International Choice Modelling Conference, International Choice Modelling Conference 2017

Font Size: 
Application of Stated Preference Choice Models in the City of Johannesburg Demand Model
Theuns Lamprecht, Evan Roux

Last modified: 28 March 2017


The City of Johannesburg (CoJ), as many other cities in South Africa, has introduced Bus Rapid Transit (BRT) as a new mode of public transport in 2009.  During the updating of the city’s Integrated Transport Network in 2014, stated preference surveys were carried out to understand users’ preferences and choices regarding transport modes, especially towards BRT.  The SP data was used to developed choice models founded in multinomial logit modelling.   The SP Model was incorporated into the city’s demand model, based in PTV VISUM software, in order to assist with decisions regarding network expansion.


This paper will demonstrate the application of these choice models in the CoJ transport demand model as well as some useful outputs generated from the model.  The incorporation of the choice models gave the transport demand model created a dynamic interaction between trip generation, model split, trip distribution and trip assignment.  The demand model follows an iterative process of determining trip characteristic for multiple modes, public transport services, and paths between trip origins and destinations, which are fed back into the mode choice models until system equilibrium is reached.


The choice models developed for the City of Johannesburg consist of two parts.  The first part dealt with the captivity of users to a specific mode, either based on their lifestyles or the affordability of alternative modes.  Based on the SP surveys, some users are “captive” to cars since these users are not prepared to consider public transport, not matter how good the service is.  Another part of the users is captive to public transport, because they cannot afford cars.  This first part of the mode split model thus separates users as car captive, public transport captive and choosers.    The choosers can afford cars, but is willing to consider alternative modes based on what each mode can offer.


The second part of the mode split model assigned a specific mode of transport to the public transport captive users and choosers.  This part of the mode choice model is heavily dependent on interaction with the transport demand model. Choice specific data is generated with the model before being fed to the choice model.  This data includes, information such as mode availability, travel time (including congestion), travel cost, waiting times, and number of transfers for each mode on a zonal basis.  Based on this information, demand is assigned to the various modes.  An iterative process is used to find system equilibrium based on service capacity, congestion and system utilisation of each mode.   The modes which are considered in this part of the model are cars, combi taxis, conventional buses, BRT and rail (Gautrain and MetrolRail).


The model is capable of assessing demand increases, network expansions, and service characteristics.


The paper will go further to point out challenges in developing and calibrating the model.




Conference registration is required in order to view papers.