Optimizing Demand-Responsive Transport (DRT) System with Predictive Modeling

AI Quick Summary

We want to validate and extend a predictive model for identifying areas where Demand-Responsive Transport (DRT) services can be successfully implemented and integrated with conventional public transport (railway, bus lines). Through a visual tool, this model allows both Public Authorities to identify regions for possible DRT applications and PT Operators to evaluate their convenience to operate DRT in a specific region. The model will be tested in a Swiss region and validated through simulation (MATSim). On-demand transportation in rural areas offers a flexible, efficient, and environmentally friendly solution to address the challenges of limited public transit options. On one hand, it provides better accessibility for customers and on the other hand, it is economically advantageous for the operators compared to a classic frequency-based PT line. To achieve these advantages, at the moment a complex, largely manual, planning of the on-demand offers is necessary.This project is dedicated to pre-select areas where Demand-Responsive Transport (DRT) services could have a higher impact. The selected area will be then validated using MATSim, a powerful transportation simulation software. The core objective is to optimize DRT planning by developing and applying a predictive model that identifies geographic areas where DRT can be successfully applied, thereby reducing the computational and time efforts normally required both for planning and for implementation of the DRT service.We build on previous work together with PostAuto AG, where the initial regression model and pre-simulation filter were developed and applied to a case study with real data from the Weinfelden district, Canton of Thurgau (a scientific publication of this work is in preparation). This project will validate and extend the models and results through a real test case in a different Swiss canton. MATSim will be used to simulate the impact of DRT services in the identified areas, providing detailed insights about how such services can be optimized for maximum efficiency and sustainability. By accurately predicting the regions where DRT implementation will be most effective, the project will contribute to a reduction in CO2 emissions and support Switzerland’s 2050 climate strategy and energy transition goals.Public Authorities and Public Transport companies have common interests in eliminating inefficiency in frequency-based bus lines (e.g. low demand) and substituting them with DRT services. Moreover, there is a potential demand for DRT service in those areas where the car is seen as the only option (e.g. bus stop too far); with its flexible routing, DRT can provide efficient transport to all the people's categories, which are "imprisoned" by cars and would like to move in an affordable and ecological way. A tool that allows identifying where and how to integrate the Public Transport offer would be of high interest both for PA (e.g. cantons) and for operators (e.g. PostAuto). PA, with the same invested money, would be able to maximize the modal split impact and reduce the vehicle-kilometers traveled in the canton. Operators would be able to maximize their profit and reduce the time-to-market required to assess and make simulations on all the many different alternative areas where the service could be activated.The project has four main contributions:Pre-simulation filter using the developed predictive model: The application of the predictive models will serve as a pre-simulation filter that reduces the amount of necessary MATSim simulation to get the areas for a successful DRT service implementation. These models are based on population density, line utilization, accessibility, area, and current modal split data as inputs. The expected outputs of the model are: Difference in modal split after implementing DRT service, DRT Total rides, PT total rides, Distance Efficiency Ratio (Total Passenger Distance Traveled/total Distance covered by vehicles), Loss/Profit. To do this, we will perform an in-deep data analysis of the geographic characteristics of the areas.Impact Assessment for a different region in Switzerland: Utilizing MATSim to simulate and analyze the impact of potential DRT implementations in the selected areas, it will be possible to validate the decisions based on the predictive model previously used as a filter. One region of interest is the district of Frauenfeld (proposal of project partner PostAuto AG).Extension of the predictive model: The newly performed simulations will validate the predictive model for DRT performance based on geographic characteristics. These will validate the model’s reliability and will allow to enrich it with data related to the newly analyzed areas.Provide a visualization platform (software) that operators can use to apply this methodology within their areas of interest.Regarding point 4, we need to pay an external developer to realize the visualization tool. This would allow us to position on the market as DRT planning consultants for operators and PAs, allowing them to use our platform to reach their targets. The funds coming from the Innovation Booster Mobility Lab will be therefore used for this activity (i.e. point 4).This project directly aligns with the goals of the Innovation Booster New Mobility Lab, which seeks to foster systemic mobility solutions that advance Switzerland’s sustainable transport objectives. By refining and validating a tool that aids in the strategic deployment of DRT services, the project will not only improve public transport efficiency but also enhance accessibility in underserved areas, contributing to a broader shift towards sustainable urban mobility.The foundation for this project was established through research and simulations in the region of Weinfelden, where MATSim was used to assess DRT service areas. This initial application demonstrated the model's effectiveness in optimizing public transport offer and reducing the need for private vehicles. Lessons learned will include the importance of precise data integration and the adaptability of the model to different regional contexts.Moving forward, the project will focus on validating this approach in another Swiss region, using MATSim to simulate and refine the predictive model based on diverse geographic and demographic inputs. Close collaboration with local municipalities and transport operators will be essential to gather data, test the model, and implement pilot projects.To ensure the success of the project, it will be important to have:Data Access and Pilot Testing: Collaboration with local municipalities and cantonal authorities to obtain geographic and transport data and to conduct pilot testing in new regions. We have already contacts with the Kanton Baselland.Partnerships with operators: Already collaborating with PostAuto.Academic Support: Further research activities, particularly in refining the predictive model and validating outcomes through MATSim simulations.The New Mobility Lab can provide vital support by offering financial resources, facilitating connections with relevant stakeholders, and providing methodologies to enhance the project’s implementation and impact. With the backing of the New Mobility Lab, this project is poised to make a significant contribution to the sustainable transformation of Switzerland’s mobility sector.

Project Idea Metadata

Project Idea Name: Optimizing Demand-Responsive Transport (DRT) System with Predictive Modeling
Date: 8/20/2024 3:46:43 PM
Administrators:
- Corrado Muratori

Project Idea Description

On-demand transportation in rural areas offers a flexible, efficient, and environmentally friendly solution to address the challenges of limited public transit options. On one hand, it provides better accessibility for customers and on the other hand, it is economically advantageous for the operators compared to a classic frequency-based PT line. To achieve these advantages, at the moment a complex, largely manual, planning of the on-demand offers is necessary.

This project is dedicated to pre-select areas where Demand-Responsive Transport (DRT) services could have a higher impact. The selected area will be then validated using MATSim, a powerful transportation simulation software. The core objective is to optimize DRT planning by developing and applying a predictive model that identifies geographic areas where DRT can be successfully applied, thereby reducing the computational and time efforts normally required both for planning and for implementation of the DRT service.

We build on previous work together with PostAuto AG, where the initial regression model and pre-simulation filter were developed and applied to a case study with real data from the Weinfelden district, Canton of Thurgau (a scientific publication of this work is in preparation). This project will validate and extend the models and results through a real test case in a different Swiss canton. MATSim will be used to simulate the impact of DRT services in the identified areas, providing detailed insights about how such services can be optimized for maximum efficiency and sustainability. By accurately predicting the regions where DRT implementation will be most effective, the project will contribute to a reduction in CO2 emissions and support Switzerland’s 2050 climate strategy and energy transition goals.

Public Authorities and Public Transport companies have common interests in eliminating inefficiency in frequency-based bus lines (e.g. low demand) and substituting them with DRT services. Moreover, there is a potential demand for DRT service in those areas where the car is seen as the only option (e.g. bus stop too far); with its flexible routing, DRT can provide efficient transport to all the people's categories, which are "imprisoned" by cars and would like to move in an affordable and ecological way. A tool that allows identifying where and how to integrate the Public Transport offer would be of high interest both for PA (e.g. cantons) and for operators (e.g. PostAuto). PA, with the same invested money, would be able to maximize the modal split impact and reduce the vehicle-kilometers traveled in the canton. Operators would be able to maximize their profit and reduce the time-to-market required to assess and make simulations on all the many different alternative areas where the service could be activated.

The project has four main contributions:

Pre-simulation filter using the developed predictive model: The application of the predictive models will serve as a pre-simulation filter that reduces the amount of necessary MATSim simulation to get the areas for a successful DRT service implementation. These models are based on population density, line utilization, accessibility, area, and current modal split data as inputs. The expected outputs of the model are: Difference in modal split after implementing DRT service, DRT Total rides, PT total rides, Distance Efficiency Ratio (Total Passenger Distance Traveled/total Distance covered by vehicles), Loss/Profit. To do this, we will perform an in-deep data analysis of the geographic characteristics of the areas.
Impact Assessment for a different region in Switzerland: Utilizing MATSim to simulate and analyze the impact of potential DRT implementations in the selected areas, it will be possible to validate the decisions based on the predictive model previously used as a filter. One region of interest is the district of Frauenfeld (proposal of project partner PostAuto AG).
Extension of the predictive model: The newly performed simulations will validate the predictive model for DRT performance based on geographic characteristics. These will validate the model’s reliability and will allow to enrich it with data related to the newly analyzed areas.
Provide a visualization platform (software) that operators can use to apply this methodology within their areas of interest.

Regarding point 4, we need to pay an external developer to realize the visualization tool. This would allow us to position on the market as DRT planning consultants for operators and PAs, allowing them to use our platform to reach their targets. The funds coming from the Innovation Booster Mobility Lab will be therefore used for this activity (i.e. point 4).

This project directly aligns with the goals of the Innovation Booster New Mobility Lab, which seeks to foster systemic mobility solutions that advance Switzerland’s sustainable transport objectives. By refining and validating a tool that aids in the strategic deployment of DRT services, the project will not only improve public transport efficiency but also enhance accessibility in underserved areas, contributing to a broader shift towards sustainable urban mobility.

The foundation for this project was established through research and simulations in the region of Weinfelden, where MATSim was used to assess DRT service areas. This initial application demonstrated the model's effectiveness in optimizing public transport offer and reducing the need for private vehicles. Lessons learned will include the importance of precise data integration and the adaptability of the model to different regional contexts.

Moving forward, the project will focus on validating this approach in another Swiss region, using MATSim to simulate and refine the predictive model based on diverse geographic and demographic inputs. Close collaboration with local municipalities and transport operators will be essential to gather data, test the model, and implement pilot projects.

To ensure the success of the project, it will be important to have:

Data Access and Pilot Testing: Collaboration with local municipalities and cantonal authorities to obtain geographic and transport data and to conduct pilot testing in new regions. We have already contacts with the Kanton Baselland.
Partnerships with operators: Already collaborating with PostAuto.
Academic Support: Further research activities, particularly in refining the predictive model and validating outcomes through MATSim simulations.

The New Mobility Lab can provide vital support by offering financial resources, facilitating connections with relevant stakeholders, and providing methodologies to enhance the project’s implementation and impact. With the backing of the New Mobility Lab, this project is poised to make a significant contribution to the sustainable transformation of Switzerland’s mobility sector.