From a modeling point of view, the new deterministic transit assignment is certainly the single most important new development in Release 9. As this new time table based assignment model was already presented in detail in the last issue of EMME/2 News (No.19, October 1997), we will only give a very brief description here.
The deterministic transit assignment is very different from the standard probabilistic EMME/2 transit assignment and is not meant to replace the latter. Rather, it provides a completely new assignment approach in cases where steady state assumptions do not hold and where exact time table information is readily available. This is often the case for long distance travel or for rural regions with only a low level of service.
While both types of assignment share the same basic data structure for the transit network, the actual data they need --both on the network and the demand side-- are very different. Also, the much finer level of detail for the input data which is required for the deterministic transit assignment makes it more difficult to apply this type of model for long range planning exercises.
The time table is already largely determined by the segment travel times obtained from the transit time functions, the dwell times at the stops, and the transit line headways defining the interval between two consecutive vehicles of the same line. In order to complement this information to define a complete timetable, it is only necessary to specify for each line the departure time of the first run (line offset) and the number of consecutive vehicle runs of this line (0=continuous operation over the entire assignment period). The line offset and the number of runs can be stored in any of the available transit line user data items or extra line attributes.
Depending on the needs of the specific application, all regularly spaced vehicle runs operating on the same route can be combined into a single transit line for which only the total volumes for all runs are accumulated, or a separate line can be defined for each run. In the latter case, separate volumes for each vehicle run (e.g. train, plane or bus) will be generated by the deterministic transit assignment.
By definition, this type of assignment is no longer static. Thus, the demand is not only specified in terms of the number of travelers who wish to travel from an origin to a destination, but for each travel party the desired time of departure or arrival must also be known. This can be done either by a fixed target time or by explicitly allowing a certain maximum acceptable earliness and/or lateness, with optionally associated early/late penalties. A special notation was developed which allows an efficient specification of the desired departure or arrival time.
Given the detailed timetable description of the available transit services, the deterministic transit assignment finds for a given travel demand (origin, destination and desired departure or arrival time) the optimal travel itinerary and assigns it to the underlying space-time trajectory. The algorithm, which essentially computes the ``shortest'' paths in the space-time network, keeps track of ``times'' and ``costs'' separately. The times are used to determine which itineraries are feasible, whereas the costs are used to find the optimal travel itinerary among the feasible ones.
This new development is implemented in the following two new modules:
The first new module, ``5.36 - Deterministic Transit Assignment'', implements the deterministic transit assignment algorithm. The demand can either be taken from individual trip data entered interactively or read from a batch input file, or consist of O-D matrices containing the demand for several time slices.
Besides the usual assignment results
(transit volumes and times), detailed information on the travel itinerary
is also available in the assignment report, as shown in the following
example:
from 1115/lyss to 1164/chur desired departure:10h00+30 trips: 1
at arr dep with --time(late 2.00)-- --cost(late 2.00)-- -distance-
node time time line/mode aux wait inv cumul aux wait inv cumul (km) cumul
1115/lyss 10h02 a 10.00 10.00 15.00 17.00 .50 .50
115/LYSS 10h12 10h15 S2206 3.00 8.00 21.00 21.00 8.40 46.40 14.44 14.94
107/BIEL 10h23 10h27 S1511b 4.00 87.00 112.0 22.00 91.35 159.8 126.6 141.6
126/ZRCH 11h54 12h10 S1651 16.00 95.00 223.0 34.00 99.75 293.5 126.2 267.7
164/CHUR 13h45 13h45 a 10.00 233.0 15.00 308.5 .50 268.2
1164/chur 13h55 total: 20.00 23.00 190.0 233.0 30.00 77.00 199.5 308.5 268.2
The second new module, ``6.26 - Plot Transit Timetables'', provides a tool for displaying timetable information graphically in the form of space-time diagrams. The user can select the nodes on the vertical space axis either by an explicit list or using the itinerary of a transit line. The following plot shows an example of this type of graphic timetable:
