Decision support systems are tools used by policy makers to quantify environmental, social and economic impacts of transport and energy services. These systems use macro economic and population information such as economic activity (GDP), household incomes and total travel demand to derive total fuel demand and atmospheric pollution.
They often take the form of complex computer programmes that forecast, simulate and optimise effects of policy options for different scenarios and include the ability to consider environmental externalities. Many transport decision support systems rely on the urban transport planning (UTP) process. The urban transport planning process, a few relevant decision support systems and their respective data requirements are considered in the following subsections.
2.3.1. The urban transport planning (UTP) process
The urban transport planning process uses a generalised model to estimate travel demand and future transport infrastructure requirements. The UTP process is discussed here because it often forms part of transport and energy decision support systems. Bruton (1978) and Dimitriou (1992) provide descriptions of the UTP process and Kohoutek et al.
(1999) described a software implementation of the UTP process, including the simulation of emissions. A brief description of the UTP process is given here.
The UTP process involves four consecutive steps: trip generation, trip distribution, modal split and traffic assignment. Trip generation models the motivation for making a trip using demographic information obtained from household surveys such as household income, number of members in the household and ages of the members in the household. Trip distribution estimates the distribution of trips between origin-destination pairs based on sizes of the respective areas and distances between them. Household and roadside surveys are used to calibrate trip distribution estimates. Trips are then divided into different modes of transport depending on levels of income, vehicle ownership and accessibility to public transport. Once the number of trips, their origins and destinations, and their modal share have been estimated, routes that are used to execute trips are determined using traffic assignment or network models. Network models are calibrated using traffic counts at selected points within road networks.
Outputs from the UTP process include travel demand (vehicle kilometres) and traffic flow (vehicles per hour) for various areas and times day. This information, together with
speed-flow relationships (determined from sampled traffic flow and speed measurements during traffic counts), provides average speeds experienced along different routes. Traffic flow, average speed and corresponding emission factors for various vehicle types provide information needed by some decision support systems and emissions inventory models to evaluate total fuel consumption and emissions for a study area (TRB, 2000;
Brand et al., 2002). Other decision support systems substitute average speed and their corresponding emission factors for other parameters such as speed and acceleration (de Haan and Keller, 2004a).
2.3.2. TRAN
TRAN is the transport sub-model of the United States of America’s National Energy Modelling System (NEMS) (DOE/EIA, 2004). NEMS is a suite of models used by the US Energy Information Agency (EIA) to estimate the demand, supply, conversion and economic impacts of energy use within the different sectors of the US economy. The purpose of TRAN is to forecast energy demand, vehicle fleet structure and emissions from transport.
The structure of TRAN, inputs from the other models in NEMS and outputs are illustrated in Figure 2.2. Inputs into TRAN include fuel prices, new vehicle sales by region, economic data (such as GDP and income distribution), demographic data (such as population and age distribution) and defence spending. Outputs from NEMS are fuel use by region, fuel efficiencies, vehicle miles travelled and emissions.
The NEMS transport model consists of seven sub-models (shaded blocks in Figure 2.2).
The modules of most relevance to this study are the LDV (Light Duty Vehicle) Module, the LDV Fleet Module, the LDV Stock Module and the Emissions Module. The LDV module forecasts market share of technologies used in new vehicles based on fuel prices, economic indicators, demographics and fuel efficiency. The module also estimates fuel economies of new vehicles. The LDV Fleet Module calculates travel demand for fleets of vehicles used by companies and utilities. Fuel demand and efficiencies are calculated using the updated vehicle fleet estimated from the market shares of new vehicles. The module also evaluates the number of vehicle sold off to private use. The LDV Stock Module calculates new efficiencies, population structure and fuel consumption of the vehicle population not owned by vehicle fleets.
Although the emissions module is currently inactive, its purpose is to aggregate transport emissions based on vehicle technologies, fuel prices, alternative fuels and efficiencies.
Emissions considered are SO2, NOx, carbon and VOCs (volatile organic compounds). A likely candidate for the emissions module of the TRAN model is MOBILE6 (TRB, 2000), or a more recent version of the same model called MOVES (TRB, 2000; Boulter, 2007), developed by the US Environmental Protection Agency (EPA). This model is discussed in Section 2.4.
Figure 2.2: NEMS Transport Model (DOE/EIA, 2004).
2.3.3. STEEDS
Scenario-based framework to modelling Transport technology deployment: Energy–
Environment Decision Support (STEEDS) (Brand et al., 2002) is a software based decision support system. Its purpose is to allow decision makers to evaluate transport energy and environment policy options without needing to integrate a variety of different models. This is accomplished by integrating several models into one system. The structure and components of the STEEDS model are shown in Figure 2.3.
Figure 2.3: The STEEDS model structure (Brand et al., 2002).
The input phase of the modelling process allows users to develop various scenarios based on parameters external to transport systems. This can include population growth, economic development and international fuel prices. Policy options are also developed at the input stage such as investment in public transport, local fuel prices and land use policies.
The modelling phase of the system uses information from scenarios and policy options combined with information about vehicle fleets, transports system and lifecycle analysis to calculate transport demand and related fuel consumption and emissions.
2.3.4. TREMOVE
TREMOVE was developed by Transport and Mobility Leuven for the European Commission to simulate transport and environment policies and consider their impacts on
transport demand, modal shares, fleet structure, emissions reduction technologies and emissions from transport (De Ceuste et al. 2006). The original version of the TREMOVE model was developed as a technical motivation for the European Auto-Oil II programme.
TREMOVE is represented in Figure 2.4.
Figure 2.4: Model structure of TREMOVE (De Ceuste et al., 2006).
The model consists of three main modules estimating travel demand, vehicle fleet size and structure and a fuel consumption and emissions. Two additional modules calculate life cycle emissions and changes in welfare due to the policy scenarios. TREMOVE consists of several sub-models, each developed within programmes supported by the European Commission, as illustrated in Figure 2.4. An understanding of the operation of the sub-models can be overwhelming and the advantage of a single integrated system such as STEEDS is emphasised.
The module of interest and relevance to later parts of this study is the emissions model for road transport, which is the COPERT III application discussed in Section 2.4.2.
2.3.5. Summary of decision support systems
The decision support systems mentioned above all have similar structures with three main sub-modules or sets of data: a vehicle stock module, a travel demand module and an emissions module.
The systems discussed have been used to model energy and emissions for national, regional and local levels. The issue in using them in the South African context is whether local data are available and in a suitable format to be used within the models. The decision support systems discussed, however, are dependant on emissions inventories and lower level emissions simulation models. Layering of models in such a way hides details and classification of the emission factors they use.
In order to understand the origin of emission factors used in the higher-level models and hidden by abstraction, emissions inventory models and fuel consumption and emissions simulations models need to be considered in more detail. It is in these models where local driving conditions and vehicle fleet structure are considered. For this reason, decision support systems from elsewhere cannot be used for South African cities without transforming the fundamental emission factors first.