The good news has just arrived that you are on the short list for the interview stage of a transportation planning job with a modeling component. Now you will probably have to make a presentation and will certainly have to answer questions regarding your approach and what products the client will receive.

Another scenario is that you have established a relationship with a potential client, but you now have to compete with other firms for the job. You want to help the client ask interview questions of you AND your competition which will make obvious the fact that you really know what you are doing and that your competition may not be as strong in their understanding or execution of the modeling process.

A third scenario is that you are a public agency conducting interviews and want to better evaluate the products that you will receive as a part of the project.

We, at TModel Corporation, did a little role playing with ourselves as the buyers rather than the sellers of modeling services. We came up with the following list of questions and answers which you can provide to this potential client or at least use to prepare yourself to answer without too much hemming and hawing.

QUESTIONS:

Q.	Will a microcomputer traffic simulation modeling program be used for the study?
A.	Yes (TMODEL2, of course).
Q.	What network data is required? How will it be entered and verified?
A.	Street descriptive and operational data include: length, speed, capacity and other data which support derivation of capacity such as functional class, area (urban, suburban, rural, etc.), one or two way, number of lanes, parking, transit use, bicycle lanes, . . . Intersection descriptive data include: capacity, control, lanes, turn movement restrictions or delays . . . All network data will be cataloged on plots, then entered and checked with the screen graphics editor, then verified with plots.
Q.	Will the short path algorithm account for turn movements which are prohibited or which have delays significantly greater than the intersection as a whole?
A.	Yes. With TM2 we can, FOR ANY MOVEMENT, use either a) a simple base delay or b) that plus a delay producing function based on V/C in the movement under consideration or c) those plus an additional delay producing function based on the V/C of the conflicting movements, both opposite and crossing.
Q.	What land use data is required? What trip generation rates will be used?
A.	SFDU's, MFDU's, categories of commercial and industrial land use for which trip generation rates exist or can be derived through studies. They should be factored to the calibration date. FUTURE land use data in the form of estimates by zone or growth rates by land use and zone.
Q.	What existing traffic (count) data do you need for calibrating the model?
A.	Directional counts for screenlines, at externals, including breakdowns by X-X versus X-I, I-X and for the latter, breakdowns by trip type as well as average travel distance and speeds. All data must be converted to specific calibration scenario data and time.
Q.	How much data is need for, and how are through (X-X), internal to external (I-X), and external to internal (X-I) movements considered and controlled in the modeling?
A.	For each link connecting to an external zone we must have directional peak hour counts. Each direction must be apportioned to through (X-X) versus non-through (X-I and I-X). For the X-X trips, we must know the other external zone from which or to which it travels so we can establish a through trip table. The X-I and I-X must then be split into trip types (HBW, HBO, NHB) and for those we must know the average travel time (or distance and speed).
Q.	How will internal movements be calibrated in the model?
A.	There are several methods from which to choose, either singly or in combination:
	a.	screenlines - set up a screenline file and analyze crossings after each run against allowable screenline crossings (NCHRP 255, pg. 49).
	b.	allowable link errors - set up an allowable link errors file to compare against after a run (NCHRP 255, pg. 41).
	c.	trip length frequency distributions - if origin-destination survey data is available, compare the TLFD curve available from run against those from survey using F-factors editor.
	d.	speed studies - if speed delay studies are available, compare representative route speed delay data from model against them.
	e.	trees - visually check the graphic tree presentations of the program against local knowledge of the most probable routes.
Q.	How accurate will the calibrated model be?
A.	This depends on the level of use of the facility. Good on high-use facilities (<15% error), poorer on local streets (perhaps > 100% error). If accuracy promised is too good, likely modeler doesn't understand what his/her and data limitations are.
Q.	With what sort of documentation of the modeling and calibration process will we be left? Will we be able to run "what if" scenarios ourselves?
A.	See document "Model Documentation." Also consider training personnel of the agency letting the contract by the contractor or by the firm producing the software.
Q.	How will you present results?
A.	Mapped or plotted. Listed data results require too much effort to interpret, understand, and use.
Q.	Can you get turn movement forecasts from your modeling? How could they then be analyzed for several alternative scenarios?
A.	Yes. NCAP is part of TMODEL2 and does the methods of the 1985 Highway Capacity Manual exactly as they are described in the manual. Once geometry of an intersection is set up, the turn counts from any run can be easily and directly imported and an analysis run to compare against that of another scenario.
Q.	How will you use the model to allocate responsibilities or costs over cumulative impacts?
A.	Select link studies can produce a trip table showing the origin and destination of all trip takers using a particular link in either or both directions. Select zone group interaction studies can show the amounts of traffic on all links attributable to any zone or group of zones.
Q.	How can you use the modeling results to help schedule improvements?
A.	If a facility is within acceptable LOS parameters at onetime and beyond at a future scenario, the time of passage can be interpolated OR intermediate scenarios can be run. Further, it is easy to run scenarios incrementing by developments according to their scheduled build time up to the cumulative group which comprise all the known developments. These intermediate scenarios may show when facilities exceed the required LOS parameters.
Q.	Can we get any other sorts of analyses from the model?
A.	Yes. Anything which requires data which is kept in the link files such as volumes, distances, speeds, numbers of lanes, etc., can be used in a formula to produce resultant data like V/C or pollution indices or lane-miles or cost-benefit analyses or whatever. These results can be mapped and summarized statistically.