A key part of undertaking an assessment of a new transport development is the calculation of the changes to the traffic flows within the transport system that will occur, and the resulting shifts and changes in pollutant loads. Traffic and transport models provide the base data in the form of traffic flows and speeds, which can be applied to an emissions database or emissions model to determine the emissions associated with the transport corridor in question.
Transport and traffic models are used for a wide range of planning and evaluation situations, ranging from regional strategic transport planning of large urban areas, to a very localised assessment of a proposed network or land-use change. The nature and extent of models are generally tailored to their use, and so the characteristics and capabilities of models vary as much as the range of situations examined.
The models used in evaluating a transport project need to be capable of producing the traffic outputs that can be used as inputs to the air discharges effects assessment. The models should also be consistent with other relevant documents, such as Transfund’s (2004) Project Evaluation Manual.
A2.1 Types of traffic model
Traffic and transport models can be categorised and labelled in several ways, as follows.
The main purpose of a demand model is to produce travel demands for use in more detailed (assignment-type) models. Such models are calibrated to generate demands from the key pressures and reasons that result in travel (eg, housing patterns, retail facilities, workplace locations, schools), and to be sensitive to those variables that influence changes in travel patterns. The model network representation and zone structure may be relatively coarse. In these models, generally all the travelling vehicles reach their destination during the modelled time period, and specific longer-term, secondary or detailed characteristics are usually not well represented.
Network assignment models
These are not capable of generating travel demands in themselves, and generally have more detailed network and zone structures. The representation of intersections, in particular, may be more specific and detailed, and the assignment procedures compatible with this level of detail.
Depending on the software used, these models will assign all demand in the modelled time period and not be capable of passing over-capacity demand to another time period. Some territorial authorities and others have developed models of this type for evaluating specific roading projects, ranging from regionally significant projects to local network improvement works.
Demand and assignment models
An example of a demand and assignment model is the multi-modal transport model developed by North Shore City Council for the North Shore, which extends onto parts of the Auckland isthmus and CBD. This model produces travel demands and assigns traffic to the network.
These may be network models with simulation capability. They can simulate some detailed characteristics of traffic behaviour, such as representation of queues, and possibly passing of demand from one modelled time period to another. These features may be critical to a local-scale effect assessment. Some software packages are capable of both simulation and non-simulated network modelling, and users can switch between simulation within the immediate study area and non-simulation outside this.
These represent the travel behaviour of individual vehicles, and can be of a network or an isolated situation. Operational characteristics such as lane-by-lane flows, queuing, weaving and merging can be modelled, and different characteristics given to different vehicle types. The use of such models is increasing given their ability to represent driver behaviour in congested and complex network situations.
Isolated models of a single element of the network can represent detailed traffic behaviour and the operational performance of the element. An example would be the model created to assess the upgrading of a single intersection. A combination of assignment and simulation or micro-simulation models is likely to be used for air quality assessments. The wider area may be evaluated with the assignment network model, and the local area or areas of the project are evaluated with the more detailed simulation or micro-simulation models, where the ability to better represent some aspects of driver behaviour is required.
A2.2 Model validation and performance
In the normal course of developing a model for a transport project, the model undergoes a calibration process and is validated against observed base-year data. Once the performance of the model is tested, the required forecasting capabilities are developed. This can be documented in a report. Transfund’s Project Evaluation Manual sets out criteria and guidelines for the development and validation of traffic models.
Generally, if a traffic model has met the Project Evaluation Manual guidelines or has been deemed fit for use by the project’s overseeing group, then it will be suitable for providing input data for the assessment of air discharges. However, it is the responsibility of those undertaking the air discharge assessment to provide the supporting information that shows the model is suitable for the assessment, setting out the basis of the model, and its validation, performance and forecasting.
In some cases, there may be a requirement to show that the model is performing adequately in a specific location where emissions are being assessed. This will be done through a validation exercise similar to that described above, and it is preferable for this to be included in the scope of works. This process will need to show that the model adequately reflects the changes between the base case and the option in respect of flows, speeds or delays.
A2.3 Measured traffic data
Traffic counting is undertaken via counting programmes and specific-purpose counts by the agencies involved in the management of the transport system. Transit New Zealand takes counts on the state highway network, and the various territorial authorities take counts on the road networks under their management. Counts are also undertaken by private organisations for their own purposes.
The standard methods for obtaining count data are as follows.
These are taken with a rubber tube laid across the width of a road and using a counting mechanism at the side of the road triggered by pulses of air created when vehicle tyres compress the tube. This method is used at mid-block sites on arterial roads. Two tubes are required to obtain counts by direction. Two tubes placed a specified distance apart across the same width of road can enable speed data and vehicle classification data to be obtained. The latter is based on calculation of the vehicle’s wheelbase from the time between air pulses. Generally, tube count data can be obtained for one-hour intervals throughout the day, with the peak periods at 15-minute intervals, although the output format varies. The accuracy of tube counting declines as the levels of congestion increase due to slow-moving vehicles not creating an air pulse. This obviously has implications for air discharge assessments.
These are taken at, or near, intersections controlled by traffic signals with an inductive wire loop embedded under the surface of the road connected to the intersection operating system, called SCATS.8 The count data can be recorded for a continuous period over some time if required. This method of counting can provide data for each loop, which may be separate for each traffic lane, and by time of day. Speed data are not available using this method. The accuracy of the counts is dependent on the calibration of the SCATS system.
Manual traffic counts
These are usually undertaken at intersections for a specific project to obtain the flows for each turning movement. The data can be segmented into different vehicle types as required for the project. The counts are often taken for the peak periods and may also be for part of the inter-peak period. The use of count data with models varies with the type of model and the time horizon being modelled.
Network models usually obtain initial base-year and forecast traffic demands from a larger network model. Count data are compared with modelled flows to measure how well the model validates. These count data or other independent count data may be used to assist in the adjustment of the demands so that modelled traffic flows better match count data. In these models, it cannot be expected the traffic flows in the base year will match observed counts to a high level of precision in all locations. If there are particular local emissions sites of concern, there may be a need to undertake specific checks on the model’s validation at these sites and to use this information when making emissions assessments in future years.
Localised models of intersections or short sections of road may obtain their input traffic flows directly from count data. This is particularly the case where there is no route choice in the model. Hence the modelled flows should replicate the input observed flows.
A2.4 Vehicle classification systems
The vehicle classification systems commonly used in New Zealand are summarised in Table A2.1, as well as estimated equivalencies for NZTER vehicle classifications.
View classification of vehicles for New Zealand (large table).
8 Sydney Coordinated Adaptive Traffic System, (SCATS) by RTA, NSW Australia. The SCATS software package is an area based traffic management intersection control system that responds to changes in traffic flow and conditions by adjusting the phasing at each traffic light cycle in real-time. Across New Zealand, more than 500 traffic signals are coordinated and managed with SCATS.