2. Overview of transport modelling
Transport models are a systematic representation of the complex real-world transport and land use system as it exists. They are powerful tools for assessing the impact of transport infrastructure options and for identifying how the transport system is likely to perform in future, which is essential for the development of an effective urban planning practice.
The development and application of transport models are fundamental to the appraisal of many transport initiatives because they:
- Provide an analytical framework to assess existing demands on the transport system and project future demands to systematically test the impact of transport and land use options
- Enable the generation of quantitative measures to provide key indicators in the business case assessment and economic appraisal.
Transport models use mathematical relationships to represent the numerous complex decisions people make about travel so that future demand can be predicted, and to replicate observed travel patterns at various levels of geography.
At the most fundamental level, transport models comprise:
- A demand model (trip generation, trip distribution, mode choice and time of travel)
- A highway assignment model (road-based public transport, private vehicles, freight and other commercial vehicles)
- A rail, bus, and ferry assignment model (public transport and freight).
Generally, the development of a transport model requires:
- A Statement of Requirements
- A Functional Specification of the transport model
- A Technical Specification of the transport model.
2.1 Statement of Requirements
A Statement of Requirements usually details the objectives of the model, the interfaces with other models, the hierarchy of transport modelling applications, transport model attributes and transport model outputs. Each of these is described in the following sections.
2.1.1 Objectives of the model
The overall objectives for a transport model refer to what the model is required to do. This can include:
- Providing the technical means for the ongoing development of procedures to quantitatively test and evaluate transport initiatives, strategies and policies
- Assessing the strategic justification for major transport infrastructure projects
- Defining the geographic coverage – initially for specific metropolitan regions, but allowing flexibility to include regional centres within the context of a state-wide model
- Extending the model to test the impacts of transport strategies on a particular location and the intensity of land use development that might occur there.
Establishing a precise statement of objectives, in relation to what is and is not required, is critical to the effective prioritisation of resources in designing and implementing the model.
2.1.2 Interfaces with other models
This component of the Statement of Requirements defines the interface relationships between the transport model and other models. This interface can provide input to corridor‑specific models and the more detailed mesoscopic and microsimulation or operational models.
The key interface attributes are that models share common information and data sources as well as core assumptions, and provide consistency within the hierarchy of transport modelling (see Table 1 below).
2.1.3 Hierarchy of transport modelling applications
Transport modelling development and applications generally fall into the five broad categories described in Table 1. In many cases, planning progresses from the formulation of land use and transport strategy to the investigation of particular projects/schemes to deliver these strategies, and to the detailed operational planning to deliver these projects/schemes.
Land use and transport interaction modelling
Examines and evaluates the impacts of transport policy and land use changes on urban form and transport
Assesses the implications of particular strategies at the metropolitan scale
Some specific examples of the matters model applications aim to assess are:
- Quantifying the effects of land use strategies on transport network performance to identify whether and what interventions may be required
- The assessment of scenarios involving pricing policies (e.g. fuel, tolls, parking charges), transport infrastructure provision and service improvements
- How to best allocate transport system capacity in the face of the demand from freight and private vehicles, public transport users and pedestrians and cyclists.
The continuum of modelling requirements, set out in Table 1, draws on different modelling techniques. These range from macroscopic (often applied for land use and strategic planning purposes) through mesoscopic to microscopic (typically applied for operational design). A broader range of demand responses (see Sections 3.2 and 3.3) is typically represented in macroscopic models together with a more aggregate representation of the transport networks (see Sections 3.6.1 and 3.6.2). In comparison. microscopic models provide more detailed representation of driving behaviour and of the performance of the transport networks (see Section 3.6.3). Activity based modelling (see Section 3.7) involves wider use of simulation techniques for strategic planning.
Selecting a modelling technique is an important part of the Statement of Requirements. The selected tool needs to be sensitive to the relevant issues of a project. Unnecessary complexities should be avoided as they will increase the cost and introduce risks. However, the objectives of the project are paramount, so while the chosen modelling technique needs to be simple, it should not be oversimplified. A good balance needs to be found.
In some instances, the blend of requirements may best be delivered using more than one model. Ensuring consistency when using a combination of transport models inevitably adds complexity. Nevertheless, individual tools and methods may provide a better focus on individual output requirements.
A particular tension rests in the need to assess value for money where funding decisions are based on user benefits (for example, travel time savings) that are derived using consumer surplus theory. The typically more limited scope and calibration requirements (see Section 5.8) of microscopic modelling techniques tend to limit their suitability for this purpose. However, the greater precision in representing the performance of particular design options and the ability to demonstrate visually how infrastructure would operate are particular advantages of microsimulation tools.
The challenge is to match the context of a project to the strengths and weaknesses of each technique. The first step is to contextualise the project by identifying its key elements. These key elements should be able to cover the schemes and data, inter-relationships of factors and objectives of the project. These elements can then become specifications of the model to be applied, such as input variables, scope and mechanisms, and output variables (shown in Figure 1), as follows:
- Input variables – should be sensitive to the proposed schemes and make use of available data. Input variables include representation of the demand and the transport network. Demand distinguishes the what, when, where and how travel is made, together with variations in travel behaviours. The transport network is defined by physical attributes (e.g. highway geometry) and traffic control (e.g. traffic signals).
- Scope and mechanisms – cover the relevant inter-relationships of factors considered in a project, including its geographic and temporal scope, traveller responses and other capabilities required of the model (e.g. optimisation).
- Output variables – should be able to represent the objectives of the project. Output variables include the relevant indicators and accuracy requirements. The likely users of the modelling outputs should be considered when preparing the model outputs.
Figure 1: Elements of the traffic system model
No clear science exists for selecting the most appropriate transport modelling technique. The selection needs to balance the requirement for rigorous analysis against the cost. It is also important to filter out case studies that do not require modelling in order to focus resources on projects that require modelling. Accordingly, guidelines are needed to facilitate the selection process.
It is not the purpose of the ATAP Guidelines to identify a particular technique. The intent is only to structure the decision-making process. In the end, a subjective judgment is required. The Guidelines set out four steps in technique selection, as shown in Figure 2.
The first step involves careful consideration of the project objectives. The high level preliminary analysis advocated in step 2 considers the nature of outcomes and impacts that may be expected, from which the analyst can form a view on the extent to which a model can quantify these outcomes and the importance of this information in making decisions on whether and how to proceed with the project. Given an initial decision that some modelling is justified, the subsequent step 3 is to consider the aspects of the transport system that require modelling. Finally, step 4 is to consider the methods that can best be applied to represent these aspects, reflecting the nature of the project and the expected outcomes.
The Guidelines provide an introduction to modelling techniques, such as land use modelling and activity based modelling, that have had limited application within Australia. Emphasis should be placed on the need to undertake modelling to a sufficient standard, regardless of the technique selected. A 'scattergun' approach of using resources to attempt too much is not a sensible strategy, and focus is needed in selecting what should be modelled.
Figure 2: Steps in technique selection
2.1.4 Transport model attributes
Desired transport model attributes detailed within a Statement of Requirements may include:
- The model is readily accessible to key decision-makers and can enable a prompt and reliable response.
- It is an integrated multi-modal model updated annually to account for new data as well as for identified errors or changes in core assumptions.
- The model provides sensitivity to changes in demographics, individual travel decision‑making, social behaviour and land use or some combination of these.
- There are requirements for model governance, accessibility, development and maintenance.
- The model has the ability to model motorised and non-motorised travel.
2.1.5 Transport model outputs
Examples of model outputs contained in a Statement of Requirements may include:
- Road, rail, bus and ferry transport patronage estimates
- Transport network performance (such as vehicle-hours, kilometres of travel and congestion indicators) and accessibility measures (such as services, employment)
- Forecasts of aggregate travel costs and benefits
- Source and extent of diversion
- Input to externalities modelling (such as quantum of emissions)
- Input for economic appraisal and business case development.
2.2 Functional Specification
The Functional Specification of a transport model describes the functions it should include, based on the scope of the transport model (see Section 2.2.1) as defined in the Statement of Requirements. It also outlines the model structure (see Section 2.2.2) as well as the functions and methodologies appropriate for the various components (i.e. trip generation, trip distribution, mode choice, time of travel and trip assignment), the inputs and outputs, and the data source.
2.2.1 Transport model scope
The scope of the transport model is defined by the policy issues the model system aims to address. These may include:
- Land use–transport interaction
- Pricing (toll roads, parking, public transport fares)
- Parking provision (cost, reduction in availability, park-and-ride facilities)
- Road network management (new roads, commercial vehicle priority, traffic management, high occupancy vehicle lanes)
- Public transport networks (extensions, service provision, fares, interchanging, cross-town routes)
- Non-motorised travel (cycling, walking).
Establishing a suitable model scope and structure for transport modelling and analysis is not a simple process. A number of modelling approaches exist, ranging from the option of using no formal transport models to the most complex microsimulation models.
The selection of the transport model structure and scope in the Functional Specifications is driven by the modelling and appraisal requirements of the jurisdiction. Generally for urban transport, each jurisdiction aims to have a single multi-modal model for its metropolitan area with broad ranging and versatile capability that can be applied to a range of studies or initiatives. This is the only practical and cost-effective way to proceed as it reduces model development and maintenance costs. It also ensures broad consistency in the use of modelling across studies and initiatives.
Of course, there will be instances where modifications to the model will be required to meet the specific needs of a study (such as creating a sub-model) or changing the resolution of the travel zone system for a study (such as focusing on walking and cycling in a particular area). Typically, the overall aim should be to minimise the production of bespoke models for individual projects.
2.2.2 Transport model structure
A broad transport model structure may include:
- A database populated by data from travel surveys across the region by various modes by time of day, together with observed traffic volumes across the road network and patronage levels on the public transport network, including current and projected land use data and demographics (population and employment)
- The inputs to the modelling process, such as parking supply, land use distribution, fares, car travel costs, traffic management measures, access restrictions, road and public transport infrastructure, and public transport service provision
- A travel demand model to derive the quantum of travel across the region, comprising trip generation, trip distribution and mode choice modules, including factors such as travel purposes and the quantum of commercial vehicle travel
- A freight model to derive the quantum of freight transported across the region sufficient to estimate the quantum of commercial vehicle travel on the road network and the requirements of the freight task on the rail network
- A transport supply model covering the road and public transport networks, including factors such as parking supply, road and public transport network capacities, travel times and travel costs
- An assignment module to allocate travel demands to the transport supply model in an iterative manner, to ensure the forecast demands are balanced with the transport supply, taking into account congestion effects
- The required outputs, such as network performance indicators including vehicle-hours and kilometres of travel, passenger-hours and kilometres, congestion indicators and tonnages of emissions
- Other information, such as emissions (NOx, CO, CO2), traffic volumes, trip lengths, trip costs and benefits and accessibility measures.
2.3 Technical Specification
The Technical Specification of a transport model usually follows the choice of modelling approach and includes the methodologies and processes developed to meet the Functional Specification. It details, among other issues, the input data, model calibration and validation data, and the format of the required outputs. Information relating to Technical Specification is covered in subsequent sections of this document.
2.4 Transport modelling process
The transport modelling process comprises a number of stages, as shown in Figure 3. These include:
- Consolidating the modelling task, which includes identifying the key transport, socio‑economic and land use issues as well as the particular problems to be modelled. This stage is also informed by the definition of goals, objectives and the appraisal criteria to be adopted
- Data collection, which is critical to transport modelling and may include highway and public transport patronage data, import/export or production/consumption volumes by commodity as well as census information and targeted or area-wide travel surveys. Usually the data collection is defined after the model scope has been specified; nevertheless, in practice a good model design would consider the existing data available
- Model estimation, calibration and validation, which is required to develop the relationships used in the modelling process and to gauge the performance of the transport model. The process involves checking and refining input data and the suitability of relationships, and comparing model outputs against observed data for the base year conditions (discussed further in Chapter 5)
- Options development, which usually includes variations of transport network options, land use options or combinations of both
- Options modelling, which might enable further refinement and development of options as well as more detailed design and appraisal. This stage usually involves an iterative process covering options development and modelling through to appraisal
- Sensitivity analysis, which varies input data and model parameter values to identify the robustness of the model relationships and the associated forecasts.
- Economic appraisal, which uses results of modelling as input to the appraisal process to assess the performance of the options against the specified goals, objectives and criteria
- Modelling report, which involves the full documentation of each of the previous stages, including the transport model details.
In some cases it is more efficient to undertake a staged process. For example, a ‘long list’ of schemes might be appraised using simplified methods or a reduced set of performance criteria to establish a short list where more detailed modelling and appraisal is undertaken. Initial consideration of options during options modelling is often based on a focused assessment of how well they address the underlying issue, with wider considerations of value for money considered in the later economic appraisal stage.
See Appendix A for more information on the transport modelling process.
Figure 3: Transport modelling process
[click to enlarge]
2.5 Alternatives to modelling
An issue of interest is whether it is necessary to use modelling or other analysis methods for specific applications. An alternative to modelling is to use sketch planning techniques. Sketch planning methods generate only rough indicators and an enumeration of factors and their potential impact on the schemes being examined. Sketch planning is commonly used to reduce the number of alternatives being considered for further analysis. During the course of a sketch planning exercise, a certain alternative may stand out clearly or all other alternatives are eliminated. In this case, it is no longer necessary to conduct further analysis.
An example application of sketch planning techniques is illustrated by Van Hecke et al (2008). In this example, sketch planning was applied to deploy certain operations technology, including detection/surveillance, incident management, traffic flow management, traveller information and a regional weather information system. Recommendation on whether to deploy any of the technologies was based on criteria employing easy to derive or acquire data, including traffic volume, peak-hour conditions, accident records, traffic generators and weather conditions. The analysis is conducted without the benefit of modelling the effect of operations technology, but it is presumed that under certain conditions it is likely that a technology would be cost‑effective. Further analysis may then be conducted on the basis of recommendations of the sketch planning methodology.
Another example of sketch planning is the use of prescribed warrants in recommending signalisation of an intersection. One methodology specifies 11 warrants where, if an intersection satisfies a certain number or combination of warrants, the intersection will be recommended for signalisation (Kell & Fullerton, 1991). The warrants include traffic volume, type of approaches (for example, major or minor road), pedestrian volume, school crossing, accident exposure and others. Modelling is not necessary to conclude whether an intersection should be signalised or not. In some cases, crude calculations would be sufficient and past projects or studies with similar characteristics could also be helpful to reach a conclusion without modelling.
Another approach is to use qualitative comparison of alternatives. Factors to consider include environmental, social, strategic planning and economic considerations. A review of advantages and disadvantages of possible alternatives may be sufficient to reach a conclusion. Qualitative comparison could even highlight important factors beyond the scope of modelling (such as aesthetics, social impacts, strategic impacts and complex behavioural responses).
Modelling is generally a time consuming and expensive exercise. Therefore, it is good practice to first conduct a preliminary analysis of alternatives using sketch planning and/or qualitative comparison. A decision to proceed to more rigorous analysis using modelling techniques should then be conducted under one of the following conditions:
- The preliminary analysis fails to identify the best course of action.
- The project requires a rigorous analysis for approval from decision-makers.
- There are significant risks involved if the recommendations provided are wrong.
These Guidelines recommend applying various alternatives to modelling and to proceed to modelling only when a need is clearly identified.