ON THE ROLE OF UNCERTAINTY ON TRAVEL COST AND DEPARTURE TIME CHOICE

We consider a dynamic network model (1) in which each commuter has a cost function which depends on travel time and scheduled delays. In a particular case with an exogenously given travel time which vary over the day, the commuter has to choose his departure time in order to minimize his generalized cost function. In that basic case, we can show that the optimal departure time may not be unique: for some sets of parameters, we can exhibit three optimal departure time. Since this result is valid for an exogeneous travel time function, we may think of a rather more realistic system. It is well know in transportation science (2,3) that many events can modify the characteristics of a trip: weather, accident... So, the same commuter, on the same road during the same time period, will have different travel times over the week. So we set up a model in which the travel time is stochastic and we consider three situations In the first one, risk averse drivers know only the distribution of events. In this case, we study the impact of unexpected variability on the user cost and we show that commuters choose an earlier departure time such as in the deterministic situation with early arrival