Segment Specific Headways in Transit Assignment

In the EMME/2 transit assignment, the headway is used to model the frequency of service for a line and directly influences the waiting time and the boarding probabilities of the attractive lines at the boarding node. It is thus a very important parameter for determining the optimal strategy.

Up to now, the EMME/2 transit assignment already offered several possibilities for defining the effective headways: actual headways, actual headways with a maximum (which may be quite dangerous, unless you know exactly what you are doing!), or any user defined transit line attribute. As the headway of a transit line is normally the same for all stops of a line, the above possibilities provide all the flexibility needed for the great majority of applications.

There are, however, some special cases in which one would like to be able to apply different headways at different segments of the transit lines. For these, Release 9 offers the new possibility to define segment specific effective headways which can be stored in any user defined segment attribute.

Here are some examples of applications which will benefit from this new development:

• In highly congested transit networks, the congestion not only decreases the comfort of the passengers inside the vehicle, it actually prevents passengers from boarding a vehicle in which there is no room left at all. This means effectively a reduction in the service frequency at stops where the vehicles are overcrowded. Segment specific effective headways will make it possible to implement congested transit assignment heuristics which are based on this type of frequency reduction.
• Long transit lines, especially bus lines running through congested urban areas, often show a phenomenon called ``vehicle bunching''. Two or more vehicles of the same line are running right behind each other -- the first one being slow and crowded as it picks up all the passengers waiting at the stops, and the second one being faster and almost empty, as most of its potential passengers have already been picked up by the vehicle in front. Such bunching results therefore in an increase of the effective headway, which usually gets bigger the further away a stop is situated from the line's layover point.
• Segment specific effective headways can be used to model situations in which the actual headway varies during the assignment period and affects different parts of the itinerary at different times, therefore resulting in different average headways along the itinerary. (A special case of this is a two-way line which operates at a different frequency in each direction -- but of course, this case should preferably be coded as two separate one-way lines.)
• The availability of segment specific effective headways can be used to code transit lines in which not all vehicles pick up passengers at all nodes, either because not all vehicles run along exactly the same itinerary (happens sometimes at the far end of a radial line running into downtown) or because it is up to the driver to decide if he can (or ``wants to'' or ``has the time to'') stop at a given node. Note however that this only holds for picking up passengers, not for letting them disembark.

As the transit assignment algorithm always handles the headway at the line segment level, using segment specific headways will not influence the computational efficiency of the transit assignment in any significant way.