IDS Seminar on ‘Paratransit Demand Modelling - Activity Based Approach’
IDS held an online seminar themed ‘Paratransit Demand Modelling - Activity Based Approach’ on 8th February 2024 from 2 pm - 3.30 pm EAT.
The Presenter was Owen Mwaura, an Urban Transport Planner and GIS Analyst pursuing a Ph.D. in Transportation Planning at The University of Cape Town (UCT) with a focus on sub-Saharan African cities' informal transport.
The Discussant was Cecília Pereira, a Ph.D. student in industrial engineering at the Federal University of Itajubá researching last-mile delivery solutions in Brazilian favelas.
Caroline Matara, a transportation engineer and lecturer at the Technical University of Kenya.
Main Points from Owen Mwaura’s Presentation on ‘Paratransit Demand Modelling-Activity Based Approach’
- The activity-based approach is a new paradigm that has been utilized by traffic and transport engineers over the last two decades to come up with better ways of representing transport. This approach has become popular within the last decade because of advancements in computation power.
- Travel behavioral models are built on the assumption that people have patterns. Therefore, when modelling travel behavior, a modeler intends to model that pattern. The individuals could be grouped by their socio-economic differences, their needs, and their activities,
- The basic assumption of the travel demand is that demand is derived which means it is generated by the need for people to satisfy certain wants or desires in their lives. Therefore, being on a transport system is not a means to an end for many individuals, it is something that they use to get them to the next location (activity).
- With the emerging practice of transportation planning, the activity models have become important. These models are designed to simulate activities depending on when they are conducted, where, how long they take, and with whom, and finally, the travel choices between those activities. This gives the planners a closer reality of the needs of the commuter.
Context of Nairobi
- The population of Nairobi has grown exponentially over the past decades. With the growing population, the demand for transportation in the City has also grown. Therefore, the present transportation system in Nairobi which was designed for a smaller population, is inadequate to meet the mobility demands.
- The prop-up of informal networks (matatus) to satisfy the mobility needs of Nairobi. The informal networks are popular due to their affordability and availability. Some of the challenges of these informal networks are unhealthy competition due to lack of regulation, they lack capacity due to poor maintenance, labor abuse, and inadequate policy and regulations.
- Owen’s question is whether to use the activity-based (simulation) model to design an informal network that could be an optimal network for the stability of Nairobi.
How the Simulation (Activity-Based) Model Works
- Disaggregate demand: to represent an individual on the model, the modeler takes the individual’s origin and destination coordinates, the time spent in each activity, the time spent in transit, and the mode of travel used.
- Sources of data:
- JICA Study done in 2013
- Iterative proportion fitting from Travel Survey data
- The simulation framework is known as mobsim and it has different modules which work in iterations. After developing the activity chains from the initial demand, the modeler needs to put those activity chains into a simulation that uses the Q-based system.
- Within the mobility simulation, one has to have the input of the infrastructure, schedule for the paratransit, and the vehicles.
- Scoring: The modeler can score the behavior of a certain agent cumulatively before they start a trip. For instance, if a matatu was full in the mobility simulation and the individual had to wait for another matatu, the waiting time is scored negatively in the system because they are not achieving anything. Similarly, if the individual is at work (activity), the score is positive since they are achieving something etc.
- Re-planning: At this stage, evaluation is done to determine what needs to be changed in the system to improve the scores e.g. the route length, route licensing, operating hours etc.
Limitations of the Mobility Simulation
- High computational intensity
- Data requirements
- Complexity and accessibility
- Computational intensity
- Model sensitivity
Comments from Cecilia Pereira - Discussant
- The simulation model developed by Owen can be replicated in other countries as well, not just Kenya.
- How possible is it to obtain data that is representative of the entire population?
- How did Owen run his simulation? His model assumes that demand will be fixed which is not always the case – how does he account for this?
The model is run for 24hours with disaggregate demand – one has to select a representative day to enhance the representativeness of the data. Lastly, the model is not deterministic, since there is a re-planning stage to cater for any eventualities.
Question & Answer Session
- How are you able to control for other variables i.e., what to simulate and what to leave?
Response: One has to do a lot of case studying, sensitivity analysis and expert review to determine what has worked and what has not worked in other places. This gives one a way forward when deciding what to simulate and how to discuss one’s results.
- How are you factoring in other factors such as pollution that are caused by the transport sector?
Response: The model has an extension that can test pollution depending on the type of vehicles (amount of pollution).
- How do you communicate your research in a way that is easy to understand?
Response: This is something that Owen will need to work on.
- Will the service providers really use the research findings? (The supply-side are very innovative)
Response: Use or communication of the model by authorities can make it easier for the model to be applied widely. Further, an entry point for the service providers would be through the SACCOs and Associations etc.
- How do you ensure that there is an equilibrium on your model i.e., demand-side and supply-side?
Response: To ensure that the model runs to an equilibrium, I will utilize genetic algorithms, meta-heuristic algorithms and the like to come up with sub-optimal solutions which I can choose from. By utilizing technology, one can achieve a certain level of equilibrium.
- How does this model capture the non-linear behaviors of individuals?
Response: The model is stochastic, not linear.
- Can Owen try to simplify his graphs so that they are easily understandable?
Response: Owen will improve the paper to make it easily understandable.