Multiclass Equilibrium Auto Assignment

While the consistency and flexibility of the EMME/2 auto assignment has always been one of its strongest points (just think of the possibilities of the additional options, the traversal computations and the variable demand assignment), so far all variants of the auto assignment in EMME/2 were based on the assumption that all travelers shared basically the same view of the network, i.e. were implementations of single class assignments. Despite this fact, in the past many EMME/2 users implemented auto assignment models which take into account more than a single user class. Examples of this include HOV modeling, as well as models which combine passenger and truck traffic. While from a practical point of view these implementations were a great success and yielded very plausible results, they all assumed, from a theoretical point of view, the dominance of one class over the other classes, since the assignment of the different user classes had always to be performed sequentially, using the results of the first assignment as the pre-load volumes on the next.

In Release 6.0, it is now possible to perform true multi-class equilibrium auto assignments, where all user classes are dealt with simultaneously, so that the route choice of each class takes into account the congestion effects caused by the route choice of all other classes.

Up to 12 classes of users can be assigned simultaneously in this new assignment variant. Each class has access to a subnetwork of the complete auto network. For the purpose of defining such subnetworks, the new mode type auxiliary auto mode has been added. Thus, when the multiclass assignment is prepared, each user class is associated with the auto mode (access to complete network) or an auxiliary auto mode (access to a restricted network). In addition, the demand matrix is specified, and an optional time matrix can be saved for each class. One user class (which can correspond to any sub-mode) has access to the full features available in the single class assignment, i.e. auto occupancy factors, additional demand, additional options and even variable demand.

The new multiclass feature of the EMME/2 auto assignment allows the consistent modeling of many different situations where more than one user class has to be dealt with. Here a few examples:

• Certain classes of vehicles (such as trucks, vehicles transporting dangerous goods, vehicles without catalyst) might be forbidden on certain types of links, or alternatively some classes of vehicles (e.g. taxis, zero emission vehicles) might be given special privileges to use certain otherwise reserved facilities.
• A very important example of the above class of problems is the modeling of HOV facilities (``car pool lanes''). The high occupancy vehicle user class has access to the entire auto network, while the HOV links are not accessible to the low occupancy vehicle user class.
• Toll roads or road pricing schemes with stratified fares and/or stratified value of time perception.
• Traffic restriction schemes for core areas forbidding thru-traffic, or restricting access only to people living in this area.