SimFLEX: a methodology for comparative analysis of urban areas for implementing new on-demand feeder bus services

TL;DR

SimFLEX tackles pre-deployment evaluation of on-demand feeder buses in areas with uncertain demand by integrating microsimulation with ExMAS ride-pooling, MSA learning, and OTP routing in a two-loop framework. It generates multiple demand realizations from macro data and stabilizes traveler behavior to compute operational and system-wide KPIs that guide area selection. The Krakow case study shows area-specific differences, with Skotniki outperforming Bronowice on feeder attractiveness and added value, while Bronowice yields larger waiting-time reductions, and results remain robust across ASC variations. The open-source, modular framework provides policymakers with a practical, adaptable tool to forecast performance and compare candidate urban areas before launching new feeder services.

Abstract

On-demand feeder bus services present an innovative solution to urban mobility challenges, yet their success depends on thorough assessment and strategic planning. Despite their potential, a comprehensive framework for evaluating feasibility and identifying suitable service areas remains underdeveloped. Simulation Framework for Feeder Location Evaluation (SimFLEX) uses spatial, demographic, and transport-specific data to run microsimulations and compute key performance indicators (KPIs), including service attractiveness, waiting time reduction, and added value. SimFLEX employs multiple replications to estimate demand and mode choices and integrates OpenTripPlanner (OTP) for public transport routing and ExMAS for calculating shared trip attributes and KPIs. For each demand scenario, we model the traveler learning process using the method of successive averages (MSA), stabilizing the system. After stabilization, we calculate KPIs for comparative and sensitivity analyzes. We applied SimFLEX to compare two remote urban areas in Krakow, Poland - Bronowice and Skotniki - the candidates for service launch. Our analysis revealed notable differences between analyzed areas: Skotniki exhibited higher service attractiveness (up to 30%) and added value (up to 7%), while Bronowice showed greater potential for reducing waiting times (by nearly 77%). To assess the reliability of our model output, we conducted a sensitivity analysis across a range of alternative-specific constants (ASC). The results consistently confirmed Skotniki as the superior candidate for service implementation. SimFLEX can be instrumental for policymakers to estimate new service performance in the considered area, publicly available and applicable to various use cases. It can integrate alternative models and approaches, making it a versatile tool for policymakers and urban planners to enhance urban mobility.

Figures (7)

• Figure 1: SimFLEX computes service performance with the following methodology. For a given service area and hub location it uses widely available inputs (such as network graph, GTFS, population distribution and OD-matrices, detailed in \ref{['subsec:method_workflow']}) it runs a series of microsimulations to obtain a wide range of performance indicators. First, we sample microscopic demand patterns for services from macroscopic models (Section \ref{['subsec:method_demand']}). For each single demand realization, we simulate the travelers learning process (detailed in \ref{['subsec:method_msa']}), when they experience system performance (with unknown travel times due to detours, here sampled with ExMAS). After stabilization (when each travelers expectations meet the realizations) we simulate extra runs to compute indicators from the stabilized system (detailed in section \ref{['subsec:method_outputs']}). This concludes a single run of SimFLEX, which can then be replicated (for different realizations of the demand), or used for comparisons (between areas, hubs, parameterizations, etc.) as discussed in Section \ref{['sec:results']}.
• Figure 2: Two candidate areas in Krakow (Poland): Bronowice (top-left) and Skotniki (bottom-left) as candidate locations for on-demand feeder bus implementation, with address points (green dots) and public transport hubs (red circles) marked. The proposed feeder system will function on demand, allowing residents to request rides via a mobile application. Small 6-seater buses will pick them up from designated stops within their district and transport them to the respective public transport hub (Bronowice Małe for Bronowice, and Czerwone Maki P+R for Skotniki) for onward travel on Krakow's tram and bus lines.
• Figure 3: Stabilization of the average expected travel times through MSA over 30 iterations for Bronowice and Skotniki areas. The initial travel times at the first iteration correspond to raw travel times estimated for solo rides, before any learning or adaptation occurs. Travel times initially vary as travelers adapt to the new feeder bus system, but gradually stabilize as the system converges. Both areas show a similar trend of initial variation followed by convergence, though the specific travel times and rates of convergence differ. The dashed lines represent the mean value of the convergence iteration, for Bronowice in blue and for Skotniki in orange, indicating the average convergence iteration around 16 (the mean over 100 demand replications).
• Figure 4: The probability distribution of feeder bus service choice in Bronowice and Skotniki for a single demand replication. The figure shows that in Bronowice, slightly more than half of travelers (57%) and in Skotniki roughly half (51%) have almost 0% probability of choosing the feeder. Meanwhile, a tenth of travelers (10%) in Bronowice and approximately a sixth (17%) in Skotniki have a 100% probability of choosing the feeder. The mean probability of feeder choice is 0.23 for Bronowice and 0.35 for Skotniki, as indicated by the dashed lines. These values highlight differences in the feeder service adoption between the two areas in this single demand scenario.
• Figure 5: The distribution of integrated feeder system KPIs across the 100 demand replications with their mean values (illustrated with dashed lines) for Bronowice and Skotniki areas. Skotniki generally exhibits higher (less negative) values for feeder attractiveness ($\Delta A$), suggesting it is perceived as more attractive compared to Bronowice. In terms of waiting time reduction ($\Delta T$), Bronowice shows a distribution concentrated at higher positive values, indicating a more significant reduction in waiting times. For added value ($\Delta V$) Skotniki has a higher mean value and a narrower data spread, implying a slightly better added value compared to Bronowice, which shows a similar spread but with slightly lower mean values. Finally, the histograms of the probability of choosing the feeder ($P_F$) indicate that Skotniki has a distribution skewed towards higher probabilities, suggesting a greater likelihood of choosing the feeder service compared to Bronowice.