TECHNICAL DECISION-MAKING PROCESS
IN CHOOSING SUB-AREA MODELING APPROACHES
By Jin Ren, PE
October 7, 1997
[ABSTRACT]
The objective of this paper is to provide some guidance to transportation modelers or EMME/2 users to make right decisions in choosing cost-effective sub-area modeling approaches. Three approaches are presented here and they are as follows: 1) Focused Modeling Approach - aggregating and disaggregating regional Traffic Analysis Zones (TAZs), and enhancing the regional base network in the sub-areas of interest; 2) Stand-alone Sub-area Modeling Approach - using external stations as linkages to the regional areas, traversal trip matrix as inputs to external stations, and also building a detailed level of sub-area network; and 3) Direct Application of Regional Model Approach - enhancing the sub-area or corridor network in the sub-area of interest to reflect a greater detail roadway network.
This paper will present illustrative examples about each modeling approach undertaken by us, such as: City of Shoreline Model based on Approach 1; City of Lake Stevens model and City of Wilsonville model based on Approach 2; and I-405 Corridor Travel Demand Modeling based on Approach 3. One common characteristic is the detailed network in the sub-area or corridor area. More often than not, their differences challenge planners and modelers to make a viable choice usually at the very beginning of a sub-area and corridor model development. The examples will demonstrate the technical decision-making process in choosing different modeling approaches to meet different modeling needs.
Understanding the different characteristics, especially the pros and cons of each sub-area or corridor modeling approach, transportation modelers or EMME/2 users will benefit in saving time and costs by choosing the right sub-area modeling approach.
Background
At the very beginning of developing a sub-area or corridor travel demand forecasting model, we usually face a challenging technical decision-making process. As there are several different sub-area modeling approaches, we need to choose the most cost-effective one to meet the modeling objectives. A trial and error approach in a sub-area or corridor modeling can be very costly, time-consuming and frustrating.
Three conventional sub-area or corridor modeling approaches are presented here:
1) A Focused Model Approach
2) A Stand-alone Sub-Area Modeling Approach
3) A Direct Application of Regional Model Approach
As transportation modelers, we need to not only familiarize ourselves with the conventional sub-area modeling approaches, but also distinguish their respective advantages and disadvantages.
Definition of Three Sub-Area Modeling Approaches
In the Focused Modeling Approach based on a regional model, regional Traffic Analysis Zones (TAZs) are disaggregated in the sub-area and aggregated outside of the sub-area in terms of the land use data and trip generation data. The regional network is used as the sub-area base network but with local-arterial enhancements in the sub-area of study. Jennifer Heisler extensively covers this approach in her 1989’s paper “Focused Model Approach for Corridors and Subareas” [1]. This sub-area modeling approach is better suited for medium-size or big-size cities where regional trips are an important part of traffic contributions to the sub-area roadway facilities. Figure 1 exhibits the City of Shoreline, Washington EMME/2 focused sub-area model network, note that the sub-area is linked to the rest of the Puget Sound Region through the regional network.
In the Stand-alone Sub-area Modeling Approach, external stations instead of aggregated TAZs are used, traversal trip matrices derived from the regional model trip assignments serve as external trip productions and attractions. External-to-external through trips can be obtained from the traversal matrices and added on top of the sub-area model assignments. A detailed level of sub-area network should also be created. The land use data within the sub-area zones is used in the trip generation process. Trip distribution, mode splits and trip assignments are carried out within the sub-area alone. Figure 2 exhibits the City of Lake Stevens, Washington EMME/2 stand-alone sub-area model network which is only linked to the region through 20 external stations. The same is true for the City of Wilsonville, Oregon model as shown in Figure 3.
In the Direct Application of Regional Model Approach, the sub-area or corridor network are enhanced to a great detail as needed, for instance, the freeway corridors with interchange enhancements, and the sub-area network is calibrated using the regional 4-step modeling processes. This kind of modeling approach is best suited for regional freeway corridor analysis. As no data other than network alternatives are used as inputs, this modeling approach is highly dependent on the validity of the regional model estimation. Sometimes, splitting the regional TAZs in the neighborhood of the corridor or sub-area may be needed to obtain finer forecasts on the corridor and its interchanges. Figure 4 shows the I-405 corridor EMME/2 model network with greater enhancement on I-405 interchange network.
Our Experience with These Approaches
We have developed a number of sub-area and corridor models for cities, counties and regional agencies using EMME/2 software and regional, local or county modeling databases. Our experience has proven to us that at the initial sub-area modeling stage, a lot of time and efforts can be minimized if different approaches are evaluated and the appropriate approach chosen. The following section explains why we used different approaches to developing several sub-area or corridor models.
On developing the City of Shoreline sub-area model, we considered several factors in determining to use a focused modeling approach. First of all, the City of Shoreline is not a stand-alone city, but a medium-size city located in the central parts of the Puget Sound Region. Realizing the strong trip linkages to the region, the City transportation planners expected the model to capture the regional traffic associated with Shoreline. So, we eliminated using a simpler stand-alone sub-area modeling approach. Looking into the possibility of direct application of regional modeling approach, we found that the Puget Sound regional model has only 18 TAZs and King County model 45 TAZs to cover the City of Shoreline sub-area. The TAZ levels can not meet the City of Shoreline modeling needs for comprehensive transportation planning. A focused model approach was considered appropriate after these careful thoughts.
We dissaggregated the regional 18 internal TAZs into 113 Shoreline TAZs based on the Shoreline land use data summarized in 113 TAZs, and aggregated the rest of 832 regional external TAZs into 128 TAZs external to Shoreline sub-area. This was a major effort in terms of relocating the centroid connectors and manipulating the regional external modeling data to fit the Shoreline sub-area model. We used the aggregated regional data, such as mode splits and external person trip productions and attractions, as input to the Shoreline external TAZs. As a result, the model took much less calibration processes as the initial PM peak hour vehicle trip assignments showed the goodness-of-fit at R-square of 0.90. We also
successfully applied this kind of approach on the sub-area model developments or updates for the City of Renton, City of Federal Way, and the Cities of Bellevue, Kirkland and Redmond.
On a sub-area model development for the City of Lake Stevens, the Snohomish County transportation planners preferred a stand-alone modeling approach as the small City of Lake Stevens is located in a separate suburb area. They provided us with the daily and P.M. peak hour traversal matrices and County EMME/2 network data input only for the City of Lake Stevens and its urban growth area (UGA). In building this sub-area, we only need to reconstruct the TAZs systems and enhance the network inside the City of Lake Stevens UGA. A lot of effort was reduced in building the stand-alone model but the network calibration process took much of our efforts.
On choosing the City of Wilsonville sub-area modeling approach, we took the following factors into consideration: first, City of Wilsonville is a small suburb city located on the edge of the Portland urban areas, similar to the location of the City of Lake Stevens; secondly, the city planners and politicians are more interested in the City of Wilsonville sub-area traffic impact than the regional traffic impact; thirdly, the Portland regional model has 1260 zone system with large modeling data sets to be manipulated; and finally, we found that the Portland Metro model has only five TAZs to cover the City of Wilsonville. We eliminated the possibilities of using any other sub-area modeling approach but chose the stand-alone modeling approach. It also took us a lot of efforts with the model network calibration process although it was relatively easier to build the model network. Above all, the model achieved our objectives in terms of traffic forecasting for network and intersections.
In the I-405 Corridor travel demand forecasting model development, we determined to directly apply the Puget Sound Regional EMME/2 model. We surveyed and coded 24-interchange network all the way along the 30-mile I-405 corridor. This major effort was expected to help with the corridor network calibration process. Using this approach, we can directly apply the regional trip tables to do trip assignments and to validate the 1990 existing I-405 corridor model. We saved a lot of time by using the EMME/2 transcript files to code the I-405 interchanges in the 2020 future alternatives. Thirteen future alternatives were tested and evaluated based on the 4-step regional modeling processes with some parameters altered, such as parking costs, HOV 2+ instead of 3+, and transit frequencies.
Advantages and Disadvantages of Respective Modeling Approaches
About these three approaches, one common characteristic is the detailed network in the sub-area or corridor area. More often than not, their respective advantages and disadvantages are what we need to consider in the technical decision-making process for a sub-area and corridor model development.
In regard to the Focused Modeling Approach, there are advantages of strong linkages to the rest of the region, “regional model compatibility in terms of land use trip generation, distribution and assignments, (Jennifer Heisler in 1989)”, application of regional transit network if needed, no external constraints to the sub-area and easier network calibration. However, this approach requires some lengthier efforts in aggregating and disaggregating zones, centroid connector relocation, and fair amount of knowledge about regional model systems.
In regard to the Stand-alone Sub-area Modeling Approach, there are advantages of smaller number of zones, suitability to isolated small suburban cities or sub-areas, simpler model network to build, and much shorter computing time, but there are disadvantages of harder calibration process, missing trip linkages to the rest of the region, and uncertainties in using regional trip distribution curves.
In regard to the Direct Application of Regional Model Approach, there are advantages of much easier calibration process, best suitability for regional freeway corridor modeling, capture of regional trips through the sub-area or corridor. This approach directly uses regional 4-step modeling processes though it might take a longer computing time for a complete model run. Various network and policy alternatives can be tested and evaluated on both regional and sub-area/corridor scales. But EMME/2 users’ thorough knowledge of regional modeling systems is required because of the complexity in handling regional modeling database.
Recommendations
At the initial stage of sub-area model development, we should conduct comparative analyses of the respective pros and cons addressed above and take into consideration of modeling objectives, resources and budgets. Then we are able to choose a cost-effective sub-area modeling approach.
Reference:
[1] Heisler, Jennifer, 1989, “Focused Model Approach for Corridors and Subareas,” published in the Proceedings of the International Conference on Microcomputers in Transportation Vol. 1.
8