Assuming the use of the doubly constrained TMODEL2 full-featured distribution and assignment model, origins are the number of trips which should originate from a land-use zone and destinations are the number of trips which should arrive at a land-use zone during the hour being modeled. Since land-use zones (also called traffic analysis zones or TAZ's) are often large enough to have some trips which start and end within the borders of the zone, these trips are called intra-zonal trips and are not assigned to the network although they do appear in the trip table on the diagonal.

The task is to use land-use (e.g. residential and employment) data, trip-generation rate data, and other traffic data (counts, through trip-tables, etc.) to develop a set of origins and destinations for each trip type (HBW, HBO, NHB) for each zone.

BASIC NOTIONS

• A trip begins any time a vehicle leaves some place and ends a) any time the driver or a passenger gets in or out of the vehicle, and b) any time driver of the vehicle stops to interact with the outside world, such as to get something to eat at a fast-food drive- through.

• Each trip in the model (whether Internal-Internal [I-I], eXternal-Internal [X-I], Internal-External [I-E], or eXternal-eXternal [X-X]) must start and end.

  Although in real life some trips will start before the hour and end during the hour or start during the hour and end after the hour, the results should not be significantly different by allowing these compensating errors.

• Each trip can only be one trip type: examples are home-based-work [HBW], home-based-other [HBO], non-home-based [NHB], or recreational or some other type.

  The reason for different trip types is that each has different characteristics (average and distribution of trip lengths), suggesting the use of different gravity model parameters (coefficients).

  The most significant change in trip type distributions in the past thirty years is the change in HBW (home-based-work) trips. They are being displaced to a certain extent by a "linked trip" consisting of, for the PM peak, the trip from work to some point of errand (a store) which is a NHB (non-home-based) trip, then perhaps another NHB trip such as from the store to pick up a child at daycare, and then a HBO (home-based other) trip from the daycare center to home. Thus, what used to be a HBW trip may now be 2 NHB trips and a HBO trip.

• The trip type of a through trip [X-X] is irrelevant because it is not distributed by the gravity model.

  THEREFORE:

• For each trip type the total origins from all zones should equal the total destinations from all zones for I-I, X-I, and I-X trips.

  IF THEY ARE NOT EQUAL FOR EACH TRIP TYPE . . .

• . . .There is a problem with the land use and/or the trip generation rates. These problems should be investigated and corrected as much as possible.

  THEN

• The remaining inequality can be corrected by BALANCING the origins and destinations. If X-I and I-X trips are determined by the modeler using count and survey data (i.e., solid data), all balancing should be done on the I-I trips using the following steps.

BALANCING PROCESS

Create the following set of files using the steps listed below.

Hierarchy of O & D files

file_3.OND: O's and D's for internal zones as calculated through the trip generation type 2 process

file_2.OND: Same as file_3.OND augmented to contain percent of X-I (origins) and I-X (destinations) for each external zone allocated to each trip type

file_1.OND: Same as file_2.OND except actual volumes have been apportioned (normalized) over the percentages

file.OND: Same as file_1.OND except with O's and D's balanced by trip type.

STEPS IN TMODEL2

0.	The basic project file name is, for this example, assumed to be PJ.ext. Therefore, most files (using our system) would also have an iteration number such as PJ07.ext. The multiple OND file iterations will have another sequence number _2 and _1, such that a typical .OND file might be PJ07_2.OND.
1.	Calculate internal O's & D's through trip generation type 2 process. The number of zones in this file will be the number of internal zones in the system. Save this as PJ07_3.OND file.
2.	The trips into and out of an external zone are determined by the traffic counts on the "external link" (the link or links connecting that external zone to the rest of the network). This full-featured D&A procedure uses only the X-I and I-X trips over these "external links" as its O's and D's. The remainder, the X-X trips, will be used as row and column totals in developing the through trip-table (.TTI). Determine for each external zone the number of O's (X- I portions of the counts) and D's (I-X portions of the counts).
3.	Decide on a percentage basis how to allocate X-I, I=X trips among the trip types. That is, for each zone, determine for the X-I trips what percent should be HBW, HBO, NHB and any other special use trips. These should sum to 100. Using the previously created file, PJ07_3.OND, extend it to include the external zones by entering these trip-type percentage numbers for each zone in the appropriate origins column. Do the same for I-X trips in the destinations columns. This file is then saved as PJ07_2.OND.
4.	Allocate the actual trip numbers into these positions. To do this, load the _2.OND file, then normalize by row the percentage values to the actual O's as determined in step 2 for the origins. Then do the same for the destinations. Save this file as PJ07_1.OND.
5.	Determine if the overall system O's and D's balance. To do this, sum the O's and D's by trip types using the CHECKSUMS menu choice. Remember, origins should equal destinations for each trip type. If they are not relatively equal reconsider the data used to this point. Are there some errors in land use or perhaps some better trip generation factors to use? Correct previous data and files through this step.
6.	Imbalances remaining must now be mathematically normalized to balance. For each trip type, there is now a total number of origins and destinations which are probably not equal. You must now decide on a single number to which both columns should total. You may pick the lower number, the higher number, the average number, or any other number you can somehow justify using. Note how much each present total must be changed (difference between the target number and the present number) to reach the new total.
7.	Because all balancing must occur only in internal zones (trips to and from externals must match I-X and X-I counts), load the _1.OND file, move into the editor, move to the first external zone and then slash out. This sets the end of file such that it thinks it now contains only internal zones. Normalize columns to totals by entering as the new totals the present totals plus or minus the differences determined in step 6. This apportions the correction over all the internal zones. Without leaving the trip generation main menu, move back into the editor. Do a "find entry" to move back to one entry after the last external zone. Slash out from this position. This brings the externals back into the data by setting a new end-of-file. Save this file as PJ07.OND. Now sum the O's and D's by trip type as in step 5. The O and D totals should now balance within a couple of trips.
8.	That is the file used to do a distribution and assignment run. If corrections need to be made for future runs, be sure to move all the way back to the appropriate step to make the correction so that the process can be used for later corrections.