understand the chemistry of pyrolysis oil, using advanced calibration and standardization and using different MS detectors like SCD and NCD as well as different GC columns.

## The Potential and Challenges of Pyrolysis Based Biorefineries

Waste biorefineries are attracting tremendous attention as a solution to convert MSW and other biomass waste into a range of products such as fuels, power, heat and other valuable chemicals and materials. Different types of biorefineries, such as an agriculture-based biorefinery, animal waste biorefinery, wastewater biorefinery, algae-based biorefinery, plastic waste refinery, forestry-based biorefinery, industrial waste biorefinery, and Food waste biorefinery etc., can be developed depending on the type and source of waste (Gebreslassie et al., 2013; De Wild et al., 2014; Nizami et al., 2017a,b; Waqas et al., 2018). These biorefineries can play a significant role to reduce waste-related environmental pollution and GHG emissions. Furthermore, they generate substantial economic benefits and can help achieve a circular economy in any country.

A pyrolysis based biorefinery can be developed to treat a range of biomass waste and plastic waste to produce liquid and gas fuels, energy, biochar, and other higher value chemicals using an integrated approach. The integrated approach helps to achieve maximum economic and environmental benefits with minimal waste production. There are many challenges and room for improvement in pyrolysis-based biorefineries, that need to be addressed and optimized to ensure maximum benefits. Although pyrolysis oil holds more energy than coal and some other fuels, pyrolysis itself is an energy-intensive process, and the oil product requires more energy to be refined (Inman, 2012). This means that pyrolysis oil may not be much better than conventional diesel or other fossil-based fuels in terms of GHG emissions, though much detailed research studies on mass and energy balance across the whole process's boundaries are needed to confirm this. To overcome these process energy requirements, more advanced technologies can be developed using the integration of renewable energies such as solar or hydro with pyrolysis-based biorefineries, to achieve maximum economic and environmental benefits.

The availability of plastic and biomass waste streams as feedstocks for pyrolysis based biorefineries, is another major challenge, since recycling is not currently very efficient, especially in the developing countries. The gases produced from pyrolysis of some plastic waste such as PVC are toxic, and therefore pyrolysis emission treatment technology has to be further refined to achieve maximum environmental benefits. The pyrolysis oil obtained from various plastic types need to be cleaned significantly before it is used in any application, to ensure minimal environmental impact. The high aromatic contents of the pyrolysis oil is good and some aromatic compounds such as benzene, toluene, and styrene can be refined and sold in an already established market. However, some of the aromatic hydrocarbons are known carcinogens and can cause serious human health and environmental damage. Serious consideration is therefore needed in this regard.

Other aspects for optimization of pyrolysis based biorefineries, such as new emerging advanced catalysts including nano-catalysts, have to be developed and applied in pyrolysis processes in order to increase the quality and yield of products, and to optimize the overall process. The market for pyrolysis based biorefinery products should be created/ expanded to attract further interest and funding, in order to make this concept more practical and successful. Similarly, more focus is needed to conduct further research and development work on enriching the biorefinery concept and to tap into its true potential. Furthermore, it is vital to conduct a detailed economic and environmental impact assessment of biorefineries during a design stage, using specialized tools such as the life-cycle assessment (LCA). The LCA can analyze the environmental impact of the biorefinery and all products by conducting detailed energy and material balances of all life stages including raw material extraction and processing, manufacturing, product distribution, use, maintenance, and disposal/recycling. The outcomes of LCA will help to determine the sustainability of biorefineries, which is crucial in making the right decision.

## CONCLUSIONS

Catalytic pyrolysis is a promising technique to convert plastic waste into liquid oil and other value-added products, using a modified natural zeolite (NZ) catalyst. The modification of NZ catalysts was carried out by novel thermal (TA) and acidic (AA) activation that enhanced their catalytic properties. The catalytic pyrolysis of PS produced the highest liquid oil (70 and 60%) compared to PP (40 and 54%) and PE (40 and 42%), using the TA-NZ and AA-NZ catalysts, respectively. The chemical composition of the pyrolysis oil was analyzed using GC-MS, and it was found that most of the liquid oil produced a high aromatic content with some aliphatic and other hydrocarbon compounds. These results were further confirmed by the FT-IR analysis showing clear peaks corresponding to aromatic and other hydrocarbon functional groups. Furthermore, liquid oil produced from different types of plastic waste had higher heating values (HHV) in the range of 41.7–44.2 MJ/kg similar to that of conventional diesel. Therefore, it has the potential to be used in various energy and transportation applications after further treatment and refining. This study is a step toward developing pyrolysis-based biorefineries. Biorefineries have a great potential to convert waste into energy and other valuable products and could help to achieve circular economies. However, there are many technical, operational, and socio-economic challenges, as

## REFERENCES


discussed above, that need to be overcome in order to achieve the maximum economic and environmental benefits of biorefineries.

## DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the supplementary files.

### AUTHOR CONTRIBUTIONS

RM performed the pyrolysis experiments and helped in manuscript write up. HK, JD, JG, and AH have carried out the detailed characterization of the process products. MR and ASA have analyzed the data and written parts of the manuscript. MAB, MR, and A-SN have corrected and edited the manuscript. ASA and IMII supported the project financially and technically.

### ACKNOWLEDGMENTS

MR and A-SN acknowledge the Center of Excellence in Environmental Studies (CEES), King Abdulaziz University (KAU), Jeddah, KSA and the Ministry of Education, KSA for financial support under Grant No. 2/S/1438. Authors are also thankful to Deanship of Scientific Research (DSR) at KAU for their financial and technical support to CEES.


aromatics using a fluidized bed reactor. Fuel Process. Technol. 91, 277–284. doi: 10.1016/j.fuproc.2009.10.009


recyclable polystyrene with dispersed catalysts. Catal. Today 62, 67–75. doi: 10.1016/S0920-5861(00)00409-0


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Miandad, Rehan, Barakat, Aburiazaiza, Khan, Ismail, Dhavamani, Gardy, Hassanpour and Nizami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Biorefining Waste Sludge From Water and Sewage Treatment Plants Into Eco-Construction Material

Wai Nam Cheng1†, Haakrho Yi 2†, Chun-fai Yu<sup>3</sup> , Ho Fai Wong<sup>4</sup> , Guoxiang Wang<sup>5</sup> , Eilhann E. Kwon<sup>6</sup> and Yiu Fai Tsang1,5 \*

*<sup>1</sup> Department of Science and Environmental Studies, The Education University of Hong Kong, Tai Po, Hong Kong, <sup>2</sup> Department of Gwangyang Research Group, RIST, Gwangyang City, South Korea, <sup>3</sup> Environmental Science Program, Division of Science and Technology, BNU-HKBU United International College, Zhuhai, China, <sup>4</sup> Faculty of Science and Technology, Technological and Higher Education Institute of Hong Kong, Tsing Yi Island, Hong Kong, <sup>5</sup> School of Environment, Nanjing Normal University, Nanjing, China, <sup>6</sup> Department of Environment and Energy, Sejong University, Seoul, South Korea*

#### Edited by:

*Mohammad Rehan, King Abdulaziz University, Saudi Arabia*

#### Reviewed by:

*Sidney Man Ngai Chan, Open University of Hong Kong, China Na Duan, China Agricultural University, China*

> \*Correspondence: *Yiu Fai Tsang tsangyf@eduhk.hk*

*†These authors have contributed equally to this work and are co-first authors*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *31 October 2018* Accepted: *14 February 2019* Published: *19 March 2019*

#### Citation:

*Cheng WN, Yi H, Yu C-f, Wong HF, Wang G, Kwon EE and Tsang YF (2019) Biorefining Waste Sludge From Water and Sewage Treatment Plants Into Eco-Construction Material. Front. Energy Res. 7:22. doi: 10.3389/fenrg.2019.00022* This study aims to investigate the feasibility of using different waste sludge and coal combustion residuals in eco-concrete block production. The compressive strength of the eco-concrete blocks produced by waterworks sludge, bottom and fly ashes were 36 MPa, which comply with the standard specifications for paving blocks in Hong Kong. The optimal mixing proportion (by weight) of different materials in the blocks, such as aggregates, cementitious materials, water, and fly ash was 1.1:1.0:0.5:0.22, respectively. The environmental and toxicological impacts of the final products were then evaluated according to the toxicity characteristic leaching procedure (TCLP). While several heavy metals (i.e., Hg, Cu, and Pb) have been identified in the specimens, the levels of these contaminants complied with Standards (US 40 CFR 268.48). Waste materials generated from water and sewage treatment processes and power plants are feasible to be used as ingredients for paving concrete block production. These products are environmentally acceptable and mechanically suitable for resource recovery of waste materials.

Keywords: resource recovery, construction materials, sewage sludge, waterworks sludge, solidification

### INTRODUCTION

In 2015, ∼15,100 tons per day (TPD) of solid waste were disposed to landfill sites in Hong Kong. However, existing landfills will be filled up by 2020 (Environmental Protection Department (EPD), 2016a; 2016b). In terms of commingling of solid waste, waste sludge is the unavoidable by-product from water and sewage treatment processes. It is classified as special waste, which cannot be easily reduced at sources (Jordan et al., 2005; Tuan et al., 2013). The amount of sludge produced could significantly increase due to future urban development, such as the North East New Territories New Development Area. The government estimates that the sludge production will increase to ∼2,000 TPD by 2030 due to the completion of the Harbor Area Treatment Scheme Phase 2B (Environmental Protection Department, 2016b). The sludge treatment facility was built to alleviate the mounting pressure faced by the landfill. The facility can process 90% of the sludge volume through high-temperature incineration. However, the remaining 10% of the waste residues are still disposed to the landfill (Environmental Protection Department, 2016b). Thus, reusing and recycling of the waste sludge are win-win strategies for mitigating the load of waste treatment facilities and reducing natural resources and energy consumption.

Several studies have demonstrated the feasibility of using sewage sludge for the production of eco-construction materials, such as cement (Monzó et al., 2004; Cyr et al., 2007), lightweight aggregate (Liu et al., 2017), and ceramic tile (Jordan et al., 2005). Utilizing sewage sludge as ingredients in construction materials not only could reduce the production cost but also provide ecological and energy saving advantages (Wang et al., 2008; Liu et al., 2011; Tyagi and Lo, 2013). Despite an estimated 10,000 TPD of waterworks sludge generated globally, few studies have focused on its resource recovery (Babatunde and Zhao, 2007). In Hong Kong, ∼60 TPD waterworks sludge is discarded to the landfill (Environmental Protection Department, 2016a). There is a need to find ways on how to effectively reuse waterworks sludge due to the shortage of landfill space (Tang et al., 2014; Tsang, 2015). Huang and Wang (2013) illustrated the possibility of reusing waterworks sludge in the production of lightweight aggregate. Power plant contractors also face the problem of disposing of coal combustion residuals. Considering different pozzolanic properties along with the issues of cost and environment, fly ash is mainly used in cement manufacturing, while bottom ash is utilized mainly in construction materials (Meij and Berg, 2001; Tsang, 2015). Several studies reported the feasibility of using bottom ash for different engineering purposes, such as blocks (Arenas et al., 2011) and concrete (Lee et al., 2010; Kim and Lee, 2011). Bottom ash has better performance compared with sand, in fact, the compressive strength of 100% aggregate replacement was even higher than that of reference blocks (Andrade et al., 2007). The coliform population of the sludge can also be reduced by the addition of coal fly ash (Papadimitriou et al., 2008). Supplementing with coal combustion residuals may

**Abbreviations:** A/C, aggregate to cement; BA, bottom ash; CEDD, Civil Engineering and Development Department; EPD, Environmental Protection Department; FA, fly ash; ICP-AES, inductively coupled plasma-atomic emission spectrometry; ICP-MS, inductively coupled plasma-mass spectrometry; LoD, limit of detection; PC, Portland cement; S/S, solidification/stabilization; SA, sludge incineration residuals; SS, sewage sludge; SSWTWs, Sheung Shui Water Treatment Works; STSTWs, Shatin Sewage Treatment Works; TCLP, toxicity characteristic leaching procedure; TPD, tons per day; W/C, water to cement; WS, waterworks sludge.

facilitate the utilization of waste sludge as ingredients in concrete blocks. Meanwhile, Wang (2005) demonstrated the use of waste materials, such as red mud and building sand, to produce nonsintered and non-clay blocks. The resulting blocks are considered as an environmentally friendly construction material because of low cement content.

Aside from paving blocks, the Chinese Government has implemented related economic policies to encourage and regulate the production and use of new wall materials (Mohurd, 2012). These policies aim to promote the innovation of wall materials and energy-saving buildings. The Ministry of Finance and the State Administration of Taxation has made clear that enterprises using effluent, exhausted gases, waste residues, and other waste materials as primary feedstock for producing wall materials can avail of income tax reduction or exemption within 5 years. Eco-construction materials development is the trend of construction materials production. In fact, there were not many researches focus on eco-concrete block produced from mentioned waste materials. And previous researches demonstrated the feasibility of using waste materials for ecoconstruction materials production. These successful experiences could be applied to eco-concrete block production. Moreover, the government also supports green product production. Ecoconstruction materials could be developed in the future. In this study, the feasibility to reuse dewatered sludge, dried sludge, and coal combustion residues as sand and aggregate replacements for paving concrete blocks production to reduce the sludge quantities for disposal or further treatment was investigated. The ratios of different waste materials were optimized to meet the standard specifications for paving blocks. Toxicity and mechanical performance were also examined to evaluate the environmental acceptance and feasibility of the engineering view of the products.

### MATERIALS AND METHODS

#### Raw Materials

Waterworks sludge (WS), sewage sludge (SS), and sludge incineration residuals (SA) were selected to investigate the possibility of producing concrete blocks due to the properties more similar to traditional constituents. WS and SS were collected from Shaiin Sewage Treatment Works (STSTWs) and Sheung Shui Water Treatment Works (SSWTWs) in Hong Kong, respectively. SA were collected from sludge treatment facility adopted incineration at 850◦C in Tune Mun, Hong Kong. WS was the aluminum-rich sludge generated in local drinking water treatment process, in which alum is used as the coagulant in coagulation-flocculation process. SS was the sludge generated from a typical secondary wastewater treatment plant. Coal combustion residuals [i.e., bottom ash (BA) and fly ash (FA)] were collected from Castle Peak Power Plant. Ordinary Portland cement (PC) equivalent to BS EN197-1:2000 sourced commercially in Hong Kong were used as cementitious materials and partially substituted by fly ash. The moisture content of waste materials was similar to typical value. Beside of waste sludge and coal combustion residuals, red mud, coal gangue, waste glass were used to produce non-sintered blocks. Red mud that is insoluble waste residual was produced during extraction and purification of aluminum oxide and consisted of fine particles and coarse granules. Coal gangue is a solid waste that generated during the process of coal mining and washing.

## Experimental Design

This study was separated into two parts (**Graphical Abstract**). In Part one, specimens with various mixing proportion (**Table 1**) were produced to evaluate the feasibility of utilizing waste materials as substitutes in concrete blocks. The leachability of heavy metals (i.e., Hg, Cu, and Pb) in the eco-concrete blocks with different mixing proportions was also determined. In Part two, the mechanical performance of the specimens was evaluated according to the standard specifications for paving blocks published by Hong Kong SAR.

The water to cement (W/C) and aggregate to cement (A/C) ratios obtained during determining the effective ratio. W/C and A/C were 0.3 and 1.1, respectively. The ratios of waste material substitution to natural material varied from 20 to 100%, while the ratios of bottom ash to sewage sludge (BA/SS) varied between 1.0 and 2.4. In Series 4, moisture was provided from the SS; therefore no tap water was added to the concrete mix. Effective mixing proportion obtained from Part one would be used for the production of specimens in Part two. Different type of waste materials would substitute the natural resources in the specimens to evaluate various waste materials, which would present acceptable mechanical performance.

Coarse aggregate and fine aggregate were replaced by WS and BA, respectively. Coal combustion FA and SA would work as cementitious materials to substitute part of cement. The coarse aggregate to fine aggregate ratio was amended to 50% to 50%. The detailed information of mixing proportion of Part two is shown in **Tables 2**, **3**. Series 5 would determine the effect of varying materials on compressive strength. In order to maximize the amount of waste materials reusing, effect of increasing A/C ratio would evaluate in series 6. Given that no standards or regulations exist to produce eco-concrete blocks in Hong Kong, and the materials used in this study were different from traditional materials in terms of physical or chemical characteristics, the W/C and A/C ratios were different from the typical ratios used. Besides, series 7 would evaluate the mechanical performance of specimens produced by waste materials. In this mix series, cement was also replaced by coal fly ash.

## Materials Preparation

All waterworks sludge used in this study was oven-dried at 105◦C for at least 2 days before mixing to obtain harden character to simulate the character of natural coarse aggregate. Sewage sludge cakes mixed with other materials without pretreatment to determine the effect of utilizing sludge cake to replace natural aggregate. Bottom ashes are porosities and would absorb water to reduce the free-water content for cement hydration. To maintain W/C, saturated surface-dry condition for aggregate was needed to obtain (Marsh, 1997). In this study, 18% by weight of water was added to bottom ash to reach the saturated surface-dry condition. For materials used in series 7, two-grain sizes of coal gangue and waste glass were selected, which were between 0.35 and

#### TABLE 1 | Mixing proportion used in the Part one study<sup>a</sup> .


*<sup>a</sup>Specimens with various mixing proportion for evaluating the feasibility of utilizing waste materials as substitutes in concrete blocks.*

#### TABLE 2 | Mixing proportion used in the Part two study<sup>a</sup> .


*<sup>a</sup>Acceptable mechanical performance for the substitutability of different types of waste material in the specimens.*

*<sup>b</sup>Coarse waterworks sludge.*

*<sup>c</sup>Fine waterworks sludge.*

0.5 mm (medium size) and over 0.5 mm (coarse size). The ratio between medium-size particles and coarse-size particles was 6:4. Coal gangue and waste glass were mixed to prepare the aggregate mixture for series 7.

#### Effective Ratios

Waste materials properties either physically or chemically were different to natural materials utilized in the traditional concrete block. The curing performance may differ to traditional materials. More cement may be needed for concrete mix production, therefore effective ratios were needed to determine. The optimal A/C ratio for waste materials concrete mix would determine before concrete block production. Typical A/C is in the range of 3:1 to 6:1 (Shackel, 1990) and the preliminary study of A/C ratio and W/C ratio for waste materials start with 4:1 and 0.6, respectively. For W/C ratio, it is not the critical factor at this part; the typical value was applied here. Preliminary study for effective ratio would proceed until specimens' cure successfully on Day 1.

#### Specimens Preparation

Hundred wooden molds fabricated according to Civil Engineering and Development Department (CEDD) standard for the concrete paving block (i.e., L × W × H: 200 × 100 × 60 mm) were used in parts one and two studies (CEDD, 2006). The concrete mixtures were prepared by the hand-mixing method and vibrate manually to eliminate the air trapped in the mixture. The specimens were removed from the molds and cure in water for 28 days on day 1. In series 7, the sizes of specimens were in the shape of a cylinder with 40 mm (diameter) and 13.2 mm (height).

#### Analytical Methods

Leachability of heavy metals was determined by Toxicity Characteristic Leaching Procedures (TCLP, USEPA Method 1311, 1992), which is common practice to evaluate leachability of waste and contaminated soils. According to USEPA and Hong Kong TCLP limits, the selected analytes included Cd, Cr, Cu, Pb, Hg, Ni, Zn, and As (Environmental Protection Department, 2011).

TABLE 3 | Mixing proportion used in non-sintered and non-clay block.


*<sup>a</sup>Red mud.*

*<sup>b</sup>Coal gangue and waste glass.*

Samples were leached with 0.5 M glacial acetic acid at liquid to solid ratio of 20 L/kg and rotated for 18 h at 30 rpm. The sample solutions after extraction were analyzed by inductively coupled plasma-atomic emission spectrometry (ICP-AES) (Perkin Elmer Optima 3300DV, USA) and inductively coupled plasma-mass spectrometry (ICP-MS) (Agilent 7900, USA). In addition, 2 mg/L standard solution was analyzed every 10 samples and after the whole sequence of determination for QA/QC. All samples were determined in triplicate (Li et al., 2017)

For other analytical methods on mechanical performance, compressive strength test was complied with local civil engineering works standard. Water absorption test was condcuted according to AS/NZS 4456.14 (n = 3). Compressive strength test was determined by automatic compression machines (36-5165/01, ELE International, UK) with compression speed 6,800 N per second and "non-standard" for size setting (n = 3). The compressive strength was calculated by determining the breaking load of the specimens. The compressive strength of specimens in series 7 would be evaluated according to GB5101-2003, which is a collection of current common sintered brick's national standards in the People's Republic of China.

## RESULTS AND DISCUSSION

#### Effective Ratios

W/C ratio and A/C ratio for waste materials utilized in the concrete mix were determined. The A/C ratio for waste materials was dramatically lower than typical A/C ratio, and a relatively high amount of cementitious materials was needed. A/C ratio for waste materials was 1.1, which can cure successfully on Day 1. The high amount of cementitious materials was needed may due to bottom ash, which is porous materials and representing of large surface area compare to natural aggregate (Singh and Siddique, 2013). In this study, The A/C ratio and W/C ratio of effective mixing proportion is 1.1 and 0.5. And the A/C ratio in this study was higher than Poon and Lam (2008) mentioned. The author recommended A/C ratio with 3 is suitable for eco-friendly concrete block production. Regarding W/C ratio, it is like the typical value, 0.6 (Jiménez et al., 2014). It may vary due to the difference of a characteristic of materials.

### Leachability of Heavy Metals

In Part one study, leachability of heavy metals in specimens was determined through TCLP to evaluate the environmental impacts of specimens produced by waste materials, including


*<sup>a</sup>Below limit of detection* = *0.001 ppm.*

different types of sludge and ashes. The results of the TCLP are summarized in **Table 4**. The concentrations of Cd, Cr, Ni, As, and Zn were below the limit of detection (LoD). However, Pb, Cu, and Hg were detected from both specimens and raw materials. Compared with the specimens, the relatively high concentrations of heavy metals were detected from raw materials. However, the concentrations of metal leached from the specimens were lower than those from raw materials samples. It may be due to the immobilization of heavy metals during cement hydration reaction.

Solidification/stabilization (S/S) is the commonly used technique in the world to reduce the toxicity of the waste materials for treatment of contaminated soil and sludge (Li et al., 2001; Valls and Vàzquez, 2002; Malviya and Chaudhary, 2006; Chen et al., 2009; ÖzverdI and Erdem, 2010). Solidification converts waste material from liquid phase to solid phase and strength development (Hills and Pollard, 1997; Chen et al., 2009). Due to this technique not only solidification but also reduce the toxicity and leachability through insolubilize and immobilize the waste materials, it become non-hazardous or less hazardous to achieve waste materials stabilization (Hills and Pollard, 1997; Malviya and Chaudhary, 2006). Cement-based S/S processes are suitable for treating heavy metal-contaminated sludge or soils because hazardous components are difficult to leach from concrete materials (Marion et al., 2005; Malliou et al., 2007; Qiao et al., 2007). Moreover, some studies demonstrated the possibility of using cement to treat heavy metals. Hills and Pollard (1997) found that the concentration of zinc in the specimen has reduced about 99% on Day 28. In addition, Antemir et al. (2010) have conducted a series test about the long-term performance of S/S and the test up to 16 years. The results indicated that the contaminants leach from the S/S soil was not excess the specific limit and the contaminant was immobilized in the solid for a long period of time. On the other hands, Marion et al. (2005) have studied the leaching behavior of paving concrete mixing with slag and reported there is no systematic correlation between the heavy metal content and the leached heavy metal from concrete. And authors comment that there is no significant risk of contamination of soil and water. Based on these previous studies, cement-based S/S processes are available to treat heavy metal.

#### Compressive Strength Test

The compressive strength test was conducted on Day 28 in Part two study. Compressive strength was calculated according to local civil engineering standard for concrete paving blocks (**Table 5**). In this part, different types of materials were utilized to replace coarse and fine aggregate. The results were indicated that the series, which were consisting of sewage sludge cake and sludge incineration residuals was failed to cure on Day 1. The texture of specimens consist of sewage sludge cake was sponge-like and soft. The strength of the cement matrix, aggregate and interfacial bond between the cement matrix and aggregate are the factors affecting the strength of concrete (Poon et al., 2004). Fail to may be attributed to the type of aggregate and interfacial bond between the cement matrix and aggregate. Whereas, the texture of specimens consists of sludge incineration residuals was cracked on Day 1. Thermal expansion may become the reason for cracking. When deformation of concrete is restrained, thermal stress will occur. Thermal stress is greater than tensile strength consequently cracking of the concrete (McCarthy et al., 2015; Siddiqui and Fowler, 2015). Another researcher also points out thermal stress is affecting durability and the life of concrete due to thermal stress generated by cement hydration greater than loading of concrete (Siddiqui and Fowler, 2015). Therefore, series mixed with sewage sludge and sludge incineration residuals were rejected and no further analysis to evaluate the mechanical performance. Dewatered sewage sludge and sludge incineration residuals may not be suitable for substitute into the concrete mix without pretreatment and admixture addition to improving the early strength of the specimens. Weak initial strength results in thermal expansion cracking due to generated by cement hydration (Hoang et al., 2016).

In series 5, only concrete mix E-BA/CW, which was consisting of bottom ash and coarse waterworks sludge, reached the local concrete paving block standard, 30 MPa (**Table 5**). On the other hands, the compressive strength of concrete mix E-BA/FW was 18.7 MPa. It may because the coarse aggregate was replaced by fine waterworks sludge. The appearance like mortar but not concrete, which is similar to the results obtained by Wang et al. (2015). Coarse aggregate presented a significant effect on compressive strength. Compressive strength was increased with increasing amount of coarse aggregate. In order to optimize the amount of reusing waste materials, the correlation between different A/C ratios and compressive TABLE 5 | Compressive strength and water absorption tests.


strength was also determined in series 6. E-A/C1.1 and E-A/C1.3 represent the satisfied compressive strength, which was 33.9 ± 1.3 and 36.3 ± 0.85, respectively (**Table 5**). When A/C ratio increased to 1.5, the compressive strength was begun to decrease. It may due to increasing of bottom ash and need more cementitious materials for concrete production. However, the previous study indicated that the compressive strength of concrete mixture, which was replace fine aggregate by coal bottom ash was greater than concrete mix consists of sand. Because of the pozzolanic effect of coal bottom ash begun affecting the chemical reaction inside the concrete and enhancing the properties of concrete after the curing period (Singh and Siddique, 2014). It would be possible to utilizing waster sludge and coal combustion residuals to produce eco-concrete blocks with A/C ratio 1.5.

Regarding the preliminary study of non-sintered and nonclay standard size block, the compressive strength of all concrete mixes was reached MU10 mentioned in GB5101-2003. The results for these concrete mixes are summarized in **Table 5**. From **Figure 1**, the trend of the difference of compressive strength for concrete mix E-CG20 and E-CG40 become stability on Day 28 and tendencies for increasing of compressive strength of E-CG30 was slightly greater than E-CG40 and E-CG20. Therefore, the compressive strength of E-CG30 would keep rising in the following days. E-CG30's tendency for increasing of compressive strength is greater than E-CG20 and E-CG40, which means that according to the average natural curing period (28 days), the compressive strength of E-CG30's blocks, would increase more significantly than the others, and the final compressive strength may be higher.

#### Water Absorption Test

Water absorption test was conducted, and the procedure was complying with local civil engineering standard. The results of water absorption test are summarized in **Table 5**. Excepting concrete mix E-BA/FW, every concrete mix was below 6% and not exceeds the standard limit. Specimens in concrete mix E-BA/CW not only have the highest strength but also have the lowest water absorption value. Compare to another study, the result of water absorption test was lower with similar coal bottom

ash substitution. Singh and Siddique (2015) obtained the water absorption was fall in the range from 4.7 to 5.6% on Day 28. In addition, the author also found that the voids of bottom ash were filled Ettringites needle and a high amount of calcium silicate hydrate (CSH) gel in the bottom ash concrete on Day 90. Concrete mix E-BA/FW may reach the satisfied water absorption value on Day 90.

## Correlation of Compressive Strength and Water Absorption

In this study, the strong correlation of compressive strength and water absorption test was found. From the **Figure 2**, when decreasing of compressive strength of concrete mix E-BA/FW and E-A/C1.5, the water absorption value of these concrete mixes were increased simultaneously. When conducted water absorption test, a lot of air bubbles occurred during the early stage of this test. It reflects that there were some spaces presences inside the concrete block. A similar result was obtained from another study, Singh and Siddique (2014) found that the compressive strength of bottom ash concrete was lower than river sand concrete at early curing age. Authors summarized the increased porosity of concrete was one of the reasons for that. Whereas, authors also found that the pozzolanic effect of coal bottom ash was begun to react with calcium hydroxide and form C-S-H gel and this gel was filled up the void of bottom ash on Day 14. Coal bottom ash could enhance the compressive strength of bottom ash concrete after the curing age and bottom ash reach the almost similar properties to river sand with more extended curing age were summarized by Singh and Siddique.

## CONCLUSIONS

In this study, different kinds of waste materials, which are the unavoidable by-products from wastewater treatment and power generation processes, were utilized to produce eco-concrete blocks. The results indicated that eco-concrete blocks consisting of waterworks sludge (in coarse size) and coal bottom ash were feasible for use in eco-concrete blocks production. The specimens reached the requirements stated in the General Specifications from the Hong Kong Government and did not exceed the toxicity characteristic leaching procedure standards. The mixing proportion of eco-concrete block production was dramatically different compared with concrete produced from raw materials. The A/C and W/C ratios were 1.1 and 0.5, respectively. Aside from eco-concrete paving blocks, non-sintered and non-clay blocks production from wastes (i.e., sludge and incinerated residues) was evaluated from an engineering viewpoint. The results showed that non-sintered and non-clay blocks produced with large fractions in waste materials reached MU10 in GB5101-2003. The proposed method will provide potentially inexpensive sources of environmentally friendly construction materials with desirable properties through the exploration of novel and advanced routes for resource recovery. It not only converts such wastes into useful materials but also helps alleviate the shortage of landfill sites. Hence, the practice of reusing waste materials is a win-win strategy that can alleviate raw material consumption during urban development in China. It not only reduces the quantity of sludge and incinerated residues for further treatment, but also provides a cost-effective and energy-efficient method for eco-construction materials production.

### AUTHOR CONTRIBUTIONS

WC: analysis and interpretation of the data/Drafting of the article/Critical revision of the article for important intellectual content/Collection and assembly of data/Administrative, technical, or logistic support/Final approval; HY: analysis and interpretation of the data/Drafting of the article/Critical revision of the article for important intellectual content/Collection and assembly of data/Final approval; CY: conception and design/Analysis and interpretation of the data/Critical revision of the article for important intellectual content/Administrative, technical, or logistic support/Final approval; HW: analysis and interpretation of the data/Critical revision of the article for important intellectual content/Collection and assembly of data/Administrative, technical, or logistic support/Final approval; GW: critical revision of the article for important intellectual content/Obtaining of funding/Final approval; EK: drafting of the article/Critical revision of the article for important intellectual content/Administrative, technical, or logistic support/Final approval; YT: conception and design/Analysis and interpretation of the data/Drafting of the article/Critical revision of the article for important intellectual content/Collection and assembly of data/Obtaining of funding/Administrative, technical, or logistic support/Final approval of the article.

#### REFERENCES


#### ACKNOWLEDGMENTS

This work was financially supported by the Research Grants Council of the Hong Kong SAR, China (No.: 18202116), the Research Cluster Fund (RG50/2017-2018R), the Dean's Research Fund (No.: DRF/SFRS-8 and DSRAF-6 SP1), and the Internal Research Grant (No.: RG 78/2015-2016R and RG 34/2017-2018R) of The Education University of Hong Kong.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Cheng, Yi, Yu, Wong, Wang, Kwon and Tsang. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Potential of Acid-Activated Bentonite and SO3H-Functionalized MWCNTs for Biodiesel Production From Residual Olive Oil Under Biorefinery Scheme

Hadi Rahimzadeh<sup>1</sup> , Meisam Tabatabaei 2,3 \*, Mortaza Aghbashlo<sup>4</sup> \*, Hamed Kazemi Shariat Panahi <sup>5</sup> , Alimorad Rashidi <sup>6</sup> , Sayed Amir Hossein Goli <sup>7</sup> , Mostafa Mostafaei <sup>8</sup> , Mehdi Ardjmand<sup>1</sup> and Abdul-Sattar Nizami <sup>9</sup>

#### Edited by:

*Saurabh Dhiman, South Dakota School of Mines and Technology, United States*

#### Reviewed by:

*Jaime Puna, Instituto Superior de Engenharia de Lisboa, Portugal Konstantinos Moustakas, National Technical University of Athens, Greece*

#### \*Correspondence:

*Meisam Tabatabaei meisam\_tab@yahoo.com; meisam\_tabatabaei@abrii.ac.ir Mortaza Aghbashlo maghbashlo@ut.ac.ir*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *22 September 2018* Accepted: *28 November 2018* Published: *17 December 2018*

#### Citation:

*Rahimzadeh H, Tabatabaei M, Aghbashlo M, Panahi HKS, Rashidi A, Goli SAH, Mostafaei M, Ardjmand M and Nizami A-S (2018) Potential of Acid-Activated Bentonite and SO3H-Functionalized MWCNTs for Biodiesel Production From Residual Olive Oil Under Biorefinery Scheme. Front. Energy Res. 6:137. doi: 10.3389/fenrg.2018.00137* *<sup>1</sup> Department of Chemical Engineering, South Tehran Branch, Islamic Azad University, Tehran, Iran, <sup>2</sup> Department of Microbial Biotechnology, Agricultural Biotechnology Research Institute of Iran, Agricultural Research, Extension, and Education Organization, Karaj, Iran, <sup>3</sup> Biofuel Research Team, Karaj, Iran, <sup>4</sup> Engineering and Technology, College of Agriculture and Natural Resources, University of Tehran, Karaj, Iran, <sup>5</sup> Department of Microbial Biotechnology, School of Biology and Centre of Excellence in Phylogeny of Living Organisms, College of Science, University of Tehran, Tehran, Iran, <sup>6</sup> Nanotechnology Research Center, Research Institute of Petroleum Industry, Tehran, Iran, <sup>7</sup> Department of Food Science and Technology, College of Agriculture, Isfahan University of Technology, Isfahan, Iran, <sup>8</sup> Department of Mechanics and Biosystems Engineering, Agricultural Faculty, Razi University, Kermanshah, Iran, <sup>9</sup> Center of Excellence in Environmental Studies, King Abdulaziz University, Jeddah, Saudi Arabia*

Application of acid-activated bentonite and SO3H-functionlized multiwall carbon nanotubes (SO3H-MWCNTs) for lowering free fatty acids (FFAs) content of low-quality residual olive oil, prior to alkali-catalyzed transesterification was investigated. The used bentonite was first characterized by Scanning Electron Microscopy (SEM), Inductively Coupled Plasma mass spectrometry (ICP-MS), and X-ray fluorescence (XRF), and was subsequently activated by different concentrations of H2SO<sup>4</sup> (3, 5, and 10 N). Specific surface area of the original bentonite was measured by Brunauer, Emmett, and Teller (BET) method at 45 m<sup>2</sup> /g and was best improved after 5 N-acid activation (95–98◦C, 2 h) reaching 68 m<sup>2</sup> /g. MWCNTs was synthesized through methane decomposition (Co-Mo/MgO catalyst, 900◦C) during the chemical vapor deposition (CVD) process. After two acid-purification (HCl, HNO3) and two deionized-water-neutralization steps, SO3H was grafted on MWCNTs (concentrated H2SO4, 110◦C for 3 h) and again neutralized with deionized water and then dried. The synthesized SO3H-MWCNTs were analyzed using Fourier-Transform Infrared Spectroscopy (FTIR) and Transmission Electron Microscopy (TEM). The activated bentonite and SO3H-MWCNTs were utilized (5 wt.% and 3 wt.%, respectively), as solid catalysts in esterification reaction (62◦C, 450 rpm; 15:1 and 12:1 methanol-to-oil molar ratio, 27 h and 8 h, respectively), to convert FFAs to their corresponding methyl esters. The results obtained revealed an FFA to methyl ester conversion of about 67% for the activated bentonite and 65% for the SO3H-MWCNTs. More specifically, the acid value of the residual olive oil was decreased significantly from 2.5 to 0.85 and 0.89 mg KOH/g using activated bentonite and SO3H-MWCNTs, respectively. The total FFAs in the residual olive oil after esterification was below 0.5%, which was appropriate for efficient alkaline-transesterification reaction. Both catalysts can effectively pretreat low-quality oil feedstock for sustainable biodiesel production under a biorefinery scheme. Overall, the acid-activate bentonite was found more convenient, cost-effective, and environment-friendly than the SO3H-MWCNTs.

Keywords: acid-activated bentonite, SO3H-functionlized multiwall carbon nanotube (SO3H-MWCNTs), waste olive oil, high free fatty acid, biodiesel production, biorefinery

## INTRODUCTION

Clay, particularly bentonites (clay mineral with different colors) are very popular in various industries (Christidis, 2013). Bentonite, has been applied as adsorbent, bleaching earth, and catalyst, as well as in eutrophication management and remediation of pollutions (Önal, 2006; Jeenpadiphat and Tungasmita, 2014; Copetti et al., 2016; El Korashy et al., 2016; Rezende and Pinto, 2016). Smectites are major clay minerals in bentonite with a 2:1 structure, or in another word, two silica tetrahedral sheets sandwich, an aluminum octahedral sheet or, a three-layer structure in which an aluminum octahedral sheet is located between two silica tetrahedral sheets. Replacement of some trivalent ions such as Al3+, by some divalent ions like Fe2<sup>+</sup> or Mg2<sup>+</sup> in the octahedral layer, or substitution of Si4<sup>+</sup> with Al3<sup>+</sup> in the tetrahedral layer generates a net negative electric charge on the surface of clay. This negative charge is balanced by either Ca2<sup>+</sup> or Na<sup>+</sup> cations on the surface of clay in Na-type bentonite or Ca-type bentonite, respectively, (Shen, 2001; Önal, 2006; Önal and Sarıkaya, 2007).

Bentonites, by virtue of montmorillonite, have high absorption capacity for exchanging some certain cations from solutions with their own molecules. Bentonite quality and properties (i.e., clay minerals loading, cation exchange capacity (CEC), porosity, selectivity, surface acidity, surface area) are very important in industrial applications and could be modified by activation methods (Önal and Sarıkaya, 2007). In another word, prior to effective application of bentonite in industry, some of its properties should be improved through activation or treatment procedures. There are four major ways to activate bentonite and improve many of its properties. Those include (i) chemical, (ii) physical, (iii) pillaring, and (iv) thermal procedures. Bentonite is activated by the application of inorganic acids (such as H2SO<sup>4</sup> or HCl) for replacing the exchangeable ions of bentonite during chemical procedure. Physical procedure involves mechanical improvement of bentonite's surface area by crushing it into smaller pieces. Similarly, pillaring procedure also improve bentonite's surface area as well as adsorption capacity. However, during pillaring procedure, these improvements are obtained through intercalating some elements or metal hydroxides

**Abbreviations:** BET, Brunauer, Emmett and Teller; CEC, Cation exchange capacity; CNT, Carbon nanotube; CVD, Chemical vapor deposition, FFA, Free fatty acid; FTIR, Fourier-Transform Infrared Spectroscopy; ICP-MS, Inductively Coupled Plasma mass spectrometry; SEM, Scanning Electron Microscopy, SO3H-MWCNT, SO3H-functionlized multiwall carbon nanotube, TEM, Transmission Electron Microscopy; XRF, X-ray fluorescence.

between the lamellar structures of bentonite creating pillars. In the last procedure, crystalline structure, or chemical composition of bentonite is thermally modified by heating bentonite at appropriate temperatures depending on its type. Amongst the aforementioned procedures, chemical treatment is the most common, and convenient method. It is worth mentioning that acid-activated bentonite could be used as solid acid catalyst in various reactions, such as esterification of oil for biodiesel production.

Carbon nanotubes (CNTs), first synthesized in 1991, have many applications as fillers, chemical sensors, hydrogen storage, electronic devices, catalyst supports, and etc., due to their unique properties including chemical stability, electrical, and thermal properties, high surface area, mechanical characteristic, etc. (Ham et al., 2004). CNTs are divided into single-wall CNTs and multi-wall CNTs (MWCNTs), arranged like a single rolled-up graphite sheet and like multiple layers of graphite rolled-up on themselves to form a tube shape, respectively (Kalamkarov et al., 2006). These structures can provide good support for various functional groups of specific features.

Oil feedstocks could be converted into biodiesel through the transesterification reaction (Aghbashlo et al., 2018b,c); however, food vs. fuel debate prevents the application of edible oils for this purpose (Hasheminejad et al., 2011; Aghbashlo et al., 2015, 2017b; Sahafi et al., 2018). On the other hand, ∼70–88% of the total biodiesel production cost arises from the cost of raw material (Haas et al., 2006). Therefore, biodiesel may be synthesized under a waste-oriented biorefinery scheme to address the aforementioned issues (Aghbashlo et al., 2018a, 2019). Under this concept, non-edible oils, such as waste cooking oils, and vegetable oil refinery waste oil that are either generated during processing oily crops into edible oil or their subsequent residuals from food processing sectors could be recycled into biodiesel (Aghbashlo et al., 2017a; Hajjari et al., 2017). For example, 0.11– 0.22 billion liters a year residual olive oil is recovered from olive cake (Pütün et al., 2005). Therefore, one prominent feedstock for the reduction of biodiesel production cost is such waste/residual oil resources such as residual olive oil. The main drawbacks of such oil feedstock are; however, their high water, and free FFA contents, lowering the yield of direct alkaline-catalyzed transesterification of oil into biodiesel due to the consumption of high amounts of catalyst and the formation of soap.

Alternatively, biodiesel may be produced from high FFA oil feedstock through acid-catalyzed (trans) esterification reaction, in which FFAs and triglyceride simultaneously react with alcohol to produce methyl ester (biodiesel) (Pan et al., 2017). The main disadvantage of this strategy is the lengthy process that requires high methanol-to-oil molar ratios. Therefore, often acids are only used as a pretreatment (esterification) method only to lower FFA or acid value to the acceptable points (0.5% or 1 mg KOH/g, respectively), prior to the commencement of the alkaline-catalyzed transesterification reaction (Montefrio et al., 2010; Hasheminejad et al., 2011). Homogeneous acids (such as HCl, H2SO4, H3PO4) have been widely applied as catalyst for esterification of high FFA oil feedstocks (Photaworn et al., 2017; Murad et al., 2018). Overall, the main shortcomings of using homogeneous acids include high quantity of alcohol consumed, and that the acid used could not be recovered and should be neutralized as well. Moreover, the corrosive nature of some of these acids, for example H2SO4, could increase the maintenance costs by damaging the equipment.

Hydrophobic solid acid catalysts could be used to overcome the above-mentioned drawbacks of the homogeneous acid catalysts. Carbonized vegetable oil asphalt, ferric sulfate supported on silica, and tolune-4-sulfonic monohydrate acid are some solid acid catalysts investigated for pretreating low-quality oil feedstock for biodiesel production (Hayyan et al., 2010; Shu et al., 2010; Dokic et al., 2012). Unfortunately, despite their comparatively high catalytic activity, they suffer from high production cost, complexity of preparation methods, or harsh operating conditions.

Having considered the disadvantages of the acid pretreatment methods discussed above, searching for more efficient, less expensive, and easier to operate strategies is inevitable. Considering that, this study was set to investigate the catalytic performance of acid-activated bentonite in converting FFAs of low-quality, high FFA-containing residual olive oil into methyl esters. This could be regarded as a pretreatment step for FFA removal prior to basic-catalyzed transesterification reaction. To achieve that, bentonite was characterized, acid-activated, and its performance in decreasing FFA content of residual olive oil through conversion to methyl esters was evaluated. Moreover, MWCNTs were also synthesized, SO3H functionalized, and their performance in FFA removal from residual olive oil was evaluated and compared with that of the acid-activated bentonite.

## MATERIALS AND METHODS

## Chemicals

Raw bentonite with the following physicochemical properties was used: surface area of 45 m<sup>2</sup> .g−<sup>1</sup> , mean pore diameter of 7.7 nm, total pore volume of 0.0898 cm<sup>3</sup> .g−<sup>1</sup> , swelling index of 15 mL/2 g, and CEC of 58 meq/100 g. Residual olive oil was purchased from an olive oil refinery and its FFA profiles and physical properties were investigated. All the chemicals used in this study were purchased either from Merck (Germany) or Sigma-Aldrich (Germany).

### Bentonite Preparation, Activation, and Characterization

A uniform sample of bentonite with the size of about 130µm was obtained by passing it through different sieves with appropriate mesh sizes. Then, aliquots of 50 g were weighted for the acid activation step.

Three different concentrations of H2SO<sup>4</sup> (i.e., 3, 5, and 10 N) were used for the activation of 50 g bentonite. The treatment was conducted in a water bath (95–98◦C, atmospheric pressure) and 250 ml H2SO<sup>4</sup> was slowly added while the mixture was agitated at 400 rpm using a magnetic stirrer. After 2 h, samples were washed (i.e., pH 7) with deionized water until reaching a neutral pH. The remaining slurry was subsequently dried in an oven at 130◦C for 2 h. The dried bentonites were kept in a desiccator until further use. The acid-activated bentonite was characterized by Scanning Electron Microscopy (SEM) (Tescan Vega3, Czech Republic), Inductively Coupled Plasma Mass Spectrometry (ICP-MS) (Agilent, United States), and X-ray fluorescence (XRF) (Spectro, Germany). The specific area of the bentonite was measured by BET technique with N<sup>2</sup> adsorption, at 77 K, using BELSORB Mini instrument (Bel Japan, Inc.) based on the ASTM 4567.

#### MWCNTs Synthesis and Characterization

MWCNTs were synthesized through methane decomposition (900◦C, atmospheric pressure, 20–50 min) over cobaltmolybdenum nanoparticles supported by nanoporous magnesium oxide during a chemical vapor deposition (CVD) process as described previously (Rashidi et al., 2007). Briefly, the reaction was performed using a flow of methane (50 mL/min) as carbon source and a flow of hydrogen (250 mL/min) as career gas at 900◦C for 30 min. Upon the completion of the reaction, the furnace was cooled under a nitrogen atmosphere and then the product was purified. To achieve that, the resultant material

FIGURE 1 | (A) SEM micrograph of the bentonite used and (B) TEM micrograph of the synthesized MWCNTs.

was first dissolved in HCl solution (18%) for a duration of 16 h at ambient temperature (25◦C ± 1) followed by filtration and several rounds of rinsing by distilled water. Subsequently, the product was dissolved in a nitric acid solution (6 M) for 6 h at 70◦C. The slurry was then filtered, dried, and heated in a furnace at 400◦C for 30 min. Then, the synthesized MWCNTs were subsequently sulfonated with 98% H2SO<sup>4</sup> (110◦C for 3 h), and neutralized (pH 7) by washing with deionized water. The sulfonated MWCNTs (SO3H-MWCNTs) were dried and kept until further use.

The prepared MWCNTs were characterized by transmission electron microscopy (TEM, CM30, Philips, Netherland) for morphology determination. For FTIR analysis, the samples were milled with KBr to form a very fine powder, followed by their compression into pellets. FTIR spectra were recorded on a Thermo Nicolet Nexus 670 FT-IR ESP (Thermo Nicolet Corp., Madison, WI, United States). The specific area of the MWCNTs was measured by same BET technique used for bentonite, but was analyzed by an ASAP 2010 (Micromeritics, United States).

## Catalytic Activity of the Catalysts

Catalytic activities of the acid-activated bentonite (5 wt.% of oil) and SO3H-MWCNTs (3 wt.% of oil) were calculated in respect to the esterification reaction with methanol-to-oil molar ratio

TABLE 1 | Inductively coupled plasma mass spectrometry (ICP-MS) and X-ray fluorescence (XRF) results for the used bentonite as well as surface area data obtained using BET technique for bentonite samples activated by different concentration of H2SO4.


*<sup>a</sup>Loss On Ignition (L.O.I.).*

of 15:1 and 12:1, respectively. The reaction was conducted in a magnetic stirred tank reactor (450 rpm) equipped with an alcohol reflux system at 62◦C. The acid value was determined at certain intervals and the pretreatment step was continued until the value reached about 0.5 mg KOH/g. Samples were withdrawn and immediately placed in an ice bath to stop the reaction, followed by centrifugation (3,500 rpm, 5 min). Then, these samples were used for FFA reduction measurements.

#### Determination of FFA Conversion

The composition of the methyl esters produced by esterification of residual olive oil and methanol in the presence of acidactivated bentonite or SO3H-MWCNTs as solid acid catalyst

TABLE 2 | Free fatty acid profile and physical properties of residual olive oil used in the present study.


was determined using gas chromatography (GC; Claus 580 GC model, Perkin Elmer Co., United States). The conversion of the waste oil into methyl esters was determined by using the following equation:

$$\text{Conversion(\%)} = \frac{\sum A \times A\_{IS}}{A\_{IS}} \times \frac{m}{M} \times 100 \tag{1}$$

where PA is the total area of the peaks, AIS is the peak of the internal standard (C17:0), m is weight of internal standard, and M is the sample weight.

Acid value of the samples was also measured according to the method of Cd 3d-63 provided by American Oil Chemists' Society.

## RESULTS AND DISCUSSION

## Characterization of the Acid-Activated Bentonite

The morphology of the natural bentonite as shown by SEM are presented in **Figure 1A**. The CEC of natural bentonite (58 meq/100 g) was, to a certain extent, lower than the expected range for smecite (80–150 meq/100), showing the presence of some impurities. The characterizations of the acid-activated bentonite synthesized in this study are presented in **Table 1**. The results of ICP-MS revealed that the bentonite used was of Na-type. Moreover, XRF data indicated that the bentonite structure was mostly composed of Al2O<sup>3</sup> and SiO2, providing good structure stability. It was expected that H3O<sup>+</sup> ions replaced some of these elements (such as Na and calcium) as a result of acid pretreatment (Rezende and Pinto, 2016). Bentonite activation was performed with three different concentration of H2SO<sup>4</sup> (3, 5, and 10 N), in order to investigate the impact of acid concentration on the surface area. Compared with natural bentonite, acid-activated bentonite displayed about 38–51% larger surface area (**Table 1**). It can be deduced from the BET data tabulated in **Table 1**,


TABLE 3 | The impact of 5 N acid-activated bentonite and SO3H-functionalized MWCNT as catalyst in the esterification reaction (methanol-to-oil molar ratio of 15:1 and 12:1, respectively), on FFA removal through its conversion into methyl esters and acid value over time.

\**Initial FFA content* = *1.3 (wt.%).*

that the activation of bentonite with 5 N solution of H2SO<sup>4</sup> triggered the largest surface area. In another word, increasing acid concentration from 3 to 5 N had a positive impact on the surface area, and consequently, on the number of hydrogen ions anchored on bentonite's surface. However, further elevation of acid concentration to 10 N must have damaged bentonite structure by collapsing some of its layers by dissolving them. In fact, the lower surface area obtained through the activation by 10 N H2SO<sup>4</sup> solution signifies that fewer hydrogen ions were anchored, and thus, less catalytic activity was achieved. On this basis, bentonite sample activated by 5 N acid solution was selected for the FFA reduction experiment.

## Characterization of the Synthesized SO3H-MWCNTs

The BET surface area of the synthesized SO3H-MWCNTs was measured at 230 m<sup>2</sup> /g, with the pore volume and diameter recorded at 0.76 cm<sup>3</sup> /g and 11 nm, respectively. This high BET surface area was because of inducing the repulsion force between SO3H and COOH groups on the MWCNTs surface (Shuit et al.,

2015), The morphology of the SO3H-MWCNTs as shown by TEM are presented in **Figure 1B**. As it can be observed, the sidewalls of SO3H-MWCNTs are not smooth due to the oxidizing effect of HNO3. FTIR data confirmed successful functionalization of the MWCNTs with SO3H (**Figure 2**). As revealed in the figure, a strong peak representing the stretch vibration of the SO3H functional group anchored on the MWCNTs appeared at around 1,000–1,080 cm−<sup>1</sup> . Moreover, a broad band in the region 3,100–3,400 cm−<sup>1</sup> was assigned to OH group caused by the purification step implemented. In addition, the strong band observed at 1,709 cm−<sup>1</sup> , and the weak band at 1,338 cm−<sup>1</sup> could be attributed to the stretching modes of sulfate groups.

## Impacts of the Synthesized Catalysts on FFA Conversion

The characteristics of the residual olive oil used in this study, i.e., FFA profiles and physicochemical properties are presented in **Table 2**. Respectively, the applications of 5 N acid-activated bentonite and SO3H-MWCNTs efficiently decreased the FFA content of residual olive oil to 0.43 and 0.47% that were less than the maximum acceptable value of 0.5% (**Table 3**). Intriguingly, original bentonite sample used as control did not show any positive impacts on FFA reduction. The results related to the FFA content reduction as determined by GC analysis (**Figure 3**), through their conversion to methyl esters are shown in **Table 3** and **Figure 4**. These findings well demonstrate the catalytic activity of the acid-activated bentonite and the SO3H-MWCNTs.

The acid-activated bentonite acted as a solid acid catalyst due to the hydrogen ions already anchored among its smectite's layers during the activation process. In another word, acidactivated bentonite acted as a proton donor catalyst expediting the esterification reaction. As seen in **Figure 4**, ∼67% of the FFAs contained in the residual olive oil were converted to methyl esters within the 27 h experimental period. It is worth mentioning that a methyl ester yield of 1.3% would be achieved from 100% conversion of the total FFA. However, the total methyl ester measured was 2.06% from 67% conversion of total available FFAs. This unexpected difference implies that some of the methyl ester yield (∼1.1 out of 2.06%) arose from acid-transesterification of triglycerides and not from the esterification of FFAs. This indicates that, although very minor, the active sites present on the surface of acid-activated bentonites could also attack the huge triglyceride molecules, and converted them into methyl esters.

Moreover, SO3H-MWCNTs was shown to possess favorable catalytic activity as 65% of the contained FFAs were converted, reaching the final concentration of 0.47%, in just 8 h of TABLE 4 | Comparison of some studies in which acid-activated bentonites and SO3H-functionlized multiwall carbon nanotubes (MWCNT) were used as catalysts in FFA esterification process.


esterification reaction (**Figure 4**). From the performance point of view, SO3H-MWCNTs were found as a better choice in comparison with acid-activated bentonite, as it led to slightly higher FFA reduction within 3-fold shorter time period (**Table 3**). This could be ascribed to its stronger acidic functional group (SO3H).

**Table 4** compares some studies in which acid-activated bentonites and SO3H-MWCNTs were used as catalysts in FFA esterification process. These studies differ in terms of the procedures used for catalyst preparation though. Shuit et al. (2015) used 10 wt.% (NH4)2SO<sup>4</sup> solution and ultrasonication (10 min), followed by a heating step (◦C, 30 min) to synthesize SO3H-MWCNTs for biodiesel production. This process was more advantageous than the procedure used in the present study in terms of synthesis time and being H2SO<sup>4</sup> free; however, about 2.1-time higher temperature was required. Moreover, the synthesized SO3H-MWCNTs had very lower surface area (92.37 vs. 230 m<sup>2</sup> /g, respecively) and pore volume (0.25 vs. 0.76 cm<sup>3</sup> /g, respectiely) making it a less efficient catalyst. The acid-activated bentonite produced herein showed about 62% higher surface area than that prepared by Jeenpadiphat and Tungasmita (2014). Both catalysts prepared in the current study, especially SO3H-MWCNTs, could decrease mass transfer limitation due to their high surface area. Nevertheless, and despite their characteristic advantages, lower FFA conversion yields were obtained by these catalysts in comparison with those listed in **Table 4**. This could be attributed to the fact that esterification process was not optimized in this study. Instead, a common mild-esterification reaction was targeted to only evaluate the catalytic performance of prepared catalysts.

## CONCLUDING REMARKS AND FUTURE PROSPECTS

This study showed that bentonite could be applied as an efficient raw material for synthesizing solid acid catalyst, through a simple chemical (i.e., acid) activation method, for using in solid acid catalyzed-esterification of high FFA oils. The application of this natural clay may encourage the production of sustainable fuels, as there is no requirement for sophisticated methods, and environmental hazardous chemicals as well. Acid activation of benonite using 5 N H2SO<sup>4</sup> was found to considerably improve its acidity as well as its specific surface area (more than half-time), and in turn, its catalytic activity for FFA content reduction of residual olive oil to <0.5 wt.%. It should also be noted that bentonite is very abundant and cheap (<20 USD/ton) and therefore, 5 N acid-activated bentonite could serve as a promising pre-treatment process for esterification of lowquality oil feedstock prior to alkali-catalyzed transesterification. Moreover, SO3H-MWCNTs were also synthesized and their catalytic performance was characterized. It showed excellent specific surface area (230 m<sup>2</sup> /g), and good pore diameter (11 nm) and pore volume (0.76 cm<sup>3</sup> /g), enhancing mass transfer of reaction. Compared with the acid-activated bentonite, SO3H-MWCNTs provided better catalytic performance. Despite more favorable economic preparation of the 5 N acid-activated bentonite and its relatively similar performance in terms of FFA reduction with the SO3H-MWCNTs, the latter catalyst proved more feasible for industrial application. This feasibility can be attributed to its considerably shorter reaction time due to more than 3.3 times higher specific surface area, significantly lowering mass transfer limitation. Further optimization of the esterification reaction conditions would be necessary for full exploitation of these two catalysts for FFA conversion into biodiesel. Overall, both of the prepared catalysts could be used for effective pretreatment of low-quality residual olive oil for sustainable biodiesel production under a biorefinery scheme.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

HR performed the bentonite catalyst production, SG and MM assisted with GC analyses, MeA assisted with some characterizations, MT and AR developed the idea and led the research. MT, MoA, and A-SN wrote the manuscript. HP as of substantial help during the revision process.

#### ACKNOWLEDGMENTS

We are thankful to Biofuel Research Team (BRTeam) for supporting this study.


strong protonic acid catalyst for biodiesel production. J. Taiwan Inst. Chem. Eng. 52, 100–108. doi: 10.1016/j.jtice.2015.02.018

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Rahimzadeh, Tabatabaei, Aghbashlo, Panahi, Rashidi, Goli, Mostafaei, Ardjmand and Nizami. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Effect of Pretreatment and Substrate Ratios in Biorefinery Employing Co-digestion of Plant Biomass and Poultry Waste

Fayyaz Ali Shah<sup>1</sup> , Naim Rashid<sup>2</sup> , Qaisar Mahmood<sup>1</sup> \* and Arshad Ali <sup>3</sup>

*<sup>1</sup> Department of Environmental Sciences, COMSATS University Islamabad, Abbottabad Campus, Abbottabad, Pakistan, <sup>2</sup> Department of Chemical Engineering, COMSATS University Islamabad, Lahore Campus, Lahore, Pakistan, <sup>3</sup> Military College of Engineering (MCE), National University of Sciences and Technology (NUST), Risalpur Campus, Risalpur, Pakistan*

The current study investigated the optimization of biogas generation during co-digestion of various plant biomasses with poultry waste at different ratios and pretreatment of plant biomass. The biochemical tests were executed at 35◦C in a thermostat. Water hyacinth was found as the suitable substrate for mono and co-digestion due to high volatile solids (VS) and total soluble contents. However, poultry waste was appropriate only if it was co-digested with other biomasses. The experiments evaluating biogas generation at different ratios of plant biomasses and poultry waste demonstrated that water hyacinth and poultry (50:50) produced 262 mL g−1VS, giant reed and poultry (80:20) produced 235 mL g−1VSwhile maize and poultry (60:40) generated 193 mL g−1VS. However, the pretreating the plant biomass with either Fenton's or Fenton's plus ultrasonic had no effect on biogas generation. The volumes of biogas generated after various pretreatment were low as compared to condition without pretreatment. The codigestion can be classified as WH:P (50:50 > GR:P (80:20) > M:P (60:40). So, these ratios can be applied at decentralized scale for better waste management and biogas generation due to balanced C:N ratio of plant biomass and poultry manure. Co-digestion can also be applied at large scale with optimized ratio in Pakistan and other developing countries for biogas generation and waste management and reduce the methane emission through landfills.

#### Keywords: anaerobic digestion, co-digestion, giant reed, maize, pretreatment, water hyacinth

## INTRODUCTION

Currently, the world is facing problems like environmental contamination and shortage of energy. Exploring non-conventional and eco-friendly sources of energy is serious consideration of many developing nations (Owamah et al., 2014). In this scenario, the need and exploration of alternative biofuels is a need of the hour. Even exploitation of methanogenesis during anaerobic digestion (AD) is quite old (Triolo et al., 2011). The advantages of AD are multi-dimensional; it not only reduces wastes from the environment but also helps in nutrient turn over along generation of biomethane as an important energy source. The industrial viability of AD demands an appropriate low-cost amalgamation of physicochemical parameters for the optimum functioning of this process (Owamah et al., 2014). For a sustainable and productive AD process, appropriate amounts of various macro and micronutrients including N, P, S, Fe, Ni, Se, W, Co, Mo, among many others is

#### Edited by:

*Abdul-Sattar Nizami, King Abdulaziz University, Saudi Arabia*

#### Reviewed by:

*Md Mofijur Rahman, Central Queensland University, Australia Mohammad Rehan, King Abdulaziz University, Saudi Arabia*

> \*Correspondence: *Qaisar Mahmood mahmoodzju@gmail.com*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *06 August 2018* Accepted: *10 December 2018* Published: *10 January 2019*

#### Citation:

*Shah FA, Rashid N, Mahmood Q and Ali A (2019) Effect of Pretreatment and Substrate Ratios in Biorefinery Employing Co-digestion of Plant Biomass and Poultry Waste. Front. Energy Res. 6:143. doi: 10.3389/fenrg.2018.00143* quite desirable (Chen et al., 2008; Demirel and Scherer, 2008; Hinken et al., 2008). The role of co-digestion, pretreatment and digester design for the enhancement of biogas have been thoroughly reviewed by Shah et al. (2015).

Co-digestion involves the use of two different substrates in a mixture at different ratios. Previously, it was iterated that methanogenesis could significantly be enhanced and optimized if feedstocks were co-digested with animal wastes (Cavinato et al., 2010). In view of few drawbacks of feed stocks for mono-digestion (Hinken et al., 2008; Pobeheim et al., 2010), the codigestion of various substrates always resulted in enhanced biogas generation. Considering the nature, strength flow rate of the substrate used and the biogas generation could be enhanced in range of 25–400% (Callaghan et al., 2002; Alatrwaste-Mondragón et al., 2006).

Researchers reported different results for the screening of diverse biomasses for methanogenesis in mono-digestion and codigestion. An increase of 400% was noted in methanogenesis when pig manure was co-digested with glycerol under mesophilic conditions (Astals et al., 2012). Co-digestion of algal biomass and fats, oils, and waste grease resulted in increased biogas production with gradual rise in organic loadings (Park et al., 2012). Stable biogas generation of 621 L/kgVS was observed at HRT of 42 days by digesting 50% slurry and whey mixture (Comino et al., 2012). The combination of waste paper, cow dung, and Eichhornia crassipes was evaluated for methanogenesis and the results were promising to reduce waste paper and enhancing biogas yield (Comino et al., 2012). In addition to increased methanogenesis, co-digestion of various substrates may present many advantages like high OLR, better biodegradability, balance of nutrients, suitable C:N ratio and lowering of toxicity of some substrates (Khalid et al., 2011; Wang et al., 2012, 2013). The preference of co-digestion was also advocated by (De Varies, 2012). AD of sewage sludge with poultry manure was also found beneficial (Bujoczek et al., 2000).

Biogas production was also found plentiful when substrates containing lignocelluloses was digested anaerobically (Zheng et al., 2014). In case of substrates containing high lignocellulose contents, pretreatment of biomasses with various techniques can be useful (Zheng et al., 2014) which reduce lignin, crystallinity but increase surface area. However, the richness of lignocellulose contents in the feedstock of AD may pose a barrier to microbial action on these substrates thus limiting the performance of AD (Fernández-Cegrí et al., 2012). In view of promising results obtained for co-digestion of different substrates, the importance of poultry waste in providing phosphorus for AD has never been realized to treat water hyacinth and other substrates. Water hyacinth is an abundant biomass that grows prolifically in all tropical and sub-tropical regions of the world. We anticipated that better biogas yield might be resulted by co-digestion of plant biomasses with poultry waste along with pretreatments. Pretreatment may be useful in view of fibrous nature of substrates which might have high cellulose and lingo-cellulosic materials. Based on above published reports, the aim of the current research was to evaluate the effects of pretreatment and co-digestion of various plant wastes with poultry litter for biogas generation.

## METHODOLOGY

## Sampling

Various substrates were collected for the experimentation from various non-contaminated places of Khyber Pakhtoonkhwa province of Pakistan. Prior to feeding into bioreactor, the plant biomasses and poultry wastes were air and oven dried at 35◦C. The plant biomasses were chopped and ground for further experimentation.

### Experimental Design

All the experiments were carried out in triplicates in a flask of volume of 500 ml for 30 days. The active volume was 100 ml with headspace of 400 ml for ratio experiment. The reactors were operated at temperature of 35◦C as given in the **Figure 1**. Inoculum to substrate ratio (I/S) of 2:1 was administered to the bioreactor. The contents of the bioreactors were purged

with nitrogen gas for 5 min to generate anaerobic conditions for microbes and subsequently sealed using butyl rubber stoppers. An outlet in the stopper was used for collecting biogas in inverted graduated cylinder and biogas was measured daily by water displacement as previously described by Shah et al. (2015). The pH of liquid in the inverted cylinder was kept acidic using 0.01 M HCl to avoid carbon dioxide absorption from biogas. A blank reactor was also run containing inoculum and distilled water, incubated at the same temperature to measure the background biogas produced from the inoculum and for pretreatment, which was subtracted from the total biogas production. All the digesters were monitored daily for biogas production. Experimental design for biomass ratio was given in the **Table 1** and pretreatment design was shown in the **Table 2**.

#### Biogas Unit

A schematic diagram of experimental set up was adapted from Shah et al. (2015). It consisted of a temperature controlled water bath at 35◦C containing digesters for the experiment. Each digester was connected to a graduated gas collector by means of a connecting tube. The experimental set up was shown in **Figure 1**. Biogas evolved was collected by downward water displacement as described above.

## Seed Inoculum

Anaerobic sludge was used from the already running experiment. Therefore, it was assumed as microbiologically adequate to treat various substrates proposed for the BMP assays. Similarly, the amount of inoculum used in the test bottles was determined based on amount of organic substrate available for degradation, i.e., inoculum-to-substrate (I/S) ratio (on VS basis), which was


TABLE 2 | Experimental design for pretreatment of plant biomass for biogas production.


equivalent to the inverse value of the food-to-microorganisms (F/I) ratio.

## Statistical Analysis

All determinations were performed in triplicates and mean values were presented in the results. Statistical comparisons of the mean values were performed by two-way analysis of variance (ANOVA) using Sigma PlotTM v.12.

## RESULTS AND DISCUSSION

## Effect of Different Biomass Ratios

A batch experiment was carried out to test co-digestion of different biomasses such as giant reed (GR), water hyacinth (WH), maize (M), and poultry (P) along with monodigestion. Poultry and plant biomass had no naturally occurring microbes and therefore, needed ample inoculation for startup of AD. The biomasses were co-digested at various ratios (50:50. 80:20, 60:40, 40:60, and 20:80) to test its potential for biogas generation. The biomasses were also characterized chemically to evaluate biogas generation through anaerobic mono and co-digestion. In case of mono-digestion, water hyacinth was the leading substrate among all the four selected substrates. The specific biogas production was the highest for WH (234 mL g−1VS) followed by GR (107 mL g <sup>−</sup>1VS), M (92 mL g−1VS), and P (49 mL g−1VS) for monodigestion. The chemical composition distinctly showed that WH had high volatile solids and soluble contents make the basis for highest cumulative biogas yield and daily biogas production as shown in **Figures 2**, **3**.

Various ratios of water hyacinth and poultry (WH: P) were presented in **Figure 4**. The 50:50 ratios of WH and P showed the highest biogas yield than rest of ratios of the same biomasses. The maximum biomethane yields from E. crassipes were 8.54 kJ g <sup>−</sup><sup>1</sup> dry biomass and 853.9 GJ ha−<sup>1</sup> y <sup>−</sup><sup>1</sup> which were 2.9 times higher than other crops (Yeong-Song et al., 2011). The total CO<sup>2</sup> emission reduced the bioenergy production of E. crassipes and replacing coal, fossil oil and natural gas with E. crassipes would reduce the CO<sup>2</sup> emission of 15.2–23.7 ton per year (Yeong-Song et al., 2011). Water hyacinth can be rich in nitrogen, up to 3.2% of DM and have a C:N ratio around 15. Water hyacinth was recommended as a substrate for compost or biogas production (Gunnarsson and Petersen, 2007).

On volatile solids (VS) basis, different ratios of water hyacinth and poultry 50:50, 80:20, 20:80, 60:40, and 40:60 produced specific biogas values of 262, 170, 164, 148, and 116 mL g−<sup>1</sup> VS, respectively. The ratio of 50:50 produced the highest biogas production which might be due to the better C:N ratio ranging between 15 and 30 which seemed to be suitable range for microbial growth (Haug, 1993). Lower amounts of biogas produced at other ratios might be due to imbalanced C:N ratio. As it shown in the **Table 1**, water hyacinth contained 21.34 ppm nitrogen and poultry contained 16 ppm nitrogen. So, the excessive N in other different ratio affected the C:N ratio thereby affected the biogas production. At very high C:N ratio, accumulation of VFAs occurs which leads to inhibition of anaerobic digestion. For all tested ratios of water hyacinth and poultry substrate, the biogas generation at 50:50 ratio

was significantly different from other ratios during 30 days of experiment.

The daily biogas production of water hyacinth and poultry at different ratios were shown in the **Figure 5**. A ratio of 50:50 had the most stable biogas production and was significantly different from rest of tested ratios until day 15. The 20:80 ratio of WH: P had less biogas production up to 10th day in comparison to other ratios. On VS basis, 40% of water hyacinth with poultry waste

had better contribution for biogas generation. Almost similar contribution on VS basis (35%) of was shown by raw sludge and food waste which enhanced the anaerobic digestion process as compared to mono-digestion (Koch et al., 2015). The codigestion of APW with DM increased the biogas and methane yield by 11.7–28.6 and 18.9–43.7%, respectively, compared with the mono-digestion of APW. However, the ratio of 3:1 for Aloe peel waste and dairy manure (APW/DM) depicted the optimal performance leading to the highest cumulative methane yield (195.1 mL g−<sup>1</sup> VS) (Huang et al., 2016). **Figure 6** showed the results of cumulative biogas production for different ratios for giant reed and poultry. For all the 80:20 and 20:80 mixing ratios

of Giant reed (GR), and Poultry (P), better results were recorded as compared with all other ratios. The biogas production for 80:20 and 20:80 were 235 and 229 mL g−1VS compared to monodigestion of GR 107 mL g−<sup>1</sup> VS. The 50:50 ratios of GR and P had least biogas production as compared to all other ratios. The pH of the 50:50 was 5.5. Lower pH might be due to the aggregation of VFAs. The C/N ratio may be a major factor behind such results; the impact of the C/N ratio on AD has been investigated at the optimum C/N ratio of 20–25 (Mshandete et al., 2004; Yen and Brune, 2007). With the higher C:N ratio, VFAs may accumulate. If C:N was lower than the appropriate range, methanogens may be inhibited by the high ammonia concentration (Gong et al., 2011; Banks et al., 2012).

The daily rate of biogas production for various ratios of giant reed and poultry substrates were presented in **Figure 7**. In the beginning of the experiment 60:40 ratio of the GR and P had the highest biogas production peak. The most stable biogas production rates were given by 60:40 from day 13 to day 23. By comparing the monodigestion of GR the overall daily biogas production was <40 mL d−<sup>1</sup> . The co-digestion of giant reed and poultry resulted in improved daily biogas production.

The biogas production at various ratios of maize and poultry substrate was presented in the **Figure 8**. For all the ratios of M and P, the 40:60 has the highest cumulative biogas production. However, 80:20 had the lowest biogas production by comparing with all other ratio. On basis of VS, 40:60 produced 193 mL g <sup>−</sup><sup>1</sup> VS, 20:80 gave 172 mL g−<sup>1</sup> VS, 60:40 gives 153 mL g−<sup>1</sup> VS, 50:50 generated 136 mL g−1VS, and 80:20 produced 121 mL g−<sup>1</sup> VS. The order of biogas production on VS basis was 40:60> 20:80> 60:40>50:50 and 80:20. The primary reason may be for production of less biogas production from mixing ratio of Maize and Poultry of 50:50 and 80:20 seemed high VFAs volatile fatty acids and reduced pH, which might have caused process instability.

The stability of anaerobic digestion was often expressed as the ratio of VFA to alkalinity. A ratio of <0.4 was generally regarded as optimal for anaerobic digestion, whereas a ratio exceeding 0.6 was regarded as indicative of overfeeding (Brown and Li, 2013). The results showed that acidogenesis was dominant over acetogenesis.

The production of long chain fatty acids causes slight inhibition of biogas production. However, a study evaluated the dairy manure and switch grass which presented best results at 50:50 ratio (Zheng et al., 2015). The positive effects of co-digestion can be accredited to manifold factors, including proportional nutrient composition, accelerated synergistic effects of microbial composition, an associated increase in buffering capacity and a decreased effect of toxic compounds on the digestion process (Wang et al., 2012). **Figure 9** presents the daily biogas production rate for different ratios of Maize and poultry waste (M:P). Initially, the 80:20 ratio had the highest biogas production. However, during the latter stages of the experiment 40:60 and 20:80 had the highest and stable biogas production than 80:20 and 60:40 ratios. The biogas production was decreasing with the passage of due to substrate utilization during the 30 day, experiment. During last 10 days, biogas production was minimal as compared to early days of the experiment. The comparison of present study with the previous ones was presented in **Table 3**.

#### Effect of Various Pretreatments

The effect of Fenton and Fenton plus ultrasonication pretreatment on biogas generation of WH was shown in the **Figure 10**. The biogas produced by WH biomass treated with Fenton's and Fenton's plus ultrasonic was lesser than

control. The reason for less biogas production in the present study from the chemical pretreatment might be due to the less target of ligno-cellulosic part of the WH. However, in literature the thermal pretreatment of the water hyacinth was quite effective for biogas optimization (Barua and Kalamdhad, 2017). The biochemical study unveiled that the cumulative methane production of hot air oven pretreated WH (3039 ± 32 mL CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS) at 90◦C for 1 h was way higher than the cumulative

methane production of untreated substrate 2396 ± 19 mL CH<sup>4</sup> g <sup>−</sup><sup>1</sup> VS on the 35th day (Barua and Kalamdhad, 2017). The cumulative biogas production from day second to day 8th the WH pretreated with Fenton's and Fenton's plus ultrasonic were not significantly different (p > 0.05). The **Figure 11** presents the pretreatment effect of Fenton's and Fenton's plus ultrasonic on daily biogas production. From the beginning, there was no significant effect of both pretreatment on biogas production with the control. However, Fenton pretreatment was comparatively better than Fenton's and Fenton's plus ultrasonic pretreatment. The less biogas production may due to the low energy application of ultrasound where no cell lysis occurred and COD solubilisation was due to exo-polymer dissolution, resulting in low methane production. However, at higher energies the disruption of the cell contributed to enhanced methane production (González-Fernández et al., 2012). The ultrasound pretreatment was recommended and effective when the lignocellulosic material to mix with water (Rodriguez et al., 2017). For daily biogas production, the control and both pretreatments were significantly different (p < 0.05) on day first; however, control and both pretreatments were significantly not different (p > 0.05) on day sixth. From day seventh to day 27th control and both pretreatments were significantly different (p < 0.05).

The cumulative biogas production from Maize with pretreatments such as Fenton's and Fenton's plus ultrasonic was shown in the **Figure 12**. On VS basis, the cumulative biogas production from maize under the mentioned pretreatments was 61 mL g−1VS and 30 mL g−1VS. Both the pretreatments were effective for biogas production up to 15 days. By comparing with control the Fenton pretreatment was more efficient than Fenton's plus ultrasonic. The cumulative biogas production for maize pretreatment with Fenton's and Fenton's plus ultrasonic with control were not significantly different (p > 0.05). However, from day 20th to day 30th control vs. pretreatments were significantly different (p < 0.05). The current ultrasonic pretreatment was carried out at 30◦C. Temperature has crucial influence in the samples used for sonication because other studies showed that it has a great influence on the changing of different properties of the liquid medium, such as viscosity, surface tension, and mainly vapor pressure, which influence cavitation (Karray et al., 2015). As the temperature of the liquid increases, its vapor pressure, and consequently the vapor pressure inside the bubble, also increases when implosion occurs. It can be seen that the ultrasonication pre-treatment results in the solubilization of carbohydrates. However, the current results for biogas production at 30◦C were not fruitful as compared to the other results conducted experiments with macro algal biomass (Karray et al., 2015). In the most recent article published by Rodriguez et al. (2017) it was recommended that plant biomass must be mixed with water for optimum results.

The daily biogas production of Maize after pretreatment with Fenton's and Fenton's plus ultrasonic was represented in the **Figure 13**. In the beginning, biogas production from the pretreated biomass was lesser. However, on the 5th day onwards, biogas production from pretreated biomass was higher than biomass pretreated with Fenton's and from control. The daily biogas production was stable on 15th day onwards until 30th

#### TABLE 3 | The comparison of current study with those previous researches.


day. **Figure 14** depicts the cumulative biogas production of giant reed with Fenton's and Fenton's plus ultrasonics pretreatment. The giant reed pretreated with Fenton's was comparatively better for biogas production than the giant reed pretreated with Fenton's plus ultrasonic. However, both pretreatments giant reed produced lower biogas than the control. Lower biogas production might be due to chemical effect of pretreatment on the microbial guilds of the batch system. The cumulative biogas production from day 1st to day 12th both pretreatment and control were not significantly different (p > 0.05).

The daily biogas production for the pretreated giant reed with Fenton's and Fenton's plus ultrasonic were shown in the **Figure 15**. In the early stages, daily biogas production for Fenton plus ultrasonic was lesser than control and giant reed biomass pretreated with Fenton. After day 3, biogas production from both pretreatment was almost at same level until day 19th. After 20th day the daily biogas production for both pretreatment were different. The daily biogas production from day 1st, day 3rd, were significantly different (p < 0.05).

The above-mentioned findings are quite practical in enhancing biogas production from various biomasses rich in cellulosic or lignocellulosic contents. There were two objectives of this investigation viz. the balancing of nutrients by the codigestion process and secondly the overcoming of toughness of lingo-cellulosic materials through physic-chemical pretreatment. The results produced are very promising and quite practical in the sense that we recommend co-digestion of WH, giant reed, and poultry waste in order to enhance

biogas yield per gram VS basis. Water hyacinth and poultry codigestion was the most promising in enhancing biogas production. Poultry waste could be a rich source of phosphorus which could satisfy the nutrient deficiency and to optimize C:N:P for an optimum methanogenesis. The codigestion can be classified as WH:P (50:50) > GR:P (80:20) > M:P

(60:40). So, these ratios can be applied at decentralized scale for better waste management and biogas generation due to balanced C:N ratio of plant biomass and poultry manure. Co-digestion can also be applied at large scale with optimized ratio in Pakistan and other developing countries for biogas generation and waste management and reduce the methane emission through landfills. Co-digestion can be applied at home scale by operating through pickle barrel digester to

play a key role against the global warming to avoid the open dumping of waste. The present study does not recommend the use of pretreatment as discussed above. The use of various chemicals especially Fenton reagent may disturb pH and chemical composition of feedstock and thus disturb optimum conditions for AD. The chemical structure of the feedstock may also be affected which resulted in lower biogas yields.

#### CONCLUSION

The batch experiments were conducted with various ratios and pretreatment of water hyacinth, giant reed, maize, and poultry substrates. The incubated biomasses were optimized at different ratios. The ratios experiments involving various selected plant biomasses and poultry wastes concluded that water hyacinth and poultry (50:50) produced 262 mL g−1VS, Giant reed and Poultry produced (80:20) produced 235 mL g−1VS,

#### REFERENCES


while Maize and Poultry (60:40) produced 193 mL g−1VS. So, these ratios can be applied at decentralized scale for better waste management and biogas generation due to balanced C:N ratio of plant biomass and poultry manure. However, the Fenton and Fenton's plus ultrasonic pretreatments of plant biomasses were not suitable for biogas optimization as compared to co-digestion.

## AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.


anaerobic codigestion of dairy, chicken manure and wheat straw. Bioresour. Technol. 120, 78–83 doi: 10.1016/j.biortech.2012.06.058


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Shah, Rashid, Mahmood and Ali. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Anaerobic Co-digestion of Catering and Agro-Industrial Waste: A Step Forward Toward Waste Biorefinery

Muzammil Anjum\*, Samia Qadeer and Azeem Khalid

*Department of Environmental Sciences, PMAS Arid Agriculture University, Rawalpind, Pakistan*

In the present study, a biorefinery system is proposed using catering and agro industrial waste for biogas and low phytotoxic digestate. Anaerobic co-digestion of catering waste with partially pre-treated (microwave 800 J/g/min + steam 121 ◦C 40 min) maize crop residues was conducted under different composition (20–50%) of feedstock. The results showed that the biogas production was increased by 2.03 times in co-digestion experiment (40% partially pre-treated maize crop residue + 60% catering waste: TCM3) as compared to catering waste alone (control). The increment in accumulative methane 116.7 m<sup>3</sup> t <sup>−</sup><sup>1</sup> was recorded in TCM3 which is due to improvement in biodegradation under co-digestion process. The post digestion byproduct (residual digestate) was evaluated for its phytotoxicity which is supplied with aerobic post treatment. The post treatment has improved the digestate quality by decreasing VS/TS ratio from 259 to 173 g/L and slightly increase the pH from 7.29 to 8.32. Seed germination assay showed that the germination percentage (G%), germination index (GI) and vigor index (VI) were relatively higher with post treated digesate as compared to un-treated digestate. In the germination test using wheat seeds, the post treated digestate (5% sol. extract) achieved higher values of GI and VI (46 and 609) whereas in un-treated, values for these indices were 14 and 62, respectively. Overall the findings of the present study identify the significance co-digestion based waste biorefinery, in order to development of value added bio-products such as biogas and biofertilizers.

Keywords: anaerobic digestion, biorefinery, bioenergy, digestate, phytotoxicity, organic waste

## INTRODUCTION

The advancement in living standard has risen the generation of organic waste at substantial rates. One of the signific sources of organic wastes in municipalities is from catering services which are generated excessively in hotels, canteens, restaurants, and the aviation industry (Ayomoh et al., 2008; Jiang et al., 2018). Catering waste may contain vegetables, fruit, meat, baked goods, dairy, and animal by-products which are not only rich in nutrients, but also a variety of amino acids, proteins, carbohydrates, and vitamins are also present. According to an estimate of UN Food and Agriculture Organization, about 1.3 billion tons of these kinds of waste are lost during the food supply chain (Jiang et al., 2018). Catering wastes are problematic to treat due their high moisture content, it can involve in the transmission of pathogen microorganisms that possibly cause diseases in humans and also pollute drinking water (Chen et al., 2017). Hence, it requires being dealt in an environmentally safe method (Zhang et al., 2005; Izumi et al., 2010; Anjum et al., 2016). On the

#### Edited by:

*Abdul-Sattar Nizami, Center of Excellence in Environmental Studies, King Abdulaziz university, Saudi Arabia*

#### Reviewed by:

*Vivekanand Vivekanand, Malaviya National Institute of Technology, Jaipur, India Qaisar Mahmood, COMSATS University Islamabad, Pakistan Mohammad Rehan, King Abdulaziz University, Saudi Arabia*

> \*Correspondence: *Muzammil Anjum muzammilanjum@gmail.com*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *07 August 2018* Accepted: *18 October 2018* Published: *08 November 2018*

#### Citation:

*Anjum M, Qadeer S and Khalid A (2018) Anaerobic Co-digestion of Catering and Agro-Industrial Waste: A Step Forward Toward Waste Biorefinery. Front. Energy Res. 6:116. doi: 10.3389/fenrg.2018.00116* other hand, agriculture is another major source of organic waste in the form of crop residues. Maize is known as one of the major crops in Pakistan, which produce thousands of tons of post harvesting residues. These residues are difficult to degrade through a biological process due to high content of lignocellulose (Surra et al., 2018).

The organic waste when disposed in landfills creates problems of leachate and biogas production that not only create nuisance in the landfill area, but also results in the loss of huge resource potential. Thus, an effective management system (biorefinary) is required to ensure the integration of organic waste management with energy recovery and reuse of byproduct formulation. Competence of biological waste treatment in this regard is well established option both in terms of scope and applicability, as it encompasses a wide variety of organic wastes ranging from food waste (Kuczman et al., 2018) and agro-allied industries (Alvarez et al., 2010) to waste activated sludge (Bolzonella et al., 2012; Bundhoo et al., 2016).

Among biological treatment anaerobic digestion has received special interest in the past few years, which results in the production of two byproducts that is biogas and digestate (Chen et al., 2018). However, some interventions are still needed for the provision of balanced nutrients and stable conditions for creating efficient digestion systems (Sosnowski et al., 2003). Recently, anaerobic co-digestion has been reported to improve the digestion process and energy production by improving the nutrient availability and organic load, while lowering the toxicity of the inhibitory compounds by dilution (Anjum et al., 2012, 2016; Serrano et al., 2013).

In anaerobic digestion most of the studies highlighting the anaerobic treatment of organic waste have made significant improvements to enhance the energy recovery as a factor of biogas production. Where mere efforts have been made to utilize anaerobic digestate which is an enriched nutrient source(Abdullahi et al., 2008; Trzcinski and Stuckey, 2011). The digestate effluents carry vital nutrients capable of promoting plant growth (Tornwall et al., 2017). Application of such high nutrient entities to the soil not only ensure the sound disposal of anaerobic byproducts, but also is an appealing solution to declining soil nutrient balance, which has identified as an important soil health problem worldwide (Galvez et al., 2012; Da Ros et al., 2018). However, the characteristic phytotoxicity limits the applications of anaerobic digestate as soil conditioner (Coelho et al., 2018). The fertilizer value of digestate can be enhanced by reducing its phytotoxic effects. Therefore, an appropriate pre-treatment is required that ensures sound application of anaerobic digestate to the soil. Aerobic "polishing" is identified as competent option for the reduction of toxic effects of anaerobic digestate through reduction in moisture, odor, carbon and pathogens in the digestate.

In view of above mention concerns the present study was conducted with the aims to: (1) partial pretreatment of maize crop residue to improve its digestibility in co-digestion with catering waste, (2) Optimizing substrate ratio in co-digestion catering waste and partial pre-treated maize crop residues to maximum digestibility and biogas production, (3) to test and prepare a low pytotoxic digestate as an organic conditioner for plants.

### MATERIALS AND METHODS

### Feedstock Material and Physico-Chemical Characterization

The raw materials used for the preparation of digestate were catering waste and crop residues maize crop residues. The catering waste samples were taken from local food restaurants and banquet halls of Rawalpindi city, while the maize residues were collected from PMAS Arid Agriculture University Research Farm located at Koont, Pakistan. The initial physico-chemical characterization of catering waste and maize crop residues are described in **Table 1**.

## Pre-treatment of Maize Crop Residues

Prior to anaerobic digestion maize crop residues were supplied with the pretreatment steps, which were selected based on a partial level of treatment with minimal energy prerequisite, and that avoids the complete degradation (Taherzadeh and Karimi, 2008). Sequential physical pretreatments, microwave irradiation and steam water were applied to the maize crop residues in order to enhance the accessibility of lignocellulose of maize crop residues for biological degradation. For microwave irradiation pretreatment a household-type DW-105-G microwave oven (Dawlance, Pakistan). Maize crop residues (50 g of each) were placed in the microwave oven for 3 min supplying energy of 800 J/g/min. Subsequently, irradiated substrate was placed in a high-pressure steam unit (Nanolytik Nanoclave 1, Germany) for 40 min at 121◦C for steam treatment (Anjum et al., 2016). After pretreatments, maize crop residues were dried in air for 6 h, and later in a digital oven for 12 h prior to use in co-digestion.

#### Anaerobic Co-digestion Experiment

TABLE 1 | Physico-chemical characteristics of substrates.

Anaerobic digestion experiments were performed in two steps. First, the optimization of feeding composition was performed and secondly the analysis of optimized co-digestion treatment was conducted during biogas production. At first set of experiment anaerobic co-digestion experiments were performed using catering waste as a main substrate mixed with partially pretreated maize crop residues at different compositions. Maize crop residues (M) were used in 20, 30, 40, and 50% of total fraction with catering waste, while similar experiments were repeated


with untreated maize crop residues. Anaerobic glass batch bottles of 500 ml capacity were used with total substrate volume of 350 and 150 ml empty head space. All mixtures of substrates were prepared on the basis of total solids. The seed anaerobic microorganisms were applied using previously digested organic waste added at a rate of 100 g/L (Anjum et al., 2012). A suitable amount of water was added keeping the moisture content of reactors fixed between ranges of the 70 and 75%. Co-digestion experiment took place for 6 weeks 42 at mesophilic temperature 30 ± 1 ◦C. To examine the degradation efficiency of co-digestion, samples were analyze at different interval changes in COD and VS.

In the second set of experiment, selected treatments (40 % maize crop residues and 60% catering waste) (TEM3) was subjected to biogas production in a lab scale, anaerobic static batch reactor and compared with control. Two similar reactor of dimensions 13 × 30 cm (diameter × width) each, having capacity of 2,500 ml were designed. One reactor was used for co-digestion (TEM3), while the second was used as control reactor having catering waste only. About 2,000 ml of the reactor volume was

filled with the substrates, while 500 ml head space was left empty for biogas accumulation. Biogas was collected in biogas collection bag connected at the top biogas valve, through the flow tube built in with an internal valve to control the one-way flow of biogas. The digestate samples were taken from the lower part of the reactor where a sampling valve was set. Both reactors were placed in an incubator at fixed temperature 30 ± 1 ◦C. The biogas was analyzed using liquid displacement apparatus.

### Aerobic Post Treatment of Digestate

Two digestate samples were collected from anaerobic codigestion of catering waste and maize crop residues. One digestate sample is supplied with an aerobic post treatment in order to reduce toxic substances exist in digestate, while second digestate sample did not supply with any post treatment. Aerobic pretreatment was performed using continuous stirred aerobic reactor supplied by continuous air diffusion from the bottom and string of 150 rpm. The temperature was maintained in range of 32–35◦C and reaction was continued for 20 days. The digestate quality was analyzed by measuring carbon content, volatile solids, pH and EC using standard methods. The digestate was dried in a digital oven for 24 h at 70◦C. The final obtained digestate materials were preserved in the cold storing unit at temperature >50◦C for further utilization in experiments.

## Phytotoxicity Test

The phytotoxicity test of both digestate was conducted by using 72 h seed germination assay. Germination assay is a rapid and widely applied method for evaluating phytotoxicity of organic product (Da Ros et al., 2018). Three dilution of each digestate material (post treated and untreated digestate) were prepared using 5, 10, and 20 g of dried digestate in 100 ml of distilled water.

Wheat seed were used to testing seed germination assay. For the germination assay, Petri dishes of 10 cm diameter were lined with watt man filter paper. Each dish received 5 ml of the digestate solution using a micropipet. Ten wheat seeds (sterilized) were placed in each petri dish and sealed with the parafilm. The petri dishes were incubated at 25◦C for 72 h in the dark conditions. The germinated seeds were counted and, root length and shoot length were measured were at regular interval during 72 h. The seed germination percentage (G%), germination index (GI) and vigor index (VI) were analyzed to estimate the phytotoxicity of digestate.

#### Anatical Methods

Chemical oxygen demand (COD) was analyzed using the closed, reflux titrimetric method applying the standard method of American Public Health Association (APHA (American Public Health Association), 2005). COD was calculated using the following equation (Equation 1):

$$\text{COD (mg/L)} = \frac{(X - B) \times M \times 8000}{\text{Vol. of sample}} \tag{1}$$

The factor X-B is ml of FAS used for blank subtracted from ml of FAS used for sample. M denotes the molarity of FAS, whereas the value 8,000 is the weight of oxygen mill equivalent 1,000 ml/l. Solid fractions i.e., TS and VS were analyzed using the standard method # 1684 (US EPA, 2001). TS was determined by measuring the difference in mass of the sample before and after drying in oven at temperature 105◦C, whereas VS was measured as an ignition loss at high temperature (550◦C) in the muffle

production rate, (B) Accumulative biogas production.

furnace. The C:N ratio was calculated by taking the total mass of carbon divided by mass of nitrogen in grams (Smith and Holtzapple, 2011; Anjum et al., 2012). pH, EC, and TDS values were recorded on Crison Multimeter (Model number: "CRISON MM-40þ"). In anaerobic digesters, the biogas production was measured as methane production through liquid a displacement methods for duration of 80 days using the method described by Elaiyaraju and Partha (2012). During phytotoxicity experiment, seed germination assay was conducted where seed germination index, germination percentage, seedling length, and vigor index were measured. The germination index (GI) was calculated using the following formula (Equation 2) (Mitelut and Popa, 2011):

$$GI = \frac{\text{G}}{\text{G}\_{\bullet}} \times \frac{\text{L}}{\text{L}\_{\bullet}} \times 100\tag{2}$$

where L<sup>o</sup> and G<sup>o</sup> are root growth and germination percentage. Data interpretation, calculation, averages and standard deviations were calculated using MS Office Excel 2010.

## RESULTS AND DISCUSSION

#### Anaerobic Co-digestion Test of Catering Waste With Maize Crop Residues Optimization of Substrate Ratio and Effect of Pretreatment of Maize Crop Residues

For optimization of the substrate ratio, the co-digestion experiments were conducted using catering waste as main substrate and maize crop residues as co-substrates (untreated and pre-treated with microwave-steam) for a duration of 42 days. The

results regarding organic matter removal efficiency (VS content basis) and change in COD are explained as under:

The results regarding VS content and organic matter removal efficiency are described in **Figure 1** which showed that the co-digestion experiments with pre-treated maize crop residues showed more reduction in VS as compared to untreated maize crop residues. The maximum reduction in VS was shown by CM4 (50% untreated crop residue ratio) where VS content was lowered to 644 g/kg from 881 g/kg after 42 days of incubation. However, maximum degradation in case of pre-treated maize crop residues was observed in TCM3 (40% ratio), with the VS reduction of 302 g/kg from 798 g/kg. The organic matter removal efficiency was achieved up to maximum 62% in TEM3 compared to the TEM4 (45%). The increased in organic matter removal efficiency is probably due to the increase in lignin loss under the influence of pretreatment of maize crop residues as reported by Yuan et al. (in press). The fact that pre-treatment enhance the degradation of lignocellulose and hemicellulose material of maize crop residues which are actually recalcitrant in nature (Surra et al., 2018). After applying pre-treatment these material partially degraded and become available to microorganisms. Overall, these outcomes revealed that the pretreatment of maize crop residues has enhanced the degradation in co-digestion process with catering waste.

The variation in COD values during anaerobic digestion of catering waste and pre-treated maize crop residues is presented in **Figure 2** COD is used as a sole indicator of optimization of anaerobic digestion process. After 42 days of the co-digestion process, the highest decrease in COD (30.2–11.5 g/L) was observed at 40% maize crop residues (TCM3), which means that insoluble COD (51%) was probably transformed soluble organic matter to final mineralized products (**Figure 2B**). In untreated maize crop residues the maximum degradation COD from 31.2 g/L (day1) to 20.4 g/L (day 42) was achieved at 40% fraction ration with catering waste (CM3). The lowest COD removal (31%) was observed in CM1 and CM4 in case of untreated maize crop residues which is even lower than catering waste alone (control) where COD removal of 46% was found. The improved COD degradation with pre-treated maize crop residue was due to the fact that pretreatment provides more organics in

solubilized form (sCOD) and thus more available to anaerobic bacteria (Anjum et al., 2018). Overall, these results showed that use of pre-treated crop residues is more promising with catering waste co-digestion where optimum mixing ration was found at 40% maize crop residue with 60% catering waste and labeled as TCM3.

#### Biogas Production

To observe the biogas potential of mixed catering waste and maize crop residues, anaerobic digestion was performed in two static batch reactors. One is for co-digestion containing both substrates in ratio as optimized in previous experiment (TCM3), while second reactor was taken as control containing catering waste alone. The results showed that the addition of pre-treated maize crop residues in catering waste, significantly increase the methane production potential as compared to that in catering waste alone (**Figure 3A**). During the anaerobic digestion process of 80 days, TCM3 co-digestion reactor showed an increasing trend in biogas production up 27–42 days where highest methane generation rate of 2.46 m<sup>3</sup> d −1 t <sup>−</sup><sup>1</sup> was

recorded. The incremented methane production was due to the influence of pretreatment of lignocellulosic biomass, where the pretreatment degrade the lignocellulose and increase the cellulose component (Rajput and Visvanathan, 2018). Furthermore, it is well reported that the co-digestion has ability to provide more balance nutrients such as optimum C/N (Wang et al., 2014) and increased P (Medvedeff et al., 2014) content which ultimately improve the growth of methanogens and biogas production.

On the other hand, control bioreactor showed the highest methane production rate of between after 54–60 days. Thereafter, the methane production rate was become stable in TEM3 codigestion reactor till 80 days. In control bioreactor the methane production rate was slightly decrease after 60 days, which is due to slow and incomplete degradation rate. The lowering in methane production in catering waste alone may be due the accumulation of volatile fatty acids, as acidic pH range was found during the anaerobic digestion process. Moreover, the inhibition in biogas production may be cause due to accumulation of ammonia within the system which cause lowering in C/N ratio below the optimum range (25–30 required for methanogens Wang et al., 2014. The accumulation of volatile fatty acids inhibit the methanogenic activity and longer the acetogenic phase, thus reduces overall efficiency of the system Chen et al., 2014; Yang et al., 2015; Jiang et al., 2018. Overall, the accumulative methane production was more in TEM3 as compared to catering waste alone (**Figure 3B**). The maximum accumulative methane of 116.7 m<sup>3</sup> t <sup>−</sup><sup>1</sup> was recorded in TCM3 reactor i.e., 2.03 times more as compared to control reactor. The addition of the co-substrate in anaerobic digestion process has influence the provision of more balanced nutrients such as C/N ratio and, organic load, thus creating more suitable conditions for microbial activity (Sosnowski et al., 2003; Zou et al., 2018).

### Effect of Aerobic Post Treatment of Digestate

In order to prepare anaerobic digestate bio-product as low phytotoxic organic fertilizer, aerobic post treatment was applied to the anaerobic digestate of co-digested catering waste and wheat crop residues (TCM3). **Figure 4A** expressed the results regarding effect on VS and carbon content due aerobic post treatment. VS in the digestate was decreased from 209 g/L to 113 g/L after 20 days where as carbon content was lowered to 17% from 24% in fresh digestate. The lowering in VS and carbon content is attributed degradation of toxic organic compound especially organic acids produced during anaerobic digestion process. The lowering in VS content in post treatment make digestate more stable (Wojnowska-Baryła et al., 2018). These means that aerobic post treatment for 20 days could lower the phytotoxic compounds in digestate and make more feasible bioproduct for soil application.

In case of pH, aerobic post treatment increases an overall pH of the digestate up 8.32 which is very favorable for soil application (**Figure 4B**). The basic reason of low pH in anaerobic digestion production of organic acid in high quantity which inhibit the digestion process (Yang et al., 2015; Zhang et al., 2018), but in the present study the pH of 7.29 showed that there were limit accumulation of organic acids and aerobic post treatment has further improved the pH of the digestate. Electrical conductivity is another important indicator of quality of digestate which mainly represents the concentration of salts and ions. EC values of 3.62 dS/m was found in anaerobic fresh digestate which is further lowered to 1.81 dS/m after aerobic digestion process (**Figure 4B**). EC can influence seed germination in contact with digestate, where lower EC is more favorable for seed germination (Coelho et al., 2018). This showed that EC of the digestate could be improved by post treatment of digestate in order to cause least negative effect in soil due to higher EC.

### Phytotoxicity Analysis of Digestate

Phytotoxicity is widely applied and promising parameter to evaluate digestate application on plants which describes the index its overall eco-toxicological impact (Da Ros et al., 2018). The phytotoxicity test of both untreated and post treated digestate at varying digestate solution (5, 10, and 20 g per 100 ml water) was performed using seed germination assay on wheat seeds. The results regarding seed germination percentage are described in **Figure 5**. Seed germination was continuously increased in case of control (Water only) where after 72 h 91% of seed were observed fully germinated. When digestate was applied G% was lowered where only 17 % of germination was found even after 72 h in D20 (20 g untreated digestate/100 ml water) (**Figure 5A**). However, a slightly higher G% (38 and24%) was found in D5 and D10, respectively after 72 h. This shows that the digested without post treatment cause swear toxicity to wheat seed which in increased as the concentration of digestate increase.

In case of post treated digestate an improvement in G% was observed with up to 85% seed germination was attained after 72 h in PD5 (5 g/100 ml) (**Figure 5B**). These finding also confirmed by the Germination Index (GI) values with a maximum GI of 46 was shown by PD5 and followed by PD10 (GI = 39) and PD20 (GI = 37) (**Figure 6**). However, the lower germination index values in un-treated digestate was observed which is probably due to low high EC values (Tang et al., 2018).

The seedling length was measured after 72 h of incubation and illustrated in **Figure 7**. Both shoot length and rood length were significantly increased with post treated digestate solution. A maximum root length of 6.01 cm and 5.19 cm were achieved by PD5 and PD10 which were 4.9 and 4.8 times higher than respective treatments with untreated digestate. Similarly, shoot length of 7.34 cm was the highest in PD5 with the slight lowing in PD10 (6.06 cm) and PD20 (5.60 cm). The improvement in seedling length is attributed to the post treatment of digestate which supply more plants essential nutrients [N, P, and K (Coelho et al., 2018)] to seedling with least effect of toxic materials.

Vigor index is one the known indicator used in analyses toxicity in plants (Shaikh et al., 2013). The vigor of seeds was described as a vigor index (VI) as illustrated in **Figure 8**. Seeds treated with post treated digestate showed the higher VI i.e., up to

#### REFERENCES


609 in PD5, where with the increase in concentration of digestate from 5 g/100 ml to 20 g/100 ml, VI was decreased to 377. On the other hand the seed grown with untreated digestate shows significantly lower vigor i.e., even with a value of zero with D20 (20 g/100 ml).

## CONCLUSION

The present study demonstrates the successful application of anaerobic co-digestion process for catering waste with partially pre-treated maize crop residues to produce biogas and biofertilizer production. The co-digestion of catering waste partially pre-treated maize crop residues at 40 % mixing ratio (TCM3) showed a significant degradability, where, 2.03 times higher methane production was achieved compared to control reactor.

Thereafter, the quality of digestate was improved by applying aerobic post treatment. The phytoxic analysis showed that improved G% (85%) while maximum GI of 46 and VI of 609 were achieved. Overall, the present study proposed a biorefinery concept in which co-digestion can be used as waste management approach for multiple outcomes such as energy recovery and biofertilizer production.

## AUTHOR CONTRIBUTIONS

MA conducted the basic experimental work and prepared the initial draft. AK supervised the work and provided technical and scientific assistance. The final draft was revised by both AK and SQ.

### ACKNOWLEDGMENTS

This work was conducted and supported by Environmental Science Lab, Department of Environmental Sciences, PMAS Arid Agriculture University, Rawalpindi Pakistan.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Anjum, Qadeer and Khalid. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Role of Biochar in Anaerobic Digestion Based Biorefinery for Food Waste

Carol W. Wambugu, Eldon R. Rene\*, Jack van de Vossenberg, Capucine Dupont and Eric D. van Hullebusch

*Department of Environmental Engineering and Water Technology, UNESCO-IHE Institute for Water Education, Delft, Netherlands*

The effect of biochar addition on the anaerobic digestion (AD) of food waste was evaluated. From the five biochars tested, Fe, Co, Ni, and Mn were leached in very small quantities (<10 mg/kg), while a high amount of K (1,510 and 1,969 mg/kg) was leached from treated waste wood and willow tree pyrochar, respectively. AD batch experiments were performed at an inoculum:substrate ratio of 1:1, at 30◦C and under agitating conditions. The results showed that the biogas volume produced by the treatments with the brewery residue hydrochar and treated waste wood pyrochar was lower than the amount of biogas produced by the control with only food waste. The food waste supplemented with 1.5 mL of trace elements yielded the highest biogas of 588 mL/g COD (CH<sup>4</sup> content−48%). Furthermore, two identical upflow anaerobic sludge blanket (UASB) reactors, i.e., control reactor and biochar amended reactor, were operated at 30◦C, at organic loading rates (OLR) varying from 3.4 to 7.8 g COD/L.day. The average COD removal efficiencies of the control and the biochar-amended reactor were 47 and 77% at an OLR of 6.9–7.8 g COD/L.day, respectively. These study results clearly indicate that the type of biochar and trace elements concentration in biochar play a key role in determining the effectiveness of the biochar in enhancing biogas production from food waste.

Keywords: anaerobic digestion (AD), biochar, trace elements, food waste, leaching

## INTRODUCTION

Food waste accounts for about 32–62% of the municipal solid waste (MSW) fraction and is expected to increase due to the increasing population and urbanization around the world (Xu et al., 2018). Food waste generation is higher in developed countries, e.g., 98 million tons per annum in the European Union (EU) countries (Xu et al., 2018) and about 43.6 million tons of food is reported to have been thrown away in the U.S.A (Zhang et al., 2007). The methods commonly used for food waste disposal include incineration, landfilling and aerobic composting (Zhang et al., 2014). However, landfilling has been banned in most of the developed countries, leaving incineration as the preferred method. Incineration is energy intensive (Zhang et al., 2014) and an expensive technology to implement, especially in developing countries, incineration possesses environmental risks such as air pollution. Incineration of food waste is also not favorable because food waste contains high level of water.

#### Edited by:

*Abdul-Sattar Nizami, King Abdulaziz University, Saudi Arabia*

#### Reviewed by:

*Muzammil Anjum, Pir Mehr Ali Shah Arid Agriculture University, Pakistan Rajendra Prasad Singh, Southeast University, China*

> \*Correspondence: *Eldon R. Rene e.raj@un-ihe.org*

#### Specialty section:

*This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research*

Received: *15 November 2018* Accepted: *04 February 2019* Published: *04 March 2019*

#### Citation:

*Wambugu CW, Rene ER, van de Vossenberg J, Dupont C and van Hullebusch ED (2019) Role of Biochar in Anaerobic Digestion Based Biorefinery for Food Waste. Front. Energy Res. 7:14. doi: 10.3389/fenrg.2019.00014*

In contrast, AD is emerging as a most efficient technology for food waste treatment and disposal. The high amount of the organic food waste fraction in MSW presents sufficient raw material for the AD process, which does not only remove the waste from the environment but is a promising source of renewable energy as well (Meyer-Kohlstock et al., 2016; Xu et al., 2018). The biogas produced from AD consists of about 65– 70% methane (CH4) and 35–40% carbon dioxide (CO2) (Xu et al., 2018), which can be converted into a compressed natural gas (CNG), and electric energy. However, the AD process is quite sensitive to disturbance due to the presence of a variety of microorganisms that are involved in the four distinct stages such as hydrolysis, acidogenesis, acetogenesis, and methanogenesis (Fagbohungbe et al., 2017).

The first step is the hydrolysis, where the macromolecules (proteins, fats, and carbohydrates) are broken down into smaller molecules (peptides, fatty acids, and saccharides). It is catalyzed by exo-enzymes called hydrolyses produced by the fermentative bacteria, as shown in Equation (1) (Kondusamy and Kalamdhad, 2014). Example of these bacteria include Bacterioides succinogenes and Clostridium thermocellum.

$$\rm nC\_6H\_{10}O\_5 \ + \rm nH\_2O \ \rightarrow \rm nC\_6H\_{12}O\_6 \tag{1}$$

The second step is acidogenesis, where the smaller molecules are converted into volatile fatty acids (VFA) such as propionic, acetic and butyric acid, and other byproduct gases like ammonia, carbon dioxide, hydrogen sulfide, alcohols, and aldehydes by acidogenic bacteria (Clostridium butyricum) as shown in Equation (2) (Kondusamy and Kalamdhad, 2014; Zhang et al., 2014).

$$\text{nC}\_6\text{H}\_{12}\text{O}\_6 \rightarrow \text{ 3nCH}\_3\text{COOH} \tag{2}$$

The third and fourth steps, i.e., the acetogenesis and methanogenesis, involve the conversion of acetic acid into acetate, which is then converted into carbon dioxide and methane (Equation 3) by acetoclastic methanogens such as Methanosarcina and Methanosaeta. The hydrogenotrophic methanogens also produce CH<sup>4</sup> by using CO<sup>2</sup> as a carbon source and hydrogen as a reducing agent (Equation 4) (Kondusamy and Kalamdhad, 2014; Lü et al., 2016).

$$\text{CH}\_3\text{COOH} \rightarrow \text{CH}\_4 + \text{CO}\_2 \tag{3}$$

$$\text{CO}\_2 + \text{H}\_2 \rightarrow \text{CH}\_4 + \text{ 3H}\_2\text{O} \tag{4}$$

Long-term AD of food waste is characterized by poor stability due to accumulation of volatile fatty acids (VFA), which is mostly caused by the deficiency of essential trace elements such as manganese (Mn), cobalt (Co), nickel (Ni), zinc (Zn), and iron (Fe) (Zhang and Jahng, 2012). Trace elements are very important in the methanogenesis step of AD that involves the action of acetyl-CoA synthase and methyl coenzyme M reductase to catalyze key metabolic steps and require sufficient amount of Fe, Ni, and Co (van Hullebusch et al., 2016), while some methanogens may require molybdenum (Mo), tungsten (W), copper (Cu), and selenium (Se) (Molaey et al., 2018). The inefficiency of the methanogenesis step leads to low CH<sup>4</sup> production. AD metabolic by-products like ammonia (NH3) also inhibit the methanogenesis step. Several solutions have been provided to improve the long-term stability of the AD process. These include, (i) co-digestion with other substrates such as cattle manure and sewage sludge (Xu et al., 2018) due to their high buffering capacity, (ii) addition of trace elements (Banks et al., 2012), and (iii) multi-digestion steps to separate the acidogenic and methanogenic phase of AD. One of the emerging trends in AD is the use of biochar as an additive for enhanced biogas production. A few researchers have reported that biochar can help to alleviate this effect when used as an additive in AD (Mumme et al., 2014; Lü et al., 2016; Wang et al., 2017).

Biochar is a carbon-rich compound that is produced through thermochemical decomposition of biomass in the absence of oxygen (Cha et al., 2016). It is produced through different processes such as pyrolysis (300–700◦C; N2; atmospheric pressure) and hydrothermal carbonization (170–250◦C; water above saturated pressure) (Cha et al., 2016). The energy required for biochar production varies depending on the type of biomass used. For wood biomass used in the present study, approximately 160 MJ would be utilized for an effective pyrolysis process as illustrated in **Figure 1** (Joseph et al., 2018).

Recent studies have shown that the addition of biochar in AD of food waste increased the biogas yield (Fagbohungbe et al., 2016; Meyer-Kohlstock et al., 2016; Sunyoto et al., 2016; Wang et al., 2017). Sunyoto et al. (2016) added pine sawdust biochar (produced at 650◦C) to AD of aqueous carbohydrate food waste made from white bread and observed increased CH<sup>4</sup> production by 41.6%. The addition of 8.3 g/L biochar to the food waste produced higher methane (from 55 to 78%), while 33.3 g/L biochar addition resulted in the lowest yield. It is hypothesized that the biochar creates a surface area for colonization by the microbial flora in the AD and acts as an adsorbent for compounds such as limonene (Fagbohungbe et al., 2016) and ammonia (Lü et al., 2016), that would otherwise inhibit the performance of the AD. Wang et al. (2017) added vermicompost based biochar (500◦C) to mixed kitchen waste and observed that the biochar acted as a buffer and increased CH<sup>4</sup> production due to 15–20% (w/w) biochar addition. Meyer-Kohlstock et al. (2016) added Holm oak residue biochar (produced at 650◦C) to municipal biowaste and observed an increase of CH<sup>4</sup> production per organic dry matter (ODM) by 5% (257–272 NL/kgODM) due to 5% (w/w) biochar addition and 3% (252–267 NL/kgODM) due to 10% (w/w) biochar addition.

Recent studies have been carried out to evaluate the role of trace elements and biochar in the AD of food waste, independently, mostly in single-stage anaerobic systems. However, these studies have not addressed the role of biochar in improving the trace elements bioavailability in AD. The role of biochar in continuous AD process has also not yet been reported. Therefore, the objectives of the present study were to: (i) evaluate the ability of different types of biochar to leach trace elements, (ii) evaluate the effect of biochar and trace elements addition on the AD of food waste in batch reactors, and (iii) evaluate the effect of biochar addition on continuous AD of food waste using an upflow anaerobic sludge blanket (UASB) reactor.

## MATERIALS AND METHODS

## Food Waste, Inoculum, and Biochar

The food waste was simulated using potatoes (30%), white bread (44%), spinach (10%), tomatoes (10%), and soya beans (6%). The total solids (TS) and volatile solids (VS) of the individual fractions of the food waste was analyzed, and the VS/TS ratio was used to select the required amounts used for this study. Shredding and dicing was done using a kitchen knife to reduce the size to ∼2 mm. Small portion of the food waste was blended using a kitchen blender (Proline, MIX55) with frequent addition of de-ionized water (total of 2 L), which resulted in a thick paste that was used in the batch experiments. Some of the paste was stored at −20◦C until further use in continuous experiments. Anaerobic granular sludge was collected from Veolia Water Technologies Techno Center Netherlands B.V. (Biothane). The sludge was stored at 4◦C until use. Before every batch test and the continuous experiment, the sludge was maintained at 30◦C for 24 h, in order to acclimatize the sludge to the experimental conditions. The biochar used was (i) treated waste wood (600◦C, ETIA, France), (ii) spruce wood (650◦C, CIRAD, France), (iii) scots pine bark (475◦C, VTT, Finland), (iv) willow tree (475◦C, VTT, Finland), and (iv) brewery residue (260◦C, VTT, Finland).

### Trace Elements Leaching From Biochar

Biochar samples were weighed and dried in an oven at 105◦C for 24 h. The moisture content was calculated using the gravimetric method for total solid analysis. The biochar samples (0.5 g) were placed in 250 mL bottles, and 20 mL of demi-water added. The bottles were agitated at 200 rpm on an orbital shaker (InnovaTM 2100 platform shaker), for 48 h at 30◦C, and samples were collected at intervals of 0, 24, and 48 h, respectively. The mixture was filtered using a vacuum pump (Knf laboport, N816.3KT.45.18) to obtain the leachate. The electrical conductivity (WTW EC meter) and volume of the leachate were measured. Concentrated nitric acid (5% per L of filtered volume) was added to acidify the leachate for trace metal analysis by an inductively coupled plasma mass spectrometry (ICP-MS, X series 2 by Thermo Scientific). The amount of trace elements leached was expressed in mg/kg, according to the calculations described in the EN12457-4 standard (Dutch Standardization Institute, 2002).

## Batch AD Experiments

Batch AD tests were carried out in 250 mL glass bottles having a working volume of 150 and 170 mL headspace, respectively. The inoculum/substrate (ISR) ratio 1:1 (VSS inoculum/VS food waste) was used for all the batch experiments. The batch experiments were carried out in three treatments as summarized in **Table 1**. All the batch experiments were performed and analyzed in triplicates. The food waste paste used in treatments A and B had an organic load of 18 g COD/L(wetbasis), while the food waste used in treatment C had an organic load of 6 g COD/L(wetbasis). The trace elements solution was composed of Fe, Co, Ni, and Zn, each with a concentration of 1,000 mg/L. The headspace was purged with pure nitrogen gas for 2 min, and the bottles were tightly covered using a metallic ring cap fitted with a rubber septum. The reactors were incubated on an orbital shaker (InnovaTM 2100 platform shaker), at 150 rpm and 30◦C, for 6 days. The daily increment of headspace gas pressure in the bottles was measured using a manometer (LEO 1 Keller, model: LEO 1/- 1 to 3 bar/81000.2) and was used to calculate the biogas volume, according to the procedure described by de Lemos Chernicharo (2007). The biogas was expressed in mL/g COD.

## Continuous AD Experiments With UASB Reactors

Two identical UASB reactors made of perspex glass were used for continuous experiments. **Figure 2** shows the schematic of the UASB reactor set up. The working volume was 2.25 L with a height of 91.5 cm and a diameter of 5.6 cm. One reactor was used as a control and the second one with biochar amendment as the test reactor. The granular sludge volume was maintained at 600 mL, which was about 30% of the reactors working volume in both reactors. The treated waste wood pyrochar was sieved to obtain 800µm particle size fractions. The pyrochar was added to the granular sludge bed in the test reactor at a concentration of


TABLE 1 | Experimental design for batch experiments.

*M, molarity; NaOH, sodium hydroxide; Na*2*HCO*3*, sodium bicarbonate; TE, trace elements solution.*

8 g/L of the reactors working volume. The control contained the granular sludge only.

The mixed food waste paste was diluted with deionized water to the desired g COD/L and the pH adjusted to 7.0–7.5 using 1 M NaHCO3. The diluted food waste was continuously fed into the reactor using a peristaltic pump (Masterflex, by Cole-Parmer; Model: 7528-30), at a rate of 2.6 L.day and an upflow velocity of 0.04 m/h. The organic loading rate of the diluted food waste was increased stepwise from 3.4 to 7.8 g COD/L.day and the hydraulic retention time (HRT) maintained at 24 h throughout the operational period. The reactors were operated at 30◦C for 40 days. On day 25 of the UASB reactor operation, the biomass in the two reactors was reduced to about 20% of the total reactor working volume. The effluent of both the reactors was sampled daily and analyzed for COD, NH4-N, pH, and VFA.

#### Analytical Methods

The analysis of different parameters was done according to the standard protocols adopted by the IHE Delft laboratory (Kruis, 2014). Total solids, volatile solids, and volatile suspended solids were analyzed using the gravimetric (oven method). The COD, total nitrogen, total phosphorous, and ammonia-nitrogen concentration were measured using a Perkin Elmer, UV/VIS Lambda 365 spectrophotometer at 600, 655, 880, and 655 nm, respectively. VFA was measured using a gas chromatograph (Varian 430-GC, CP-8400) as well as the CH<sup>4</sup> composition (Scion 456-GC). The trace elements were analyzed using microwave digestion method with concentrated nitric acid and the concentration measured using an ICP-MS (ICP-MS, X series 2 by Thermo Scientific).

## RESULTS AND DISCUSSION

#### Feedstock Characteristics

The initial feedstock characteristics play an important role in AD. The selection of the mixed food waste recipe used for this research was inspired by research done on AD of kitchen waste (Qiang et al., 2013; Sunyoto et al., 2016). The food waste and inoculum were first characterized before the start of the AD and **Table 2** illustrate the properties. The C:N ratio of the mixed food waste was 21.07, which was in line with the findings of Kondusamy and Kalamdhad (2014) who reported that a C:N ratio of 20–30 was sufficient for the AD process. The VS/TS ratio (organic content) was 97%, which indicates a very high potential for microbial treatment as reported by Xu et al. (2018). Trace elements concentration of important elements like Fe, Co, Ni, and Zn should be present in sufficient supply for a successful AD. The trace elements concentration in the food waste was very low and could not be detected by the ICP-MS used for the analysis. The minimum detection limits were (µg/L): Fe 200; Co 0.5; Ni 3.0; Zn 20; Cu 10; Mn 100; Na 6,000 and K 3,000. Hence supplementation was necessary for this food waste. The granular sludge used as the inoculum contained 842.5, 2.43, 8.02, and 186.35 µg/ (g d.w.) of Fe, Co, Ni, and Zn respectively which may have played a role in enhancing the AD process.

The characteristics of the biochar used for the AD are summarized in **Table 3**. The brewery residue hydrochar had a 4% ash content, which was lower than that of the treated

TABLE 2 | Characteristics of the mixed food waste and the inoculum used for the AD process.


*-, Not analyzed; (d.w.), dry weight basis; B.D.L., below detection limit; g/L, gram per liter; g/g, gram per gram;* µ*g/g, microgram per gram; %, percentage; COD, chemical oxygen demand; NO*3*-N, nitrate-nitrogen; NO*2*-N, nitrite-nitrogen; C/N, carbon/nitrogen.*

TABLE 3 | Characteristics of biochar used for the AD process.


waste wood pyrochar, which had 20%. This was in line with the hypothesis from literature by Kambo and Dutta (2015) who reported that hydrothermal carbonization produced a product with a lower ash content compared to slow pyrolysis. The ash content is an indicator of the amount of alkali metals that remained in the biochar after production (Kambo and Dutta, 2015). Poerschmann et al. (2015) reported a 6.3% ash content from HTC of brewer's spent grain, which is another term used to refer to the brewery residue. The pH of the brewery residue hydrochar was 6 while that of the treated waste wood pyrochar was 8. Poerschmann et al. (2015) also reported a pH of 6.9 of the produced hydrochar. The pH is important in that the biochar can provide buffering to the AD when used as an additive.

#### Trace Elements Leaching From Biochar

The essential trace elements (Fe, Co, Ni, and Zn) were leached out in very small rates as illustrated in **Table 4** in all the biochar used in the leaching test. This can be explained by the fact that they were also in almost negligible amounts in the initial biomass. Only Na and K ions were leached out in high amounts. The Na and K ions released into the leachate were responsible for the increase in the EC in the leachate of all the biochars tested as illustrated in **Figure 3**. The treated waste wood pyrochar with the highest EC had the highest Na and K amount while the brewery residue hydrochar had the second lowest EC and Na, and the least K within 48 h. This finding corresponds to Kambo and Dutta (2015) who reported that pyrolyzed biochar contained higher ash content, which signifies a high amount of alkali metals in the end product.

The percentage of the leached trace and macro elements from the biochar about the original biomass composition was also calculated. The data for the brewery residue hydrochar and the treated waste wood pyrochar was not available for the initial biomass composition. The scots pine bark leached out 85% of the amount of Ni and 45.4% K. Cu was not detected in any of the biochar. The willow pyrochar leached 38.6% Na and 68.3% K, while the spruce wood pyrochar leached 33% Zn and 30% Fe.

#### Effect of Biochar on AD in Batch Reactors Treatment A: pH Adjustment With 1 M NaOH

In the batch reactors treated with the brewery residue hydrochar, the addition of small doses of hydrochar, i.e., 0.2 g (332 mL/g COD) and 0.1 g (332 mL/g COD) increased the biogas production by 35 and 30%, respectively, when compared to the control which produced 215 mL/g COD (**Figure 4A**). The addition of a higher dose (1.2 g) produced the same biogas volume as the control (215 mL/g COD). This indicated that the addition of small amounts of the brewery residue hydrochar was enough to enhance the volume of the biogas produced. However, the CH<sup>4</sup> percentage in all the treatments was only 7–10%.

In the batch reactors with the treated waste wood pyrochar, the addition of 1.2 g (262 mL/g COD) pyrochar increased the biogas production by 18%. The lowest production was by the addition of 0.3 g (191 mL/g COD) pyrochar, which was 11% lower than the control (**Figure 4C**). This indicated that a higher amount of the treated waste wood pyrochar was required to enhance the volume of the produced biogas. The CH<sup>4</sup> percentage in all the treatments was between 15 and 27%. The biogas volume was higher in the treatments with the brewery residue hydrochar, but the CH<sup>4</sup> percentage in the biogas was higher in the treated waste wood pyrochar treatments.

The pH of the treatments with both the brewery residue hydrochar and the treated waste wood pyrochar decreased from 7.0 to below 4.5. This was caused by the rapid accumulation of VFA. Acetic (3.2–9.8 g/L) and propionic acids (1.4 to 2.1 g/L) were the most dominant in these treatments (**Figures 4B,D**). The pre-treatment of the food waste by blending made it highly biodegradable; hence, the rate of VFA production rate exceeded



*BR, brewery residue; SR, spruce wood; SPB, scots pine bark; WIL, willow tree; Mn, manganese; Fe, iron; Co, cobalt; Ni, nickel; Cu, copper; Zn, zinc; Na, sodium; K, potassium; B.D.L., below detection limit.*

the consumption rate (Ren et al., 2018). The methanogenesis step was inhibited by the high VFA concentration leading to a low CH<sup>4</sup> percentage in the biogas. This was in line with the findings of Sunyoto et al. (2016) who used pine sawdust biochar on AD of white bread without pH buffering and observed accumulation of VFA in their cultures that led to low pH of about 4.9. The hydrochar and the pyrochar did not provide buffering against the high VFA concentration as reported in the literature by Wang et al. (2017).

#### Treatment B: pH Adjustment With 1 M NaHCO<sup>3</sup>

1 M NaHCO<sup>3</sup> was used as a buffer to adjust the pH to 7.0. Two brewery residue hydrochar and treated waste wood pyrochar doses were used for this set instead of the seven used in the treatments adjusted with NaOH (treatment A). The doses that produced the highest and lowest biogas yield in treatment A were used for this set, together with two controls (With biochar only and with food waste only). The 0.2 and 1.2 g brewery residue hydrochar, and 0.3 and 1.2 g treated waste wood pyrochar doses were used.

The treatment with 0.2 g hydrochar dose produced 325 mL/g COD which was an increase in biogas production by 3% when compared to the control which produced 315 mL/g COD (**Figure 5A**). This volume by the 0.2 g dose was 2% lower than the amount produced with 1 M NaOH pH buffering. The 1.2 g hydrochar dose (317 mL/g COD) increased the biogas production by only 0.6%. However, the volume produced by the 1.2 g dose in the buffered treatment was 32% more than in the treatment with NaOH (treatment A). The biogas produced by the control with hydrochar only (12 mL/g COD) could be attributed to the digestion of the organic carbon still present in the hydrochar as reported by Kambo and Dutta (2015). The CH<sup>4</sup>

percentage in the biogas was 23 and 24% for the treatments with the hydrochar and control, respectively, an increase from 7 to 10% reported in the treatment with NaOH (treatment A). The treatment with 1.2 g treated waste wood pyrochar dose (328 mL/g COD) increased the biogas production by 4%, but the volume was 20% higher than in the treatment with NaOH (treatment A). The CH<sup>4</sup> percentage in the biogas with the treated waste wood pyrochar treatments was 24%, a slight increase for the 0.3 g dose that had 16% in the treatment with NaOH (treatment A).

The pH in all the biochar treatments and the control without biochar, reduced from 7.0 to 4.0 due to the accumulation of VFA in the reactors. Acetic acid was produced in high concentrations (more than 8 g/L) in all the treatments with the hydrochar, pyrochar, and the control without biochar. Butyric acid was the second highest concentration with more than 3 g/L in all the treatments. The highest concentration of propionic acid was 2.6 g/L in the 0.3, and 1.2 g treated waste wood dose treatment while 0.2 g brewery residue hydrochar dose had the lowest concentration of 1.7 g/L. The 6.9 and 7.2 pH in the hydrochar control without the food waste and the treated waste wood pyrochar respectively was an indicator that the organic load from the food waste was contributing to the high VFA production in the reactors (Hobbs et al., 2018).

The addition of the 1 M NaHCO<sup>3</sup> pH buffer enhanced the volume of biogas produced in all the treatments, but the high VFA concentration quickly reduced its effect. At high organic load, a higher concentration of the pH buffer is required (Gao et al., 2015). The hydrochar and the pyrochar could not provide sufficient buffering to counter the effect of acidification from the VFA.

#### Treatment C: pH Adjustment With 4 M NaHCO<sup>3</sup>

The treatment with 4 M NaHCO<sup>3</sup> was aimed at addressing the effect of the hydrochar and pyrochar on AD on a reduced organic load of the food waste. The food waste paste was diluted by reducing the amount of food waste from 76 to 25 mL and then adding 51 mL of de-ionized water to top up the volume to 76 mL, which was the required amount for the 1:1 ISR (**Table 1**). The same hydrochar and pyrochar doses used in the treatment with 1 M NaHCO<sup>3</sup> (treatment B), were used in treatment C, together with a control without biochar, food waste with 1.5 mL mixed trace elements supplement (1,000 mg/L of Fe, Co, Ni, and Zn) and a combination of biochar doses with trace elements supplement. The NaHCO<sup>3</sup> buffer concentration was increased to 4 M in treatment C.

In the batch reactors with the brewery residue hydrochar, the food waste with trace element supplement produced 588 mL/g COD which was an increase in biogas production by 9% when compared to the control without biochar, which had a biogas production of 538 mL/g COD as illustrated in **Figure 5B**. The biogas production by the control was 60 and 41% more compared to the treatment with NaOH (treatment A) and 1 M NaHCO<sup>3</sup> (treatment B) treatments, respectively. The 0.2 g hydrochar dose biogas production (435 mL/g COD), was 24–25% higher than in the treatments with NaOH and 1 M NaHCO3, respectively. The 1.2 g hydrochar dose production (438 mL/g COD) was 51–28% more than in treatments A and B, respectively. The production was also higher than that of the 0.2 g hydrochar in treatments A and B. The 0.2 g hydrochar dose with trace elements supplement produced 4% more biogas compared to

FIGURE 5 | (A) Cumulative biogas production during the treatment with different doses of the brewery residue (BR) hydrochar and the treated waste wood (TWW) pyrochar, with 1 M NaHCO3, (B) Cumulative biogas production during treatment with the brewery residue hydrochar and the treated waste wood pyrochar (FW, food waste; TE, trace element; BR, brewery residue; TWW, treated waste wood), with 1 M NaHCO3, (C) VFA production in the treatments with different doses of the brewery residue (BR) hydrochar and the treated waste wood (TWW) pyrochar (incubation period: 6 days), with 1 M NaHCO3, and (D) VFA production in the treatments with the brewery residue hydrochar and the treated waste wood pyrochar (incubation period: 6 days; FW, food waste; TE, trace element; BR, brewery residue; TWW, treated waste wood), with 1 M NaHCO3.

the hydrochar treatments without the supplement, and 22% lower than the treatment with the trace elements supplement without biochar addition. However, the hydrochar treatments still produced lower biogas when compared to the control without hydrochar.

The 0.3 g treated waste wood pyrochar dose treatment production (447 mL/g COD) was higher by 57–37% when compared to the treatment with NaOH (treatment A) and 1 M NaHCO<sup>3</sup> (treatment B) treatments, respectively. The 1.2 g pyrochar dose production (405 mL/g COD) increased by 35–19% compared to treatments A and B. The 1.2 g pyrochar dose with trace elements supplement produced 9% more biogas than the treatments without the supplement but was 24% lower compared to the treatment with the trace elements supplement without biochar addition. This could be due to the adsorption of the trace elements ions by the hydrochar and the pyrochar as reported by Fagbohungbe et al. (2017) and Li et al. (2017). The trace elements were therefore not available for the enzymatic activity of the microorganisms. The CH<sup>4</sup> composition in the biogas was 48% in the control and food waste with trace element supplement treatments, which was higher than in the treatments with the hydrochar and the pyrochar (36–38%). The biogas volume and CH<sup>4</sup> composition improved with the high buffer concentration in this treatment compared to the treatment with NaOH (treatment A) and 1M NaHCO<sup>3</sup> (treatment B) treatments. The addition of 4 M NaHCO<sup>3</sup> provided good buffering capacity in the batch reactors as reported by Gao et al. (2015), wherein the authors used 1,000 mg/L NaHCO<sup>3</sup> in the AD of kitchen waste and reported a 48% improvement in CH<sup>4</sup> yield.

Even with a pH buffer, the pH of the treatments with the hydrochar and pyrochar reduced to between 4.8 and 5.0 due to VFA accumulation (**Figures 5C,D**). The propionic acid concentration was higher (2.6 to 3.9 g/L) in all the hydrochar and pyrochar treatments. This explains the rapid decline in the pH because the accumulation of propionic acid consumes the highest alkalinity in the AD, thus inhibiting the degradation of acetic acid, which was also very high (2.3 to 5.3 g/L). The high propionic acid concentration in the treatment with 4 M NaHCO<sup>3</sup> (treatment C) could also be due to Na<sup>+</sup> toxicity to the propionic utilizing microorganisms than the acetic acid utilizing ones (Gao et al., 2015). The pH buffer used contained a higher concentration of Na+. The control and the food waste with trace elements supplement had very low total VFA production of <0.7 g/L. Hence, the pH remained around neutral. The high VFA accumulation and low final pH observed in the three treatments are in line with the findings of Wang et al. (2017). The authors used kitchen waste and vermicompost biochar for the batch AD and after incubation recorded 0.4 g/L acetic acid, 5 g/L butyric and 1 g/L propionic acid at pH 4.9. The higher biogas volume and CH<sup>4</sup> percentage in this treatment (C), for both the brewery residue hydrochar and the treated waste wood pyrochar, shows that the methanogenesis step was relatively successful compared to treatments A and B, due to the good buffering capacity provided by the addition of 4 M NaHCO3. However, the

#### TABLE 5 | Oxidation-reduction reactions in anaerobic digestion.


*Modified from de Lemos Chernicharo (2007).* 1*G*0, *Gibb's free energy; kJ, kilojoules.*

hydrochar and pyrochar treatments did not enhance the biogas production and composition when compared to the control.

From the trace elements leaching experiments, the hydochar and pyrochar used in the batch AD released negligible amounts of the trace elements into the batch reactors. This explains the inability of the biochar to enhance the biogas production as there was a deficiency of the required trace elements for utilization by the methanogens. The conversion of the organic matter in the food waste in the AD process takes place following the oxidation and reduction reactions as illustrated in **Table 5**. The oxidation reactions have a 1G<sup>0</sup> > 0; hence, the degradation of propionate and butyrate cannot occur under standard conditions and the reactions cannot shift to the right due to high amount of hydrogen (H+) ions. There was more hydrogen production in the batch experiments evidenced by the reduced pH due to high VFA concentration at the end of the incubation period, which inhibited the conversion of propionic and butyric acids into acetic acid, and CH4. The continuous removal of hydrogen from the AD system would ensure efficient completion of the process (de Lemos Chernicharo, 2007). This can be achieved by the addition of alkalinity (electron acceptors) in the form of bicarbonate to complete the methanogenesis step. The results from the three batch AD treatments indicate that even with the addition of a higher buffer concentration, the pH of the treatments with the brewery residue hydrochar and the treated waste wood pyrochar reduced to levels that inhibited the methanogenesis step, due to high VFA concentration. The pH of the biochar did not influence the AD process, but it led to the acidification in the reactors.

The NH4-N in the batch reactors of the three treatments (A, B, and C) was below 200 mg NH4-N/L. It is reported in the literature that concentrations in this range can be beneficial to the AD process, while more than 3,000 mg NH4-N/L can cause complete inhibition of methanogenic activity at any pH level (Rajagopal et al., 2013).

### Effect of Biochar on AD in a Continuous UASB Reactor

The reactor with biochar (treated waste wood pyrochar) amendment produced the highest cumulative biogas (32 L/g CODremoved) with an average of about 0.86 L/g CODremoved per day. The highest volume was within the first 3 days with a production of 20 L/g CODremoved. This may have been due to the low amount of particles in the influent caused by some particles settling in the influent pipes. The diluted food waste was therefore easily digested owing to the high activity of the granular sludge at the beginning of the operation. The biogas production remained steady for the rest of the operation time. The control reactor cumulative biogas production was 18 L/g CODremoved with an average of 0.45 L/g CODremoved per day. The biogas production in the control was also steady during the operation time with the highest production of 5 L/g CODremoved on day 32 (**Figure 6**). On day 29, the pH of the control reduced to 5.8, which reduced the activity of the granular sludge. This led to the accumulation of organic matter in the sludge bed, and once the pH was restored to 7.0, the activity and COD removal recovered causing a sharp increase in the biogas volume.

The COD removal efficiency (RE) was higher in the reactor with the biochar amendment than in control (**Figure 7**). For both reactors, the RE increased to 75% up to day 4 when the organic loading rate (OLR) was 3.4 g COD/L.day. The OLR was increased to 4.4 g COD/L.day on day 5. The RE of both reactors decreased to 33 and 25% for the control and the reactor with biochar amendment, respectively, between day 6 and 9. This was due to a design problem of the influent pipes, which led to clogging of the pipes hence the diluted food waste could not reach the granular sludge bed. This led to a sharp decrease in the pH (**Figure 8**) and reduced the activity of the microorganisms. The influent pipes were replaced with a smaller size, which allowed efficient pumping of the influent. The pH was adjusted to 7.0 using 1 M NaHCO<sup>3</sup> and the reactors recovered on day 10. Apart from day 8 and 9, the pH of the two reactors remained stable (around neutral) up to day 40. The OLR was maintained at 4.4 to 5.2 g COD/L.day from day 5 to 29. The average RE of the control was 38–59% for the reactor with biochar amendment during this period. The CH<sup>4</sup> percentage in the biogas was 28% for the control and 63% for the biochar amended reactor in this period.

During the operation, the granular sludge bed was rising to the top of the reactors (**Figure 9**). The produced gas was trapped within the sludge bed causing it to float, and because the diameter of the reactor was very narrow (5.6 cm); the granules could not fall back to the bottom without manual manipulation. Therefore, the biomass was reduced to about 20% of the reactors working volume on day 25. After the reduction, the granules were able to rise and fall back.

The OLR was later increased to 6.9–7.8 g COD/L.day on day 30–40. The average RE for the control was 47–77% for the reactor with biochar amendment during this period. The average CH<sup>4</sup> percentage in the biogas during this period was 39% for the control and 65% for biochar amended reactor. The high CH<sup>4</sup> percentage in the biogas produced by the biochar amended reactor was in line with the findings of Latif et al. (2012) who reported a 63% CH<sup>4</sup> composition in the treatment of liquidized food waste at an OLR of 12.5 g COD/L.day, and a COD removal efficiency of 75%, but without biochar amendment. The control (R1) is more comparable to the findings of these authors and the CH<sup>4</sup> percentage, and COD removal efficiency was 38–37% lower, respectively. The results indicate that the COD, RE and CH<sup>4</sup> percentage increased in both reactors when the OLR was increased to 6.9–7.8 g COD/L.day. The treated waste wood biochar amendment in the test reactor enhanced the COD removal and increased the CH4% by 37–47% respectively when compared to the control. The high carbon content in the biochar

FIGURE 9 | Photograph of the floating granular sludge during the operation period.

enhanced the C:N ratio of the diluted food waste paste hence the better performance of AD in the test reactor.

The VFA production in control was slightly higher than in the reactor with the biochar amendment during the operation period (**Figures 8**, **10**). The total VFA concentration in control was below 500 mg/L up to day 24 when it started increasing up to a high of 1,660 mg/L of acetic acid on day 34. This corresponds to the fluctuating pH level of the reactor. The total VFA concentration in the reactor with the biochar amendment was below 350 mg/L during the operation period. This indicated that the AD process in the biochar amended reactor was relatively stable. A total VFA concentration of between 50 and 250 mg/L would indicate a stable AD system (Ren et al., 2018). The VFA concentration in both reactors was below the inhibitory 2 g/L level as reported by Kondusamy and Kalamdhad (2014). The NH4-N concentration in both reactors was below 200 mg NH4-N/L and was below the inhibitory levels as reported by Rajagopal et al. (2013).

## Practical Implications of This Research

The results of this research show that biochar addition can be used to enhance the AD process especially in the UASB reactor. However, some of the challenges faced in this research can be stated as follows:


The effect of the biochar of this food waste in the batch reactors should further be investigated using different inoculum: substrate ratios, and trace elements concentration to establish the right ratios in case of future applications in large-scale applications to avoid VFA accumulation. The applicability of the UASB reactors in the AD of food waste requires research to establish the suitable design due to the high particulate matter present in the food waste. The reactors should be designed with an optimum height: diameter ratio than the ones used in this research to create more space for the granules to rise and release the gas with ease and fall back. The operation of the UASB at a large-scale level should be automated to ensure continuous pH monitoring and buffer dosing, which will prolong the longevity of the operations. The relationship between the initial biomass trace elements concentration will need to be established and the bioavailable amount at the end of the biochar production for AD applications. A database of suitable biomass and production temperature/process should be put in place to act as a guide for future AD applications. Biochar production at the industrial level for use in AD applications requires careful planning involving the integration of different policies. These include Waste Management, Renewable Energy, European GMO policy, Clean Development Mechanism (CDM) of the Kyoto Protocol, United Nations Framework Convention for Climate Change (UNFCCC), and Economic Policies. Some of the standards that need to incorporate the biochar include; the REACH (Registration, Evaluation, Authorization & Restriction of Chemicals) regulation (Regulation 1907/2006/EC), use of fertilizer regulation, soil remediation and protection regulations, and regulation of biodiversity (Van Laer et al., 2015). The International Biochar Initiative (IBI) has developed standards for biochar properties and testing of biochar, while the Delinat Institute defines on top of the biochar properties also the required properties of the feedstock used for the production of the biochar more in detail (Van Laer et al., 2015). The available standards are meant for the application of biochar in soils, hence, there is an urgent need to develop and integrate standards that apply to the use of biochar for AD processes.

## CONCLUSIONS

The biochar used in this research leached very small quantities of trace elements because the initial biomass used for their production also had low quantities of Fe, Co, Ni, Zn, Cu, and Mn.

## REFERENCES


The high amount of K leached from the treated waste wood (1,510 mg/kg) and the willow tree pyrochar (1,969 mg/kg), and relatively high amount of Na and K from the other biochars caused an increase of the EC in the leachate. Even with the addition of a buffer, the pH of the biochar treatments reduced to levels that inhibited the methanogenesis step (pH 3.8–5.0), at both higher (18 g COD/L) and lower (6 g COD/L) organic load of the food waste in the batch reactors. The biogas volume produced by the treatments with the brewery residue hydrochar and treated waste wood pyrochar was lower than that produced by the control with food waste only (538 mL/g COD). Trace elements supplementation enhanced the biogas production of the food waste without biochar addition (588 mL/g COD) by 8.5%. The combination of the 0.2 g brewery residue hydrochar and 1.2 g treated waste wood pyrochar with trace elements supplement in the batch AD enhanced the biogas production compared to the treatments without the supplement, but the biogas volume and composition was still lower when compared to the control. The treated waste wood pyrochar enhanced the COD removal efficiency and CH<sup>4</sup> composition in the test UASB reactor by 37–47% respectively, compared to the control. This shows that addition of treated waste wood pyrochar can enhance the performance of a UASB reactor for the AD of food waste.

## AUTHOR CONTRIBUTIONS

CW performed the laboratory experiments, data analysis, and wrote the manuscript. ER assisted in designing the laboratory experiments, as well as in writing the manuscript. JV and CD offered technical guidance, supported in data analysis and worked on the completed manuscript. EH provided overall guidance, supervision, and scientific knowledge support for this research work.

## ACKNOWLEDGMENTS

CW thanks UNESCO-IHE Institute for Water Education and the Dutch government for financially supporting the MSc programme through the Nuffic fellowship. The authors thank the UNESCO-IHE laboratory staff members, ETIA (France), RAGT (France), VTT (Finland), CIRAD (France), and Veolia Water Technologies Techno Center, Netherlands B.V. (Biothane) for providing analytical support, the bio/pyrochars, and inoculum used in this study.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wambugu, Rene, van de Vossenberg, Dupont and van Hullebusch. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Fuzzy Cognitive Map-Based Modeling of Social Acceptance to Overcome Uncertainties in Establishing Waste Biorefinery Facilities

Konstantinos Kokkinos <sup>1</sup> , Evangelia Lakioti <sup>2</sup> , Elpiniki Papageorgiou1,3 , Konstantinos Moustakas <sup>4</sup> \* and Vayos Karayannis <sup>2</sup>

<sup>1</sup> Department of Computer Science, University of Thessaly, Lamia, Greece, <sup>2</sup> Department of Environmental Engineering, Western Macedonia University of Applied Sciences, Kozani, Greece, <sup>3</sup> Department of Electrical Engineering, University of Applied Sciences of Thessaly, Larissa, Greece, <sup>4</sup> Unit of Environmental Science & Technology, School of Chemical Engineering, National Technical University of Athens, Athens, Greece

#### Edited by:

Abdul-Sattar Nizami, Centre of Excellence in Environmental Studies, King Abdulaziz University, Saudi Arabia

#### Reviewed by:

Giuseppe Mancini, Università degli Studi di Catania, Italy Alok Satlewal, Indian Oil Corporation, India Ta Yeong Wu, Monash University Malaysia, Malaysia

> \*Correspondence: Konstantinos Moustakas konmoust@central.ntua.gr

#### Specialty section:

This article was submitted to Bioenergy and Biofuels, a section of the journal Frontiers in Energy Research

Received: 07 August 2018 Accepted: 04 October 2018 Published: 26 October 2018

#### Citation:

Kokkinos K, Lakioti E, Papageorgiou E, Moustakas K and Karayannis V (2018) Fuzzy Cognitive Map-Based Modeling of Social Acceptance to Overcome Uncertainties in Establishing Waste Biorefinery Facilities. Front. Energy Res. 6:112. doi: 10.3389/fenrg.2018.00112 Sustainable Waste Biorefinery Facilities (WBFs) represent multifactorial systems that necessitate the organization, cooperation and the acceptance of different social stakeholders. However, these attempts have become targets of environmental, social and legal oppositions despite their obvious economic benefits. The variety of ambivalent and heterogeneous external effects of such projects result in either local support or opposition to the facility, which in turn becomes a critical factor affecting facility location decisions, and subsequent success of a WBF. Research has shown that simple surveys do not sufficiently measure social acceptance of such endeavors, and in most cases, local community factors dominate other external valuable impacts. In the current study, a novel Fuzzy Cognitive Map (FCM) modeling approach is proposed in order to analyze the socio-economic implications and to overcome multiple uncertainties occurring in sustainable WBF development and implementation. The primary investigation relates to the factors that influence the development of organic or chemical treatment of waste by the local communities and the competent authorities. The determination of concepts involved in the FCM modeling depends on a hybrid approach where both experts' opinion and statistical results from questionnaires distributed to stakeholders participate in the concept circumscription, thus identifying the centrality of each node in the model. Several steady state and dynamic analysis scenarios show the influence of driver concepts to receiver concepts on the social aspect FCM constructed.

Keywords: waste biorefinery facility (WBF), social acceptance, uncertainty, fuzzy cognitive map (FCM), modeling

## INTRODUCTION

Our society well-being remains heavily dependent on non-renewable fossil sources that provide heat, electricity, pharmaceuticals, transportation and food production. The major problem is though that our reliance on non-renewable energy sources has important negative environmental impacts, such acidification of oceans and global warming. To safeguard present livelihoods and offer development pathways for non-industrialized societies, energy sustainability is needed. Thus, the requirement to use renewable resources and to minimize waste becomes apparent. For that reason, an increasing number of individuals, companies, organizations and governments turn toward the development of biorefineries and biofuel production (Halder et al., 2012). This rapid growth is predominantly supported by major associations, such as central and local agencies whose primary goal is to promote policies, regulations and R&D initiatives that will lead to the increased production and use of sustainable biofuels. For the biorefinery industry the selection of the installation location is of great importance, as this is the major parameter impacting the initial capital of the project, the transportation costs for the raw materials and the produced fuel, which in turn become secondary factors affecting the retail price of the product. It must be noted here that a major role affecting the decision on the installation location is the acceptance of the local community at a first level and the approval of the local competent authorities at a second level. The immediate social acceptance drastically diminishes installation time delays and indirectly decreases the installation costs. On the other hand, companies believe that they are the ones that should decide where to locate the plants, arguing on the assumption that producers are economic free agents, evaluating sites as if all counties are contenders for their business, weighing the availability of feedstocks along with their infrastructure needs, operating without ties to localities, and being subject to enticement from policy incentives (Walter and Gutscher, 2011; Tigges and Noble, 2012; Fortenbery et al., 2013; Nizami et al., 2017a,b). Regardless of any minor disadvantages in relation to bioeconomy, the vast majority of researchers support the broad participation of various stakeholder groups and citizen associations, in order to "chart a strategic plan" to promote crucial importance of Waste Biorefinery Facility (WBF) growth. To be more specific, the following important impacts necessitate the urgency to open up the discussion about WBFs into the general public and include all affected stakeholders and major citizen groups (McGuire et al., 2017):


Certainly, there are also negative effects emanating by WBF growth. However, appropriate management of such inverse impacts ends up in promoting new sustainability standards and a better policy making framework than the existed one. On the other hand, high levels of public acceptance are essential if local authorities plan substantial increases in new renewable energy power plants for a certain area, as such projects may impose both positive and negative externalities on the local communities concerned. Apart from the increase of local jobs created and the stimulation of the local economy, such attempts positively affect local tax revenues, while at the same time purchases of local goods and services are also boosted. Nevertheless, the most common phenomenon occurring is a social syndrome usually referred as "Not In My Back Yard (NIMBY)" (Dear, 1992; van der Horst, 2007),reflecting residents' opposition to a proposed development in their local area. This characteristic phenomenon often carries the connotation that local residents are only opposing the development of waste biorefinery plants because they are close to them, and that they would tolerate or support such installations if they were built further away (Haddad et al., 2009; Lambert, 2009; CTV Kitchener, 2012; Lakioti et al., 2017). Characteristic cases are the ones mentioned in Selfa (2010), which investigate the opposition to biofuel facilities by local communities in three different locations in Kansas and Iowa. Lambert (2009) also mentions a good practice by a consulting company, which has helped tremendously various WBF corporations to address the NIMBY wars triggered in 135 locations. However, more and more researchers lately tend to discard the NIMBY-thesis due to the limitations imposed relatively to the human attitudes in obscuring the actual reasons for rejecting any kind of land use and particularly any WBF related land use (Devine-Wright, 2009; Wolsink, 2012).

Similar thesis relative to land use (Brion, 2015; Ciment, 2015) is the "Locally Unwanted Land Use" (LULU) syndrome. This basically argues that there is a land use creating externality costs on those citizens living within close proximity and these costs include potential health hazards, poor aesthetics, or reduction in home values. Therefore, the development of such facilities, with an increased level of hazards, need to be created for the greater benefits of the society as a whole, in order to get global acceptance and supportiveness.

The design and control of engineering systems such as waste biorefineries is a complex decision-making process in which multiple conflicting objectives of social, economic, and environmental nature must be taken into account. The key sustainability factors that drive the development and management of WBF or change the status quo of an existing waste management system can be summarized and categorized into: (a) environmental (climate change, land use, depletion of natural resources etc.), (b) social (local demographics, public resistance/resilience, NIMBY-LULU, public participation in making policies etc.), (c) competent authorities (institutional and administrative policies for WBF development, regional and municipal politics and legislation, structural development etc.), (d) economic (development funding, structural efficiency in installation minimizing costs, pricing models, secondary materials market etc.), and (e) technological (collection and transfer system, treatment technologies, waste stream composition and change) (Bovea and Powell, 2006; den Boer and Lager, 2007; Beigl et al., 2008; Graymore et al., 2008; Rehan et al., 2018).

The decision and policy making relatively to the development and sustainability of such large-scale waste biorefinery projects is done by stakeholders that influence these operations and sometimes their opinion is conflicting with others. Stakeholders include each of people or groups who are affected by technology developments or can affect that. Many of the barriers to access a successful performance of the projects could be the result of lack of acceptance of such projects by these social groups (Hosseini et al., 2018). From a socio-political point of view, acceptance of projects by core stakeholders and political actors has critical importance. When challenge is programming and increasing pub1ic participation incentive, cooperation with stakeholders and policy makers and also analysis of the acceptance level is important. For that reason, the decision-making model that will be adopted should provide quantifiable metric enabling systematic comparisons of the effect of different compromise solutions on the dissatisfaction of the stakeholders. With that model, it must be possible to make a thorough analysis relatively to the fairness of the decision taken and how decisions affect the opinions of specific stakeholder communities. Because of the diversity of stakeholder perceptions and attitudes, the model should also be able to analyze if a certain stakeholder or subset of stakeholders fully dictates the nature of a decision and therefore dissatisfaction of the rest of the stakeholders occurs. Stakeholder opinions along with polling procedures are the main tools that can be used for the development of such decision and policy making models, as they alleviate the conflicts created due to the way the information is presented and processed by the stakeholders (Cucek et al., 2013; Geraili and Romagnoli, 2015; Liew et al., 2015; Dowling et al., 2016).

Even though there is not an exact consensus on all environmental, economic and social impacts of building and managing WBFs, in the current paper we outline how an understanding of the processes of bridging cognitive and sociopolitical legitimacy can be used in the formation of a Fuzzy Cognitive Map (FCM) as a decision and policy making tool for the management of WBFs. Unlike all previous attempts of modeling such a problem, this study primarily depends on the local society stakeholder perceptions, opinions, attitudes and expectations to model the installation of new sites in rural and suburban areas.

The structure of the paper is as follows: In the next section, all the material and methods related to WBFs and FCMs are presented. More specifically, the integration of stakeholders in the WBF management decision process, the multi-criteria analysis and the scenario-based analysis methodologies are reviewed. Also, the characteristics of the FCM including the most popular activation and transformation functions as well as the algorithm of FCM construction from experts and the development of FCM from learning and training techniques are quoted. In section three, the application of FCMs in analyzing Social Acceptance for WBFs, developing a decision making FCM for a case study of WBF development in Thessaly, Greece, is discussed. Furthermore, due to the amount of concepts and the FCM complexity generated, the social part of the FCM is focused, by discussing its steady and dynamic state analysis and how the driver concepts influence the social acceptance. In the final section, conclusions and future challenges on the issue are highlighted.

## MATERIALS AND METHODS

## Integration of Stakeholders in the WBF Development Decision Making Process

Analyzing waste biorefinery development and management decision making has been a challenge in the recent years along with other similar problems (Lin and Yang, 2018; Stadler et al., 2018). Most of the researchers initially based their analyses on science, consensus, structured and multicriteria decision making approaches according to Gregory et al. (2012). With a focus on methods to involve citizens and other stakeholders in decision-making processes Van Asselt and Rijkens-Klomp (2002) identified eight different methods such as focus groups, scenario analysis, scientific stakeholder workshops, policy exercises, participatory modeling building, citizens' juries, consensus conferences, and participatory planning. However, when it comes to make decisions, the integration of stakeholders along with the opinion differences created between them and the experts in co-producing knowledge forces the analysts to use more widely approved methodologies. Among these methods, multi-criteria analysis, scenario analysis and rationality theory analysis are the ones to bring stakeholders and experts together in triggering learning processes that can make science more sensitive for societal problems (Mielke et al., 2016).

#### Multi-Criteria Analysis

The basic idea behind Multi-Criteria Analysis (MCA) is to set up measurable parameters that affect the process under study and to identify well established units for these impacts along with a clear ranking methodology. This formalized procedure allows decision makers to select the most desirable alternative among all available choices. Whenever there exist trade-offs between various alternatives, MCA clarifies values associated with decisions and evaluates these different alternatives including additional criteria. More specifically, the alternatives are further classified and clustered into criteria, which are weighted separately and then recombined back together onto the initial grading scale. Munda (2004, 2008). MCA methodologies utilize predominantly quantitative data. On the other hand, when the incorporate of stakeholders' perspectives into the decision process is attempted, their input usually consists of qualitative information making MCA methods insufficient to integrate this input. To overcome this problem, researchers synthesize traditional social research techniques (surveys, discourse based evaluation, narrative analysis, and value integration methods) with MCA for addressing the integration of stakeholders' perspectives. Thus, alternative MCA methodology variants have been developed to overcome the aforementioned challenge (Stirling, 2006; Yatsalo et al., 2007). For our case, the most important variant is the one that corresponds to Social Multi-Criteria Evaluation (SMCE), as emphasis is given mostly to stakeholder engagement in making decisions and also on the operational framework useful to answer the following question: how can the local society and the local competent authorities integrate a plurality of technical aspects and social views into its ex-ante impact assessment in a coherent and transparent manner for the development and management of WBFs. According to Scolobig and Lilliestam (2016) the phases of SMCE are: (a) institutional analysis, (b) alternatives and criteria, (c) assessment of the criteria, and d) identification of the most desirable alternative. Finally, the most critical parameter in this impact assessment technique is to produce a decision policy based on the following three main objectives:


#### Scenario-Based Decision Making

In general, the Scenario-Based methodology simply presents all the alternative scenarios evaluating the impacts they impose in decision making without the inclusion of a decision-making directive. After evaluating the impacts of each scenario, the decision maker should be able to make a decision related to the problem at hand, evaluating the less expensive alternative. A critical task in this process is the involvement of stakeholders in suggesting different scenarios for evaluation, or evaluating the existing ones, or setting scenario boundary conditions together with the decision makers (Kok et al., 2015; Trutnevyte et al., 2016). Furthermore, as we already explained, the difficulty stands on the fact that all the qualitative information coming from the stakeholders' input needs to be converted into quantitative. Therefore, it is useful to incorporate Fuzzy Logic techniques to achieve this defuzzificaton process. To summarize, the key steps of the engagement of the stakeholders in scenario-based approaches (Scolobig and Lilliestam, 2016) are: (a) the elicitation of stakeholders' perspectives, (b) quantification of inputs to feed the model parameters and (c) the scenario ranking and identification of actions to enact a particular scenario.

#### Plural Rationality Decision Making

In this case, decision and policy making is based on a solution as a result of a compromising process among multiple "voices" through extraction of multiple stakeholder perspectives. This methodology is based on the theory of plural rationality, which suggests that all stakeholder perspectives are organized and relate to five different issues, namely the hierarchical, the egalitarian, the individualistic, the fatalistic, and the autonomous. These categories share common attitudes and views of stakeholders. Thus, for this method, the key steps that need to be taken are: (a) the elicitation of stakeholder perspectives, (b) the creation of technical-policy alternatives, (c) the construction of focus groups to discuss these alternatives and prioritize the actions needed to be taken, and finally (d) a thorough discussion to reach a decision.

#### Experts and Stakeholders

The increasing number of decision making models that can incorporate both expert opinions and a diverse collection of stakeholder perspectives creates inevitably a confusion in selecting the best analytic-deliberative-adaptive methodology for developing and managing WBFs. For this reason, we choose to base our approach on the inclusive definition of all the concepts made primarily by experts. At the same time, the model scrutiny will also utilize systematic and transparent procedures including as wide a range of stakeholders as possible. However, it must be noted that priority must be given to the quality of the enquiry for those responsible for waste biorefinery management and policy formulation for this kind of projects. For such systems, the efficiency of decision-making depends largely on the ability of decision-makers to analyse the complex cause and effect relationships and take productive actions based on the analysis. More specifically, the dynamics of large scale complex WBF reveal different components which affect each other, and these cause and effect relations show system behavior. All aforementioned arguments guide us to use system conceptualization graphs to understand all of the system aspects, such as FCMs.

## Fuzzy Cognitive Mapping (FCM)

FCM is a qualitative or rather semi-quantitative and dynamic method to structure expert knowledge that aims to capture a person's perception of a particular issue in a diagrammatic format. FCM graphs provide both the modeler and the interviewee with an informal structured process having the ability to give additional beliefs, insights and concepts about a certain domain. Furthermore, the interrelations and interdependencies of these concepts are also revealed, providing information about how the change of one issue can affect the others. The selection of the FCM methodology to investigate the development and management of WBF is due to following advantages according to Ozesmi and Ozesmi (2004): (a) FCMs are well structured to imprint the complexity and the reason for the waste biorefinery model, (b) FCMs are capable to represent the quantitative and qualitative information obtained from the stakeholders' opinion, overcoming the lack of quantitative reliability of data due to uncertainty, and (c) FCMs are suitable to illustrate the effects of factor changing for the whole systems even though they are not able to make quantitative predictions. Therefore, FCMs are allowed to predict the effects of the policy taken under the "what-if " scenario, having assumed that, since the real world is complex, knowledge can be obtained from the perception of people involved into a certain issue (Kosko, 1986, 1987).

#### Basic Concepts for FCMs

More rigorously, an FCM is a graphical representation of a system used to illustrate the cause and effect relations between nodes, thus giving us the opportunity to describe its behavior in a simple and symbolic way. In FCM graphs, nodes represent concepts, and arcs represent the perceived relationships between

these concepts (Axelrod, 1976). These relationships in a FCM are logically imposed by connecting concepts via semantic or otherwise meaningful directed linkages showing the causality between them (Novak and Cañas, 2008). According to Gray et al. (2014), this causality in representing cognition by weighted arcs in a structural map is derived from constructivist psychology, which suggests that individuals interactively and collectively construct experience/knowledge developing internal associative representations that help to catalog, interpret, and assign meaning to environmental stimuli and experiences (Raskin, 2002). Another view of the FCMs is from the Neural Networks and Fuzzy Logic point of view. For that reason, learning techniques and algorithms can be borrowed and utilized to train the FCMs and adjust the weights of the concept interconnections.

In order to ensure the operation of the system, FCMs embody the accumulated knowledge and experience from experts who acquaint how the system operates under different conditions. For experts to conclude on which concepts to integrate, stakeholders give firstly feedback prior to knowledge extraction. Especially for decision policies made, this knowledge extraction is succeeded transforming all linguistic variables into numeric values via a defuzzification process. This produces a set of concepts denoted as C<sup>i</sup> (i = 1, 2, . . . n) (graph nodes) with their interrelations denoted as w<sup>i</sup> (graph directed edges). After the defuzzification, concepts are assigned a value within the range [0, 1] and weights are assigned values in the range [−1, 1] to capture negative and positive causality. A positive value of the weight wij indicates that an increase (decrease) in the value of concept C<sup>i</sup> results to an increment (decrement) of the concept's value C<sup>j</sup> . Similarly, a negative weight wij indicates that an increase (decrease) in the value of concept C<sup>i</sup> results to a decrement (increment) of the concept's value C<sup>j</sup> , while a zero weight denotes the absence of relationship between concepts C<sup>i</sup> and C<sup>j</sup> , respectively. **Figure 1** depicts a typical FCM.

Taking into consideration the interrelations between the concepts of a FCM, the corresponding adjacency matrix can easily be formed. Every concept C<sup>i</sup> in the graph has a value A<sup>i</sup> that expresses the quantity of its corresponding physical value derived after the defuzzification described above. The value A<sup>i</sup> of Ci is computed in each simulation step and it basically indicates the influence of all other concepts C<sup>j</sup> to C<sup>i</sup> (inference). The most popular inference rules are: (a) Kosko's inference, (b) Modified Kosko's inference, and (c) Rescale inference, as shown in the following three activation functions, respectively.

$$\begin{aligned} A\_l(k+1) &= f\left(\sum\_{j=1, j\neq l}^N \mathbf{w}\_{jl} \times A\_j(k)\right) \\ A\_l(k+1) &= f\left(A\_l(k) + \sum\_{j=1, j\neq l}^N \mathbf{w}\_{jl} \times A\_j(k)\right) \\ A\_l(k+1) &= f\left((2 \times A\_l(k) - 1) + \sum\_{j=1, j\neq l}^N \mathbf{w}\_{jl} \times (2 \times A\_j(k) - 1)\right) \end{aligned}$$

Also f(.) is the threshold (transformation) function that can be: (a) bivalent, (b) trivalent, (c) sigmoid or (d) hyperbolic, as shown in the following four equations, respectively.

$$\begin{aligned} f\left(\mathbf{x}\right) &= \begin{cases} I \ x > \mathbf{0} \\ \mathbf{0} \ x \le \mathbf{0} \end{cases} \\ f\left(\mathbf{x}\right) &= \begin{cases} I & \mathbf{x} > \mathbf{0} \\ \mathbf{0} & \mathbf{x} = \mathbf{0} \\ -I & \mathbf{x} < \mathbf{0} \end{cases} \\ f\left(\mathbf{x}\right) &= \frac{1}{1 + e^{-\lambda \mathbf{x}}} \\ f\left(\mathbf{x}\right) &= \tanh(\lambda \times \mathbf{x}) \end{aligned}$$

where λ is a real positive number (λ > 0) that determines the steepness of the continuous function f and x is the value Ai(k) on the equilibrium point. It should be noted that the sigmoid threshold function ensures that the calculated value of each concept will belong to the interval [0, 1]. When the values of concepts can be negative and their values belong to the interval [−1, 1], the hyperbolic tangent function can be used instead.

#### FCM Development Using Expert Knowledge

Experts can use their knowledge in the area under study to develop an FCM by firstly identifying the main concepts involved and secondly indicating the causal relationships among these concepts. The final step is the calculation of the causal relationships strengths using either crisp numeric values within the range [−1, 1] or using linguistic variables and values that at second stage are defuzzified into numeric. Furthermore, experts can improve an existed FCM by collectively analysing the key characteristics of the system under study and reevaluating the structure and the interconnections of the graph using fuzzy conditional statements or fuzzy rules. The algorithm used for the development of a FCM is depicted below (Groumpos and Anninou, 2017):

	- wij ≥ 0 means that an increase or decrease in concept C<sup>i</sup> causes the same result in concept C<sup>j</sup> .
	- wij ≤ 0 means that an increase or decrease in concept C<sup>i</sup> causes the opposite result in concept C<sup>j</sup> .
	- wij = 0 means that there is no relation between concepts C<sup>i</sup> and C<sup>j</sup> .

The degree of influence between the two concepts is indicated by the absolute value of wij. During the simulation, the value of each

concept is calculated using the following rule:

$$A\_{\vec{i}}(k) = f\left(k\_I A\_{\vec{i}}(k-1) + \sum\_{\substack{j=1,\,j\neq i}}^N \mathbb{w}\_{\vec{j}\vec{i}} \times A\_{\vec{j}}(k-1)\right),$$

#### Learning Process in a FCM

In general, the manual process for developing an FCM described in section FCM Development Using Expert Knowledge can be used only when there is at least one expert with expertise on the problem under study. However, in some situations, a FCM could not be constructed because of the following reasons:


For such extreme cases there has been developed a systematic way of constructing the FCM using a learning process, i.e., a process that will automatically determine the weights of the FCM, which best fits the decision-making and prediction problems. Diverse learning methods have been proposed by many researchers and they are based on the same techniques that used to train neural networks. These techniques are of three basic types namely: (a) Hebbian-based, (b) Population-based, and (c) Hybrid. Hebbian oriented learning requires data to be used alomg with a learning formula that regulates the adjustment of the cognitive map weights. The initial map before FCM learning in this case is established using experts' knowledge. Population-based learning algorithms on the other hand use evolution strategies to learn FCMs from data such as particle swam optimization (Koulouriotis et al., 2001), real-coded genetic algorithms (Stach et al., 2005), and the well-known big bang-big crunch algorithm (Yesil and Dodurka, 2013). Finally, hybrid learning approaches make use of both the effectiveness of Hebbian learning and the global search capability of evolutionary-based algorithms. Even though the hybrid approaches can learn FCMs with high accuracy, these approaches still suffer from the problem of converging to undesired states for the output concept values (Zhu and Zhang, 2008).

In the case of unsupervised learning algorithms, Hebbianbased methods use available data and a learning formula that is based on several modifications of Hebbian law, to iteratively adjust FCM weights. Typical Hebbian-based methods have been reported in the literature (Dickerson and Kosko, 1994; Huerga, 2002; Papageorgiou et al., 2003, 2004; Konar and Chakraborty, 2005; Stach et al., 2007, 2008).

In the case of population-based algorithms, such as simulated annealing, evolution-based and particle swarm optimization, the basic idea is to use the available input datasets in order to discover models that mimic the input data. It should be noticed that this method uses an objective criterion or a function to be optimized, thus making the method computationally intensive (for the genotype development and management). The primary goal is to find a near-optimal weight matrix based on the functional characteristics of the FCMs. Typical population-based methods have been published by many researchers (Koulouriotis et al., 2001; Papageorgiou et al., 2004; Mateou et al., 2005; Petalas et al., 2005; Stach et al., 2005; Alizadeh et al., 2007).

Finally, the hybrid method is implemented combining the previous two learning techniques using the coupling of differential evolution algorithm and nonlinear Hebbian learning algorithm, by using both the global search capabilities of evolutionary strategies and the effectiveness of the NHL rule (Papageorgiou and Groumpos, 2005). This hybrid learning module was applied successfully in real-world problems. Moreover, through the experimental analysis, the results showed that this hybrid strategy was capable to train FCM effectively, thus leading the system to desired states and determining an appropriate weight matrix for each specific problem (Zhu and Zhang, 2008).

### APPLICATION OF FCMS IN ANALYZING SOCIAL ACCEPTANCE FOR WBFS –STEADY AND DYNAMIC STATE ANALYSIS

#### Description of the Case Study

The knowledge relating to FCM was applied in a case study considering a hypothetical development of a WBF. This study was supposedly applied to the district of Thessaly, Greece. Thessaly region is located in the central part of Greece (Coordinates: 39.6103◦ N, 22.0476◦ E) and has one of the two largest plains in Greece covering a total area of 14,037 km<sup>2</sup> . Also, there exist four major cities generating a huge amount of municipal solid waste among other types of waste making the area an ideal candidate for developing a WBF. The hypothetical location chosen was near the existed one, i.e., near the Parapotamos of Tempi municipality. The reason the location chosen were nearby the real location was not to create any opposition and prejudice in answering questionnaires by the local stakeholders and residents. Furthermore, we assumed that the heat and cooling customer was a local community. The main goal for the study was to assess the community attitudes toward the development and management of a WBF, measuring the degree of this acceptability. Apart from the site selection we also tried to keep the other characteristics of the hypothetical study as close as possible to the real case and only altering the ones that deal with the social issues. That is, our polling methods avoided to deal with: (a) the selection of energy conversion technologies and processes, (b) the product management policies, (c) the technological improvement, and (d) the economic and environmental sustainability. Even though we included a plethora of concepts relating to environmental, health, governmental, economic and technological issues, we primarily concentrated on the social concepts affecting the acceptance proposition or opposition of WBFs.

In order to draw the FCM, a diverse team of stakeholders and experts were interviewed to express their perceptions toward the environmental, health, social, governmental, economic and technological aspects relating to the development and management of WBFs. The diversity of the group imposed the inclusion of stakeholders such as local citizens, consumers, competent authority's government employees, institution representatives, farmers etc. Two experts were also interviewed, one with socio-economic and the other with chemical process engineering expertise. The experts were showed similar research works on the subject such as of Lopolito et al. (2009) and Sacchelli (2014). This is the reason that some of the concepts involved in the creation of the FCM were similar to the ones used in these two works. The total number of stakeholders raised to 23 and complemented by the interviews of the two experts. In total, the experts have identified 43 concepts, as it is explained in the following subsection focusing on the social acceptance of WBF operations. We must say that the number 43 resulted after the merging of similar concepts that were practically the same but with different linguistic description/labeling. It should also be noted that some concepts were dominant for most of the stakeholders and experts, but others were depended of their area of interest and expertise. The age of the involved participants in the study was between 29 and 59 years. The weight of the answers was calculated differently for the stakeholders and the experts though. More specifically, 40% of the weighted average of the answers was due to experts and the rest of 60% to the stakeholders following a widely known practice in developing and drawing FCMs. Relatively to the interviewing process, at first we described to the stakeholders the most relevant concepts to WBF and also the causal relationships between concepts using mostly natural language, so they could understand, digest and share among themselves the new information. Once the participants were given an assessment example of a similar problem, they were able to assign negative or positive causality. The values assigned to causality (weights) were of the fuzzy range [very high, high, medium to high, medium, medium to low, low, very law] and similarly on the negative scale. After the collection of interviews, the results were defuzzified and entered in the online software "Mental Modeler" (2018)<sup>1</sup> to draw the FCM, which is depicted in **Figure 2**, and calculate the following:


### FCM Steady State Analysis

**Table 1** shows the general statistics of the FCM and **Table 2** the concept individualized properties described above. The complexity of the resulting FCM comes from the large amount of connections, but this is due to the nature of the concepts identified. It should be noted that we categorized these concepts into five classes, namely: (a) Environmental-Health, (b) Social, (c) Government-Laws, (d) Economics, and (e) Technological. An apparent similarity and increased interdependence of the concepts belonging to the same class can be observed. However, there is also an increased number of interconnections between concepts of different classes. This is because of the correlation of environment and health with societal issues and how these affect the decisions of local citizens when it comes to altering their lives.

It must be noticed that experts and stakeholders from the local community (citizens, consumers) referred to the social issues (Risk perception, Public Trust, Well-being of citizens, Inclusiveness) as the most crucial factors influencing the operation and management of such facilities, and regarded the technological and economics issues as secondary. On the other hand, experts and stakeholders coming from competent authorities such as government employees, institution representatives and chemical engineers gave extensive weight to the potential consequences of using WBF products and by-products and also to the resulting economic benefits (new investment and business plans, profit analysis for the

<sup>1</sup>www.mentamodeler.org (Accessed July 7, 2018).

TABLE 1 | General FCM statistics.


companies involved etc.). Respectively, the average number of connections was 158, ranging from 123 to 209 according to different participants. For the merged FCM shown in **Figure 2**, a density of 0.115725 is deduced, with average Connections per Component raised up to 4.860465.

Another structural measure of FCM is the Hierarchy Index which is defined by the following equation and depends on the out degree (od) of each concept (node) in an FCM of N concepts (MacDonald, 1983).

$$h = \frac{12}{\left(N - 1\right)N(N + I)} \sum\_{i} \left[\frac{\left.\sigma d\left(\mathbf{v}\_i\right) - \left(\sum \mathbf{o} d\left(\mathbf{v}\_i\right)\right)}{N}\right]^2$$

When h is equal to 1, the mapping is completely hierarchical, and when it is equal to 0, the mapping is completely democratic (Ozesmi and Ozesmi, 2004). The hierarchy index of the FCM was calculated to be 0.137, making it very close to 0 and therefore highly democratic. Of course there is diversification of the hierarchy index in the case of the individual FCMs as opposed to the merged one. Obviously, the collective case increases the number of nodes, as it affects the in-degree and out-degree of these nodes, making the FCM more democratic and its system's steady state more resistant to the alterations of individual concepts. The concept with the highest centrality, as expected, was the "Acceptance" with a very high score of 24.83. Furthermore, the most central concepts directly affecting the Acceptance concept were the following, in decreasing order of their complexity: WBF Product Certification 11.32, Risk perception 11.08, Equity of decision making processes 6.64, Uncertainty toward the settlement of the new industry 6.51 and Well-being of citizens 6.09. This depicts the direct association of the social issues (primarily) and the environmental issues (secondly) to affect this decision.

To the best of the authors' knowledge, there does not exist any relevant research that could establish the current state and values of the environmental, health, social, governmental, economic and technological oriented concepts involved in the creation of the FCM depicted in **Figure 2**. For this reason, we consider the steady state of the FCM model as the initial scenario to start our analysis. In order to make evaluations and draw conclusions, we examined the worst case and the best case scenario as compared to the steady state. The worst case scenario is set up with all driver concepts to have the value of 0.1. On the opposite side the best case scenario is set up with all driver concepts to have



the value of 1. In **Figure 3** we depict the analysis where in the worst case scenario a decrease of about 1% is shown in the "Social Acceptance" receiver concept when it is compared to the original steady state scenario which was set as the base for the analysis. In the same figure, we also observe the biggest decrease in the "WBF Product Certification" concept rising up to 18% with the "Intervention of non-local businessman" to follow having an overall 11% decrease. These two values show

that the relation of the driver concepts for this FCM to the "WBF Product Certification" and the "Intervention of nonlocal businessman" concepts is driven by what we call negative causality. The meaning of this great decrease of concept C7 and C14, respectively is that a small increase of the driver concepts causes the highest irrelevance to them. Therefore, these two concepts are the ones that they are the least affected in positive changes of the driver concepts in the FCM. The values of 18 and 11% decrease, respectively are not significant but the fact of the greater negatively causality is of great importance. The highest increases appear on the "Uncertainty toward the settlement of the new industry" and "Equity in sharing WBF product profit" scoring a 9 and 2% increase, respectively.

On the opposite side, the best case scenario is set up with all driver concepts to have the value near the value of 1. That means that we set up our model so that driver concepts are the ones that primarily affect the FCM as opposed to the rest of the ordinary concepts. Similarly in this case and after setting up all the driver concepts in the steady analysis to the value of 1 we observe the difference on the ordinary and receiver concepts giving special priority on the final receiver concept of the "Social Acceptance" (see **Figure 4**). We observe a 10% increase in the "Social Acceptance" receiver concept when it is compared to the original steady state scenario which was set as the base for the analysis. Also several ordinary concepts behave in a similar way with representative the "Intervention of non-local businessman" with an increase of 10% and the "Consumer/citizen awareness" with an increase of 9%. In the inverse side concept C15 in **Figure 4** which is the "Risk perception" is decreased as expected with a decrease of 12% and concept C24 (Uncertainty toward the settlement of the new industry) performs accordingly with a decrease of 7%. The significance of the highest decrease of concept C15 in this Figure is similar as in **Figure 3** for the negative causality between the driver concepts and the concepts C15 and C24, respectively. The highest the increase of the driver concept makes the concepts C15 and C24 to be insignificant for the social acceptance receiver concept. The values of the decrease are as much important as the fact of signifying the least relevant ordinary concepts C15 and C24 in the model. For this case also, it was expected that the biggest increase of the driver concepts will decrease the risk (for example the bigger the Mass media publicity the smaller the risk reception and similarly this holds for the rest of the driver concepts).

## Dynamic Scenario Analysis of the Social Issues

The size and complexity of the FCM conducted in sections Description of the Case Study and FCM Steady State Analysis however make the dynamic simulations needed for scenario analysis of the model less efficient. Such simulations on the extreme case (all concepts used as variables with semi-continuity on the values assigned between 0 and 1) run for long time periods to converge. Furthermore, the combinations of fixed value and variable value concepts are of exponential nature in this case, thus making us unable to explore the dynamic interactions between all concepts in the merged FCM. For this reason, we developed a second FCM based only on the concepts related to the social issues, trying to discover how only these aspects affect the social acceptance of WBFs. The graph achieved is shown in **Figure 5**, and from now on, it will be denoted as Social\_FCM.

For this case, the total number of concepts including "Social Acceptance" rises up to 18, with the total number of associations between concepts to be 60. For this case again, the only receiver concept is the "Social Acceptance," with 7 Driver components and 10 Ordinary. The hierarchy index for this case was calculated to 0.0023, making the FCM slightly "more democratic" than the previous one. The general statistics of the Social\_FCM are shown

FIGURE 4 | The effect of the driver concepts in the full FCM for the best case scenario compared to the steady state.

in **Table 3** and the degrees and centralities of all the concepts in **Table 4**.

For the resulted FCM, we run a significant amount of simulations to explore the nature of the associations between the related concepts in the map and the importance of one concept (defuzzified concept value). We focused on the convergence of the FCM after a certain number of iterations, using the clamping process as stated by Kosko (1986). According to clamping, a

#### TABLE 3 | General Social\_FCM Statistics.


TABLE 4 | InDegree, OutDegree, Centrality and Type of Concepts in the Social\_FCM.


subset of the concepts is selected to be studied, i.e., to change values as opposed to the rest, in order to understand their effect in reaching the equilibrium end states. We may use all the ordinary and driver concepts for this process, although the receiver concepts cannot be included. However, it is intuitively logical to clamp only the driver concepts, as they primarily participate on the influence of the social acceptance receiver concept. According to this process, each of the concepts is increased or decreased accordingly, and the resulting concept value vector is compared to the steady state vector of the FCM (Vasslides and Jensen, 2016). For our case, the changes are compared to the initial Social\_FCM steady state. **Figure 6** shows an indicative simulation of the Social\_FCM, with all 18 concepts ranging from 0 to 1 along with their final converged values, and the corresponding curves of concept activation levels per each iteration. Simulations are run for all cases of activation functions available in combination to the available transformation functions, as described with the equations in section Basic Concepts for FCMs. Furthermore, any other combination and mixture of variable and ordinary concepts can easily produce individualized effects of specific concepts to Social Acceptance.

Similarly to the previous subsection, we consider the steady state of the Social\_FCM model as the initial scenario to start our analysis. Again, in order to draw conclusions for the effect of the driver concepts to the receiver and ordinary concepts of the Social\_FCM, we examined the worst case and the best case scenario as compared to the steady state. The worst case scenario is set up with all driver concepts to have the value of 0.1 as in the full FCM use case before. On the opposite side, the best case scenario is set up with all driver concepts to have the value of 1. In **Figure 7**, we depict the analysis where in the worst case scenario a decrease of about 3% is shown in the "Social Acceptance" receiver concept when it is compared to the original steady state scenario that was set as the base for the analysis. In the same figure, we also observe the biggest decrease in the "Public Trust" concept rising up to 8%, with the "Equity of decision making processes" to follow having an overall 16% decrease. The highest increases appear on the "Risk perception," "Uncertainty toward the settlement of the new industry" and "Equity in sharing WBF product profit," recording a 16, 8, and 8% increase, respectively.

It must be noted that **Figure 7** does not depict absolute concept values, but only the increase or the decrease of these values relatively to the corresponding values of the steady state. For the case of the worst scenario explained previously, the concept values after convergence, as well as the curves of activation levels as they relate to the number of iterations, are shown in **Figure 8**.

On the opposite side, the best case scenario was run using the same activation function to be comparable with the previous scenario. For the best case, all seven driver concepts are set to value 1 and we study the decrease or increase of the ordinary and receiver concepts relatively to the steady state. It should be noted that, on **Figure 9**, we have: increase on the "Stakeholder participation," decrease of the "Risk perception," increase on "Public trust," decrease on "Bias," increase on "Equity of decision making processes," decrease on "Uncertainty toward the settlement of the new industry," decrease on the "Equity in sharing WBF product profit," and finally an increase on the "Social acceptance." It should also be noticed that these increases or decreases are relatively to the worst case scenario.

All increases and decreases are judged logical expect the concept of "Equity of decision making processes." This and only misbehavior of the FCM is minor for two reasons: firstly the concept affected is ordinary and secondly the receiver concept is not affected. Similarly, on **Figure 10** the curves of activation levels as they relate to the number of iterations are shown.

FIGURE 6 | Indicative dynamic simulation of the Social\_FCM with the converged values of the social concepts and the corresponding curves of concept activation levels per each iteration.

### CONCLUSIONS AND FUTURE CHALLENGES

Waste biorefinery facilities (WBFs) development and sustainable waste management are still in infancy in the Southern Europe countries, due to limited allocated assets given, resulting in ischemic infrastructure and maintenance facilities. On the other hand, waste management is an enormously critical aspect because of the environmental, economical, technological and social consequences. For that reason, such large scale operations necessitate the coalition of large social bodies, stakeholders of diversified interest and competent authority policy makers.

Our methodology was two fold in relation to identifying the most important aspects that affect the social acceptance of WBF. Firstly, a Fuzzy Cognitive Map (FCM) was developed, which captures the interoperation of environmental, health, social, technological and economic concepts that influence the social acceptance of WBFs. Since the complicated structure of the resulted FCM did not allow any definite conclusion regarding the most influential concepts to the social acceptance, in the second phase we modeled the impacts of just the social aspects to "social acceptance" for WBFs. For this purpose, of smaller FCM was used, to show treads and general directions in identifying the most determinant factors affecting the public opinion. Our analysis was also twofold: a) Steady State, where the FCM resulted as an amalgamation of stakeholder and expert knowledge, identifying five major classes of issues and b) Dynamic scenario analysis, after extracting only the social components, resulting in a much smaller FCM. For the second case, we were able to highlight the effect of the driver concepts, comparing the worst case and best case scenarios with the steady state analysis results. Various simulations were tried also, with a combination of activation and transformation functions selected as well as a variety of value levels between the worst and best case scenarios for a subset of ordinary and driver components.

More specifically, the analysis revealed that concepts such as "Uncertainty toward the settlement of the new industry," "Risk perception," "Stakeholder participation," "Consumer/Citizen awareness," "Equity of decision making processes" are the most influential factors to "Social Acceptance." This was what the dynamic scenario analysis has verified for the Social\_FCM part of the FCM that was originally developed. However, there are some shortcomings to this study, which should be taken into account. First, the representation of the model and its dynamic behavior was focused only on the social aspects. This happened because the preliminary analysis of the full FCM resulted in a complicated structure increasing tremendously the state space of the FCM. For that reason, around 25 issues/concepts were cut from the original FCM overcoming the inability to provide dynamic scenario analysis with so many concepts. This concept reduction process also diminishes several indirect interrelations of the social concepts with the concepts of the other classes (environmental, economic, technological etc.). Thus, these interrelations and their impacts are not taken into consideration and their result in increase or decrease of the "Social Acceptance" cannot be revealed in the second version of the FCM. Secondly, the impact of the number of experts/stakeholders deployed to the developing task of the FCMs was not examined. There are studies that highlight the fact that, the greater the expert body the more stable the version of the

#### REFERENCES


FCM produced. Also, the improvement of the conceptual model describing the WBFs social acceptance is shown to be dependent of learning methodologies used.

Based on our discussion above and the input received from stakeholders, there are several recommended areas of focus as industry moves toward the optimization of WBF application. However, the most important issue is to design a competitive waste-biorefinery concept by systematic participation of all the factors that affect the decision and policy making in WBF application. The technology knowledge must be transmitted to various stakeholders in the wider society and the final decision must integrate not only the technological and economic advantages but also health and social acceptance. For that reason, social acceptance must always be present alongside lifecycle assessment and economic value and environmental impact analysis tools for the WBF application. The bottom line is that it is critical to engage local communities in the development and management of WBFs in order to elicit the view of lay people, as a means to amend burdens relating to social factors and to avoid the "NIMBY" phenomenon. We believe therefore that the enlargement of the present study with further investigation in this research area can provide a valuable tool for competent authorities and policy makers in developing WBFs.

### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.


perspective. Glob. Environ. Chang. 12, 167–184. doi: 10.1016/S0959-3780(02) 00012-2


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Kokkinos, Lakioti, Papageorgiou, Moustakas and Karayannis. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.