The “Minibus” Contribution

Paratransit is an informal, market-oriented, self-organizing public transport system. Despite the signi cance of this transport mode, it is mainly unsubsidized, relying on collected fares. Paratransit systems can be categorized by route pattern and function, by driver organization, type of stops and fare type. Most case studies covered by the Neumann (2014) thesis indicate that paratransit services are mainly organized as route associations operating 8-15 seater vans on xed routes. Most of the services run in direct competition to a public transport system belonging to a public transit authority. Such a service—minibuses with xed routes, but without xed schedule—is o en called a jitney service. The minibus module of MATSim is based on the most common characteristics, with the understanding that the jitney/minibus service is only one of many possible paratransit services.

The minibus model is integrated in the multimodal multi-agent simulation of MATSim. In the model, competing minibus operators begin to explore the public transport market, o ering their services. With more successful operators expanding and less successful operators going bankrupt, a sustainable network of minibus services evolves. In Neumann (2014), the model is veri ed through multiple illustrative scenarios, analyzing the model's sensitivity to di erent demand patterns, transfers, and interactions of minibuses and a formal operator's xed train line.
The minibus model can be applied to two di erent transport planning elds. First: in the simulation of real paratransit targeting the inner workings of di erent paratransit stakeholders' relationships, the model can create "close-to-reality" minibus networks in a South African context. Neumann et al. (2015) gives an in-depth presentation of the module application and South African paratransit in general. Given the informal and emergent nature of minibus paratransit in developing countries, routes, schedules and fares are usually not published; they can only be captured in the tacit knowledge of operators and frequent users. Applying the minibus model has proven valuable in gaining a better understanding of how routes evolve. Instead of imposing routes and schedules on the MATSim model, as is usually the case for formal transit, the modeler observes and gets the paratransit routes as an output from the model. As each operator aims to maximize their pro t, the resulting network o en favors the operators' business objectives, instead of the connectivity and mobility of the mode's users. This model feature accurately captures route-forming behavior in the South African case, where commuters are o en required to take multiple, longer trips instead of direct trips.
Second, the same model provides a demand-driven approach to solving a formal transit authority's network design problem; it can be used as a planning tool for the optimization of single transit lines or networks. For more details on the second form of application, see Section 17.3.
For further reading: Neumann (2014) provides an understanding of the underlying principles of paratransit services, namely minibus services, its stakeholders, fares, route functions, and patterns. Furthermore, it contains an in-depth description of the minibus model, its theoretical background, and its application to illustrative scenarios, as well as real world examples. The website of MATSim also hosts latest implementation documentation at http://matsim.org/doxygen.

Network Planning or Solving the Transit Network Design Problem with MATSim
A public transport system's success depends primarily on its network design. When transport companies try to optimize a line using running costs as the main criteria, they quickly nd that demand must be taken into consideration. The best cost structure is unsustainable if potential customers leave the system and opt for alternatives, like private cars. The basic problem to solve: nd sustainable transit lines o ering the best possible service for the customer. More speci cally, • the customer's demand side asks for direct, uncomplicated connections, and • the operator's supply side asks for pro table lines to operate.
Informal public transit systems around the world, o en referred to as paratransit, are examples of market-oriented, self-organizing public transport systems. For an in-depth coverage of paratransit, see Section 17.2, with references. Despite the signi cant and increasing importance of this transport mode, it is mainly unsubsidized and relies only on collected fares. Thus, the knowledge of paratransit-and its ability to identify and ll market niches with self-supporting transit services-provides an interesting approach to solving a formal public transit company's network design problem.
The "Minibus" Contribution 113 The minibus module of MATSim provides a demand-driven approach to solving a formal transit authority's network design problem; it can be used as a planning tool for the optimization of single transit lines or networks. In the Neumann (2014) thesis, the model was applied to two different planning problems of the Berlin public transit authority BVG (Berliner Verkehrsbetriebe).
In the rst scenario, the model constructed a transit system, from scratch, for the district of Steglitz-Zehlendorf. The second scenario analyzed the Tegel airport closure impact on BVG's bus network. Apart from Tegel itself, the rest of the bus network was una ected by the airport closure. The resulting minibus model transit system resembled the changes BVG had scheduled for Tegel's closure.
In conclusion, the minibus model developed in the thesis automatically adapted supply to demand. The model not only grew networks from scratch, but also tested an existing transit line's sustainability and further optimized the line's frequency, time of operation, length, and route. Again, the optimization process was fully integrated into the behavior-rich, multi-agent simulation of MATSim, re ecting passenger reactions, as well as those from competing transit services and other road users. Thus, the minibus model can be used, along with more complex scenarios, like city-wide tolls or pollution analyses.