Estimation of User Benefits of Road Investment Considering Induced Traffic with Combined Network Equilibrium Model in Tokyo Area

Estimation of User Benefits of Road Investment Considering Induced Traffic with Combined Network Equilibrium Model in Tokyo Area


T Maruyama, N Harata, K Ohta, University of Tokyo, JP



It is well recognized that new road investment produces induced traffic, so it is necessary to consider the induced traffic in the appraisal of road investments. Benefit estimated by the fixed OD matrix method will be biased, although the method has used traditionally. An easiest treatment of induced traffic is elastic OD matrix model, but it is difficult to assume reliable parameter of elasticity in the model and the result varies largely according to the parameter. In view of drivers' behavioral side, induced traffic is explained mainly by change of route, change of mode, change of travel destination, and increase of trip frequency when new road is open. In our paper, we used the 4-Level Nested Logit model, that is, travel choice, mode choice, destination choice and route choice model to express these behavioral responses. The level of services in NL model variesaccording to network congestion, so we have to consider the equilibrium conditions. We employed Logit type Stochastic User Equilibrium model. We formulated the mathematical optimization problem that is equivalent to the whole of 4-level NL model and SUE model. The consistent results are given by solving this optimization problem. There are no internally inconsistent problems in our demand model such as conventional 4step travel demand model. Our model is the Oppenheim (1995) type model, but ours model differs from his model because we formulate models for each traveler's trip purposes in travel choice, mode choice and destination choice phase. We applied this combined model to the Tokyo Metropolitan Area. Parameters in NL model are estimated for each trip purposes and each time periods with Person Trip survey in this area. We used multi-mode and large scaled transport network data using GIS platform. The road network has more than 22,000 links and railway network has more than 4,900 links. The network congestion is expressedby conventional link performance functions in road links and discomfort functions in railway link. The discomfort functions explain the travelers' discomfort in the crowded train, which is needed to express the railway highly congestion in peak period in Tokyo area.

We estimated user benefit of Tokyo Outer Ring Road (Tokyo Gaikan Expressway), which is actually planned in Tokyo Area with this model. Some people change their travel mode from railway to automobile because of increased accessibility of the road system according to our model. This modal sift worsen the congestion in road network, and it partly relieve the congestion in railway network especially in the peak periods. So the induced traffic caused by the modal split partly raises total social welfare in this case. The increased accessibility brings another induced traffic that consists of longer distance trips by changing destination and increased trip frequency. The benefits estimated by our combined model considering induced traffic are compared with the fixed demand model. In our case, the benefits by the conventional fixed demand model are shown to give overestimated results.


Association for European Transport