AUTHOR=Yeomans Julian Scott , Kozlova Mariia 

TITLE=Extending system dynamics modeling using simulation decomposition to improve the urban planning process

JOURNAL=Frontiers in Sustainable Cities

VOLUME=Volume 5 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/sustainable-cities/articles/10.3389/frsc.2023.1129316

DOI=10.3389/frsc.2023.1129316

ISSN=2624-9634

ABSTRACT=Abstract. Urban planning problems frequently require the need for decision-making in situations containing considerable sources of uncertainty. Many social phenomena in this type of urban planning are modelled within system dynamics and/or multi-agent modelling frameworks. These approaches are commonly characterized by outputs in the form of a function of time. However, because standard sensitivity analysis techniques require the outputs to be scalar for the significance measures to be calculated, the majority of studies in this modelling domain lack sensitivity analysis and, consequently, important insights into model behaviour. Monte Carlo simulation methods have been used in a wide array of urban planning settings to incorporate uncertain features including time. Simulation-generated outputs are commonly displayed as probability distributions. Recently simulation decomposition (SimDec) has been used to enhance the visualization of the cause-effect relationships of multi-variable combinations of inputs on the corresponding simulated outputs. SimDec partitions sub-distributions of the Monte Carlo outputs by pre-classifying selected input variables into user-defined states, grouping combinations of these states into scenarios, and then collecting simulated outputs attributable to each multi-variable input scenario. Since it is a straightforward task to visually project the contribution of the subdivided scenarios onto the overall output, SimDec can reveal previously unidentified connections between the multi-variable combinations of inputs on the outputs. SimDec is generalizable to any Monte Carlo method with negligible additional computational overhead and, therefore, can be readily extended into any urban planning analysis that uses simulation models. This study demonstrates the efficacy of adapting SimDec for the sensitivity analysis of urban system dynamics modelling to improve the planning process.