Additional Options for Auto Assignment

Release 3.0 will add new functionality to the auto assignment (modules 5.11 and 5.21). Since the new features are based on a different use of the already existing "additional demand matrix" and "additional volumes", it falls well that the term additional options for auto assignments will be used to collectively refer to these new options.

During the regular equilibrium auto assignment, the additional demand matrix is added to the auto demand matrix, after the latter has been converted into vehicles using the (optional) auto occupancy matrix. The additional demand matrix is optional, and indeed, many applications do not use it. The additional volumes are used to model a fixed background volume on the auto network which is independent of the auto route choice. Additional volumes are computed in the preparation phase of the assignment (module 5.11) and originate from a user link data item and / or the number of transit vehicles on the auto network, converted to auto equivalents. The results of the auto assignment are: auto volumes on links and turns, auto times on links and turns and auto times per OD pair. The figure below illustrates the regular EMME/2 auto assignment.

While the results obtained by the regular assignments are those that are essential and always needed, there are circumstances in which it would be desirable to have means to compute other, more specialized results. Examples:

• A true distance matrix which reflects the average OD distances of the assigned trips, not merely the shortest possible distances (such as are obtained by a 0 iteration assignment with a single volume delay function "fd1=length").
• Cost or toll matrix containing a different kind of OD cost or attribute, such as road tolls, fuel consumption or emissions - computed according to the paths actually used by the equilibrium assignment.
• Select link analysis, i.e. a partial equilibrium assignment of only those trips that use a particular part of the network. The result of the assignment is a) a demand matrix containing only those trips that do use the selected facilities and b) the corresponding link and turn volumes. Note that the path choice must be based on link and turn times that depend on the total auto volumes, not just those trips that do use the selected facilities.
• A partial assignment, i.e. the partial volumes that corresponds only to a part of the assigned auto demand need to be computed separately. The part of the demand that is of interest may be given explicitly as a fixed matrix or be defined implicitly by a "path rule", such as the rule to distinguish local trips (path < 6.4 miles) from regional trips (path >= 6.4 miles) used in the article "The Use of EMME/2 in a Functional Classification Study" by A. Cotugno, D. Walker and N. Kramer, in EMME/2 NEWS September 1987.

All these results require no change in the assignment itself (they do not change auto volumes or times), rather they need to be gathered in parallel to the ongoing auto assignment. The newly introduced "additional options" constitute a new and versatile framework to do just this, i.e. to account for certain quantities while the auto assignment is carried out.

To make this possible, the "additional demand matrix" and the "additional volumes" are used differently than in the regular auto assignment. Instead of adding the additional demand to the auto demand and assign both into the auto volumes, the additional options allow the additional demand to be assigned separately, yielding the additional volumes as a result. This is done through a "slave" assignment which uses the same paths that were generated for the auto assignment. An additional path attribute may be computed for each path, by combining the additional link attribute (a user link data item or the link length) with a user specified binary operator. On these path attributes, a lower and upper threshold can be specified to define a path rule. Only paths that yield a path attribute satisfying the threshold condition are be considered active, i.e. the corresponding demand is assigned to the additional volumes. The "additional OD attribute", finally, may be used to store the computed path attributes, the active path attributes weighted by the additional demand or the active additional demand.

Note that each of the options can be activated individually, thus limiting the computational overhead to the minimum necessary.

The block diagram below illustrates the additional options for the auto assignment:

Let us now look at how these additional options may be used in practice. The following table shows the use of the additional options to implement the special analyses mentioned above.

ANALYSIS	ADDL DEMAND	ADDITIONAL LINK ATTR.	PATH OPER.	PATH THRESH.	ADDL OD ATTR.	RESULTS
DISTANCE MATRIX	auto demand	link length	+	none	path attr.	distance matrix (at equilibrium)
COST/TOLL MATRIX	auto demand	link cost in user data item	+	none	path attr.	cost matrix (at equilibrium)
SELECT LINK	auto demand	user data: 1=selected 0=else	.max.	L:1.0 U:1.0	active demand	demand and volumes for trips using the selected facilities
PARTIAL ASSIGNMENT with fixed demand	partial demand	none	none	none	none	partial volumes
PARTIAL ASSIGNMENT w. path condition (example used here: regional trips)	auto demand	link length	+	L:6.4 U:999	active demand	regional demand regional volumes (local volumes = VOLAU-VOLAD)

Finally, we conclude with an application example that shows the use of the additional options for performing a select link analysis and a partial assignment for the Winnipeg standard network.

First, a link select analysis for one of the downtown bridges is performed, yielding the volumes of only those trips that actually use the bridge, as well as the corresponding trip matrix. Note that UL3 has been prepared using module 2.41 to contain all zeroes, except for the bridge (links 414-973 and 973-414), where it contains 1. The assignment is prepared as follows:

 5.11  PREPARE SCENARIO 3000 FOR ASSIGNMENT

 Select: Type of assignment
         1= auto
         2= transit
         3= bimodal
         4= end
          <1

 Select: Source for additional auto volumes
         1= no additional volumes
         2= auto equivalent of transit vehicles
         3= user data on links and turns
         4= transit vehicles and user data
         5= assign additional demand
          <5

 Note that when the additional demand is assigned to additional volumes, the
 parameter VOLAD will assume the value zero during function evaluations.

 Demand in persons
 Enter: Matrix=<mf1
 mf01: wod76d/r   1976 observed auto demand (vehicles)

 Vehicle occupancy in persons/veh (optional)
 Enter: Matrix=<

 Additional demand to be assigned to additional volumes
 Enter: Matrix=<
 Additional demand defaults to auto demand (vehicles)

 Matrix to hold travel times (optional)
 Enter: Matrix( mf )= <mf4
 mf04: autim0     auto times scenario 1000
 Change header information? <no

 Select: Source for additional link attribute
         1= user data UL1
         2= user data UL2
         3= user data UL3
         4= link length
         5= none (no additional path attributes calculated)
          <3

 Enter: Operator to compute additional path attributes=<.max.

 Enter: Lower, upper threshold for active paths=<1,1

 Matrix to hold additional O-D attributes (optional)
 Enter: Matrix( mf )= <mf8
 mf08 not initialized.
 Do you want to initialize it?<y
 Enter: Name (6 char) = <sel-lk
 Enter: Description (max 40 char) = <SELECT LINK DEMAND FOR BRIDGE 414-973
 Enter: Default value= <0
 mf08 header modified
 mf08 initialized

 Select: Type of additional O-D attribute
         1= additional path attributes
         2= active additional path attributes
         3= active path attributes weighted by additional demand
         4= active additional demand
          <4

 Enter: Max. number of iterations ( 15)= <15
 Enter: Stopping criterion for relative gap (  .50%)= <1
 Enter: Stopping criterion for normalized gap (avg time difference)
        (   .50 min)= <0.50

The resulting additional volumes show where the cars crossing the bridge come from and where they go to (plot generated with module 6.12):

Next, a new scenario is created by copying the original one and deleting the two links that form the bridge. An auto assignment is performed to on this scenario to find the new auto volumes and times that reflect the closing of the bridge. At the same time, a partial assignment is performed with the trips that were using the bridge before.

 5.11  PREPARE SCENARIO 3000 FOR ASSIGNMENT

 Select: Type of assignment
         1= auto
         2= transit
         3= bimodal
         4= end
          <1

 Select: Source for additional auto volumes
         1= no additional volumes
         2= auto equivalent of transit vehicles
         3= user data on links and turns
         4= transit vehicles and user data
         5= assign additional demand (additional options assignment)
          <5

 Note that when the additional demand is assigned to additional volumes, the
 parameter VOLAD will assume the value zero during function evaluations.

 Demand in persons
 Enter: Matrix= <mf1
 mf01: wod76d/r   1976 observed auto demand (vehicles)     (96-05-20 09:26)

 Vehicle occupancy in persons/veh (optional)
 Enter: Matrix= <

 Additional demand to be assigned to additional volumes
 Enter: Matrix= <mf8
 mf08: sel-lk     SELECT LINK DEMAND FOR BRIDGE 414-973    (96-06-04 08:17)

 Matrix to hold travel times (optional)
 Enter: Matrix( mf )= <

 Select: Source for additional link attribute
         1= user data UL1
         2= user data UL2
         3= user data UL3
         4= link length
         5= none (no additional path attributes calculated)
          <5

 Enter: Max. number of iterations ( 15)= <15
 Enter: Stopping criterion for relative gap (  .50%)= <1
 Enter: Stopping criterion for normalized gap (avg time difference)
        (   .50 min)= <0.50

The resulting additional volumes, displayed using module 6.12, show the volumes generated by the detoured cars. (Note that, while the flows displayed here correspond exactly to the cars that used the bridge before, the trips that did not use the bridge before might have changed their path as well -in response to the changed conditions- as a comparison of the auto volumes using module 6.23 would show.)