5. STEP 3: TRANSPORTATION IMPACT TOOLS

This chapter describes the third step in the analysis framework, which is the analysis of transportation impacts. Given the type of project (as described in Step 1), the transportation analysis estimates the effects in terms of modal performance and cross-modal diversion, and the implications of those changes for shipper logistics. This analysis provides a basis for the economic analysis in the subsequent step. It is organized into five parts:

1. Initial screening to define transportation efficiency benefit;
2. Mode-specific performance analysis for freight;
3. Modal diversion analysis for freight;
4. Treatment of carrier and shipper cost; and
5. Final analysis and presentation of results.

5.1 Phase 1. Identification of Transportation Efficiency Benefit

The classification of project types in Step 1 (Chapter 3) leads to an initial identification of the relevant types of transportation impacts and their measurement. In general, these can be classified into five types of impacts:

1. Travel time improvements. Savings in the average travel time for existing and projected shipments given their origins and destinations. This reflects changes in average speed and schedule frequency of service, and also accounts for effects of predictable congestion bottlenecks and transfer delays. Improvements in travel time are also used to evaluate improvement in shipper access to markets (for buyers/suppliers), border crossings, and other modal facilities (seaports, airports, intermodal rail terminals).
2. Cost savings benefits. The net reduction in freight transportation cost associated with transportation system performance improvements. Typically methods of estimating cost savings account for changes in cost of driver time, vehicle (or vessel) operating cost implications, and changes in effective vehicle capacity (cargo volume).
3. Reliability benefits. The reduction in nonrecurring delays associated with improvements that reduce the frequency of traffic accidents, vehicle breakdowns, and other nonpredictable situations that lead to variation in travel time. These benefits can be measured in different ways depending on how the improved reliability is achieved, and who will benefit from among carriers and users. Examples of typical metrics would include hours of nonrecurrent delay (which may be calculated through regression models that relate nonrecurrent delay to traffic volumes, roadway classification, or other network characteristics), buffer time, or on-time performance percentage.
4. Cargo capacity benefits. These benefits are most commonly related to projects that increase net effective capacity and throughput at seaports, airports, or intermodal terminals, in light of existing capacity constraints.
5. Safety benefits. These can represent an environmental externality, but for carriers and shipper, they may also translate into changes in business costs for breakage, vehicle repair, insurance, and/or employee down time.

For any given project analysis, some combination of these direct transportation impacts will be measured and in most cases, the nature of impacts, lack of data or availability of models will result in not measuring all five benefit categories. There are two sets of tools that can be applied to predict and calculate the magnitude of major impacts of these types. They are 1) mode-specific performance models, and 2) cross-modal diversion models. Using these models (discussed in Section 5.2), it is possible to calculate how travel times, vehicle costs, reliability, capacity, and safety levels can be affected by a proposed project. Different models may be necessary depending on the specific modes involved.

The types of metrics that can be developed by applying modal performance and diversion models are summarized in Table 5.1. It is important to note that impacts are distinguished by mode of freight travel, and the analysis may need to also include passenger modes that share facilities (highways, tracks, harbors, airports) with the freight modes. In addition, it should be noted that some of these effects (such as changes in cargo tons, vehicle volumes, time and costs) come directly from transportation models, while others (such as market delivery and terminal access times) need to be derived by the analyst using available model statistics.

Table 5.1 Typical Data Used to Describe Transportation Characteristics/Impacts of Large-Scale Projects

Note: Form of freight distinguishes bulk, break bulk, container, truckload, and less-than-truckload shipments.

5.2 Mode-Specific Performance Analysis

Types of Models. For each mode of transportation (highway, rail, air and marine), there are models that analyze the demand for use of routes and terminals, the functional capacity of those routes and terminals, and resulting changes in their performance when faced with changing patterns of demand. The utility of this class of models for impact analysis is their capability to simulate how improvements to network or terminal facilities can improve performance in terms of increasing speeds, reducing bottlenecks and schedule delays, reducing operating costs, and improving safety.

The models for highway and rail systems tend to focus on network performance, since the major capacity constraints for those modes are the travel network links. Likewise, the models for air and marine transportation tend to focus on terminal or port performance, since they are the major capacity constraint for those modes. As a practical matter, highway network models are more available than models for the other modes. Models for rail, air-freight, and water-side maritime operations are rare (see Chapter 10 for further discussion).

Depending on the type of project, all of the various modes can be relevant for analysis, and any of the network or terminal performance models may be relevant for use in project evaluation. However, in the context of large-scale freight projects, it is most useful for this guide to focus on special challenges posed by projects that involve multiple modes of freight. In fact, nearly all large-scale freight projects are multimodal and involve some combination of rail-truck, port-rail or port-truck or airport-truck interface. All of the case studies described in Chapter 8 also illustrate these same modal combinations and interactions.

Thus, in the interest of brevity, we have relegated the overview of network and terminal performance models to the Chapter 10 Toolbox. That chapter discusses the availability of individual performance models for highway, rail, aviation, and marine transportation. The rest of this section focuses just on the most common problems facing multimodal freight facilities, which focus most on network performance for railroads and reliability for road systems.

Major Challenges for Network Model Analysis. A critical question in this type of analysis is whether local congestion impacts have any benefit to users other than the limited effect they have on carrier costs in a local market. The analysis needs to first determine how the effects are translated into long-haul travel time savings. Examining the origin-destination (O D) characteristics of freight movement using the facility is a way to begin to gauge the extent of the network’s influence. For example, the proposed Cross Harbor freight rail tunnel in New York City (see Chapter 8 case study) would impact freight rail trips originating in Atlanta and Chicago. Also, facility-related trips that require intermodal linkages add another layer of complexity in understanding end-to-end goods movement.

Another issue is idiosyncrasies from network reconfiguration. This issue focuses on whether a local congestion hot spot once alleviated would lead to major re-routing of traffic in a larger network. This can be detected by a standard travel network model as long as the area is large enough. The other issue is whether cleaning up one bottleneck just creates a new bottleneck downstream. Some have argued this was the case with the Alameda Corridor. There are new techniques being developed that prioritize bottlenecks based on overall system performance impacts that may be helpful in this regard.

Traffic Simulation and Highway Network Analysis

Most of the examples provided in this guidebook conduct highway network impact analysis using travel demand models (these are the most commonly encountered tools for evaluating travel time savings and related network effects). However, in the case of projects that deal with freight bottlenecks, traffic simulation models may be a necessary tool because they take into account queuing behavior that can build over time at a bottleneck. This type of queuing will affect reliability and overall travel time over long periods of time as capacity of the system is significantly exceeded and may provide an indication that the facility is saturated and cannot accommodate more growth. The implication this has for modal diversion, diversion of traffic to another geographic area, or increased overall cost of freight movement may be far greater in these situations than would be indicated from the results of a travel demand model. While traffic simulation can be expensive, there are simple simulation tools that are being used to provide some initial indication of this type of bottleneck. In cases where the investments are significant and the bottleneck is bad enough, use of more sophisticated simulation tools may be warranted.

Special Issues for Railroad Networks. Special challenges are involved in examining railroad network performance. Since highway facilities are publicly owned, the responsibility for analyzing their performance falls to state and metropolitan transportation agencies across the nation. As a result, the available tools for evaluating highway facilities, including their corridors serving marine and air ports, are commonly available. However, railroad facilities are usually privately owned, and thus most public agencies have far less familiarity with data and tools for evaluating rail system performance. That also makes it particularly critical to improve methods for assessing the public benefit for improving freight rail facilities. Accordingly, it is useful to focus particular attention on challenges for evaluating freight rail networks serving urban ports, terminals, and intermodal truck-rail facilities.

Railroad simulation models are used to evaluate track configurations, signal systems, and operating plans. These models generally mimic train dispatcher logic and are used to evaluate infrastructure and/or operational changes. A common use is to evaluate the running of passenger and freight trains over the same track to identify bottlenecks and capacity constraints. Most models produce schedules, string line displays,⁴ and various performance measures permitting comparison of alternative scenarios. These simulation models do not provide a direct measurement of capacity, but are used to identify potential capacity problems.

This class of models is designed to simulate the decisions made by train dispatchers. They do not, in general, contain optimization or other decision-making components. They do follow a set of fixed rules governing train priorities and a train performance calculator to model train physics (acceleration and deceleration). By providing track configuration, signal systems, and operating plans as input, an experienced user can evaluate the outputs to determine bottlenecks and conflicts. Adjustments are made to the inputs to resolve these conflicts (typically adding and/or lengthening a siding, double tracking, or adjusting train schedules).

In addition to the simulation models, there has been recent interest in developing parametric rail capacity models. These models develop capacity curves for various operating characteristics and, based on the operating plan profile of a rail line, identify areas with capacity constraints. They are much less data intensive than the simulation models. Parametric models can help identify capacity “hot spots,” which would then need to be further explored with a simulation model.⁵

Railroad operation and impact models tend to be very data and labor intensive. They are used internally by the railroads and for large-scale projects and mission critical analysis. Because of the effort and cost of these specialty models, they are more appropriate for a detailed design phase than a preliminary benefits phase. There is a need for simpler, sketch planning rail models to answer a few questions at a more general level, such as: 1) how many trains will run through my town? 2) Will this project improve freight rail service? 3) Will other investments be needed to fully achieve the benefits?

Special Issues for Highway Reliability. For freight movement, schedules and travel time reliability are important, particularly for time-sensitive shipments that tend to travel via truck or air-truck combination. Methodologies that are being used and developed for transportation analysis have been focused primarily on highway network systems, although there are logistics process models from the industrial engineering/operations research fields that are used in the rail industry to predict the cost implications of reliability changes on production processes.

If reliability is defined as variability of travel time, very few of the methodologies above address true reliability estimation. This topic is currently being addressed for highway networks in NCHRP 7-15 and sketch planning methods may be developed as a result of this work that will prove valuable in the future. Actual variability in travel time may be important for economic impact evaluation for several reasons:

• Carriers may react to travel time variability by planning for mean delay (or some threshold of delay). In this case, knowing the distribution of incident-related delay is as important as knowing the cumulative total incident related delay. The planned-for incident related travel times can be used to estimate carrier costs regardless of average recurrent congestion conditions calculated by traditional planning travel demand models. The same value of time can then be used to begin calculating the economic impact of recurrent congestion.
• Shippers (freight system users or customers) may have a window of on-time performance before intermodal connections are missed and production processes are affected. Knowing how often this window is missed may be important in logistics process models.

Methods that are generally available for reliability analysis of highway networks are not true reliability predictors. Rather, they estimate cumulative incident-related delay, often as a function of volume/capacity ratios. The economic impact of travel time variability is effectively taken into account by valuing incident related delay at a much higher level than recurrent delay. A system that does this explicitly is the benefit/cost component of the ITS Deployment Analysis System (IDAS) developed for the Federal Highway Administration (FHWA), which values nonrecurrent delay at three times the value of recurrent delay.

A number of studies have been conducted over the years that attempt to estimate the impacts of incidents, but since they employed techniques that do not directly predict incidents the results are limited when analyzing the effects of system improvements.⁶ Most of these techniques are based on traffic engineering methods and often rely on microscopic simulation methods.

A useful approach for estimating travel time variability for highway mode is also included in the FHWA’s 1998 study on Sketch Methods for Estimating Incident-Related Impacts. The method computes vehicle-hours of incident-related delay for freeway corridors based on defined characteristics including: number of lanes, free-flow speeds, V/C ratios, accident rates, incident duration factors, and presence of recurrent bottlenecks. The method was developed using a combination of macroscopic simulation methods, queue analysis methods, and stochastic procedures, and it represented a precursor to IDAS. As applied in several of the economic impact case studies presented in Chapter 8, the lookup tables of nonrecurrent delay impacts have been used as a post-processor with travel demand models.

It should be noted that none of these methods takes into account the performance benefits of truck-auto separations and truck tollways.

Safety Impacts. Though the motivation for large-scale freight projects is commonly more efficient movement of goods, another motivation and benefit can be improved safety and reduced accidents. This is particularly important when projects help to reduce intermodal interactions (such as reduction in road-rail grade crossings, or diversion from congested areas). For example, freight rail improvement projects that help divert truck traffic to the rail system can result in fewer highway accidents. Or, reducing at-grade rail-highway crossings can also improve safety at the same time as it improves efficiency (as illustrated in Chapter 8 case studies, including the Alameda Corridor and Chicago CREATE). Relevant issues regarding the modeling of safety benefits include:

• Statistical methods to estimate changes in accidents of varying severity;
• Numerous user benefits models have algorithms that estimate accidents based on factors such as facility class, travel speed, capacity, and traffic volume; and
• Accident impacts typically arise in the context of highway network analyses when reductions in truck VMT occur or improvements are made to the functional class or geometry of a roadway.

Despite these methods, there are few other predictive models available and most other modes are usually not evaluated in terms of accidents for large-scale projects. Consequently, few analyses contain detailed evaluations of accident reductions.

5.3 Modal Diversion Analysis

Types of Cross-Mode Substitutions. Any improvement to facilities and services of one transportation mode can have implications on demand and performance of other modes. This can occur insofar as there are substitutions between air and sea for overseas shipping or between truck and rail for domestic surface shipping. Of course, substitution is also possible between other modal combinations for specific short-, medium-, and long-distance shipping. This can include short sea barge shipping in place of rail for heavy cargo, or air in place of trucking for container cargo, as well as other combinations.

Two classes of tools are available for analyzing modal choice and diversion among freight shippers: statistical models of market shares and total logistic cost models. Both calculate how shippers change their mode choices in response to changes in the various service features and costs of modal options (most frequently rail and truck modes). The availability and features of such models are discussed in the Chapter 10 Toolbox. In the interest of brevity, the discussion here illustrates the most common modal diversion issue facing large-scale multimodal freight investments, and that is the impact of rail-related investments on truck and other highway vehicle traffic.

Truck-Rail Diversion Issues. Experience suggests that the estimates of net reduction in trucks on the roadways due to rail improvements is often viewed with skepticism by public officials because of the complexity of the issue, the risks involved, and the impacts these estimates have on public benefits. It raises many additional questions. If the public invests in a freight rail line, will the railroad improve service and/or lower costs to attract new business? If so, will the shippers respond by diverting traffic from truck to rail? How will changes in shipper logistics patterns and costs ultimately impact the consumer, who paid for part of the rail improvement through taxes?

A mode choice, or diversion, model for truck and rail choices is used to determine the extent to which mode shares change, given a change in any of the transportation service attributes. Mode choice for freight shipments is based on three primary factors: goods characteristics; modal characteristics; total logistics costs and supply chain design. The factors impact the feasibility of freight rail diversion in different ways:

• Goods characteristics. Shipment size, package characteristics, shipment shelf life, shipment value, shipment density; some goods are simply not suited to rail carriage (e.g., pharmaceuticals), while others are not suited to highway carriage (e.g., coal).
• Modal characteristics. Capacity, trip time, reliability, equipment availability, customer service and handling quality, modal cost; these characteristics, some of which can be changed through rail freight investment, interact with the goods characteristics to determine the feasibility of movement by rail.
• Total logistics costs and supply chain design. Even if the goods are well suited to rail transport, and rail service is available, the design of the logistics chain may be such that trucks provide more economical and dependable service. In cases where truck travel times are significantly faster than rail, the inventory carrying costs of expensive goods, or an environment requiring short lead time, can overcome higher transportation rates.

The importance of modal diversion analysis for major investment projects is that such impacts will also affect carrier costs for both modes and it will change the calculation of cost savings for shippers. The treatment of those costs is discussed next.

5.4 Treatment of Carrier and Shipper Costs

The first component of economic analysis of transportation investments requires an assessment of carrier response to transportation improvements. For most analyses, transportation investments will reduce operating costs of carriers by introducing or improving infrastructure (ports, roads, etc.) that carriers use. In many cases, analyses focus on the (positive) impacts of transportation investments relative to “do-nothing” scenarios under which existing infrastructure degrades and/or cannot accommodate expected growth in demand.

Link Between Carrier and Shipper. For purposes of discussion, assume that a proposed transportation investment will lower operating costs of carriers by reducing congestion delay. As such, the first component in an economic analysis must focus on the link between a reduction in operating costs for carriers and a reduction in prices that carriers charge (and shippers pay).

There are two extreme cases. In the first, there is a monopoly carrier that does not pass along any of the reduction in operating costs in price reductions, thereby, raising its profits by the amount of operating cost reductions. In this case, the national economic impact will be reduced to spending generated by the increase in profits for company owners or shareholders; the local impact will be determined by the proportion of owner/shareholder profits that stay in the local economy. These local and national economic impacts do not necessarily reflect net gains. Calculation of net gains would require comparison of the cost of public investment with private gains by carriers. If the cost of public investment is greater than the private gains, then the net national impact will be negative. If the cost of public investment is less than the private gains, then the net national impact can be expressed as the difference between public and private gains. (Note that in this case, efficiency gains are a product of the investment itself, which has greater benefits than costs, rather than to the behavior of carriers or shippers.)

In the second extreme case, perfect competition causes carriers to pass along the entire reduction in operating costs in the form of price reductions for shippers (and receivers). In this case, the national economic impact will be a function of productivity gains to local carriers, which will face higher demand for their products and thus could achieve economies of scale;⁷ productivity gains that accrue to freight users, who can now produce a given amount of output for fewer inputs; and any business relocation and economy of scale impacts these productivity gains generate. Local economic impacts will be driven by increased demand and output for carriers and increased business activity associated with lower costs and increased output at existing shipping firms in the local area and any relocation gains that accrue.

It is unlikely that the first of extreme cases occurs very often in the real world. When faced with falling operating costs, even pure monopolies should lower prices (though far less than firms in competitive markets) and increase output (though far less than firms in competitive markets) in order to maximize profits.⁸ Firms in perfectly competitive markets will employ marginal cost pricing. Thus, to the extent that transportation improvements reduce marginal (rather than fixed) costs of carrier operations, the reduction in operating costs will be wholly reflected in price reduction to freight users in perfectly competitive markets.

Although the market for freight services is not perfectly competitive in all (geographic) markets, the default assumption in analyses of transportation investments is usually that cost reductions for carriers are passed onto freight users. This assumption is made because of the levels of competition thought to characterize freight markets in the wake of international competition (e.g., between Canadian and American ports) and deregulation efforts in air, rail, and trucking services. Empirically, however, it is difficult to determine a priori the effect of a carrier cost reduction on prices faced by shippers and receivers. However, a recent examination of rail freight rates found that rates vary by rail line/location and commodity being shipped (GAO, 2002). The latter study presented evidence that on some lines, “railroads did not pass on all cost reductions to customers in the form of rate reductions” and concluded, among other things, that a range of factors, including local competition in freight services (rail and nonrail), influences rail freight rates.

Given the importance of carrier pricing to estimates of the economic benefits of transportation projects, it is important that analyses include thorough consideration of likely responses of carriers to changes in operating costs brought on by transportation investments. It does not appear, however, that methodologies to address this question have been fully developed and unfortunately, economic models commonly used in large-scale transportation project analyses cannot usually be used to examine the likelihood of or size of price reductions associated with a reduction in operating costs for carriers.

Research on this topic suggests that freight pricing issues can be sensitive to context, so analysts should consider whether the level of competition in freight markets is reasonably competitive before (and after) the project investment. If so, it is likely that the assumption that freight users capture all or most of the benefit of cost reductions for carriers is valid. If the local freight market will not be competitive even after the project investment, analysts must consider how accurately to capture the links between transportation investment, reduction in operating costs for carriers, and prices paid by shippers. Nonetheless, given the long-term impacts associated with freight transportation investments, it is likely that even if cost reductions aren’t immediately passed along to freight users, that over time, shippers and receivers will experience a benefit given the generally competitive freight transportation industry.

5.5 Final Analysis and Presentation of Results

Multiple Measures and Perspectives. The final results of the transportation analysis should provide a series of findings that are summarized in Table 5.2.

Table 5.2 Example Portrayal of Findings from Transportation Analysis

These results must be consistent with three needs:

1. First, analysis should portray how proposed changes in network and terminal facilities will affect system performance by mode. That may include changes in average travel times, flow volumes, shipping costs (per ton or TEU), reliability and/or safety. This is represented by the first row in Table 5.2.
2. Second, they must show the volume of freight (tons or TEUs) by type that is projected to be subject to these transportation performance improvements. That must account for baseline forecasts and any modal diversion as well as any changes in activity levels due to elimination of capacity constraints. This is represented by the second row in Table 5.2.
3. Third, they must portray how the changes in system performance translate into direct benefits for freight shippers who are the users of the freight transportation systems. That may include transport costs or savings passed on by carriers as well as logistics and delivery access impacts. These impacts are represented by the third row in Table 5.2, and they actually form the basis for economic impact modeling as discussed in the next chapter.

There are several other notable elements of Table 5.2. One is that the results are shown by mode. That is needed so that the final analysis of benefits (discussed in Chapter 7) can distinguish the incidence of impacts by mode as well as the associated public-private cost and benefit allocation. Another notable feature is that impacts on shipper cost (given projected freight origin-destination flow patterns) are complemented by separate measures of other changes in system throughput and market access (or connectivity) impacts. All of those other forms of transportation impact can then be assigned additional benefits using procedures described in Chapter 6.