An algorithm is presented for transit passenger OD estimation. The algorithm, like some other existing OD estimation techniques, generates estimates based upon passenger boarding and alighting counts at each stop along the route. The algorithm is distinct in that it does not only estimate an OD matrix for the vehicle trip from which the boarding and alighting counts were taken. Rather, it further estimates the passenger alighting probabilities at every stop on the route and these are more apt to remain fixed across transit trips. Therefore, when coupled with projected boarding counts, the alighting probabilities better characterize OD patterns on the route. These probabilities, moreover, are estimated in such way as to reflect the passengers’ latent tendencies to travel to and from ‘‘major activity centers’’ where trip-making is induced. The algorithm is therefore a more general-use method than is the estimation technique proposed by Tsygalnitsky. Since the algorithm does not require a seed matrix, and since the number of iterations required for generating estimates is specified a priori, the algorithm is easier to apply and more computationally efficient than the balancing method of OD estimation.

Long bus queues can form at a busy bus stop where multiple routes converge. These queues and the delays that they impart to buses can often be mitigated by altering the stop's design or the manner in which it is operated. Queueing models are formulated to this end. The models estimate the maximum rates that buses can discharge from a stop that is isolated from other stops and from nearby traffic signals, while maintaining targets for service quality. Model inputs include the stop's number of serial berths, and the processes by which buses arrive and dwell at the stop. The models can be used for a variety of purposes that include: choosing a stop's suitable number of berths or determining when it is useful to alter the rules governing how buses are to dwell in those stops. Application of these models indicates that much of the previous literature on the subject of bus queueing at stops is flawed. And because the analytical solutions derived for the present models describe the special operating features of serial bus berths, the models can be applied to other serial queueing systems common in transport. These include taxi queues, Personal Rapid Transit systems and toll plazas with tandem service booths.

Traffic congestion is a growing problem in Nairobi, Kenya, resulting from rapidly increasing population and the crowding of motorized traffic onto a limited street network. This report includes analysis of the traffic conditions in Nairobi, the expected effects of further growth in demand, and a set of recommendations for how to improve the performance of the street network. Data describing motorized vehicle traffic was used to build a simulation model of Nairobi’s street network considering cars and matatus. This model was used to analyze traffic conditions at the city-scale under existing conditions and future growth scenarios. The results provide insights for improving the network performance and support recommendations for Nairobi. City-scale analysis of the street network was conducted with the use of the macroscopic fundamental diagram (MFD) which relates the number of vehicles circulating on the street network to the rate at which trips reach their destinations. The results of simulations with different demand patterns show that there is a consistent MFD relating vehicle accumulation to network flow in Nairobi’s central business district (CBD). Therefore, detailed knowledge of demand is not necessary to understand how the network performs, because the MFD depends on the properties of the street network itself. Monitoring and controlling the number of vehicles in the network is sufficient to maintain traffic flow on the city’s streets. As traffic demand grows in the future, the streets will quickly become more congested, so measures should be taken to improve the system. The first recommendations seek to control the accumulation of vehicles in the network so that traffic flow is maximized according to the MFD. One method is to meter the rate at which vehicles can enter the CBD in order to control accumulation so that everyone can reach their destinations sooner. Metering can be effective in the morning when more vehicles are entering the CBD from outside, but during the evening there are many internally generated trips which will tend to jam the network anyway. Policies that reduce the peak travel demand by shifting trips to public transport or spreading the demand across more time can reduce traffic congestion in the evening. A second set of recommendations expand the shape of the MFD itself by increasing the capacity of the streets in the network which is largely dependent on how intersections operate. Traffic circles (roundabouts) are common in Nairobi, but signalized intersections can have greater capacity. Converting intersections will also reduce the congestion effects when queues spill back into upstream intersections. Capacity can be further increased by adding redundancy to the network. An analysis of dedicating lanes to buses and matatus on radial arterials shows that queues in the remaining lanes will grow longer. In the morning, these queues grow away from the center, so matatus experience reduced travel times, but in the evening, the queues back up into CBD increasing delays for everyone. The simulation study provides an illustration representing Nairobi approximately, so results are relevant and qualitatively useful. Further data could be collected to estimate the real MFD for Nairobi and provide more accurate quantitative values. Although Nairobi’s streets are congested and bound to get worse, the network performance can be improved by making strategic investments in the transport network.

The maximal rates that buses can discharge from bus stops are examined. Models were developed to estimate these capacities for curbside stops that are isolated from the effects of traffic signals. The estimates account for key features of the stops, including their target service levels assigned to them by a transit agency. Among other things, the models predict that adding bus berths to a stop can sometimes return disproportionally high gains in capacity. This and other of our findings are at odds with information furnished in professional handbooks.

Local bus operators can reduce costs and better serve their customers by matching service frequencies to those of the regional trunk lines that pass through their jurisdictions. For an entire region, trunk line and local bus operators should coordinate their schedules and the carrying capacities of their fleets. Metropolitan Planning Organizations (MPOs) are in a position to encourage this through their purse-string powers and ability to provide financial incentives. For future investments, facilities should be designed to allow quick and efficient transfers not just between the same types of vehicles (e.g. cross-platform transfers for rail systems), but also between different types of vehicles across different systems (e.g. small feeder buses and commuter rail lines).

Before implementing these changes, several issues should be considered. One, additional costs will likely be incurred for technologies that improve scheduling and for upgrading stations. Also, improved bus-rail coordination can lengthen headways, thus reducing the frequency of services. Additionally, not all local feeder bus riders are destined for trunk line stops. Thus, the benefits of coordination should be weighed against the additional waiting times experienced by those riders.