APC currently uses a traditional trip-based three step model (i.e., no mode choice) on the TransCAD 5.0 software platform. The model was most recently updated, by a consultant and APC, in 2010 with a 2010 base year. Approximately half of the work to update the model was completed by APC.
When the 2010 base model was built, it was expanded from a 600 zone system to a 1033 zone system, with 47 zones representing external access points. There are approximately three to five blocks per TAZ. The model has the following land use and employment categories:
To project where people will live, APC used the official estimates developed by the state of Louisiana to estimate population in the coming years. APC reviewed the estimates and modified the projections based on recent trends in changing demographics and migration patterns.
APC calculated how many women were of child bearing age and how many elderly people lived in each zone. Based on these numbers, APC estimated how many people would be born and how many people would pass on. These are the same figures used by life insurance companies to estimate longevity. AMPO also looked at how many move in and move away from Lafayette using their age and gender as guidelines. Using birth, deaths, in-migration, and out-migration, the APC developed growth rates for each TAZ.
APC also reviewed Year 2010 aerial photographs to locate undeveloped land located in close proximity to new transportation projects that may be redeveloped into new subdivisions. In addition, APC hired a consultant to track down the location of each and every business, more than 150,000 businesses in all. The consultant used the employers' addresses and geocoded each business location. APC looked at the kind of business each employer was doing and then looked at the probable growth rates of each of these kinds of business. The method also entailed looking at how far away people travel to do business and shop in the Lafayette area. A factor was developed for each employment sector and then multiplied to find how big the business sector might be in the coming years. These analysis were used to develop one set of land-use forecasts for Year 2020, 2030, and 2040.
The model includes the following internal trip purposes:
It also includes the following external trip types:
The model uses cross-classification trip production models for the home-based and non-home-based trip purposes. Trip rates, developed using the 1997 Baton Rouge Personal Transportation Survey, vary by household size are applied at the zonal level. These trip rates were refined as needed during the calibration process. Productions and attractions are distributed using a singly-constrained gravity model based on free flow skims by holding trip productions constant.
Commercial vehicle trips represent four-tire commercial vehicles, including delivery and service vehicles. Commercial vehicle trips are generated through a linear regression equation that relates zonal employment and households to trip productions and attractions. The trip attraction models are linear regression equations that relate zonal employment, households, and student enrollment to trip attractions. Productions and attractions are distributed using a singly-constrained gravity model by holding trip productions constant. Commercial Vehicle trip models were derived using the Quick Response Freight Manual, September 1996.
In order to build the EI and EE trip tables, roadside travel surveys conducted from 2000 were used and updated to the current traffic counts through extrapolation and the Fratar procedure to obtain trips crossing the study area boundary. The EI attraction equations used in this model were derived from the Baton Rouge, LA 2032 Metropolitan Transportation Plan Update. In addition, EI trips were also separated into auto and truck trips based on the vehicle classification counts at external stations. The year 2000 traffic counts at each external station were forecast to 2010, 2020, 2030, and 2040 by developing a growth factor based on historical traffic counts at the external stations. The total traffic at each station was then divided into EI and EE trips with the assumption that there would not be a significant change in the distribution from the base year.
The AMPO model is a daily model. There is one set of skims for all time periods (i.e. separate skims are not developed for the peak and off-peak periods). Diurnal factors are used to divide trips into four time periods (i.e. AM Peak, Mid-Day, PM Peak, and Night). The AMPO model includes a User Equilibrium (UE) assignment with the Bureau of Public Roads (BPR) volume-delay function. Feedback is not included between the model components.
Validation of the AMPO Travel Demand Model proceeded from consideration of its areawide performance to the relative distribution of traffic by roadway functional classification and ADT range. In the final stage of the validation process, the accuracy of the model with respect to specific routes and roadway groups was analyzed. At each level, an appropriate degree of accuracy was defined in terms of the maximum tolerable deviation from base-year vehicular volumes (i.e., estimated annual average daily traffic) and Root Mean Square Error (RMSE).
Overall, the cumulative model volume for all network links associated with the DOTD traffic count locations of 4,860,904 vehicles differed from total model estimated ADT of 4,945,530 by 1.7 percent compared to an allowable error limit of five percent. Model estimated volumes are slightly higher than the observed traffic counts on Interstate routes while model volumes on all non-Interstate routes match observed counts well within acceptable deviations. An analysis of historical traffic counts along Interstates indicated a high degree of fluctuations. Due to this reason, the model estimated volumes on Interstates were considered to be valid. APC concluded that AMPO model performs well within the established limits of acceptable deviation from base-year estimated volumes.
The DOTD counts provided to APC consist of tube counts at 50-60 locations. Data is provided in 48 hour increments with 50-50 directional split, and includes axial adjustment factors by functional class, weekday adjustment factor, and seasonal adjustment factors.
The primary reasons that APC pursued a model peer review were to:
APC also had specific questions that they requested the Peer Review to answer. The report provides answers to most, but not all, of the questions:
