Transportation planning is undertaken at many levels starting from strategic planning to project planning and also over various geographic scales. During the last forty years urban transportation planning has undergone many changes and continues to evolve. But many of the basic notions developed in the early years still exist. The basic urban transportation planning process usually consists of the three interrelated major components: the pre-analysis phase, the technical phase and the post-analysis stage. The pre-analysis stage involves identification of problems or issues, formulation of goals and objectives, data collection and generation of alternatives. The problem definition needs to be broad enough to accommodate considerably broader set of possible solutions.

The technical phase involves mathematical descriptions of travel and travel related behavior, used to predict the consequences of each alternative transportation plan being evaluated. It consists of three major components: the land use-activity system model, the urban transportation model system and the impact prediction models. The land use-activity system models comprise of the spatial distribution of people, activities, and land use within an urban area and these are now integrated with transportation models to asses its impact on travel. They help predict urban activity patterns and generally use regional population and employment as input and distribute these totals spatially over a region. The Urban Transportation Model System (UTMS) consists of models commonly used to predict the flows on the links of a particular transportation network as a function of a land use activity system that generates travel. The sub-models are trip generation, trip distribution, mode choice and trip assignment. The UTMS predicts the quantity and quality in terms of travel time of flow on the links of a specified transportation network, given land use-activity system as input.

Assessment of alternative options needs estimates of a broad range of impacts. These include construction and operating costs, energy consumption, and air quality. The impact prediction models need the UTMS as inputs.

The post analysis phase starts with the output of the technical analysis, which comprises of predictions of the impacts of alternative plans and policies. This phase involves evaluating the impacts, both economic and non-economic, of the alternatives analyzed; selecting the alternative to be analyzed; programming, budgeting and implementing the alternative chosen; and monitoring of the system performance.


Urban and Metropolitan Area Modeling

The classical modeling approach called the Land-Use/Transportation System (LUTS) or Urban Transportation Planning Package (UTPP) in urban transportation attempts to provide the broader system view (1). It attempts to model both the estimation of demand and routing of traffic through a transportation system. This process is often modified to meet local data availability, study needs and modeling preferences. It consists of intercity passenger modeling and supply or network modeling. LUTS/UTPP can be carried out at different scales, at regional or local levels. The basic process remains the same but the level of detail and accuracy needed for traffic forecasts and network definition are different. The level of detail and the scale of analysis need to match. The basic structure of the LUTS/UTPP approach is illustrated in Figure 1.

The classic urban transport model

The classic transport model has been developed after years of development and experimentation. The structure of the model is based on the practice of the 1960s but has remained more or less unaltered despite the major improvements in modeling techniques since the 1970s. The approach starts by considering a zoning and network system, and the collection and coding of planning, calibration and validation data. These data would include base-year levels for population of different types in each zone of the study area as well as levels of economic activity including employment, shopping space, educational and recreational facilities. These data are then used to estimate a model of the total number of trips generated and attracted by each zone to the study area (trip generation).

The next step is the allocation of these trips to particular destinations, in other words their distribution over space, thus producing a trip matrix. The following stage normally involves modeling the choice of mode and this results in modal split, i.e. the allocation of trips in the matrix to different modes. Finally, the last stage in the classic model requires the assignment of the trips by each mode to their corresponding networks, namely, private or public transport. It is not always that travel decisions are actually taken in this type of sequence; a contemporary view is that the ‘location’ of each sub-model depends on the form of the utility function assumed to govern all these travel choices. This model is very narrow in its scope and does not analyze a wide range of transport problems and schemes. But it at the same time provides reference to contrasting alternative methods. It forms the basic conceptual framework for transport models (2).

During the 1940s urban transportation dealt with specific problems like congested bridge or intersection. It was very myopic in its scope. In 1944, the Bureau of Roads conducted the first "origin-destination" survey, which involved collecting data to understand the observed traffic volumes. This was the first attempt to understand the underlying traffic generating process. In the 1950s and 1960s, urban transportation studies were synonymous with regional studies. The important factor in the development of the analytical tools that formed the basis of early urban transportation planning studies was the emerging availability of digital computers capable of handling relatively large quantities of data. These computers allowed planners to analyze urban travel patterns on a region-wide basis and encouraged efforts to develop mathematical equations describing these patterns. Other factors that contributed to the growth of urban transportation planning in the 1950s include rapid urban population growth, growth in car ownership, increasing movement of population to suburban areas and increasing federal involvement in funding urban development while requiring comprehensive planning.

The Detroit Metropolitan Area Traffic Study in 1953-55 and the Chicago Area Transportation Study (CATS) in late 1950s were pioneer studies using the emerging analytical techniques. The Detroit study employed a process that included data collection and goal formulation, development of forecasting procedures, and testing and evaluation of alternatives. Such work was matched by similar applications of "systems approaches" in other areas of economic and social inquiry and, in particular, to land-use modeling; examples were the development of spatial analysis/locational theory and the Lowry Model (1).

These studies were followed by a number of others in the late 1950s, including those in Washington, DC, Baltimore, Pittsburgh, and Philadelphia. They used the computerized procedures developed during the CATS study and had the objective of forecasting future trip-making patterns and producing a long range, region-wide transportation plan. The massive data collection exercises were followed by lengthy analysis time on mainframe computers. Although there were adequate financial resources to gather appropriate data, computer technology was not adequate for analysis. The emphasis in these studies was on planning a highway system that would cater to the growing automobile travel in urban areas. This "systems approach" to transportation planning spread to Europe in the early 1960s and a Land-Use Transport System (LUTS) was carried out in London in the 1960s. Between 1963 and 1967, the Bureau of Public Roads published a large number of manuals dealing with the technical aspects of the planning process and the procedures developed in the 1950s and 1960s were thereby codified and institutionalized. These procedures were exclusively oriented towards analysis of long-term, capital-intensive expansions of the transportation system, mostly in the form of highways.

In 1976, federally sponsored work lead to the development of the Urban Mass Transportation Administration/Federal Highway Administration UTPS software package. UTPS was adopted by many agencies in the US and also influenced later commercial packages. This package was initially oriented to large mainframe computers, used link-coded networks with distinct modeling limitations and had limited graphics capabilities. The computational tools included capacity constrained network models, environmental impact models and stochastic equilibrium models.

In the early 1980s, the LUTS programs adapted mainframe computer program packages for microcomputer use. At the same time computer hardware improvements tried to make these complicated, multi-component systems such as the comprehensive transportation models better packaged and more user-friendly. These resulted in the adoption of menu systems, user-sensitive layering of software, graphical interfaces and database/toolbox modular approaches. This is an approach to software design that starts with, rather than ends with, the viewpoint of the user. In some cases these programs created a comprehensive CTM separated into modules with some user-oriented features, including menus and interactive modes. In other cases (e.g., Tranplan), programs were run in batch mode, normally using a dedicated database with fixed format and a separate graphics package, and, sometimes, an on-screen editor. Computational tools included disaggregated demand models for passenger transport, shipper-carrier freight models, and transportation system management models. There were better calibration techniques, also.

The late 1980s saw larger and faster microcomputers with interactive menu systems and graphics communications networks with other computer. It had a toolbox and GIS approach. There was use of knowledge-based expert systems, interaction with other planning databases. This has lead to more open exchange of databases, project definition and evaluation. The tools included new network solution algorithms, demand management models and methods of combining databases to estimate origin and destination matrices and traffic generation. Their database management features include flexible format data inputs and automatic cross-indexing of network characteristics, simplified plotting commands and other user-friendly graphics-based features.

Urban and regional network analysis models have undergone more development in both theoretical basis and software packages. Features are constantly being added and improved in all packages (1).

Intercity Regional and National Models

Intercity or national transportation planning has not kept pace with the technology of urban packages. Intercity models have concentrated more towards broader policy and economic issues. Though recent developments have user friendly intercity modeling packages for policy evaluation with mapping functions but they lack the more sophisticated graphics interface available in urban CTM packages.

The first major innovation in predicting intercity passenger travel was in the Northeast Corridor of the United States in the 1960s (3). This study introduced techniques for evaluating transportation performance and also created interest in activity shift models for forecasting changes in industry location as a function of transportation changes. In the late 1960s and 1970s, a similar type of modeling was done with the Harvard-Brookings model (4). The major contribution was that it had a detailed performance and cost model of each mode and a macroeconomic model to the standard traffic assignment model, all integrated into the package along with feedback. Kresge and Roberts’ basic modeling structure (see Figure 2) is still used today with improved cost-performance, modal choice and assignment conceptually realistic, had massive data requirements and model calibration problems. This was models (1). The integration provided in these models represented actual linkages in a more realistic manner-linking transportation with regional economic development and also the linkages between physical parameters and operating costs. But this approach, though evident from the applications of the Harvard-Brookings model. Later models have avoided this approach and had a modular approach.

Very little intercity modeling was done during the 1970s other than intercity airline passenger model developed for the US and a freight-oriented model of the US Waterways transportation system (5). Both of these models were structured for mainframe computer use with limited graphics and user interaction features and relatively simple system optimizing algorithms. Many such packages were made in the late 1960s and 1970s for military logistics purposes, short term transportation studies and analysis of food aid (1).

Other than North America and Europe, the countries using CTM technology were Brazil, Argentina, Bolivia, Paraguay and Uruguay in South America, and also Egypt and Indonesia. In the 1980s, modeling developed, both theoretically and technically. Improved models considered producer-shipper-consumer interactions and the introduction of microcomputer-based models made it more accessible. Like the urban model packages, these became more user friendly and encouraged the introduction of interactive menu-driven packages and the use of database management software in the new toolbox and database approach (see Figure 3). These packages pioneered modular programs with totally independent database management function and flexibility to add software tools in a "toolbox" for modeling and evaluation purposes. They sometimes provided user-friendly basis for policy evaluation. But they lacked the sophisticated graphics interface available in urban CTM packages (1).

Critics of the LUTS approach note that though models have become increasingly sophisticated, their results are unreliable. Other points of concern indicate that travel behavior is not based on reliable laws and the nature of design of models for understanding a system is not in essence the same as required for operational planning. Moreover, there are added problems of data collecting and lack of support by institutions responsible for maintaining CTM.

In spite of all the criticisms, the CTM provides a systematic, logical framework for analysis and there are few alternative methods to CTM for system level analysis. CTM serves the purpose to get any relative values rather than absolute values for forecasting traffic. The criticisms are based on practical problems rather than theoretical ones. And many of the practical problems could be handled with the use of available technology and new technology.


Micro-economic theories of land-use

These theories represent the beginning of transport and land-use modeling. The micro-economic models have common features that explain land use as a result of market mechanism, in which individual households and firms compete for space, generating an equilibrium pattern of land rent. At the same time, equilibrium prices allow for optimum allocation of land to households and firms, and these, in turn maximize their utilities. Von Thunen’s model is basic for spatial economic theory and it has been extended in several ways. This model explains the effect of transport costs on the location of activities and functioning of the land market. The model has also been extended to include demand economic system works. De la Barra credited Wingo and Alonso, who incorporated the element of budget constraint in the extended version of the Von Thunen model. Christaller and Losch were cited for their explanation of the way in which multi-center regions are formed, each commodity gives rise to its own pattern of location, a network of market areas. Each pattern in turn is conditioned by others and forms hierarchies of patterns and transport networks (6).

Spatial interaction models

The first spatial interaction models were mainly based on a gravitational analogy, derived naturally from the aggregate approach. Instead of looking at individual behavior of an urban area, these models were more interested in the behavior of the different urban areas and the relationship between them. One of the pioneers in this approach was Hansen , who, using the gravitational analogy, concluded that the location of residents was a function of accessibility to employment. Then Huff interpreted the basic gravity model in economic terms and probabilities. Lowry used economic base principles and introduced a multiplier to provide a more comprehensive explanation of the urban structure. Lowry’s work was then improved by Rogers and Grin using matrix methods. The original gravity formulation of the spatial interaction model was then replaced by Wilson's work on entropy maximization. This approach assumes that choices are perfectly random and then introduces a rational (cost) restriction.

Random utility theory and discrete choice models

This is considered as a bridge between the models of microeconomic theory and spatial interaction approach. This allows integration between the theories or principles of the former and discrete aggregated formulation of the other. Discrete choice analysis uses the principle of utility maximization. A decision-maker is modeled as selecting the alternative with the highest utility among those available at the time a choice is made. An operational model consists of parameterized utility functions in terms of observable independent variables and unknown parameters, and their values are estimated from a sample of observed choices made by decision makers when confronted with a choice situation. The early applications of discrete choice models were made for the binary choice of travel mode. Some of these studies focused on the estimation of a "value of time," the trade-off between travel time and travel cost implied by a travel demand model. This value has been used to assign a monetary value to the travel time-savings in the evaluation of alternative transportation projects. Other researchers emphasized the development of policy-sensitive models for predictions of the market shares of alternative modes. Discrete modeling methods in the 1970s were oriented towards mode choice models with more than two alternatives, and applications to other travel related choices such as trip destination, trip frequency, car ownership, residential location, and housing. Studies on the choice of mode for travel to work have used different types of data from widely differing urban areas, and developed more comprehensive model specifications with socioeconomic variables, and tested the forecasting accuracy of the models with data before and after transportation changes (7).

Models of land use impacts on transport

The connection between land use and transport changes cannot be ignored and long-term effects of transport policy may be of considerable potential importance. The estimation and prediction of these effects will depend on the development of reliable, quantitative models that enables two-way interaction between land use and transport. During the past decade or so a number of such models have been developed and have been used for policy testing and planning. Validation of these models is difficult because of the long time scales over which their mechanisms operate, and the Transport and Road Research Laboratory has initiated an international collaboration of seven countries to try to assess the plausibility of nine models by a comparative analysis of their structure and performance.

The International Study Group on Land-Use/Transport Interaction compared the behavior of the models when applied to a set of more than 40 standardized tests involving changes in population growth and composition, changes in the distribution of employment and shops, changes in travel costs and the transport network, and different sequences of transport investment. The different models were AMERSFOOT, CALUTAS (Computer-Aided Land Use-Transport Analysis System), DORTMUND, ITLUP (Integrated Transportation and Land-Use Package), LILT (Leeds Integrated Land-Use/Transport Model), MEP (Marcial Echenique and Partners), OSAKA, SALOC (Single Activity Location model) and TOPAZ (Technique for Optimal Placement of Activities in Zones). These models were put through three tests where in the first two the effects of changes in travel speeds and costs respectively on land use patterns are examined. In the third the effects of redistribution of employment are considered.

The effects of increased speed on land use were generally quite small except in a couple of places. Within relatively unresponsive land use patterns overall, there were disturbances in some individual sectors of population and employment, and there were substantial movements at a more disaggregated spatial level, than in the extremely aggregate comparisons used in the study. The effect of increase in cost provides a good case of the importance of using a fully interactive land use-transport model to examine transport policy. The third test involved radical relocation of employment and this showed a remarkably small effect on travel costs, time or energy. It seemed that people would still travel as much as ever to maximize their choice of employment. The process of testing the models faces the problem of identifying whether the different responses of the models are due to differences between the cities or due to different mechanisms within the models themselves. But the process does help in identifying the characteristic behavior of individual models, to judge their plausibility and find out aspects that need improvement (8).

In the traditional four-step travel demand modeling process, the number of trips made by a household is modeled in terms of household size, income, and other socio-demographic variables and any effect of accessibility is usually not taken into account. But some theories suggest that trip rates must vary with accessibility, i.e., location, land use, or transportation service level does play a role in affecting the number of trips and there are some empirical studies which does support this theory while others do not. The independent effects of land use and accessibility variables on household trip rates were tested for using data from Florida travel surveys. After controlling for socio-demographic variables, residential density, mixed use, and accessibility do not have significant, independent effects on household trip rates. Conventional trip generation models, which generate person trips by vehicle (not by all modes), discounting the effect of accessibility, may not be as inaccurate as it is considered to be in theory. But land use and accessibility variable may have some effect on household trip rates, indirectly through their effect on automobile ownership (9).

Models involving urban character variables

In addition to land use and transportation system variables, it has been suggested that accounting for urban size class, activity concentration, and geographic clusters can enhance model transferability. Based on the nationwide major city characters, several models were developed to estimate trip frequency and trip length. The sample data covered 151 urban areas or 57.6 percent of U.S. SMSAs. The developed models indicate that urban size class and geographical cluster are highly correlated to trip frequency; and urban size class, urban activity concentration, and geographic cluster all affect trip length (10).

Studies show that urban character bears significant influence on travel demand, and the model transferability varies with demand measures and model specification. The specific findings include: (a) the type of activity concentration has a significant impact on trip frequency for non-metropolitan areas and trip length for both metropolitan and non-metropolitan areas; (b) the influence of urban area size on mode choice and trip length is significant for metropolitan areas; (c) the impact of geographic characteristics on travel demand can be ranked in order to trip length, trip frequency, and mode choice; (d) metropolitan trip frequency models are more transferable than their non-metropolitan area counterparts, while the transferability of trip length models of both metropolitan and non-metropolitan areas is very low; and (e) in non-metropolitan areas, non-work trip frequency models are more transferable than work trip models (11).

Individual choices to general equilibrium

This approach presents, under a unified, coherent behavioral framework, integrated, static models, under average or steady conditions, which predict all major dimensions of travel behavior, under a spatial structure of analysis. The basic problem of individual route choice under uncontested conditions, for both car and transit passengers, was solved at the individual level, in the deterministic case (when link utilities are known with certainty), from application of "Minimum Cost Route" and its probabilistic version developed by Speiss. In the stochastic car case, the route choice problem was solved from the application of the "STOCH" algorithm. In all cases, the concept of the "representative traveler" was applied to retrieve aggregate route demands providing network assignment as the solution to the representative traveler’s utility maximization problem. The stochastic transit case, both the individual and the aggregate levels, is left as a combination of both the methods.

Next, other dimensions of travel demand, including destination, mode choice, route choice (still under the absence of congestion), are combined. This is the expansion of the ‘travel consumer’ framework. Route choice modeling is then done with more realistic assumption of network and destination congestion. Under the assumption of various travel externalities, the various aspects of travel demand, destination, mode choice and route choice are modeled. A combined equilibrium model of urban personal travel and goods movements in which commodity flows are generated by the need to support a given urban activity undertaken by individual travelers involving consumption of a given commodity. An explicit, full representation of the interacting behaviors of travelers and commodity suppliers and shippers within the framework of spatial competition is presented. Passenger and freight flows take place concurrently on a common congested network, which is used also for general travel. It has been shown that under general conditions, the model has a unique solution and the algorithm for obtaining the solution was described. The model was then extended to the case of multiple trading levels, and of multiple commodities. Finally, the supply side, the demand responsive determination of optimal transportation and equilibrium between travel supply and demand addressed. It also formulates the network design problem for the car and the transit under congested as well as uncontested conditions (12).

From general equilibrium to dynamic planning models/evolutionary models

The assumption of this behavioral model is that supply and demand are not in perfect equilibrium because costs of moving and switching jobs are high and traffic assignment may not be in perfect equilibrium because knowledge is not perfect. Here the demand in a given year depends on the demand of the previous year. The model redistributes a fraction of work trips each year associated with relocation of a household or taking a new job, and changes in distribution associated with growth or decline are considered. The modeling framework considers equilibrium and evolution as two poles with two interim combinations of the methods, depending on decision time horizon evaluated, whether day-to-day or year-to-year. Day-to-day decisions include route choice, mode choice, departure time choice, and non-work trip destination choice. Year-to-year decisions include relocation or work trip (re)distribution for a fraction of commuters, automobile ownership and trip (re)generation. The decisions are not mutually exclusive, so endogenous year-to-year decision reflect changes in the day-to-day decisions. In addition, the system variables like network, land-use and demographics vary annually. The new model component here is the decision to relocate. The model components include trip generation, distribution, mode choice, departure time choice, route choice and intersection control. This approach has the ability to use observed data more easily and thereby limit modeling to changes in behavior-it adds more realism in the concept of the model. It also has provision of a framework to extend and integrate with land use models and makes available additional information to policy-makers (13).


Stockholm Metropolitan Region

A package of large-scale investments in the transportation infrastructure is currently being proposed for this region. It contains new investments in the railway and subway systems as well as new links in the road system. Four different approaches of appraising this kind of investment program have been done and they are related to one another. The first study uses a network-based mode-split/assignment model with a fixed trip matrix. The second study is complementary, as its aim is to also trace the impacts on the spatial distribution of population and jobs by applying an integrated transportation and land use model. The third study looks at the long-run effects of the investments on regional economic growth within the framework of regional production functions. Fourth, an alternative approach is used, in which benefits from the investments are assessed through their estimated influence on aggregate land values (14).

San Francisco Bay Area

A regional travel forecasting model system update using the 1981 Bay Area travel survey and the 1980 census Urban Transportation Planning Package was done for this area. The demand model development process is characterized as a six-step process involving development of component models and the subsequent packaging into an aggregate forecasting system. The MTCFCAST-80/81 forecasting system involved re-estimation of all model components. Simplifications to the original MTCFCAST system were introduced where warranted; the structure of the mobility and work trip models was tampered with the least. In contrast, the work-trip mode choice model was expanded to distinguish between two-occupant and three-plus-occupant carpools, in support of travel forecasting for high-occupancy-vehicle lane projects. Continuity is seen as the key to maintaining and updating regional travel demand model systems (15).

Tiete-Parana Valley in Brazil

Multicommodity Multimodal Network Design model was used as a planning tool for determining investment priorities for freight intercity networks. The pilot application of the model to assess its efficiency when dealing with large networks at the Tiete-Parana Valley in Brazil indicated that the model was able to simulate accurately the flow of commodities on a large, real network. A new implementation of the solution algorithm within a parallel processing framework was being developed to turn this model into a practical tool.

The basic concept of the MCMND model was based on the assumption that this model would be used as an analytical tool in the process of planning transportation infrastructure investments at a strategic level in developing countries. This calls for a complex level of detail for the representation of the transport services provided in the medium to long-term planning. The multimodal network was characterized as a set of nodes, physical arcs, and logical arcs. The physical arcs connect nodes representing cities, rail yards and stations, river ports, other types of transfer facilities and sometimes the intersection of two different road segments. Logical arcs result from the expansion of zone centroids as well as transfer nodes located at the intersection of arcs representing different modes. The logical network starts with the transformation of each zone centroid into two logical nodes, for demand and supply. Then a logical node is constructed for each mode that enters or leaves a node. Finally loading and unloading arcs are added to link supply and demand logical nodes to each logical node-mode combination, as well as the intermodal transfer arcs. The demand for transportation services was assumed to be fixed and exogenous to the model. Mode choice in shipping freight was modeled in combination with traffic flow assignment, using the assumption that goods are shipped at minimum total generalized costs. A simple interface was designed to link the MCMND model to TransCAD. This interface consists of a file interchange scheme between the model’s solution algorithm and TransCAD. It simplifies the data input-output process and also allows users to visually evaluate the distribution of freight flows over the network, the addition of new links, location of transfer facilities and ultimately the effect of alternative design options. A new path-based stochastic user equilibrium assignment algorithm was proposed to distribute trips over the multimodal network according to a logit-type model (16).

Quad County, Washington State

A study conducted for the Quad County Regional Transportation Organization in Washington State demonstrates that traditional urban transportation planning techniques can be applied in perhaps a new way to perform detailed analysis of freight movements. To accommodate for more refined analysis of goods movement, the traditional modeling process required a number of modifications. The first step of the process included the traditional modeling approach to forecasting vehicle trips, the second step involved detailed assessment of agricultural freight moved and the third step was the approach used to model special generators, in this case, primarily recreational travel. Distribution patterns of the agricultural goods were noted and detailed information on crop distribution helped develop an input-output matrix that was converted into a trip table by making it conform to the transportation zone system. Goods movement was converted from tons to truck, rail or barge trips. The trips were assigned to a highway network developed and microcomputer based transportation planning software was used. The model was used to evaluate the future changes in traffic volumes and goods movements by mode. The basis for these forecasts was the estimates of activities within each land use sector. Growth related projections were then converted into estimates of vehicle and freight movement using the transportation model. External trip growth was also estimated. This study highlights the point that rural transportation issue focuses on farm-to market and roadway conditions than on capacity issues. It is also an example of how quick-response methods (similar to those described in NCHRP Report 187) were applied to enable the project to be completed within 6 months at a lower cost. In fact, it highlights the fact that truck and tonnage can be modeled using techniques similar to those developed for urban transportation vehicle models (17).

Metropolitan Region of Dortmund in the Federal Republic of Germany

The International Study Group on Land-Use/Transport Interaction (ISGLUTI) conducted a study for the metropolitan region of Dortmund in the Federal Republic of Germany. Three land-use/transport simulation models were applied to the Dortmund region: the DORTMUND model, the LILT model and the MEPLAN package. The three models briefly characterized and their ex-post forecasts are compared with the actual development of the region. The study compares how the three models respond to a common set of assumptions and policies from the fields of land-use control, traffic management and transport investment. The differences in model response give insights into the validity of the theoretical foundations and internal structures of the model (18).

Southeastern Wisconsin

A method was developed by the Southeastern Wisconsin Regional Land Use Transportation Study (SEWRPC) for the estimation of modal split land use plans and application of this method in plan preparation. The method was developed specifically for regional planning purposes and has greatest applicability as a broad, area-wide transportation planning tool. The modal split mathematical model assumes that the variables, which presently influence the level of transit utilization, will do so in much the same manner in the future. On the basis of the tests performed on the model, it was concluded that the model replicated the actual transit utilization pattern within the region with accuracy. In the regional, district and zonal levels, the model was found to estimate satisfactorily the transit utilization rate and number of transit trip productions for the four trip purpose categories’ home based work, home based shop, home based other and non home based (19).

Toronto Area

The main objective of the Toronto Area Regional Model Study (TARMS) was to develop a traffic forecasting model to be used as a long range planning tool for the study area. The TARMS model had three major components: trip generation, trip distribution and modal split. In order to establish more precise and reliable relationships between trip making and land use, the trip generation process has been carried out by trip purpose. The TARMS model is in two parts: a 24 hour model to simulate daily person trips, by both private vehicle and public transit for long range planning purposes, and a P.M peak model to simulate trips between 4.30 and 5.30 P.M. for system purposes (20).

Auckland Strategic Planning Model

The Auckland Strategic Planning Model (ASP) is a new generation interactive land-use transport model designed to investigate strategic futures for Auckland over a 30 year planning horizon. The ASP model has enabled for the first time the real integration of transport planning with both land-use planning and environmental management within Auckland. The ASP model provides consistent projections of future urban activities for input into transport models for Auckland. It also represents the interactions between land use policies, transport policies, infrastructure investment, and development controls and their impact upon urban form and the transport system. The ASP model consists of a location model for households and employment, a transport model, a regional demographic model, a model of regional employment growth and an evaluation module (21).

Introduction of the Information Feedback Loop

The Urban Transportation Modeling System (UTMS) is used to forecast travel demand in response to changes in land use patterns, roadway characteristics, and socioeconomic factors. This demand is measured by the volume of traffic that flows through a system of streets and highways. Regional and local area models are developed to respond to different issues, though they share a common pool of information regarding the physical characteristics of the network, as well as the demand for travel. Traditionally, the sharing of information between regional and local models has been a one-way flow. Through the use of traffic assignment software, parts of UTMS have become automated. One of the newest automated processes is the extraction of a sub-area from a larger regional model. This extraction process is important to the local planner because it maintains a link from the regional model to the local model and allows the planner to extract an already distributed trip table rather than build one from scratch. This sub-area extraction process, as practiced, is a one-way information flow. Network and travel demand information from the regional model is extracted and used as part of the development of the local area model. The regional model is calibrated and its information is passed down to the sub-area model. There is a need for the "information feedback loop" to be inserted into the process, to improve the regional model. The sub-area model information is looped back to the regional model and used in the regional calibration. This improvement benefits both the regional and local levels. The enhanced process was applied to a case study in northern New Jersey. The results showed that the new methodology improved the calibration of the regional model, particularly in the vicinity of the sub-area focus model. This improved calibration process is the key to developing sub-area focus models with properly distributed trip tables. The new methodology would work at all levels of the modeling process. This would mean that planners collect data specific to their area and replace these new attributes back to the regional model, attempting to gain a better calibration for their specific area. Once the local planner processes this information, the data could be channeled back to the state DOT. Modification can be then made into the statewide modeling chain and translated into new link attributes or new coefficients for production and attraction equations. This process would mainly contribute towards consistency and greater frequency between calibration updates and thus lead to better forecasting (22).


The most important problem in transport modeling is the lack of coordination between theoreticians and practitioners. Practitioners need answers to problems in the time period available for study, i.e., usually in the short run. They generally tend to adopt a pragmatic modeling approach reflecting the limitations of data, time and resources available for the study. On the other hand theoretically sound models might be difficult to implement, though they might guarantee stable results, consistency and add confidence in forecasting. Theoretically sophisticated models can be too complex and this implies that heuristic approaches, rules of the thumb and ad hoc procedures are sometimes preferable. Model output needs interpretation and this is only possible if reasonable understanding of the basis for such model is available. Good publications bridging the gap between the practitioner and the academic are an urgent need. Use of model criteria depends on the nature of the problem, the level of detail needed and the context of the problem. So the aim of the modeling approach should be to use good, sound models as far as possible, even if it means sacrificing some level of detail. The best balance between theoretical consistency and expediencies each particular case and decision-making context has to be found.

The Atlanta case study brings out the drawbacks of the traditional urban transportation process (23). The lack of specific goals that influence the nature of estimation methods employed is often a drawback in aggregate analysis. The policy sensitivity of the procedures highlighted in this case is inadequate to suggest the possible impacts of a specific transport design on the attainment of the goals. In most cases there may be direct trade-off where one goal is achieved at the expense of the other. In this case, the focus on circumferential, limited access highway construction and improvements enhances the mobility of the suburban areas but at the expense of reduction of accessibility of central-city residents to the economic opportunity that has been fostered by the outer belt. The goal in this case was provision of accessibility to all of Atlanta’s resources, energy conservation, minimization of undesirable environmental impacts and effective transportation to the handicapped and elderly. There is an apparent lack of connectivity between goals and methods.

In analyzing trips generated, the number of trips leaving an origin zone is modeled as a function of automobile ownership; household size and accessibility are not considered. This implies that changes in the trip frequency are independent of changes in the transportation system. The same problem lies at the level of trip attractions. In the trip distribution model (the gravity model) in this case, had the impedance factors based on trip duration as measure of automobile travel, so changes in provision of improved transit service and associated impacts cannot be accurately assessed. So, in this case planners are not in a position to predict accurately the spatial distribution of travel demand under the new equilibrium where Atlanta’s travel times have changed through changes in the transportation system. The basic flaws that were outlined in this study was the requirement of better goals, linkages between forecasting methods and policy sensitivity and the ability of planners to assess the effects of transportation alternatives on goals.

The behavioral approach to transportation modeling, which reflects disaggregated or individual demands, involves assumptions about individual objectives, opportunities, constraints etc., as in microeconomic analysis. It grounds travel demand analysis in an explicit, rigorous framework in which individual travel behavior is explained as a rational outcome of an explicit decision making process of the individual travelers, under specific conditions. This approach though has a sound theoretical base, but it is very contextual, i.e., this model might not be applicable to a whole lot of situations, since the basic assumptions on individual choice behavior might differ. There is need for a more universal approach, which encompasses individual travel behavior but at the same time can be modified and applied to almost all situations. The issue is of flexibility of models. This also brings in the issue of "feedback" into the modeling process and the effect of it on the results. This would in reality mean a large number of iterations to converge. This is really essential to adjust the models to changing times and its effect on the various aspects of transportation. This implies incorporating different other impacts on the transportation planing process, integrating the transportation modeling process with that of other models of regional impact analysis, environmental impacts, energy impact etc. The other very important issue is the distribution of impacts of transportation investment spatially, and the issue of equity. The modeling process should incorporate the cost-benefit analysis into the process, in other to get the best alternatives and assess the impact of the competing alternatives. This would give a broader view of the impact of the transportation planing process to the planner to help in decision making.


The transportation planning process has made considerable progress in recent years in terms of technology used and also theoretically. Some deficiencies in the earlier modeling approaches have been taken care of in some models. But a universal, integrated model, in which all these deficiencies are taken care of, is yet to be developed. This "ideal" model would need to have an integrated system of sub-models, ranging from economic models, regional impact models, environmental impact assessment models, cost-benefit assessment models and so on.


Weisbrod and Weisbrod (25) compared the various techniques for economic impact analysis of transportation projects. Such analysis guides decision-makers of public investments and ensures that they recognize both positive and negtaive economic effects of potential projects. Direct economic benefits to residents affected by transportation projects include increased income from sales to non-residents, lower costs of products and services, and increased opportunities for work and recreation becauses of greater accessibility. Likewise, businesses further benefit from greater product availability, in terms of costs and quality, greater access to labor markets, and lower costs of delivering finished goods. Indirect impacts may be growth of business suppliers, while induced business impacts may be additional business from greater numbers of local employees. There could be further induced changes in business locations and population shifts, both of which will affect land use patterns, wealth, environment, "quality of life," and the revenues and costs of governments in the area affected.

The elements of impact of transportation projects are summarized by Weisbrod and Weisbrod as spending (on construction, maintenance and operations) and user benefits (in terms of time, cost, and safety). These have interactions with:

1. Growth of economic activity (sales, jobs, wages, value added)

2. Overall growth of economic activity (includes multiplier effects)

3. Land development (land use, property values)

4. Fiscal impacts (government revenues and costs)

5. Environment and quality of life impacts

The interactions in the different levels of impacts lead the authors to be concerned with "double counting" of benefits, ignoring the potential for some business gains to be offset by losses elsewhere. While the analysis focuses on economic benefits (leading to increases in income), it is recognized that projects may also have non-economic or social benefits (the "quality of life" discussion).

Stepwise approach

In their guide for planners, Weisbrod and Weisbrod begin with formulating a statement of the transportation project, in terms of mode (freight vs. passenger, infrastructure, vehicles, services), service area, type of change (upgrade, expansion, maintainence, or new mode), and purpose (ease congestion, link existing activities, future demand, new development, quality of life). Then comes the purpose of the analysis, which may be to measure impacts, provide public information, produce the benefit-cost analysis, or conduct a research study of users. The types of analysis will vary, depending on its purpose, as will the "base case," which may assume either a change from existing conditions, a continuation of existing conditions, or if the base case is a time prior to the existing conditions.

Selecting the appropriate geographic study area "causes more error or confusion in economic impact analysis" than any other subject, according to the authors. Analysts should consider the sponsoring agency’s jurisdiction, the area the project influences directly, the distributional impacts of "socially desirable" goals and "dis-benefits," and external consequences (25, page 13). The time periods for analysis of the effects of transportation projects vary from one year to the lifespan of project.

Impact measures

Four different categories of impact measures are listed. User impacts include the money cost of travel, travel time, safety, and intangibles such as comfort and reliability. Total user benefits would be the combination of these impacts. User impacts may be used for impact assessments and benefit-cost analysis of a proposed new transportation service.

Economic impacts are measured by employment, personal income, property values, business sales volume, value added ("personal income plus business profits"), or business profit. Only one measure should be used from these alternatives. Several of the economic measures are used in impact assessment of new service. In addition, employment, personal income, and business sales volume are acceptable for public information about new or existing transportation services. Benefit-cost analysis for new services may use personal income, properthy values, or value added as measures.

Government fiscal impacts are the combination of public revenues and expenditures. The fourth category is a combination of other societal impacts, including air quality, and other environmental and social conditions. Both of these impact measures may be used in analyzine the impace of proposed services (25, page 15).

Benefit measures for benefit-cost analysis

The traditional transportation system efficiency measure is user impacts, a summation of travelers’ cost and time savings (expressed in monetary terms, estimating a value of time), and safety benefits. Since benefits to all users (including those passing through) are included, the benefits to residents may be overstated. Benefits are understated to the extent that nonusers are not included although they see an improved quality of life from having a greater selection of goods and services, or increased property values).

The most common measure of economic benefit is personal income. It includes changes in wage incomes earned by both users and nonusers within the study area. Weisbod and Weisbod consider it a reasonable measure if most of the affected workers live in the study area. However, since some of the increased business income may be paid outside the area, the true impacts are underestimated. Likewise, the benefits to time savings are undercounted unless they lead to changes in business activity.

Value added (gross domestic or regional product) may be "the most appropriate measure of impact on overall economic activity in a geographic area" (25, page 16). It does include business profit that may be paid or reinvested to owners outside the region, thus overestimating the true economic impact. Personal income may be preferred as a more conservation measure of benefits to area residents. Both value added and personal income underestimate the benefits reflected by increased property values which may be induced by transportation projects that deliver cleaner air or a greater selection of opportunities for education, recreation, social interaction, shopping, and jobs.

A hybrid measure of societal benefit might be formed by the combination of business travel benefits (overall income generated) and nonbusiness travel benefits (in terms of "willingness to pay"). The authors warn that obtaining valid data on the latter is "potentially problematic" (25, page 17).

Methods of analysis: transportation models, economic models, direct measurement

Three types of analytic tools may be applied. These are transportation system models, economic models, and direct measurement techniques. Transportation system models simulate and forecast trip generation and routing, modal split, and travel times and costs due to the effects of transportation services and facilities. Economic models are grouped into three types: input/output models, general equilibrium simulation models (which include input/output impacts and changes in business productivity, competiveness, and growth becauses of changes in travel costs), and business attraction and tourism market models (which forecast the effects of transportation linkage enhancements). Direct measurement techniques analyze historical data to measure the effects of transportation facilities and services.

The first type of transportation system model is supply side modeling, which produces either a "full simulation model" or a "sketch planning model" of a portion of the transportation network.. Alternative scenarios describe capacity, projected volumes and trip distributions, plus performance. Outputs will be users’ travel times and costs, links (routes), and nodes (terminals or points of transfer). Net changes are measured in terms of vehicle-miles of travel, passenger-miles of travel, vehicle-hours of travel, and passenger-hours of travel (VMT, PMT, VHT, and PHT, respectively).

Pratt and Lomax present a somewhat expanded list of performance measures for multimodal transportation systems. Travel time was redefined as desired door-to-door travel time, but also divided into segment or trip length. One of the authors’ major conclusions was that planners should employ both mode-specific measurements and multimodal measurements, rather than rely on the latter. Time and distance affect average speed (overall), and average and desired travel rates. Segment volumes are also defined for both vehicles and persons. The performance of intermodal terminals is measured by time or difference in travel time, by delay rate, total delay, relative delay rate, and delay ratio. The authors advise planners to match objectives with performance measures and not let concerns for data govern the development of measures.

Compared to the shipping patterns of the various modes that are identified in supply side analysis, the unique shipping patterns of of different types of businesses are explained in demand side analysis. Outputs will be trip volumes, both current and projected, origin-destination pair, for alternative scenarios. Traffic patterns, volumes and times, by link and node, are estimated by software such as EMME2, MINUTP and Tranplan (25, p. 18).

Input/output models and macroeconomic models are the two basic types of economic models used for transportation project analysis. Input/output models generate multipliers of incomes, jobs, and output for each dollar of project spending. Examples of software based on input/output models are IMPLAN, RIMS-II, and PC I/O Model. Econometric and general equilibrium models are in the macroeconomic simulation category. The REMI model and a model by DRI-McGraw Hill are examples of available software. Economic models would be used to estimate how a project may affect jobs and income, based on the changes in travel times, travel costs, and business costs that are estimated by the transportation models. The integration of the two models may need to be reiterated if the economic model forecasts significant shifts in jobs or population resulting from the proposed transportation project.

More a feedback measurement than a planning step, direct measurement of impacts surveys travelers and businesses to document impacts of a transportation project. Secondary data is also used for tracking changes in income, employment and property values.


Economic impacts: project costs, user benefits and economic benefits

Two categories of project costs are addressed in calculating a project’s economic impacts. In addition to the costs of construction and property acquisition for highways, other rights of way, and facilities, the costs of ongoing operations must be included. For example, maintenance and periodic rehabilitation of highways cost between $6,000 and $10,000 per lane-mile (25, p. 20). User (or travel efficiency) benefits, in terms of cost savings, time savings, and safety, should be calculated for "existing trip" users and "diverted trip" users, that is, those who would switch to the new service or facility. A third group, "induced trip" users, are new trips because of the project; their benefits per trip are typically estimated as one-half the exiting user benefits. While benefits are assumed positive, they may be negative during construction and for congestion.

Past experience has provide some typical figures for valuing benefits. Vehicle operating costs are stated on the basis of dollars per 1,000 miles, and vary by vehicle and average speed. Time savings are also translated into dollar values which differ by vehicle type (tractor-trailer or single unit truck, or automobile) and by trip purpose (work-related or nonwork). Safety improvements are likewise accounted by estimating fatalities, injuries and property damage per 100 million VMT, and converting those to dollar figures using published average costs. Such estimations may appear overly simplified but they do add a degree of consistency of analysis, necessary when results from several projects are to be compared.

Economic benefits are grouped into three classes. The first is jobs directly resulting from construction and ongoing activities connected with the tranposrttion project. Second is the direct project benefits, in terms of travel and logistics efficiencies, economies of scale, and availability of goods and services. The third category is indirect and induced impacts, also known as the muliplier effects. For example, the "job multiplier" would be the number of new jobs in the area per each new job in a specified industry in that same area. Published multiplier values are between 1.5 and 3, depending on whether local, state or national impacts are being considered (25, page 24).


Output considerations

Results may be stated as single-year impacts or a stream of benefits over a longer time period, e.g., the life of a project. Weisbrod and Weisbrod caution users against double- and under-counting of impacts, and to be clear in stating dollar amounts in either constant or nominal terms; they recommend contant or deflated reporting as being better understood by the public (25, p. 25). Likewise, job impacts need to be clearly stated as to their time horizon: single year vs. recurring over time.


Benefit-cost analysis

Both benefit-cost (B/C) analysis and the net present value (NPV) method use the present value of benefits and costs, but B/C ranks projects by the ratio of benefits to costs, while NPV ranks them by magnitude. Thus, a small project with a very high B/C ratio would be rejected in favor of a larger project that had a positive NPV but a lower B/C ratio. Because of uncertainty with both benefit and the cost estimates, planning agencies may require threshold of B/C ratios greater than 1. Selecting the discount rate is described as "an important and controversial policy issue, reflecting political values" (25, p. 26). Using the criterion of "opportunity cost of capital," alternative measures include the cost of borrowing money (interest rate paid by the publis agency), the real rate of return possible in the private sectorthat the money could have earned in the private sector (the interest rate paid in the private sector), or the more conceptual rate expressing peoples’ preference for money now instead future returns. Discount rates typically range between 4 and 8 percent, with 7 percent recommended as matching private sector returns.

Other considerations not included in B/C analysis include feasibility and cost-effectiveness. The resources, in terms of money and technical abilities, need to be available for the project to be feasible, regardless of its calculated B/C ratio. For projects without benefits expressed in monetary terms, a cost-effectiveness measure may be applied. For example, espressing costs in amount per unit or per person served will provide an unbiguous measure and an objective more clearly stated than a goal such as "Improve existing transportation service."

An application of benefit-cost analysis in an intermodal setting was described by Jayawardana and Webre (26). Development projects at both private and public port were eligible for support from a state’s Transportation Trust Fund. An overall ranking of projects used the benefit-cost ratio as the most important criterion, with the highest ratio project receiving 100 points on a 200-point scale. Additional points were given for technical feasibility (maximum of 45 points), economic impacts, in terms of jobs created or saved (20 points), environmental impacts (10 points if no adverse effects; 15 points if a project would enhance the environment), and management of the port (20 points for greatest return on investment). The added criteria, some of which may be viewed as already being included in the benefits or costs, were to accommodate a range of missions from a diverse group of public and private sector "stakeholders," including legislators and representatives from various modes and deep-water and shallow-draft ports.


Transportation planning is undertaken at many levels, from strategic planning to project planning, and over various geographic scales. During the last forty years, urban transportation planning has undergone many changes and continues to evolve. But many of the notions developed in the early years still exist. The basic urban transportation planning process usually consists of three interrelated major components: the pre-analysis phase, the technical phase and the post-analysis stage (14). The pre-analysis stage involves identification of problems or issues, formulation of goals and objectives, data collection and generation of alternatives. The problem definitions need to be broad enough to accommodate a considerably larger set of possible solutions. The technical phase involves mathematical descriptions of travel-related behavior, used to predict the consequences of each alternative transportation plan to be evaluated. It consists of three major components: the land use-activity system model, the urban transportation model system and the impact prediction models. The land use-activity system models are comprised of the spatial distribution of people, activities, and land use within an urban area. These are now integrated with transportation models to asses its impact on travel. They help predict urban activity patterns and generally use regional population and employment as inputs and distribute these totals spatially over a region. The Urban Transportation Model System (UTMS) consists of models commonly used to predict the flows on the links of a particular transportation network, as a function of a land use-activity system that generates travel. The sub-models are that of trip generation, trip distribution, mode choice and trip assignment. The UTMS predicts the quantity and quality in terms of travel time of flow on the links of a specified transportation network, given the land use-activity system as the input. Assessment of alternative options needs estimates of a broad range of impacts, including construction and operating costs, energy consumption, and air quality. The impact prediction models basically need the UTMS as its input. The post analysis phase starts with the output of the technical analysis that comprises of predictions of the impacts of alternative plans and policies. This phase involves evaluating both the economic and non-economic impacts of the alternatives analyzed; selecting the alternative to be analyzed; programming, budgeting and implementing the alternative chosen; and monitoring of the system performance.


  1. Simon Lewis, Peter Cook and Marcelo Minc, "Comprehensive Transportation Models: Past, Present and Future," Transportation Quarterly, Vol. 44, No.2, April 1990, pp. 249-265.
  2. Juan de Dios Ortuzer and L G Willumsen, Modelling Transport, 2nd ed., John Wiley and Sons, West Sussex, England,1994
  3. Richard E Quandt and William J. Baumol, "The Demand for Abstract Transport Modes: Theory and Measurement," Journal of Regional Science, 6, No.2, 1966, pp. 13-26.
  4. Robert Kresge and Paul Roberts, Techniques of Transport Planning, Vol. 2, Systems Analysis and Simulation Models, Brookings Institute, Washington, Transport Research Program 1971.
  5. Adib K. Kanafani, Transportation Demand Analysis, McGraw Hill Inc., USA, 1983
  6. Tomas De La Barra, Integrated Land Use And Transport Modeling, Cambridge University Press, Cambridge, 1989
  7. Moshe E. Ben-Akiva; Discrete choice analysis: theory and application to travel demand. Cambridge, Mass.: MIT Press, 1985.
  8. P. H. Bly and F. V. Webster; "Comparison of interactive land use and transport models," Transportation Research Record 1125, 1987, pp. 29-38.
  9. R Ewing, M. DeAnna and S.-C. Li, "Land use impacts on trip generation rates," Transportation Research Record 1518, 1996, pp. 1-6.
  10. Agyemang-Duah Kwaku, William P Anderson and Fred L Hall; "Trip Generation for Shopping Travel," Transportation Research Record 1493,1995, pp. 12-20.
  11. F.-L. Ou, J. C. Yu,"Effect of urban character on transferability of travel demand models," Transportation Research Record 874, 1982, pp. 5-12.
  12. Norbert Oppenheim, Urban travel demand modeling: from individual choices to general equilibrium. New York: Wiley, 1995.
  13. David M. Levinson, "Evolutionary Transportation Planning Model: Structure and Application," Transportation Research Record, 1493,1995, pp. 64-73.
  14. C. Anderstig, "Appraising large scale investments in a metropolitan transportation system," Transportation, 19 (3), 1992, pp. 267-283.
  15. H. P. H. Kollo and C. L. Purvis, "Regional travel forecasting model system for the San Francisco Bay Area," Transportation Research Record, 1220, 1989, pp. 58-65.
  16. C. F. G. Loureiro and B. Ralston, "Investment selection model for multicommodity, multimodal transportation networks," Transportation Research Record 1522,1996, pp. 38-46.
  17. C. E. Prem and P. Yu, "Applying urban transportation modeling techniques to model regional freight and vehicle movement," Transportation Research Record, 1518, 1996, pp. 22-24.
  18. M. Wegener, R. L. Mackett and D. C. Simmonds, "One city, three models: Comparison of land use transport policy simulation models for Dortmund," Transport Reviews, 2 (11), 1991.
  19. E. Weiner, A modal split model for Southeastern Wisconsin. Southeastern Wisconsin Regional Planing Commission.
  20. Toronto Area Regional Model Study: Trip generation. Toronto: Ontario Department of Transportation.
  21. P. D. Winder, "Integrating planning for transport, land use and the environment: The Auckland Strategic Planning Model," Conference Proceedings: 17th Australian Road Research Board Conference, Gold Coast, Queensland, 1994, Vol. 17, Part 7.
  22. K. B. Winslow, A. K. Bladikas, K. J. Hausman and L. N. Spasovic, "Introduction of information feedback loop to enhance urban transportation modeling system," Transportation Research Record 1493, 1995, pp. 81-89.
  23. Ronald Mitchelson and James O. Wheeler, "Analysis of Aggregate Flows: The Atlanta Case," in Susan Hansen (ed.), Geography of Urban Transportation, 2nd ed., 1995.
  24. Pratt, Richard H., and Timothy J. Lomax. "Performance Measures for Multimodal Transportation Systems." Transportation Research Record 1518, 1996, pp. 85-93.
  25. Weisbrod, Glen, and Burton Weisbrod. "Assessing the Economic Impact of Transportation Projects: How to Choose the Appropriate Technique for Your Program." Transportation Research Circular 477, October 1997.
  26. Jayawardana, Jay, and D. J. Webre, Jr. "Louisiana Port Priority Program: An Application of Benefit-Cost Analysis to Project Appraisal." Transportation Research Record 1511, 1995, pp. 26-33.

Multimodal Investment Analysis: Phase 1 Contents

CTRE is an Iowa State University center.

Address: 2711 S. Loop Drive, Suite 4700, Ames, IA 50010-8664

Phone: 515-294-8103
FAX: 515-294-0467

CTRE Communications: Marcia Brink
CTRE Webmaster: Michele Regenold


Iowa State University--Becoming the Best