Smart Growth and Urban Goods Movement (2013)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

34 C H A P T E R 7 7.1 Introduction This section reviews the status and application of urban freight-forecasting models to account for the impacts of smart growth. This review is limited to urban models that forecast trucks because the majority of urban goods movements are by truck, and because existing urban-level freight-forecasting tools in use do not have the capabilities to address freight movement by other modes. While there are existing and planned statewide freight models, which have a valuable role in providing input into urban models, these models are at a resolution that is impractical for the evaluation of smart-growth activities. 7.1.1 Municipal Freight Models Given freight’s importance to an urban economy and trucks’ environmental and roadway- congestion impacts, the existing interest in planning for and forecasting freight activity is not surprising. Urban forecasting models that account for some aspect of freight are relatively com- mon in large urban areas, with many of the modeling programs being operated by MPOs. The Transportation Research Board’s Special Report 288: Metropolitan Travel Forecasting—Current Practice and Future Direction surveyed MPOs about travel modeling and noted: “Truck trips are modeled in some fashion by about half of small and medium MPOs and almost 80% of large MPOs.” But it went on to state, “The treatment of commercial and freight travel is one area in which most travel-forecasting models need substantial improvement. The development of better models is hampered by a lack of data on truck and commercial vehicle travel both within and beyond the metropolitan area” (Special Report 288). An important overview of the state of freight modeling is provided in NCHRP Synthesis of Highway Practice 384: Forecasting Metropolitan Commercial and Freight Travel (Kuzmyak 2008). This synthesis identified urban freight-forecasting methods in professional practice and pre- sented the results of a survey of organizations with active urban freight-modeling programs. The report provided supplemental case studies that highlighted more innovative freight-forecasting methods and approaches, and it also discussed promising paths for urban freight forecasting based on existing research. NCFRP Project 25, a recent effort that examines freight trip generation and land use, includes a number of tables that summarize a review of national and international freight-generation and freight-trip-generation modeling applications (The results of this project were published in NCHRP Report 739/NCFRP Report 19 [Holguín-Veras et al. 2012]). The review found that a number of the planning-level models have truck components, and it forecast vehicle trips at the municipal level. Some of these models potentially could be used to evaluate some smart-growth Urban Truck Freight Models

Urban Truck Freight Models 35 activities. However, one major challenge is that these models are very diverse, using a wide range of data sources and with different independent variables. Any accounting for smart-growth evaluation would require processes unique to each model. 7.1.2 Trucks and the Four-Step Modeling Process NCHRP Synthesis of Highway Practice 384 noted that almost all MPOs and urban areas that model freight are actually forecasting trucks using an adaptation of the traditional four-step process common in passenger forecasting. The four steps and the model’s adaption for truck forecasting are as follows (Virginia Department of Transportation n.d.): 1. Trip generation (the number of trips to be made). For trucks, this can be an estimate of pro- duction or consumption linked to economic activity as represented by zones. Truck trips from internal locations or going or coming from locations external to the study area can be factored in at this point. A number of studies have found this linkage between land use and truck trips to be weak and in need of better data (e.g., Fischer and Han 2001). 2. Trip distribution (where those trips go). Truck data are often integrated into the overall model during this step by the use of a zone-to-zone trip table (origin-destination matrix) that accounts for truck travel between zones. For a truck model, the external and internal trips are added, and flows are often sorted by truck size or type. This process creates a correspon- dence between actual and forecast link counts. Validating this step requires truck classification counts and survey data. 3. Mode choice (how the trips will be divided among the available modes of travel). The mode- choice step is not commonly used for urban freight models because most goods move on trucks without a modal change. 4. Trip assignment (predicting the route trips will take). Trucks, along with passenger vehicles, are assigned by type or class to the roadway network using the shortest path or the lowest-cost travel times, often by time of day. Reviews of these adapted four-step freight models point out some notable limitations, a number of which can be expected to reduce the models’ ability to accurately account for the impacts of smart growth. One significant limitation is that the four-step process fails to account for the trip and tour (chaining) behavior of truck activity in urban areas. Passenger movement can be reason- ably modeled by capturing single trips in response to a few purposes such as work, home, or shop- ping. Freight movement is much more complex than human travel because multiple actors create a purpose for the goods to move (brokers, warehouses, trucking, and consignees), and truckers respond to these needs, which means that they must work within the limitations of the roadway network. In response to this complexity, and in support of efficient travel, many truckers make multiple tours with multiple trips in each tour, but existing four-step models do not account for this travel behavior. In terms of smart growth, this means many truck models might not be able to capture the intricacies of drivers’ responses, at the level of urban streets, to smart-growth-driven network changes. For example, this type of model may do a poor job of capturing the impact of the growth in large consolidation and distribution centers and their impact on the patterns of urban truck travel (Kuzmyak 2008; Donnelly et al. 2010). 7.1.3 Accounting for Smart Growth in Four-Step Passenger-Transportation Models A number of studies have presented techniques to incorporate the effects of smart growth into passenger-oriented four-step travel-demand models (Purvis 2003; Cervero 2006; DKS Associates

36 Smart Growth and Urban Goods Movement 2007; EPA 2010). These studies recognize that many MPO modeling practices “have very little sensitivity to smart-growth land use or transportation strategies” (DKS Associates, 2007, p. 3). This study specifically noted the following model limitations that could limit sensitivities to smart growth: • Trips not related (e.g., does not recognize “trip chaining”) • Consideration of only vehicle trips • Limited or no transit-modeling capability • Limited or no modeling of walking and bicycling • Fixed vehicle trip rates by land-use type • Development design (building, street, and sidewalk layout) not reflected in traveler choices • Zonal aggregation of decision-maker characteristics • Focus on travel during peak periods • Travel analysis zones often too large • Land use not affected by travel patterns Beyond trip chaining, several of these limitations hamper a four-step model’s ability to account for smart growth. In particular, given the importance of parking, curb space, and other street-level issues found in the focus groups, the lack of ability of many models to account for development design is relevant. In general, the methods to adjust for these smart-growth-related limitations in passenger models can be categorized as follows: • Post-processors that run after forecasts are completed • Stand-alone pre-processors for aggregate application of smart-growth trips and vehicle miles of travel elasticities • Built-in changes or enhancement of the forecasting models • Integrated land-use, economic, and transportation models Each of these techniques requires intervention in the modeling process. Modelers in the San Francisco Bay Area assumed that smart growth would decrease the overall average trip length of vehicle traffic and increase transit use and non-motorized travel, which would shift the mode choice to a higher non-automobile share (Purvis 2003). To account for these impacts, they adjusted the socioeconomic databases, adjusted the model’s modal networks, and modified zone-to-zone travel times, distances, and costs. This study also proposed a peer group review panel to sign off on such changes to the models. A study by Cervero (2006) highlighted a number of examples using post-processing to account for smart-growth impacts not accounted for directly in models. He noted that a four-step model’s traffic-analysis zone (TAZ) structure is too gross to pick up many of the impacts of smart growth. He suggests the use of both post-processing and direct models. The post-processing involved “tweaks” to adjust model output for smart-growth factors such as increased transit and pedestrian travel. Direct modeling, which is an off-line, stand-alone model, can be tailored to estimate travel for specific smart-growth areas. One advantage of off-line models is that they can be compared with standard model results and used as a “first cut” to enhance or direct the four-step model output. Similarly, the NCHRP Report 684 (Bochner et al. 2011), which attempted to capture trip estimates for mixed-use developments, suggests an improved methodology for internal zonal trip generation from mixed-land-use neighborhoods. The modification suggested in the report would “include the effects of proximity (i.e., convenient walking distance) among interacting land uses to represent both compactness and design.” If used as an input into a model process, the new input would likely reduce local overall automobile trips.

Urban Truck Freight Models 37 The Environmental Protection Agency (EPA) supported a 2010 effort to “accurately predict the impacts of mixed-use studies” and suggests . . . the potential vehicle trip reductions from Mixed-Use Developments (MUDs) were signifi- cant enough to demonstrate that conventional trip-generation methods could exaggerate roadway impacts . . . (Fehr and Peers brochure, undated). The resulting trip-generation tool accounted for more internal zonal trips, more walk and tran- sit trips, and shorter trip lengths. This spreadsheet tool was designed to update or replace the trip-generation rate that had traditionally been used, which was derived from the Institute of Transportation Engineers manual (2001), and it reduces the number of vehicle trips. The techniques used in passenger-oriented, four-step models to account for smart-growth impacts can be modified to account for truck travel. Table 4 highlights these adjustments. The following summarizes techniques that could help to capture the impact of smart growth on truck mobility. 1. Access, parking, and loading zones Parking restrictions. Typical adjustment of the four-step model would occur by changing intra-zonal travel times for trucks. Locations in which it is difficult for trucks to park could receive a “penalty” added to a terminal cost. Such improvements to a model would require information about truck dwell times at locations where it is expected that truck trips are longer due to parking constraints. 2. Road channelization, bicycle, and pedestrian facilities Accessibility by bicycling, walking, and transit. Intra-zonal travel times could be adjusted for slower truck travel. Empirical data demonstrating slower truck travel, as well as the extent of the slower travel, would need to be obtained for all areas where it is expected that trucks travel slower than they otherwise would because of intermingling and conflicts with other modes. If available, bike and transit volumes could be factored into the modal mix. Smart Growth Impact Passenger Four Step Model Adjustment Freight Comment Addional local travel by non automobile models Smaller transportaon analysis zones, non motorized mode choice Requires beer truck origin/desnaon data to account for greater conflicts with trucks. Addional bicycle and pedestrian travel Expand model mode choice, Assign bicycles to network pedestrian and bicycle networks Mode choice does not account for trucks. Conflicts with trucks not accounted for in models. More use of transit More trip purposes Neutral outcome, possible curb space conflicts (limited relevance in regional models because of intra zonal nature). Constrained parking Incorporate pricing in model steps; change level of automobile ownership Reduced load/unloading opportunies for trucks (limited relevance in regional models because of intra zonal nature). Addional linked trips related to more local travel opportunies Acvity based and tour based modeling Good for truck models. Table 4. Potential smart-growth adjustments for four-step models.

38 Smart Growth and Urban Goods Movement 3. Land-Use Mix Consolidated freight facilities. Special trip-generator zones such as ports or major warehouse areas could be added to a model to reflect freight-oriented land use and consolidated ware- house areas or facilities. 4. Logistics Time and size restrictions. The impact of time-of-day or vehicle-size restrictions on a firm’s logistics decisions can be reflected in a model network structure (such as road links limited to smaller trucks) and a trip-assignment step (time-of-day travel time limitations). 5. Network System Management Operational efficiencies on a transportation network due to better traveler information or metering can be addressed by modifying a model’s networks, volume-delay functions, and other model parameters. 7.2 Alternative and Future Modeling Approaches Alternatives to the four-step model are both already in use and being developed. These models can be placed into two broad categories—activity-based and commodity-based (NCHRP Synthe- sis of Highway Practice 406 [Donnelly et al. 2010] and NCHRP Synthesis of Highway Practice 384 [Kuzmyak 2008]). Both styles of models have been used because they are seen as an improvement over the traditional methods of forecasting freight. In a number of cases, they also may be better at representing the impacts of smart growth. 7.2.1 Activity-Based Models A number of activity-based models have been applied or are under development in the United States (NCHRP Synthesis of Highway Practice 406). This family of models, also known as trip- based or tour-based models, uses a demand-based approach. Unlike the traditional four-step model that uses single trips as the basic modeling step, these models forecast flow based on travel demand derived from activities that people (or goods) need to perform. Travel is based on the activities households or individuals wish to complete during a day and is modeled in terms of tours. This is a significant modification to the four-step approach. Activity models offer a more effective approach to modeling smart growth because they rec- ognize that trips made by truckers are not independent of one another but rather are often con- nected for efficiency or convenience. 7.2.1.1 Calgary’s Tour-Based Commercial-Vehicle Model For freight modeling, a notable example is the Calgary, Ontario, Canada, tour-based commercial-vehicle model (CVM) (Kuzmyak 2007; PB Consult Inc. 2007). The model was originally developed for passenger travel, and trucks were forecast by the scaling of truck flows for counts. The CVM is a combination of three distinct models. Taken together, these three market segments are estimated to account for about 10% of the total travel in the region. The elements include the following: • A tour-based microsimulation model of internal commercial trips that captures travel for business purposes, such as delivering goods or performing services • A fleet-allocator model that models travel business for vehicle fleet-management purposes, such as taking vehicles off-line for repair or returning them to the business establishment at the end of the day, as well as travel by establishments that use a large, coordinated fleet that tends to service an area rather than specific demands; examples include mail and courier ser- vice, garbage hauling, newspaper delivery, utilities, and public works • An external trip model that captures travel to and/or from outside the region, as made by medium and heavy commercial vehicles

Urban Truck Freight Models 39 The tours simulated by this model are derived from travel-diary surveys conducted at 3,000 businesses. The numbers of trips in tours are decided by an aggregate trip-generation module, and then each trip is completed using a random (Monte Carlo) process. This model has a stop duration module that could also be modified to account for the smart-growth impacts such as limited curb space and significant interaction with non-motorized travel modes. Tours are also given start times, which allows flexibility in responding to time-of-day restrictions. 7.2.1.2 Atlanta Regional Commission (ARC) Model The ARC model is a trip-based model with modules that are designed to work backward from truck-count data to create a zone-to-zone matrix of trips. This approach emphasizes truck counts. The model also has specific truck zones with truck stops, warehouses, distribution centers, and so forth. The purpose of identifying truck zones was to capture the higher truck-trip-per-employee rates that are likely to occur in those areas (as noted earlier, employment-based trip generation has been shown to be ineffective for truck trips). This model is activity based on and uses dynamic traffic assignment. 7.2.2 Commodity-Based Models Freight movement is a derived demand related to the need to move commodities, not vehicles, in our economy. Critics of traditional freight models suggest that a commodity-based (as opposed to a trip- or vehicle-based) freight model is structurally superior. One major limitation to this family of models is a notable lack of commodity flow data—there are not any fully commodity- based urban freight models currently in use (Kuzmyak 2007).