Table 1 and Figure 1 show the number of small WWTPs in Upper Austria. Overall, the total number of plants increased by about 130 between 2016 and 2019. Currently, there are more than 2′500 small WWTPs in operation. Among them, there are about 870 SBR plants, 630 CAS plants and 535 TWs. There are still about 300 small WWTPs that are only primary treatment plants. These plants are three-chamber septic tanks and no longer state-of-the-art. However, they were implemented before the nitrification requirement was introduced in the 1990s and have a long-term permits. Authorities aim to reduce the number of non-state-of-the-art plants with time. Between 2016 and 2019, the number of plants with only primary treatment was reduced by 80 plants. Other technologies implemented with smaller numbers are trickling filter (about 100 plants), RBC plants (37), MBR plants (26) and filtration systems (27). Nowadays, the most popular technologies for small WWTPs are SBR plants and VF wetlands, with 170 and 60 implementations between 2016 and 2019, respectively. CAS plants and trickling filters were among the first technologies implemented in the 1990s (Langergraber et al., 2018) and are decreasing in number. This might be attributable to replacement by other technologies or to the fact that some small WWTPs are closed down once public sewer networks are expanded.

Based on the capacity of the receiving waters, authorities can require more stringent effluent concentrations for small WWTPs. If this is the case, a polishing stage, most of the time a VF bed following the technical system, is applied. Thus, for the evaluation of the treatment performance, not only the main treatment stage is considered. 33% of all SBR plants (290 out of 873) and 9% of CAS plants (54 of 628) are implemented with a VF bed for polishing (Figure 2).

Tables 2, 3 show the BOD₅ and COD effluent concentrations for different technologies, respectively. All technologies can achieve low median values of BOD₅ effluent concentrations (below 10 mg BOD₅/L): the number of exceedances of the BOD₅ threshold value are very low (below 10 for each technology). The lowest median values of COD effluent concentrations are found for VF wetlands, SBR and CAS plants with VF beds as polishing stage and MBR plants. For these technologies, median values were between 20 and 25 COD mg/L, for all other plant types between 35 and 45 COD mg/L. Most values exceeding the threshold are found for SBR and CAS plants, trickling filters and RBC plants.

A similar result can be found for NH₄-N as shown in Table 4. The highest number of exceedances of limit values was measured in soil filter systems (11.1%) and in trickling filter and submersible systems (each approx. 6%), while the lowest number of exceedances of limit values was noted in SBR systems with a downstream planted soil filter (0.8%) and in constructed wetlands (1.5%). Systems with a downstream planted soil filter or with a planted soil filter as the main treatment stage (constructed wetlands) also had the lowest median value of NH4-N effluent concentration. The median value was 0.24 mg/L for activated sludge plants with a downstream planted soil filter, 0.37 mg/L for constructed wetlands, and 0.68 mg/L for SBR plants with a downstream planted soil filter. MBR plants had a significantly higher median NH4-N effluent concentration of 1.78 mg/L than, for example, SBR plants and activated sludge plants using the continuous flow principle with 1.00 and 1.09 mg/L, respectively.

Figure 3 shows a violin plot of NH₄-N effluent concentrations for various technologies. A lot of the near-zero values were measured particularly in VF wetlands and both CAS and SBR plants with VF beds. These are also the three technologies with the lowest median values and the lowest lower quartiles. Most values measured for these technologies are clustered below 2.5 NH₄-N mg/L. The higher median value of SBR plants with VF bed can be seen in the violin plot as the second bump in the density shape around the upper quartile. In contrast, the data for trickling filters and filtration plants are quite evenly distributed between very low effluent concentrations and the threshold.

Figure 4 shows the correlation of NH₄-N effluent concentrations with wastewater temperature. In the case of VF wetlands, the dependence of the treatment performance on temperature is less pronounced, which in turn indicates a more stable and robust operation of VF wetlands compared to technical wastewater treatment plants (SBR, CAS, trickling filters and RBC). In SBR and in CAS plants, the NH4-N effluent concentrations are only slightly higher than in VF wetlands; in trickling filter and RBC plants, somewhat higher NH4-N effluent concentrations are generally to be expected. When a polishing VF wetland is used, the influence of temperature on the treatment performance is almost eliminated.

Table 5 shows the size of the dataset available and the median values of the COD effluent concentrations for four technologies depending on the age of the system. The oldest plant in this list is a trickling filter plant that was built 29 years ago. However, only a few trickling filters have been built in the last 10 years. In contrast, there are many SBR plants (with and without VF beds) and VF wetlands that were built during this period. Thus, there are many values from external monitoring reports of young plants. In the following diagrams, median values were calculated only in cases where values from more than five plants were available, in order to avoid outliers due to individual plant performance.

Figures 5, 6 show the COD effluent concentrations of different technologies depending on the age of the plant, whereby Figure 6 compares the technologies with VF beds. VF wetlands and MBR plants showed the lowest COD effluent concentrations. Higher COD effluent concentrations were generally measured for RBCs and trickling filter plants. However, no significant deterioration in treatment performance over time was observed for any technologies. There is a tendency for the median values to fluctuate more from 1 year to the next as the plant ages. The fluctuations of the median values tend to be higher for technical plants when compared to VF wetlands. Larger fluctuations could be an indication of less stable performance, but can probably also be attributed to the smaller number of older plants.

Figure 6 compares SBR and CAS plants with and without VF beds and VF wetlands. Technical wastewater treatment plants with VF beds produce COD effluent concentrations similar to VF wetlands. Thus, the additional treatment step after the technical system significantly increases the treatment performance in terms of lower COD effluent concentrations and more stable treatment performance over time. Again, a greater fluctuation of the median values can be observed as the age of the plant increases. The behaviour of the BOD₅ effluent concentrations is similar to that of the COD effluent concentrations (not shown).

Figures 7, 8 show the same data for the comparison of the median values of the NH₄-N effluent concentrations. The comparison of the different plants yields results similar to the COD values (Figures 5, 6). Technical wastewater treatment plants show greater fluctuations in the median values of the NH4-N effluent concentrations in subsequent years when compared to VF wetlands. On the one hand, this indicates that nitrification is the most sensitive process in wastewater treatment plants. On the other hand, the robustness of VF wetlands is clearly evident, especially with regard to reliable and robust nitrification performance. CAS and SBR plants have lower NH₄-N effluent concentrations and less fluctuations comparted to other technical systems. Higher NH₄-N effluent concentrations of MBR plants compared to CAS and SBR plants are in contrast to the results for COD where MBR plants show better performance. The higher COD removal of MBR plants can be explained due to the removal of particulate COD during membrane filtration. For trickling filters, the high fluctuations of NH₄-N effluent concentrations indicate less stable nitrification performance when compared to their COD treatment efficiency. With an additional VF beds after SBR and CAS plants, NH₄-N effluent concentrations of SBR and CAS plants can be improved and similar performance of VF wetlands can be reached.

Table 6 shows the results of the treatment performance analysis (using NH₄-N effluent concentrations) over the operation time of every single treatment plant. In total, data from 493 SBR plants were available; and of these, 369 plants had five or more measured values included in the analysis. For only 5% of the SBR plants, a trend in the treatment performance could be observed, whereby about half of these 5% had a positive trend (i.e., increasing effluent concentrations and thus deteriorating treatment performance over time) and the other half a negative trend (i.e., decreasing effluent concentrations and thus improving treatment performance over time). For all technologies with a higher number of implementations (SBR, SBR + VF bed, CAS and VF wetland), about 10% of the single plants evidenced a trend within a 95% significance level. These trends were both negative and positive, with a tendency toward higher percentages of negative trends (improving treatment performance) for VF wetlands. However, due to the low percentage of single plants showing a clear trend, this behavior is more likely attributable to specific aspects of these single plants than to a general trend for different technologies.

Karczmarczyk et al. (2021) evaluated the performance of compact plants in Poland. For 2019, they presented a number of more than 230′000 implemented systems in Poland. For their study, they selected 23 plants from a list of municipalities that received funding for construction from a Regional Fund. Effluent samples from these 23 plants (including CAS and SBR plants as well as fixed-bed reactors) have been taken and analysed for BOD₅ whereby more than 50% of the plants did not fulfil the Polish effluent standard of 40 mg BOD₅/L. Based on their measurements, Karczmarczyk et al. (2021) estimated the environmental impact an annual costs of on-site wastewater treatment systems for Polish conditions.

In the study of Hellström and Jonsson (2004), 14 single-house treatment plants using different technologies have been investigated in Sweden. Systems tested include package plants, urine separating systems, blackwater separating systems and chemical precipitation as a supplement to sand filter beds. In addition to removal of organic matter, also nitrogen and phosphorus removal was targeted. Weiss et al. (2008) performed an environmental systems analysis of four on-site wastewater treatment options with focus on phosphorus removal under Swedish conditions. However, they did only use literature data for the performance of the systems.

Kegebein et al. (2007) evaluated the performance of four technologies (i.e., trickling filter, RBC, CAS and a moving bed biofilm system) using measured data from lab-scale experiments with defined loading of the systems. Main objective of their study was to test the behaviour of the plants during and after non-loading periods.

Performance of on-site wastewater treatment systems in Western Australia has been evaluated by Gunady et al. (2015). However, this evaluation was not based on measured quality data but on the evaluation of the reasons for non-performance of the systems. Identified reasons were inadequate installation, inadequate maintenance, poor public awareness, insufficient local authority resources, ongoing wastewater management issues, or inadequate adoption of standards, procedures, and guidelines.

Studies on the performance of small WWTPs ≤50 PE, on-site and household wastewater treatment systems are very scarce. Compared to our study, the number of measured field data used for the performance evaluation in other studies was very small (e.g., Hellström and Jonsson, 2004; Karczmarczyk et al., 2021). Other studies such as Kegebein et al. (2007) aimed to monitor the treatment performance during specific loading scenario, i.e. non-loading periods.

Reasons for non-functioning plants in Western Australia (Gunady et al., 2015) have been addressed in the Austrian case by developing design standard, proper management of small WWTPs on the authority side (e.g. permits required for every plant), proper monitoring schemes (self and external monitoring) as well as the training courses for operators od small WWTPs (Langergraber et al., 2018).