In terms of travelers and trips, a simulation of all of Switzerland, with more than 10 million trips, is comparable to a simulation of a large metropolitan area, such as London or Los Angeles. It is also comparable in size to a molecular dynamics simulation, except that travelers have considerably more ``internal intelligence'' than molecules, leading to complicated rule-based instead of relatively simple equation-based code. Such multi-agent simulations do not run well on traditional vectorizing supercomputers (e.g. Cray) but run well on distributed workstations, meaning that the computing capabilities for such simulations have virtually exploded over the last decade.
This paper describes the status of ongoing work of an implementation of all of Switzerland in such a simulation. The whole simulation package consists of many modules, including the micro-simulation itself, the route planner, and the feedback supervisor which models day-to-day learning. A single destination scenario is used to verify the plausibility of the replanning set-up. A result of a morning peak-hour simulation of all of Switzerland is shown, including comparisons to field data from automatic counting stations. These results are shown to be better than VISUM assignment model results when compared to the same field data. This is in fact somewhat surprising, since the OD matrices were adapted by a VISUM module to make the assignment result match the counts data as well as possible.
However, the really big advantage of the multi-agent approach is that it is theoretically justified even under dynamic and congested conditions, and for that reason is extensible even under those condtions. This makes it possible to integrate aspects such as dynamic activity-based demand generation into the framework. Our expectation is that this new technology will allow to introduce many important aspects, such as time-dependent elastic demand or analysis of multi-functional land-use patterns, into the methodology while maintaining or even improving the level of realism.