Decision Makers' Guidebook
What do we mean by optimisation?
Strictly optimisation means finding the best solution to a given set of transport problems, or the best strategy to meet a given set of objectives. In practice, cities will not often be free to implement the combination of policy instruments which is theoretically best for them, either because they do not have overall control on all policy instruments (for the reasons given in (Section 3) or because they face barriers of finance or acceptability (Section 10). In practice, therefore, optimisation involves identifying the best solution within a given set of constraints.
Why should we use optimisation methods?
Traditionally, cities and their consultants have attempted to determine the best strategy through a process of identifying a possible solution, testing it (Section 12), appraising it (Section 13) and then seeking improvements. These improvements could either be straightforwardly to increase performance, or to overcome barriers such as lack of finance or limited public support. However, this process can be inefficient; time will be wasted on testing inappropriate strategies, and there is no guarantee that the best strategy will be found. Thus the benefits of optimisation are both in developing more effective strategies and in doing so more rapidly. In an early example in Edinburgh, an initial study used some 70 model runs to develop a “best” strategy; a subsequent study using optimisation methods found a combination of policy instruments, after 25 model runs, which increased economic efficiency by a further 20%.
Optimisation is thus a very elegant way of choosing the best strategy. Even if we do not often want to automate the decision making process in this way, experience shows that it produces interesting new strategies that would not otherwise have been thought of.
How does optimisation work?
Formal optimisation is a relatively new concept in the analysis of integrated land use and transport strategies. We describe it further in the PROSPECTS Methodological Guidebook, and in a more recent report on the generation of optimal strategies for UK cities. It involves maximising a quantified objective function within a given scenario, and subject to a given set of targets and constraints, by using a given range of land use and transport policy instruments.
How are objectives represented?
At the heart of this policy optimisation process lies the definition of the objective function, which is a quantified measure of the policy-makers’ objectives and the priorities between them. The objective function should be consistent with the appraisal framework (Section 13), and can thus be based on either a Cost Benefit Appraisal or a quantified Multi-Criteria Appraisal, in which weights are assigned to the individual objectives. The value of the objective function for each set of instruments and their associated levels is derived by running a land-use transport interaction model (Section 12).
How are scenarios and constraints reflected?
Scenarios can be selected based on the principles in Section 11. Often the strategy is optimised against one scenario, and the optimal strategy is then tested for robustness against other scenarios. In due course methods may permit optimisation to be pursued for all scenarios, with techniques of appraisal under uncertainty being used to minimise the risk of poor performance under more demanding scenarios.
Constraints can be dealt with in two ways. Political barriers can act as a constraint on which instruments may be considered and within which ranges; for example parking charge increases of above a given level may be considered unacceptable. Financial barriers and outcome targets can be incorporated within the optimisation process; for example a restriction on capital investment could be used to rule out those strategy options which exceeded it. In either case the optimisation can be repeated without the barrier to demonstrate the benefit of removing it. This can help in making the case for changes in legislation (Section 10).
How are policy instruments selected?
Policy instruments can be chosen from the list in Section 9. In due course, new approaches to option generation may help to suggest which policy instruments should be considered. A formal optimisation process is most useful in considering a package of strategic instruments which are expected to have a significant impact on the city. They will reflect the key strategy elements in Section 11. Most strategic instruments have some level which may be varied (e.g. a price) which can be optimised. The diagram shows an optimum for a range of levels of fares and frequencies. Some, such as discrete road and rail projects, are either included or not. Once an optimal set of strategic instruments has been selected, other second order elements of the strategy (Section 11) may be added in ways which enhance the overall policy.
When are optimisation methods appropriate?
When a city is assessing a relatively small number of policy instruments, or simply assessing one new proposal within a given strategy, formal optimisation is unlikely to be needed. However, where the number of options is substantial it will often be much quicker and less expensive to use a model in conjunction with an optimisation method than to use the model alone. Where there are several scenarios to consider, or constraints whose impact needs to be assessed, optimisation can prove even more valuable.