Click for next page ( 90


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 89
Methods for Detailed Analysis G-89 Shippers can provide information on the cost and schedule reliability characteristics of their shipping services. If data can be collected or estimated for a typical group of shippers, then a logistics cost model can be used to estimate the cost and service characteristics of the competing modes, and that information can be applied to estimate resulting changes in mode shares. The database can include actual and/or hypothetical data. The advantage of using actual data is that the study will be more realistic and more believable; the disadvantage is that it may be very time-consuming and costly to collect the data. The advantage of using hypo- thetical data for typical cases is that the study can produce some results very quickly; the disadvantage is that it may be difficult to ensure that the typical case is truly representative of actual conditions. For bulk shippers, it may be possible to identify a small number of customers currently using truck who would be excellent candidates for using mini-unit trains. If the shippers cooperate, it will not be difficult to obtain the relevant information concerning the commodity, the customer, and the modal options. For containerizable freight, there will be many more potential customers, and a survey will be more difficult. 5.4 Estimate Truck to Rail Diversion 5.4.1 Overview This step estimates project effects on freight traffic diversion, i.e., the expected level of freight movement likely to be shifted from congested roads to new, better, or expanded rail services. It builds on the analysis of logistics cost and service quality features and tradeoffs identified in the preceding step to identify the potential for a project to allow some customers to save cost by shift- ing from truck to rail. 5.4.2 Components The analysis of freight modal diversion involves two elements: Mode Choice and Modal Share Analysis to estimate changes in rail and truck modal shares asso- ciated with proposed project investments; and Sensitivity Analysis to estimate the extent to which small refinements in the proposed project can make rail more attractive than trucks. 5.4.3 Background The rail and truck shares of freight trips are the result of decisions by many different shippers. Even within a single company, there may be different transportation requirements for various shipments involving different origins and destinations. For some of these shipments, rail or intermodal could be the obvious choice, but for others, truckload or LTL could be preferred. Hence, shippers and their customers are likely to select multiple freight modes. Policies can be established regarding when it is appropriate to use each mode. There may even be traffic man- agers who do not ship by rail because of bad experiences, no matter how long ago and no matter how compelling the economics of using rail. Over time, customers' overall use of rail often changes, partly in response to changes in freight service, but also in response to changes in how they manage their supply chains. Generally, modal choice models work on an aggregate level that ignores the idiosyncrasies of individual firm decisions. Instead, such models work by estimating the impact of cost and other shipping changes on the overall share of shipments moving by each mode, given a particular commodity mix.

OCR for page 89
G-90 Guidebook for Assessing Rail Freight Solutions to Roadway Congestion 5.4.4 Factors Any analysis of existing freight mode split or future freight modal diversion is necessarily based on consideration of six key factors as follows: The mix of commodities moving to, from, or through the study area or corridor; Existing rail and truck mode shares for those commodities and industries; The availability of rail options for commodities now traveling to/from the area by truck; Carrier service and cost features for rail and truck options (discussed in Section 5.2); User logistics costs associated with rail or truck options (discussed in Section 5.3); and Additional taxes, fees, or subsidies that affect decisions about rail or truck choices. 5.4.5 Methods The analysis of modal diversion can be viewed from two perspectives: (1) from an individual case perspective, in which a mode choice model identifies the best and most likely mode choice for a given type of business, commodity, and origin-destination combination, or (2) from an aggregate perspective, in which a modal share model estimates the overall portion of shipments moving by each mode, given a mix of business types, commodity types, and origin-destination characteristics. In fact, a common approach spans both perspectives by applying a mode choice model for a representative set of individual cases and then developing a weighted sum of those cases to estimate aggregate mode shares. In most individual situations, one mode will clearly be the best, so it will be expected to cap- ture all of the freight. Still, in many situations, two or more of these models will be close. For pol- icy analysis, it is generally more realistic and more informative to assume a mode will get some of the freight if its total logistics costs are close to the other modes. The modal shares for these cases can be estimated by comparing the total logistics costs for rail, rail-truck intermodal, and truck. Various techniques can be used to estimate modal shares given the total logistics costs for each mode. The math can become complicated, but the logic is simple: if the total logistics costs are about equal, then the two modes should be predicted to each get about half the freight; as the total logistics costs for one mode increase, then its share should go down; if the total logistics costs for one mode are much higher, then it should not be expected to carry any of the freight. Two approaches are commonly used to calculate these shares and the effect of proposed projects on them. They are discussed below. Approach 1: Use Logit Models of Discrete Choice Decisions Logit models, which have been extensively used in modeling mode choice for commuters, are statistical models that allocate mode shares based on a comparison of the "utility" (estimated overall benefit) of each available mode of transportation. The basic form is as follows: Mode share (A) = e-U(a) / (e-U(i) ), for all modes i In this equation, U(i) is the utility associated with mode i. For freight analysis, the total logis- tics cost has most commonly been used as the predominant measure of utility, so that this for- mulation can easily be used with the logistics cost model. Approach 2: Use Statistical Analysis of Logistics Cost Variation A second approach is to assume that the estimates of total logistics costs are the expected values of a random variable that is normally distributed with a known variance. The mode split can then be thought of as being the probability that the logistics costs of the mode are in fact lower than the logistics costs of the other options. If the estimates of logistics costs are very good and if the analysis includes all of the variables used by the shipper, then the standard deviation of the total logistics costs

OCR for page 89
Methods for Detailed Analysis G-91 will be small (this is the kind of analysis that a shipper will perform--identify the best option and use it). If the estimates of logistics cost are less precise and if it is unclear that all important elements have been properly included, then the standard deviation of the total logistics costs will be larger (this is the more usual case for a researcher or a planner). The difference in the estimated costs can be compared to the standard deviation of the costs in order to estimate the probability that one cost will be lower than the other. While this requires complex mathematics, spreadsheets typically have a function that will return the probability that a is less than b, under the assumption that a and b are the expected values of normally distributed random variables with a known standard deviation s (in Microsoft Excel, the desired probability is calculated as NORMDIST((a-b)/s,0,1,true)). Policy Analysis Policy analysis involves re-estimating the truck and rail mode shares, using either of the above- cited techniques, while varying the assumed values of costs and service levels associated with those alternatives. This tests the sensitivity of the results to variations in the assumptions. It is useful to show how changes in mode characteristics (e.g., rates or service quality) will affect the split of mode shares. For example, public subsidies of or investments in rail could be reflected as a change in serv- ice, a changes in rates, or a change in loading/unloading costs, depending on the investment. Public investment in rail-truck intermodal could be represented by adding an intermodal option or by changing the characteristics of the intermodal option. Public actions that increase costs to highway users (such as tolls) could be reflected in the truck characteristics. To conduct this type of policy analysis, it is necessary to create a database to represent the profile or mix of shippers and shipments that move freight over a corridor, through a city, or within a region. The database needs to have information related to a sample of origin/destination movements to which the logistics cost model can be applied. The database will therefore need to have customer, commodity, and carrier characteristics for a representative set of movements. Using these data, the logistics costs and then the mode share can be estimated. The effects of a proposed project, change in operations, or a new pricing strategy must be translated into changes in the commodity, customer, or carrier characteristics, so that the logistics costs and estimated mode shares can be re-estimated. Two types of studies can be done, one using actual data and the other using representative but hypothetical data. The advantage of using actual data is that the study will be more realistic and more believable; the disadvantage is that it can be very time-consuming and costly to collect the data. The advantage of using hypothetical data is that the study can produce some results very quickly; the disadvantage is that it can be difficult to ensure that the hypothetical data are com- pletely realistic. Examples of both approaches are provided in the collection of Project Resources, cited in Chapter 6. For bulk shippers, it might be possible to identify a small number of customers who are currently using truck and who would be excellent candidates for using mini-unit trains. If the shippers coop- erate, it will not be difficult to obtain the relevant information concerning the commodity, the customer, and the modal options. For containerizable freight, there will be many more potential customers, but it will still be possible to conduct a survey to obtain representative information concerning commodity, customer, and modal characteristics. For either situation, it will also be possible to use data representing a hypothetical set of customers. This approach can be useful because it allows rapid assessment of the relative merit of various changes in the freight system. 5.4.6 Required Resources To study the potential for freight traffic diversion, it is necessary to develop profiles of com- modity mix, shipper/customer types, and carrier price and service characteristics. Then a modal