Rail Freight Solutions to Roadway Congestion--Final Report and Guidebook (2007)

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.

107 6.1 Introduction The purpose of this chapter is to review for planners and analysts the range of data sources that are useful for assessing rail freight solutions to highway congestion. It describes each type of data source in turn, explains why it is useful, and tells how to collect or acquire the data. This chapter also assesses the adequacy and limitations of the currently available data and suggests ways in which information could be further enriched. Its central focus is data sources that can support the modeling framework presented in the Guidebook. It is meant as a companion to that framework and generally will not repeat treatments (including data points and rules of thumb) that may be found there. There are various levels of knowledge and increasing degrees of detail and sophistication in collecting data. What is appro- priate can depend on the size of the project or opportunity, its phase of development, and the capabilities and resources of the decision makers. To come to grips with the options for rail, there is a need to understand how well the system is function- ing, where it could do better, and where there are opportuni- ties. A motivated and informed group of public officials, freight carrier officials, and chamber of commerce people may be able to provide workable answers to the central questions, if they can be brought together to look at them. In this sense an “expert system” can substitute for data in some stages, dimen- sions, or magnitudes of projects. Ultimately, data are a means to an end: they are vital, but there can be alternate ways to reach an objective, both in the kinds of information utilized and the kinds of approach taken. The Guidebook’s framework moves progressively through analyses of increasing complexity, and data should be thought of as following in its path. Data of several sorts may be needed to support freight planning studies: • Traffic flows, depicting freight by lane and mode—for identifying trends, traffic distances and densities, and diversion possibilities; • Traffic volumes on infrastructure—for determining truck contribution to highway demand, overall traffic activity, and rail requirements; • Congestion on highways—which is the specific problem being addressed, and information will be needed in areas like level of service, recurring and non-recurring problems, and temporal variations; • Freight customer characteristics—such as who is shipping or receiving what commodities within the area of interest, annual volumes, service sensitivity, loading/unloading needs and capabilities, rail access, and modal usage; • Commodity characteristics—value per pound, product density, perishability, storage requirements, equipment needs, and so forth; and • Carrier characteristics—such as actual or typical service, cost parameters, capacity indications, and asset ownership. The different data needs are diagrammed in Figure 6-1. Commodity flow data document various aspects of ship- ments, which go from a particular origin by a particular mode to a particular destination. Commodity flow data will identify the type of goods, but will not directly give important attri- butes of the commodity, such as value, density, shelf-life, and special packaging requirements. Nor will the flow data provide direct information about the shipper or the receiver, although data will give some indication. Attribute and customer infor- mation, which are pertinent to mode choice, must be obtained from another source. State and local transportation planning typically is less well supported for freight than for passenger planning. Local movements are the predominant form of passenger travel, with most travelers beginning and ending the day at home. Planning procedures have developed that include sophisti- cated network models, frequent surveys of travelers, well- funded data collection efforts, and large planning groups at the state or regional levels. Freight is much different. Freight travel covers a broader region, with trip lengths an order of magnitude longer than passenger trips. Surveying is difficult C H A P T E R 6 Data Sources

because a diverse range of industries is involved, with a signif- icant portion of mode and route decisions not made locally and shaped by shipment staging and supply chain structures. Trips typically are one-way, displaying seasonal as well as diur- nal patterns to flows. Planning procedures on the whole are not well developed, nor are the data sources as good as those available for passenger planning. The private sector is much more dominant in freight transportation, so that knowledge about the freight infrastructure, freight flows, and the charac- teristics affecting freight demand is seldom automatically accessible to public agencies. Nevertheless, a great deal of information about freight and freight transportation is available. This chapter discusses var- ious sources of data and the procedures that planners might use to obtain relevant input. It is beyond the scope of this report to present detailed strategies for assembling compre- hensive databases suitable for all levels of freight planning. Instead, this chapter considers the primary data sources and discusses each of them. It does not demonstrate how disparate sources of data could be joined together or manipulated to derive insights regarding freight transportation or economic development. To understand data joining and manipulating techniques, other NCHRP reports on freight and economic planning1 should be reviewed, particularly NCHRP 8-43 which treats freight planning in the statewide context. Many of the data sources discussed are readily available, either as part of the data sets that State DOTs, economic devel- opment agencies, toll authorities, and other public agencies already collect or as part of a commercial data service. However, there are many incremental ways in which State DOTs could further leverage their freight data streams. With some effort and outreach, public agencies should be able to collect or assemble data and gain analytical insights that may not be immediately at hand. 6.2 Practicalities Together with the Guidebook, this report considers how the public sector can work more effectively with the rail industry to allow the rail system to carry more intercity freight. If there is a public/private partnership for a study, then some of the problems of data collection will disappear. The railroads are well aware of the strengths and weaknesses of their own services and facilities; freight customers know why they use trucks instead of rail and what it would take for them to shift freight to rail; highway officials know where roads are congested and where heavy trucks are most com- mon. The challenge is for the various parties in such a study to combine their knowledge and expertise in order to (1) identify areas where rail solutions may be effective; and (2) evaluate specific options for improving rail mode share. For freight, the questions, data requirements, and solutions are different from those commonly used in the four-step planning process for transportation planning, but it is possi- ble to assemble groups of knowledgeable people who can, as a group, identify workable strategies. One way to begin is with Freight Advisory Councils, which are becoming common fix- tures in state and urban jurisdictions and are employed by the federal government as well. They offer a proven and available method for making realistic assessments of the public plan- ning options and for opening doors to other stakeholders who can contribute requisite data and participate in project opportunities. These points suggest a basic and pragmatic orientation that planners should remember. Information is required to meet the objective of roadway congestion relief. Data are one way to 108 1. Commodity flow data 2. Traffic flow data 3. Commodity characteristics 4. Rail facility inventory 5. Rail engineering cost 6. Establishment data 7. Service performance data Figure 6-1. Rail Freight Diversion Data Needs. 1NCHRP 20-29 Development of a Multimodal Framework for Freight Transportation, NCHRP 2-19(2) Economic Development Toolbox, NCHRP Report 456: Methods to Assess Social and Economic Effects of Transportation Projects, NCHRP Report 463: Economic Costs of Congestion, NCHRP Synthesis 290: Economic Effects of Transportation Investments.

supply this information; direct observation and professional judgment are others. Data are especially useful as inputs into analytical models, but when they encounter limitations—and they often will—estimations can be adopted (as the Guidebook demonstrates). A combination of all these methods can be and usually is the way that projects get done. This means that while better data are desirable, a data challenge often can be reframed as an information challenge and solved in another way. For example, each of the remaining sections of this chapter addresses a specific kind of data and begins with a short sum- mary of why such data are needed. In each case, planners may only require general information, such as “where are the most congested highways where heavy trucks account for a signifi- cant portion of VMT?” A set of observant truck drivers who worked throughout the region could answer this question, probably with details concerning the time of day. The indi- viduals who report traffic conditions for the local radio sta- tions or who monitor operations for the DOT could also answer this question. Alternately, public agencies ordinarily have databases that incorporate traffic counts and estimates of congestion levels for segments of major roadways; from those, a planner can derive a list of segments with a level of service of D or worse where heavy trucks make up at least 15% of the traffic. Both kinds of sources produce practical information— yet it is helpful to note that it is not necessary to create a data- base to get a reasonable answer to the initial question. In general, any MPO or state transportation agency should be collecting data related to freight, including such things as potential rail customers, truck usage of highways, trends in truck traffic, and congestion levels on major roads. If they lack such data, they should initiate data collection efforts. However, the lack of such data should not be taken as an impediment sufficient to defer freight planning efforts. As the Guidebook describes, analyses can vary widely in their data intensity, depending on the scope of the problem and its stage of devel- opment, and there are a number of ways to get at them. It is usually possible to proceed on some basis, if not with detailed material then with the information and insights that can be provided by carriers and their customers, along with rules of thumb and whatever data is available to the public agencies. This chapter therefore is not a checklist of data sets required to begin a freight study. Rather, it is a guide to possible data sources that a planner will use flexibly, imaginatively, and dif- ferently, according to the needs of the problem and the options open at the time. It emphasizes alternative ways of developing information, because this is a practical and productive approach. Previous chapters of this report have cited almost two dozen rail freight projects in North America that have assembled information by the means described here, leading to investments of public funds across the range from small to very large. The types of data most commonly collected and used for assessing rail freight solutions are • Commodity flow data, • Traffic count data, • Commodity characteristics, • Maps & inventories of rail infrastructure and service, • Railroad engineering cost data, • Shipper characteristics & needs—establishment data, • Modal service and cost parameters, • Trend Data—traffic & economic projections, and • Institutional and privacy factors. The remainder of this chapter reviews eight types of data and concludes with a look at the institutional and privacy fac- tors that can affect the accessibility of information and the rules governing its use. The eight types correspond to the seven needs diagrammed above, with the addition of trend and forecast data. For each type, four sets of considerations are discussed: • What is the problem? What kind of data would be useful? • Are there readily available sources for the data? • How can the data be collected? • Levels of accuracy and precision. 6.3 Commodity Flow Data What is the problem? What kind of data would be useful? Commodity flow data are needed for two reasons: (1) to understand what type of freight is causing congestion; and (2) to determine whether such freight can, in fact, be feasibly diverted with a suitable rail freight service. When used in con- junction with other data, the information can also be used in determining a suitable rail freight service program. Commodity flow data also are an important driver for many types of forecasting activities. As a measure of freight activity levels from an economic perspective, they give insight into not only how much freight is moving, but also what type of freight is moving, which will begin to imply why the flow exists. The reason for freight movement can be important in predicting whether shipments will continue and whether they can be expected to grow. Are there readily available sources for the data? For rail traffic information, the key source is the Carload Waybill Sample (CWS), issued annually by the Surface Transportation Board. The Waybill Sample is a statistically based stratified random sample of shipments terminated by U.S. rail carriers. All carriers terminating 4,000 or more car- loads per year are required to report and 62 railroad systems thus are captured, encompassing all Class I and II roads, and 109

the more prominent short lines. (Carriers smaller than 4,000 annual loads may be sampled when they act as haulage agents for larger railroads, and the latter appears as the carrier of record on a shipment.) The full (and confidential) Waybill Sample file contains highly detailed information on the origin, destination, com- modity, and volume of each sampled movement. Intermodal and unit train traffic can be separated from single or small block carload, and the rail carriers handling the traffic are identified. State DOTs have access to this data source for activity in their state, subject to certain requirements on the confidential handling of the information. MPOs may petition their State for access and usually can gain it. A public edition of the Waybill also is available, with far less detail released but without privacy restrictions; in addition, there are commer- cial versions of the public data that interpolate some of the missing detail and can make it easier to use. A separate, semi- commercial source exclusively for intermodal traffic data is the Intermodal Association of North America (IANA), which publishes monthly flow volumes by trailer/container type between large geographic regions. These data have less speci- ficity but more currency than the CWS. For truck data in the public domain, the U.S. Bureau of the Census publishes the Commodity Flow Survey every 5 years, as part of its economic census. It provides a sound, basic body of information on the flow of goods between U.S. markets by mode, with highway freight traffic separated into for-hire and private fleet volume and multimodal activity identified. (The CFS also has information on shipments by other modes, although for rail the CWS offers more detail and usually is more current.) Flows in the CFS are released by state and major metropolitan markets, which can be adequate for some applications and are a good scaling tool for sketch planning. (The Guidebook presents an example of this.) The most recent CFS at this writing was completed in 2002. In addition, the FHWA’s Freight Analysis Framework con- tains a database of 1998 commodity flows for several classes of truck (and other modal) traffic, on a state-to-state basis. FHWA has begun the update of this information and issued a 2002 data set based on the CFS that (1) interpolates traffic flow information covered but not released by the CFS; and (2) supplements CFS coverage with modeled information, much of it covering truck activity. The geographic units in this FAF II data are the 114 regions utilized by the CFS. Truck data are available from commercial sources as well. Commodity flows between counties, based on proprietary samples and incorporating public information, are updated annually for various classes of truck fleets and equipment types. Data of this kind have been used in a variety of rail studies, including complex truck diversion analyses. Such databases usually incorporate the STB Waybill to facilitate modal comparisons; a typical example appears in Figure 6-2. Finally, truck trip tables often are a part of state and urban transportation models and can be adapted for other kinds of analysis. While commodity information sometimes is absent, the tables can be reflections of local survey work or an amal- gam of public and commercial data with local observations. How can the data be collected? Intercity commodity flow data is collected through sur- veys, exchanges, or legal reporting requirements. The STB Waybill sample is a mandated collection from rail carrier records, and the CFS as an aspect of the Census also is a legal obligation on freight shippers, who are its respondents. There are no similar reporting requirements for motor car- riers, and data sharing from this source has been accom- plished through voluntary surveys or through commercial trading of information. Data procured through a survey of selected shippers and receivers can be effective for locally based shipments, but will not pick up overhead traffic. Intercept surveys can detect overhead truck traffic as well as that locally based, although a comprehensive set of information can be chal- lenging to obtain. Both methods can be time-consuming and expensive, yet many studies have been successfully completed that relied on the use of survey data. Respon- siveness is a further obstacle. Because of growing privacy concerns and aversion to government monitoring (as well as the explosion of surveying and tracking in modern soci- ety), cooperation in surveys increasingly is difficult to obtain. One solution is the use of an intermediary (such as an external contractor) as a way to ensure confidentiality. Data collected by an independent third-party and blended with other information before delivery to the public agency can avoid subjection to the Freedom of Information Act and encourage participation by transportation stakeholders in data compilation efforts. Combining and cross-referencing information derived from multiple sources is one of the key techniques used by planners to assemble truck information. One or more forms of data collection—such as the shipper interviews, truck driver surveys, or intercept data mentioned above—are used with commercial or national information to produce a local picture against a regional backdrop. Data assembly is done within the processes of an urban or statewide travel demand model or processed more narrowly to answer the immediate needs of project or program development. Numerous state, urban, and project studies have been approached in this way, for overall freight planning as well as for rail initiatives.2 110 2The Metroplan Orlando (FL) Freight Goods & Services Mobility Strat- egy Plan and the New Jersey Portway Extensions project are just two of many.

Levels of Accuracy and Precision Commodity flow data is available at different levels of granularity. Geographically, the data can be collected at the state or county level or even by zip code or railroad freight station. Time-wise, the data might be compiled daily, monthly, quarterly, or annually. Different resolution is also possible along dimensions of vehicle types, commodity detail, and so forth. Daily or monthly data will capture diurnal variation and the effect of seasonality, whereas annual data will capture effects of longer term business cycles. Getting accurate sam- ples with shorter time periods or finer geographic granular- ity tends to be more difficult, because a larger sample size is required to estimate flow data with higher precision. Commodity flow data may be quoted to the ton or vehicle unit, but whatever its degree of historical accuracy, the cur- rent picture is subject to change. Variability from month to month or year to year can be significant, due to economic, seasonal, and random effects. However, when considering rail freight diversions, although a reliable base volume is going to be important, fine precision often is not as critical. A 10-percent error in the average base volume may only have a small impact on the viability of the service, compared with other factors such as the quality of performance or the con- dition of pre-existing rail freight infrastructure. 6.4 Traffic Count Data What is the problem? What kind of data would be useful? The first step in the analysis is to find sections of the high- way system that (1) are congested and (2) have a significant portion of heavy truck traffic. Facility-level traffic count data are needed first and foremost to assess the level of congestion on existing roads and highways and to determine whether or not trucks are prevalent in the congested area. Detailed facility- level traffic data will provide a base case for the development and evaluation of suitable strategies and solutions. The greater 111 Figure 6-2. Illustration of Waybill Database.

the congestion and the greater the proportion of heavy trucks, the more likely that a rail freight solution may be feasible. If traffic counts are available for multiple years, they may illustrate trends in congestion and in truck traffic. Since traf- fic counts do not show the origin or the destination of the traffic, this type of data is not sufficient for modeling flows over the network, although they are commonly used for cal- ibrating trip tables and flow model constraints. Moreover, detailed facility data (including breakouts at the equipment- type level) can be very helpful in determining the impact of proposed solutions and developing relief performance meas- ures. Similarly, volumes and trends for rail facilities are important if there are concerns for railroad capacity or to understand capital investment needs. Are there readily available sources for the data? Captured in the Federal Highway Performance Monitoring System (HPMS), state DOTs maintain databases of historical and current facility levels-of-service (LOS) between A and F, which is an indicator of roadway traffic flow and degree of congestion (traffic tie-ups are at level F). DOTs mainly derive this type of data through continuous or spot placements of automated loop count equipment, whose raw data streams are informative but require some processing to remove anomalies prior to use. The counts specifically measure the passage and timing of vehicle axles; algorithms then are used to provide an interpretation of vehicle type and other details, which may be stored in a DOT database. Counts have the additional virtue of depicting temporal patterns: time-of-day and day-of-week volume variations as well as seasonal fluctuations when traffic is monitored often enough. How can the data be collected? Traffic count data in fact are collected in a number of dif- ferent ways. The commonplace loop count data are regularly compiled by states along major routes and at strategic inter- sections to determine the passage of traffic. More advanced systems will differentiate between automobiles, light trucks, and heavy truck classes. Other advanced methods of moni- toring traffic exist and can enrich the data substantially beyond interpreted counts. Weigh-in-motion stations, traffic cam sites (as shown in Figure 6-3) joined with the appropri- ate data-extraction software, and aerial photography all can be used to identify vehicle volumes by type, assess the degree of congestion, and characterize and understand some of its causes. In toll facility territory, data from collections can also be used to develop enriched traffic count data, especially with advanced billing and information systems such as EZ-Pass. Truck counts by vehicle size and time of day can be taken from bridge and turnpike records. However, past attempts to use advanced information (such as operator identification and histories) for traffic demand management or transportation planning have been met with resistance due to valid privacy concerns on the part of the users. With toll territory appar- ently set to expand aggressively and nationally, mechanisms and opportunities for utilizing these data are apt to expand in parallel, and a legal framework similar to that applied to the STB private waybill sample might finally allow the more 112 Figure 6-3. Traffic Camera Can Also Generate Flow Data.

detailed data to be used in planning. Another issue with col- lecting traffic count data is that counting devices may be owned and operated by different entities; state DOTs tradi- tionally controlled inductive loops on non-toll routes, while turnpike authorities controlled tollbooths, and traffic cams may be privately owned or operated by a state contractor. Railroad counts can be constructed for current and histor- ical years from the CWS, with the aid of a routing model. Waybill records have indicators for the path traveled by ship- ments, and a model can turn these into a complete picture of linehaul traffic, provided carrier operating preferences are observed. Railroads also prepare and can make available track density figures; many state DOTs have access to these. Pickup and delivery traffic at intermodal transfer centers can be esti- mated from the CWS, although railroads may be willing to offer lift figures as well. Volumes at other kinds of terminal are harder to derive because they are a function of blocking practices and train configurations. There are models that can estimate this information, but if it is important to have it, it is probably best to request it directly from carriers. Levels of Accuracy and Precision Rail counts of the sort just described are reasonably detailed and accurate. On the highway side, traffic count data are available at different levels of granularity. Geographically, the data may be collected simply for a given highway segment, although more sophisticated systems will differentiate among northbound/southbound, turning or passing vehicles, vehi- cles using exit ramps, and different classes of trucks. Some systems will also convert vehicle counts into rates at different resolutions, e.g. counts per hour, per day, and per year. One conventional output is the quantification of Average Annual Daily Truck Trips (AADTT). Vehicle identification by means of interpretative algorithms presents two kinds of difficulty: (1) the conversion of axle obser- vations to truck counts may be off and (2) the definition of “trucks” may include light vehicles (and even pickups) that have almost no susceptibility for rail. The more advanced systems do a much better job of isolating the heavier freight that rail can remove from the roadway, but, thus far, there is much less of such data available. In addition, when using data at the hourly level or daily level, the usual caution about spurious accuracy from small sample sizes applies; if the truck count during a given hour on a given day is 30, the truly representative value might actually be somewhere between 15 and 50. An average of counts during the same hour over a number of days will give a narrower confidence interval of the normal range. Rail freight diversions may make the most noticeable reductions in the number of trucks or observable reductions in congestion, in specific circumstances. For example, there will be locations where a large portion of the highway traffic is attributed to a few bulk truck trip generators, such as ports or major manufacturing plants. Also, in certain locations on the highway network, through trucking may fill one or more lanes of highway, especially where multiple routes converge at or near major cities or geography causes traffic to be fun- neled along a coast or mountain range. In that sense, the accuracy of the traffic count, which gives an idea of overall levels of congestion, is less critical than the accuracy of the distribution of vehicle types that measure proportion of total congestion for which trucks or heavy trucks are responsible. 6.5 Commodity Characteristics What is the problem? What kind of data would be useful? Commodity characteristics are important, because certain types of commodities are more suited to rail than others. Bulk commodities, lower value general merchandise, and com- modities that are shipped in large quantities are typical targets for rail. Commodity price data can also be used to assess the impact of changing freight flows on the economy and on economic development. Conversions of commodity weight information to price information (e.g., dollars per ton) are useful in this connec- tion. The prices of commodities may not simply depend on the physical goods, but also on their location and other fac- tors like packaging or extent of processing—for example, paper may be more valuable in consuming markets than in production regions, due to added transportation costs and localized demand-supply equilibrium. In addition to price, other commodity characteristics may be important in considering rail freight diversion feasibility. These can include such factors as equipment requirements, storage needs, loading and unloading demands, perishability, and product density. Are there readily available sources for the data? Several federal publications have been good sources for key elements of such data. Until its discontinuation in 2006, the Vehicle Inventory and Use Survey (VIUS)3 supplied equip- ment types and payload (loading) factors for broad categories of commodities hauled by truck; the same things can be derived in even greater detail for rail commodities from the CWS. The CFS contains commodity values overall and by mode; in addition, the Bureau of Transportation Statistics produces the Surface Transborder Commodity Data, which 113 3VIUS data were collected on a 5-year cycle. The last collection was in 2002, and the data were released substantially later. Thus under the tra- ditional cycle, the 2002 VIUS would remain current through about 2009.

contain flows of NAFTA goods and their declared values at the border crossings. The Bureau of Economic Analysis also produces various input-output tables and accounts, which can be used to derive the value of goods traded per ton when combined with a commodity flow database and a matrix to map such flows to specific industrial sectors. Other sources on commodity pricing are available from var- ious industry associations and government departments, such as the U.S. Departments of Agriculture and Energy, and the Western Wood Products Association. Some web news services and investment information services also carry up-to-the- minute as well as historical commodity price data for selected commodities on their websites; however, getting it to a form usable for rail freight assessment may represent significant work. Tonnage-to-volume and tonnage-to-value conversion matrixes also can be an element of commercial freight flow databases, whose equipment type classifications help to address storage and equipment requirements as well. Example metrics based on the FHWA Freight Analysis Framework, represent- ing an amalgam of federal and commercial resources, are given in Tables 6-1 and 6-2. How can the data be collected? The information sources described above are readily avail- able and adequate for most planning purposes, making the collection of original data unnecessary. Moreover, some pub- lished elements can be sufficient proxies for others that are harder to come by: equipment and commodity payload char- acteristics can stand in for product density, for example, and also shed light on storage and handling aspects. When more current or specific price data are needed, it can be possible to compile it from web and other reference sources into a data- base with modest effort. However, results should be scrutinized to be sure that values are reported on the same basis: some figures may relate to wholesale, retail, delivered bulk, or spot- market prices, and others to costs of production. Collecting such data from empirical observations or through calls to vendors probably is impractical except as cross-checks; alter- nately, local chambers of commerce, economic development agencies, or economic research consultants may have some pre-existing data points that they use for internal purposes. On a limited basis, for very specific freight flows and eco- nomic sectors (e.g., cement, coal, building materials, wood, 114 Source: derived from FHWA Freight Analysis Framework STCC2 Description Tons/Load STCC2 Description Tons /Load 20 Food and Kindred Products 18 17 17 13 21 19 14 17 9 31 Leather Products 21 Tobacco Products 32 Concrete, Clay, Glass,Stone 14 11 20 14 13 11 15 16 8 22 Textile & Mill Products 33 Primary Metals 23 Apparel Products 34 Fabricated Metals 24 Lumber & Wood Products 11 yrenihcaM 53 11 erutinruF 52 36 Electrical Equipment 26 Pulp or Paper Products 37 Transportation Equipment 27 Printer Matter 10 stnemurtsnI 83 28 Chemical Products 39 Misc. Manufactured Goods 29 Petroleum or Coal Products 22 41 Misc. Freight 30 Rubber & Plastics 50 Secondary Traffic Source: derived from FHWA Freight Analysis Framework STCC2 Description $$/Ton STCC2 Description $/Ton 1 Farm Products 27 Printer Matter 8 Forest Products 28 Chemical Products 9 Marine Products 29 Petroleum or Coal Products 10 Metallic Ores 30 Rubber & Plastics Coal11 31 Leather Products 14 Non-Metallic Minerals 32 Concrete, Clay, Glass, Stone 20 Food or Kindred Products 33 Primary Metals 21 Tobacco Products 34 Fabricated Metals 22 Textile & Mill Products 35 Machinery 23 Apparel Products 36 Electrical Equipment 24 Lumber & Wood Products 37 Transportation Equipment 25 Furniture 38 Instruments 26 Pulp or Paper Products 39 Misc. Manufactured Goods 230 37000 690 200 1600 15000 100 770 2400 9300 19000 9400 11000 3500 470 1000 140 30 10 850 6900 4100 6500 210 3170 910 Table 6-1. Tonnage to truckload volume conversion by commodity type. Table 6-2. Value to tonnage conversion by commodity type.

and other such bulk materials), shippers and producers may be willing to provide rough price data for planning purposes. If special equipment or storage requirements apply, these will become evident during the course of dialogs with the shippers. Levels of Accuracy and Precision High degrees of accuracy in commodity value data are not critical to developing a successful rail freight diversion scheme, and the information resources described in this sec- tion normally are adequate indicators. If a scheme can be shown to be possible and likely to deliver a positive return on investment, it is unlikely that short-term changes in com- modity value will overturn it, and other features like handling characteristics ordinarily do not shift very much. It is not usu- ally prudent to pursue plans or schemes where the diversion hinges on having a low estimate for commodity values. In most cases, relatively modest changes to the plan, particularly in infrastructure or operating requirements, can strengthen its business case substantially. Two exceptions that planners should keep in mind pertain to long-term market trends. New entrants or new production sources, especially in commodity markets where transporta- tion is a significant component of delivered cost, can cause an otherwise viable rail service to become uncompetitive. Usu- ally the traffic pattern then will change completely, with the commodity production moving elsewhere instead of just switching mode (although that may happen, too), but this certainly can disrupt the return on a rail investment. Modifi- cation to logistics practices are a second way the ground can shift: the move to low-inventory, high-speed supply chains, for instance, favors smaller shipment sizes and tends to reduce commodity payloads over time. 6.6 Maps and Inventories of Rail Infrastructure and Service What is the problem? What kind of data would be useful? Knowledge about the location, design, condition, and uti- lization of rail facilities is basic information for strategies and policies aimed at increasing the role of rail to relieve congestion. Several questions should be asked at the beginning of any study: • Does the rail system have the capacity to handle more freight? • If not, what are the limitations and where are the key bottlenecks? • Do the railroads have plans (and capabilities) to expand the system to meet traffic growth? • Are rail terminals well located in terms of handling addi- tional freight? • How important are grade crossings (rail-highway and also rail-rail) in terms of delays to highway traffic and to rail traffic? • How well are the facilities performing? Information about the current system is necessary in order to determine how much and what kinds of changes might be needed to improve its performance or increase its capacity. The essential question is where a public investment or a program of investments can be made that will make rail transportation more attractive or more available and induce a traffic shift. The answer will begin with access: the location of prospec- tive shippers along the network, their connection to it through sidings or transfer terminals, and the distances involved. Public initiatives here may be able to establish or improve the conditions of access, shorten distances, or even encourage a different pattern of location among shippers. The next part of the answer will consider the physical condi- tions that affect service: track speeds and geometry, terminal functions and design, network connections and circuity, and grade crossings. In addition are the network features through which performance is bound up with capacity, including such elements as double tracking, siding profiles, and signaling. Public initiatives here will seek out the sensitive components, in order to make them targets of a set of investments that may enable system performance and competitiveness to rise. Capacity itself is the third part of the answer and perhaps the most complex. Its obvious importance is to ensure that if rail performance improves, the network can accommodate the diverted traffic—or, if rail performance already is high, that the network can be marketed for additional volume and can accept growth. Basic elements of capacity include features of line (e.g., tracks, siding lengths and frequencies, speeds and limitations, weight restrictions, and train controls), yards (e.g., total and receiving tracks, track lengths, and humps), and terminals (chiefly track length and storage). Public investment at least nominally is able to address any of these elements and expand the traffic volume available to rail. By no means is it necessary to have all of these pieces assem- bled in order to evaluate the prospects for rail. An overview of the line and terminal network, the kinds of traffic it serves, expert but subjective views of capacity, and performance indi- cators like train speeds may be sufficient to get started. Greater specifics then can be sought where they seem most warranted by conditions and opportunities. Are there readily available sources for the data? The Carload Waybill Sample and a compendium of oper- ators and networks like the railroad Official Guide are ways 115

planners can start looking at the systems in their districts. Going further, statewide rail plans setting forth an inventory of freight rail infrastructure are in existence around the coun- try, with varying degrees of depth, detail, and currency. In some cases, the state plan will also describe the operations of railroads within its jurisdictional boundary briefly, giving an insight into what kinds of service might be available and how intensively and in what manner the infrastructure is utilized. Class I and smaller railways report a range of information on their web sites, including schedules and, in some cases, per- formance figures. The web site of the Surface Transportation Board also carries current and historical Class I performance measures, with data like train speeds and cars on line. Access information can be obtained directly from large shippers if it is a question of sidings; rail carriers also will have this for facilities with recent activity or where sites are known to have been constructed off line. Transfer terminals and the kind of traffic they support will be published and more or less readily available. Engineering charts kept by owning railroads and public authorities will contain detailed information on types of signaling installed, location of infrastructure, and the state of infrastructure. Large railroads keep computer-based asset registers that will contain similar information. However, neither the plans nor the computer database may be totally up to date, unless the maintenance of way and signal depart- ments make a routine effort to maintain it. Third-party mapping companies or GIS solution providers, such as DeskMap Systems, Delorme, or ESRI, often will have databases of rail infrastructure covering entire regions, with some more complete than others. However, unless the company specializes in rail operations, it is unlikely to have information such as signaling systems and location of yards, sidings, interchanges, and switches. Equivalent rail networks for carload freight and intermodal also can be obtained from the Oak Ridge National Laboratory, although (as with other sources) the network may not be entirely current. In some communities, digital or aerial mapping of rail infrastructure would have already been carried out for specific projects. Those are often the most accurate source for the condition of local rail infrastructure. Capacity assessments for the most part will not be ready to hand, unless either the rail carrier or a public agency has con- ducted a local study of the network. Capacity assessment can be conducted with models, and with railroad cooperation in the assembly of input data, but this detailed exercise rarely is appropriate in the early stages of project evaluation. The most practical initial measure probably is professional evaluation by persons familiar with the operation—railroad personnel or sometimes their customers—whose subjective views neverthe- less can be well informed and directionally or entirely correct. When dealing with previously abandoned lines, local his- torical and railroad societies may produce publications detailing the status of local rail lines, and some will include detailed civil surveys. Independent producers have produced detailed U.S. rail atlases, some of which are more accurate than those provided in generic GIS sources. How can the data be collected? Maps, or GIS databases and routing networks, can gener- ally be acquired from third-party providers. Some agencies may also have internal teams who develop the data or will have done so in the past for rail plans. Railroad carriers ordinarily can provide much of the information needed if they feel moti- vated to do so, deriving it from various sources—operations databases, asset registers, and their own capital planning team. Service plans can usually be obtained from the railroad or a knowledgeable intermediary such as an intermodal marketing company or publishers of railroad freight information and schedules. Where there is a need for information not presented on typical rail network maps, railroad engineering departments represent effectively the sole source of information. Track maps can be found from third-party sources, but these can become outdated and do not contain often vital signaling capacity information. When the question concerns aban- doned lines, or an uncooperative railroad, approved field vis- its and dialogue with knowledgeable personnel (such as retired employees) can be useful to obtain information. Levels of Accuracy and Precision For existing and operational infrastructure, railroad oper- ating and engineering departments are the authoritative and most accurate source of information. Elements like yard and mainline condition and utilization can be reliably defined and can be substantive indicators of performance and capacity. Public agencies planning rail freight schemes with capital components based on upgrade of rail infrastructure must ensure that railroad carriers are part of the dialogue and plan- ning process. Planners considering operational changes in ways the railroad infrastructure is used should also contact the railroad operating department to assess the feasibility of the plan being proposed and identify any infrastructure upgrade or additional maintenance costs that may be incurred by changes in operations. Conditions shape project specifica- tions and investment requirements, so dependable figures are important. One method of checking carrier-supplied numbers is to have them reviewed by an experienced, inde- pendent party who is able to judge magnitudes, calculations, consistency, and overall reasonableness. For planning purposes with abandoned infrastructure, cost assumptions can be made based on information gathered from maps, aerial photographs, asset databases, and reference figures 116

to the extent that they are available. These “planning-only” numbers should not be used in cost-benefit calculations, as the physical condition of the plant may be substantially different from the planners’ assumptions, leading to inaccuracies in service restoration cost. Field visits can be another direct and accurate way to deter- mine infrastructure conditions, if undertaken by knowledge- able personnel. Such visits can also be relatively cost-effective, given the time required to research and reconcile different reference sources or to reach out to engineering departments and other stakeholders. They are one of the fundamental ways that short-line investors evaluate properties, and it is usually helpful for planners as well to have on the ground exposure to facilities in order to develop a practical under- standing of issues. 6.7 Railroad Engineering Cost Data What is the problem? What kind of data would be useful? The investment costs in proposed engineering projects obviously have to be quantified and generally have three major components: materials, construction labor, and equipment. In addition, there may be other outlays associated with a cap- ital improvement scheme, such as land acquisition, design, permitting, management and planning. The most detailed cost estimation falls into the domain of engineers, but with intelligent use of data points and a grasp of the physical requirements, planners can develop good estimates of the cost of projects. The Guidebook presents various figures and contextual information to help understand the range of costs associated with different kinds of rail investments. These will not be repeated here, but they generally employ two types of data. First, costs from past construction contracts (and actual costs once construction is completed) give an idea of what the cost would be if a similar project were carried out—for example, the addition of a siding or a spur. Second, financial factors such as the cost of railroad materials, lease rates for equip- ment, and labor rates can be used to estimate expenditures by enumerating each activity. The first method offers a view of the way various project components may total up, and the second allows for dissimilarities and gives a way to proceed if comparisons to analogous projects are not obtainable. Are there readily available sources for the data? Again, a series of factors and applications appear in the Guidebook that can serve as a resource for project evalua- tions. For additional specifics, there are many alternatives. Costs of track and other materials are available from vendors, industry associations, and some independent publishers. Labor rates can be found in past cost estimations and con- tracts or from trade unions. Some reports will cite a standard cost per mile of track given a set of assumptions; this type of number is useful for planning purposes, although it is impor- tant to be aware that changes in the assumptions can lead to different costs. Similarly, when using costs derived from past construc- tion, it is important to understand the conditions under which the work was done. Constructing a railway from scratch can be cheaper than upgrading an existing one if the upgrade requires the use of restrictive work-windows between trains. Installing a new siding on a heavily traveled main line will cost more than the same siding on a branch line. The amount of earthwork required and foundation sta- bilization can vary greatly from site to site (also, depending on the line speeds and load ratings required from the new track), resulting in very different costs and schedules. If signaling work is required, it should be understood that its cost estimation is difficult without some preliminary design work. Most of the cost involved in commissioning new signal- ing relates to specialist labor for installation and testing and the solid-state equipment to be installed. Costs tend to be dissim- ilar from contract to contract. Moreover, seemingly routine work such as moving an existing signal head from one location to another could be a minor or major expense, depending on the amount of other work required as a result of the change. How can the data be collected? Research into the kinds of primary and secondary sources cited above will yield the requisite data. Another alternative is to turn to civil engineers with rail project experience; many will have estimation methods that allow a cost projection to be done in a few hours. For more detailed cost estimation, an on-call contract with an engineering consulting firm is a way to assemble anticipated expenses before a formal project bid is released. Levels of Accuracy and Precision Cost estimation is vital for project evaluation, financing, and job management. This means accuracy is essential, and the need for precision will increase as a project moves toward programming. The methods presented in the Guidebook and touched on here are capable of producing sensible estimates whose reliability is appropriate to the stage of project devel- opment. Any engineering project faces an assortment of con- tingencies touching on anything from market cost changes to permitting and job management, and rail (like highway) projects are certainly subject to them. Allowing for this, infor- 117

mation resources nonetheless are sufficient to the needs for precision and accuracy. 6.8 Shipper Characteristics and Needs—Establishment Data What is the problem? What kind of data would be useful? Railroads or their intermediaries ultimately must be able to determine which companies might be willing or able to shift some of their freight from truck to rail. Planners will want to engage with some of them on subjects ranging from access to service design and divertible volume. Candidates would include companies originating or terminating large amounts of freight, port authorities, and national corpora- tions known to ship substantial volumes through the region. Many of the relevant companies will be well known, because of their importance to the local economy; others, particularly shippers of low-cost bulk materials, may have a low profile and generate significant tonnages with a modest number of local employees. The geographic dispersion or clustering of important businesses also is essential to understand, because of its effect on operating density. The available databases about commercial establishments are useful for a number of reasons. On a macro level estab- lishment data are used to assess economic geography. Estab- lishments are classified in terms of the Standard Industrial Classification (SIC) or North American Industrial Classifica- tion System (NAICS) codes. Based on these codes, the nature of the state’s economy can be understood and the corridors where rail freight solutions have leverage can be identified. On the micro level, establishment data are used to create a list of potential stakeholders to interview and to organize them into logical groups based on their characteristics and likely freight needs. Typical business databases contain not only the physical location and the name of the establishment, but also the number of employees and an indication whether the firm is a subsidiary of a larger corporation and, in some cases, the input-output relationships (i.e., the industrial codes of any upstream and downstream industries, as well as non-core production activities). SIC or NAICS codes can usually be translated into commodities to determine what types of goods are being shipped. Establishment databases by themselves are decidedly helpful, but when joined to other information resources discussed in this chapter, they help create a potent analytical system to determine freight needs and traffic. Are there readily available sources for the data? Several commercial databases are available, each with different coverage and pricing options. A basic list of establish- ments is often within reach from the local chamber of com- merce or phone book or from web-based equivalents. Some state governments also keep internal or public establishment databases as part of a census or other research support activity. Three of the main vendors providing data in the private sector are Dun & Bradstreet, InfoUSA, and Harris InfoSource; all are able to supply data at the level of detail described above. Other vendors, such as ZipInfo, offer less detail, but may represent a cost-effective solution. These data normally do not reveal the existence of rail access, yet normally are geocoded. GIS analy- sis of establishment data alongside a reasonably detailed rail network will show the proximity of businesses to rail lines, and this can be used for a first approximation of access. How can the data be collected? There are different approaches to collecting establishment data. A simple, if laborious, approach is to work through the business telephone directory, especially if it can be organized by geography. Another is through field visits—if the search is to find all businesses abutting a given rail branch line, field visits can actually be a cost-effective way to conduct research and may generate much more information than any database (Figure 6-4 illustrates this). A third approach is to use maps, charts, zoning records, and aerial photographs, combined with other reference material, to locate large industries near the rail line. If a comprehensive database is not available from a com- mercial vendor, information can be extracted by joining data from the local chambers of commerce, zoning records, local knowledge, and postal or phone book address records. Zon- ing records will help locate industrial activity, and sometimes SIC or NAICS codes of businesses can be ascertained either from the name of the establishment, from a chamber of com- 118 Figure 6-4. Some Potential Rail Freight Shippers’ Activities are Self-Evident from a Field Visit.

merce database, a quick phone call, through locally knowl- edgeable persons, or a short site visit. Railroads will have information about line access, at least for recently active cus- tomers; for inactive ones, phone calls may be required to define status because a former siding may have been paved over. Sometimes a site visit is the only way to ascertain the industrial activity and freight requirements at certain brown- field sites. The importance of fieldwork should not be underesti- mated. No database fully replaces it, and sometimes fieldwork is simply a matter of driving by, observing signs, and taking digital photographs of commercial activity. Levels of Accuracy and Precision The accuracy of establishment data in general is good for the existence of activity, reasonable for employment levels and business mix, and less good for business levels. Surveys are utilized to obtain the data and some information is con- sidered confidential by the respondents; furthermore, there is no integrated mandatory reporting process for commercial establishments, except for financial data on publicly held companies (which do not report site-specific data in any case). Analysts need to (1) be careful that employment esti- mates are particular to the local address and (2) watch for misleading codes suggesting that manufacturing takes place in a location really dedicated to services. Commercial activity is also highly dynamic. Some industries that are transportation-intensive (such as building materials, scrapping, and some chemicals) tend to be cyclical in nature, and business levels can be tied to discrete contracts. A plant may shut down or start up again in a matter of months, or pro- duction locations may shift. Thus, maintaining up-to-date establishment information requires ongoing effort, and data- bases should be renewed to ensure currency. Commercial databases are useful for systematic planning and identifying opportunities. Nevertheless, for development of a specific rail-freight initiative whose success may hinge on several major customers, locally knowledgeable persons can be a great resource, and early contact with major shippers should be considered a vital part of the planning process. 6.9 Modal Service and Cost Parameters What is the problem? What kind of data would be useful? Modal service and cost parameters are used to assess whether a rail freight solution is in fact feasible from a ship- per’s point of view. If shippers cannot reduce their overall private logistics costs by moving to rail, either a different incentive will have to be provided, or they will continue to ship by truck. On the service side, shippers must be able to manage the logistics chain so that their business activity is compatible with typical rail performance. Except for pre- mium intermodal and some other operations, rail shipments may be slower, require longer lead times, and perform less reliably than trucks. The business may be able to adapt, but the service it can expect to receive should be understood. An extensive treatment of logistics cost factors appears in the Guidebook. Here, it is sufficient to say that performance indicators, operating costs for both truck and rail, and infor- mation relating to inventory and handling expenses all are useful in a comparative modal assessment. Are there readily available sources for the data? Shippers ought to possess accounting records of logistics costs, including the cost of transportation, warehousing, and value of inventory in transit. Without shipper contact, it is still possible to calculate a likely range of costs using standard cost functions for trucking, generic commodity dollar val- ues, and estimates of the cost of storage. The most difficult step sometimes is in approximating the significance of inventory in transit, since business decisions affecting tran- sit time requirements can be linked to the strategic value associated with tight channel control and point-of-sale response. For railroads costs, commercial products will estimate the cost of railroad shipments between intermodal terminals or freight stations. Most of these models are based on the Uniform Rail Costing System (URCS) methodology devel- oped by the Surface Transportation Board (STB) and its predecessor, the Interstate Commerce Commission; the STB makes available a URCS-type cost model as well. The AAR also publishes a quarterly Rail Cost Adjustment Factor (RCAF), as part of its Railroad Cost Report (RCR). For a general idea of costs, a simple cost function with a cost per mile could be used. Most rail users and rail service market- ing companies will have such rules-of-thumb, and a variety of them are presented in the Guidebook. The key trucking costs for rail comparisons are full truck- load, which will also serve as a profile for linehaul costs in LTL. Up through 2005 there had been good information from which these could be derived in the M-1 financial reports, which larger motor carriers were required to submit to the federal government. The discontinuance of reporting in that year meant that trucking costs eventually would have to be estimated from engineering factors, although the his- torical figures would offer a reasonable template to work from for a fair period of time. Truckload service characteris- tics are reasonably well known and are shaped by distance, 119

travel speeds, the number of drivers, and hours of service reg- ulations. Overnight trucking service with a single driver is typically difficult for rail to divert, as is the premium team service where two drivers alternate shifts. Longer distance trucking service that involves a layover for a single driver (thus, with a dock-to-dock average speed lower than about 50 mph) can often be diverted with rail intermodal. Beyond intermodal, rail service tends to compete on characteristics other than speed, such as costs, safety, size of shipment, and other factors. Rail service characteristics can vary with the type of rail service purchased, proximity to major yards and mainlines, train frequencies, and other system-wide factors. Thus, pre- dicting the service level in a given rail lane is much more dif- ficult than for trucking. If the rail freight diversion proposed relies on existing services, then the railroads would usually have a fairly good time estimate for the shipment. If new service is being planned, then the sponsor may have more flexibility over cost and service levels—with the caveat that truck-equivalent service levels tend to be more expensive except in high-volume service lanes. Generally, the best way to validate proposed service levels is through careful opera- tions planning, followed by test runs designed to determine the feasibility of the operating plan and its impact on other railroad operations. The major Class I railroads (and the two Canadian majors, CN and CP) are required to report service performance lev- els to the STB on a weekly and quarterly basis. Although these numbers are available from a website maintained by the AAR4, the highly aggregated performance data are of limited value for predicting service levels within particular service lanes. Nonetheless, they are a good indicator of broad service trends (e.g., whether the probability of a regular shipment arriving on time is increasing or decreasing). How can the data be collected? There are two major types of service performance data: (1) empirical results, which must come from the carrier, the shipper, the agent, or another interested party who has kept historical records such as the sources mentioned above; and (2) performance simulations, which can be estimated with knowledge of current operating practices, plans, and infra- structure conditions by either a consultant or the carrier’s operating managers, but must be validated by actual service performance or test runs. Ultimately, the data must be obtained from one of these sources. Unlike passenger rail, it is generally costly and difficult to ascertain rail freight per- formance by direct field observation, because of the long variability of run-times and the difficulty of tracking the operations without using one of the railroad’s proprietary information systems. There are also two major types of cost data: (1) accounting data, which may be available from shippers or carriers willing to make them public or share them through an intermediary conducting a study on behalf of a public agency; and (2) cost model data, which are calibrated by a knowledgeable party based on known expense and operating factors. Price data are rarely possible to observe directly and therefore must be obtained through modeling, interviews, and other coopera- tive methods. Levels of Accuracy and Precision For typical rail freight diversion applications, service times need to be known to within one day, or perhaps half a day. Service time precisely to the hour is usually less important than the reliability factor. Under unconstrained conditions, a train may be able to move from siding to sid- ing in a standard number of hours; however, for a feasible service plan, the number of intermediate switching moves and the probability of delay at each location must be accounted for. Even for bulk commodities, a missed deliv- ery can lead to problems at the receiving plant unless a suf- ficient stockpile is maintained—which drives up the total logistics costs. Thus, errors in reliability estimations may lead to excessive costs being incurred by the shipper, result- ing in a seemingly promising operation becoming an uneco- nomical one. For intermodal diversion, time performance can be espe- cially critical, since the truck-like performance it aims for is often associated with low levels of inventory. Nevertheless, typical services still are discussed in terms of morning, after- noon, or evening delivery, instead of a specific hour within which the shipment must arrive. While there are premium intermodal products that do guarantee certain time windows and cut-offs, those normally are geared to the requirements of a particular customer or group. Operating costs are an important factor in determining whether services can be sustained. Prior to investment in expensive infrastructure, comparative analyses of modal costs should be conducted. The cost savings of moving from truck to rail need to be significant in order to allow an annual contribution toward paying off the infrastructure. If the cost savings are not significant, then even if the infrastructure is constructed, the traffic may not materialize. Thus a com- pelling case is required before an investment decision is made—but having made such a case, and given the magni- tude of infrastructure costs, minor errors in rail operating 120 4The website at http://www.railroadpm.org/ features such performance measures as Total Cars On Line, Average Train Speed, Average Termi- nal Dwell Time, and Bill of Lading Timeliness. Performance measures of shortlines have not generally been available.

costs are unlikely to change the fundamental conclusions in a freight diversion project. 6.10 Trend Data—Traffic and Economic Projections What is the problem? What kind of data would be useful? In planning, trend data are sometimes used to illustrate a future scenario and to convince the stakeholders that changes are needed now to prepare for the future. Congestion tends to worsen with economic growth, and if rail freight investment can keep ahead of growth while highway investment remains stagnant, railroads will become comparatively more attractive to some shippers. Trend data are therefore needed to illustrate the effect of both highway and railroad congestion if nothing were to be done, and the payoff from taking action. In general, trend data fall in two broad categories: (1) eco- nomic trends and (2) traffic trends. Economic trends serve to suggest how fast the economy might grow in future and can be used to infer how costs, service levels, and other attributes of freight transportation may change over a long planning horizon. Traffic trends serve a shorter term purpose—if con- gestion is growing by a certain percentage per year on one highway route, it can be conjectured that the congestion will continue to grow at a similar rate until the facility becomes comparatively less attractive versus substitutable facilities or versus alternatives such as supply source substitution. Are there readily available sources for the data? Economic forecasting is a specialized discipline, and fore- casting data are made available both by governmental agen- cies and commercial vendors. Past economic trends can be found in various reports made available by the Department of Commerce,5 Bureau of Economic Analysis, and the Eco- nomics and Statistics Administration, in addition to private economic research resources. However, the federal-level data may not contain enough regional detail, and state-level data should be consulted. Many states have official projections of population and other economic drivers, and some have invested in forecasts directly aimed at transportation or rea- sonably pertinent to it. In addition, a number of regional eco- nomic models are available6 in the marketplace. The previous chapter of this report presented a cross sec- tion of trend and forecast information and cited relevant sources that may be consulted. For the tracking of traffic con- gestion, it displayed data from the Texas Transportation Insti- tute, whose annual Urban Mobility Study7 is the standard compilation of developments across the nation. TTI indexes and ranks traffic congestion problems for the 85 major U.S. urban areas, and its data can be compared and extended in time series. However, forecasting future traffic congestion based exclusively on its current trend is not advisable beyond about 5 years; to understand the extent of long-range conges- tion, long-term economic trends should be used. For more information on forecasting future freight con- gestion, a good source is the NCHRP report 8-43: Guidebook on Statewide Freight Planning. Although this manual does not specifically deal with rail freight, using the methodologies demonstrated therein to understand where future congestion and bottlenecks may occur could be helpful. Once these potential hotspots are identified, the methods in this Guide- book can help planners decide if a rail freight diversion scheme is apt to alleviate the likely problem. How can the data be collected? The economic and traffic trends rely on numerous data sources, and it is generally not cost-effective to duplicate the data collection effort. Economic trends require data about trade activity, which is collected by the Department of Com- merce through business reporting requirements. Traffic trend data and projections may be based on Highway Perfor- mance Monitoring System (HPMS) and automated data col- lection devices. The source data are publicly available. Levels of Accuracy and Precision Economic and traffic trends are usually reliable, if their data are sound and their dynamics are accurately understood. Projecting from trends is another story, because of the under- lying presumption that past events will continue on a logical course toward a future conclusion, which is not always the case. More sophisticated forecasting tries to anticipate course changes and the interaction of trends, and while inevitably imperfect, it will give a better result. Econometric forecasts of this type can be purchased from a number of sources, and banks and news services like the Wall Street Journal offer comparative performance ratings for vendors. In many cases, predicting economic growth itself is not as important as predicting political decisions. Lack of highway investment is one catalyst for rail freight investment; how- 121 5See http://www.commerce.gov/ and http://www.bea.doc.gov/ for more details on the types of data provided. 6Examples include REMI from Regional Economic Models, Inc.; REDYN from Regional Dynamics, Global Insight (DRI*WEFA); Fair Model (Yale University)—as well as many consultants who produce forecasts based customized versions on one or more different models. 7See http://mobility.tamu.edu/ums/congestion_data/ for more details about this study.

ever, if congestion becomes too severe, citizens may demand highway or mass-transit improvements. When planning rail freight investments, many such factors should be taken into account and weighed through a scenario analysis. Planners should prefer not to rely on a single set of traffic or economic assumptions being completely correct or base the viability of a specific rail freight plan on a single scenario. The best rail freight plans will view an investment case under a range of development assumptions and test its success across them. 6.11 Institutional and Privacy Factors To develop a successful rail freight diversion scheme or other rail freight solutions, three basic types of data are needed. The planner should have an understanding of (1) the markets in which freight travels and levels of demand; (2) the supply cost of providing freight services and infrastructure to meet that demand at appropriate levels of service; and (3) the economic trend data that reflect how the supply and demand, the associated congestion, and the area’s economic development can be expected to change in the near and fur- ther future. There are a series of sources for satisfying each, with options that can be scaled to the size or phase of a proj- ect or program, from small or preliminary to very large or well advanced. State DOTs, MPOs, and other organizations should make an active effort to make freight data collection part of their regular data collection efforts. In some cases, data collected for passenger facility performance monitoring and/or for opti- mization of facility maintenance strategies can be used to pro- duce informative freight data streams. Alternately, it may be possible to add features to a data collection program that will partially feed freight planning applications. Freight activity is also heavily connected to economic activities; thus, as part of an area-wide economic development or re-development effort, data streams might have already been collected that could assist freight planning. Since ownership of these data could lie outside the domain of DOTs, it is important to estab- lish contacts in other public organizations with overarching responsibility for economic development and become famil- iar with the information they may have available. Examples of such organizations include • Local economic development agencies (e.g., the Boston Redevelopment Authority); • Local port authorities (e.g., the Delaware River Port Authority); and • Multi-state agencies (e.g., the Tennessee Valley Authority). Developing a data program and encouraging working rela- tionships with entities that may become sources of information introduces institutional and privacy issues. Some of these issues are explored below. Privacy Concerns Private-sector carriers in both trucking and rail are rightly concerned that their competitors might use information about the flows on their network (and by inference, about their customers) to their own advantage. The negative effects can include customer poaching, disruption of density, and loss of network balance. This type of competition also may result in destructive price wars that can harm individual car- riers or delay reinvestment by an industry. Shippers of freight have similar concerns. In addition, rail carriers may worry that any new reporting of market data begins an unwelcome return of government oversight, such as prevailed prior to the Stag- gers Act. For reasons such as these, the STB waybill sample is pro- tected by law. Decisions on using its detailed version are reviewed by the Federal Railroad Administration, and state- level governments have access only in a controlled fashion. On the occasions when a private enterprise is permitted to make use of these data, strict guidelines must be adhered to. In most cases, the data processing must be done by an inter- mediary, who then must use the data only for the specified purpose and destroy it after the work is completed. When pri- mary data from motor carriers have been tapped for some public studies, it has been done voluntarily, instead of on the compulsory basis that applies to the rail waybill. Nevertheless, restrictions and protections have been built in for the benefit of cooperating truck lines: information has been aggregated, intermediaries have been employed to avoid subjection to the Freedom of Information Act, and reuse has been prohibited. If state DOTs and other governmental organizations expect to develop the trust of industry in conducting plan- ning studies and sharing data and plans to mutual advantage, these privacy concerns must be taken very seriously. Demon- strating a good understanding of the issues and why privacy is necessary, honoring the commitments, having a codified policy on how data may be used and distributed, and never using data in less than good faith will go a long way toward building a successful and fruitful relationship with industry partners. In joint planning, it is always important to achieve a win-win outcome; the industry cannot ‘win’ if the data pro- vided for planning purposes are not treated with care and caution by trusted agencies. Financial Data Publicly held companies are required to report certain financial data to the Securities and Exchange Commission 122

(SEC), for example, on the Form 10-K.8 However, 10-K information typically is not very useful to the transportation planner, as it is rolled up for the whole corporation (most likely a multi-state enterprise), and there are ways to report the information such that it is difficult to understand the company’s cost structure. In addition, some very large freight carriers are held by private entities, who are under no obliga- tion to disclose financial results to the SEC. On the other hand, railways in the United States are required to submit R-1 reports annually, which set forth a substantial body of financial and operating statistics, some of it like conventional balance sheet and income statements, and some of it quite different and oriented, for example, to operating assets. (Dis- continuance of the comparable if less detailed M-1 reporting for motor carriers was noted above.) Railroad capital programs normally are published annually and can both be helpful and unhelpful to the public planner. The capital budgets will be defined in terms of number of ties to install, bridges to rebuild, and sidings or track miles to add. In addition, ongoing projects may have special line items that highlight the investment that railroads are planning using their own capital. However, it is generally difficult to extract specific cost numbers from such documents. Moreover, pub- lic agencies rarely are invited into the strategic planning process at private railroads, so public planners may believe that they have little influence. Still, many Class I railroads have a government relations department. Taking a proactive approach to railroad capital planning at a state level can yield fruitful results. In several states, there are standing funds available for railroad infra- structure upgrades, which can be a good way to become engaged in railroad capital planning. With a stake in the process, it becomes much easier to acquire financial data needed for planning and budgeting on the public side; also, planners will develop a better understanding of whether rail freight diversion plans can work or not and how much they may cost. Railroad Capacity and Reliability Public planners are aware that it can be difficult to per- suade railroads to release seemingly ‘spare’ capacity on their tracks that is not currently in use because, once an operating agreement is entered into, it will be difficult for the railway to remove that traffic, replace it with more profitable business, and not cause a public-relations problem. Without removing existing traffic, infrastructure upgrades typically are required when additional capacity is needed. These can be time- consuming and costly, especially in metropolitan areas. Thus, spare capacity on a not-yet-congested portion of railroad is still an expensive commodity, even if infrastructure upgrades are not immediately required for new traffic. Public agencies wishing to use capacity on private railroads must understand that not only do they have to cover the operating cost of the train, they must offer a premium to out-bid any future use of that capacity the railroad may have planned. A pragmatic solution to this problem is to have the public agency upgrade a piece of private railroad infrastructure at public expense, in lieu of premium payment for a spare train path. In some cases, loading a network with additional traffic can cause sometimes-subtle effects that lead to increased costs. For example, spare capacity may be required at strategic points about the network to prevent cascading congestion when long-distance traffic is delayed. The cost of this capac- ity is usually borne by the railroad. Cascading congestion can be extremely expensive, requiring many more crews and power units to move the same amount of freight compared to an uncongested network. Increasing traffic can dramatically increase the cost of recovering from such an incident and is a cost that public planners should be aware of when aiming to use apparently untapped capacity. Data Collection is a Cost In addition to the concerns discussed above, two further issues may give carrier management pause in respect to shar- ing data: • Rate of Return on Data Collection Activities. Developing a relationship with public authorities and finding new freight with public support can be profitable activities for railroads in the long run. Even so, rail managers may think they lack the current resources to manage a data collection exercise or may doubt that new business is going to arise from the effort. Even in a business development environ- ment, managers will be reluctant to do extensive data col- lection or grant high priority to the proposals of public planners unless the prospects of rewards are substantial. When requesting data, it is helpful to state upfront what the rewards might be—for instance, by showing that investment funds will become available through a certain feature or channel. Railroad partners may be more likely to engage in data collection if such data are made a part of the application for a specific grant or if the data are being offered on the understanding that public officials will pur- sue available funds and take over some of the development work based on the data. • The Litigation Threat. Freight carriers, like other corpo- rations, have a healthy respect for the legal system, and some of their caution with information release may stem from the lack of clearly codified limits on how data may be 123 8See http://www.sec.gov/answers/form10k.htm for more details. The Securities and Exchange Commission is at http://www.sec.gov/.

handled. The U.S. Census, and the STB waybill sample achieve successful data collection in part because there are clear laws on how the public may use and disseminate the data. Confidentiality is guaranteed and exemptions plainly exclude certain data-mining activities. Steps such as those outlined in the discussion of privacy issues will allow public planners to assuage concerns about litigation exposure. The establishment of clear contracts limiting the application of data for planning purposes and the use of vetted intermediaries to process it help to create a trusted framework for information exchange. 6.12 Data Environment There are special issues concerning the electronic data environment in railroad and motor carriers alike that are worth understanding. Some data systems are legacies from development early on in the computer revolution, when each carrier sought to acquire IT capability for its own internal financial planning and operations purposes. As such, data formats occasionally predate the concept of relational data- bases and data mining and are driven by transactions far more than analysis. Because the systems are intended mainly for internal use, there may be limited standardization on what kind of data are kept, how they are kept, and what for- mat they are kept in. Public planners should understand that data simple to generate in an environment powered by latter- day data centers are not necessarily easy for every carrier to compile, despite their best intentions. With that in mind, it is important to be patient and flexi- ble when requesting data that may require downloading from legacy systems. It is possible that carriers in these environ- ments will have to expend substantial effort to find the data being sought by public planners. Once the data are found, they might be available only as a line-printer output, requir- ing optical character recognition software to translate into machine-readable form. It is likely that carriers would want to further process such data before handing it to the public planner, in order to elide commercially sensitive information, and this imposes an expense on the carrier. Offering com- pensation for such expenses or maintaining a confidential data-processing expertise in house can be ways to ensure that data collected by the private sector for private purposes are not lost as a planning resource. 124