The MUSIC (Management of traffic USIng traffic flow Control and other measures) project has demonstrated that a novel approach to traffic signal control, alone or in combination with other measures, may be used to meet a variety of traffic management goals including:
- To reduce the travel times of public transport vehicles;
- To improve pedestrian facilities/comfort; and
- To reduce delays and stops experienced by vehicles and travellers.
The project has also disseminated the overall approach, the particular optimisation procedures currently adopted, the modelling and real-life results, and proposed future developments; so that the MUSIC approach may be further developed and utilised throughout Europe.
The MUSIC approach to designing traffic signal timings
The MUSIC approach to designing traffic signal timings may be applied at low cost to any city or town that has an existing network model and traffic signals. The approach has the following six stages:
Stage 1: Agree measurable objectives that quantify some relevant aspects of the local transport policy;
Stage 2: Translate data from an existing network model;
Stage 3: Use an off-line optimisation procedure implemented in software to create new time-of-day traffic signal timing plans which aim to meet the measurable objectives while attempting to take some correct account of travellers' future choices;
Stage 4: Test the new timing plans in the existing network model;
Stage 5: Implement the traffic signal timing plans on-street; and
Stage 6: (optional): Conduct "Before" and "After" studies to assess performance against the objectives agreed in Stage 1.
The six-stage MUSIC approach to designing signal timings is a form of decision support system. Technical barriers have prevented effective large-scale implementation in the past.
Novelty
The MUSIC project is novel because:
- The measurable objectives are explicitly defined in terms of transport policy objectives and timing proposals produced are assessed against the objectives;
- Traffic signal settings are used to achieve a variety of these traffic and demand management goals; and
- Proper systematic account has been taken of future traveller behaviour, both in a computational model and in reality, over a whole town rather than in either a very small network or a single corridor.
Usually signal timings are chosen so as to minimise an objective function (often total travel time) on the assumption that the traffic or transport pattern does not change. The MUSIC project sought to deliver a much greater variety of goals while allowing properly for future changes in the transport pattern.
Theoretical basis
The theoretical basis for the current optimisation procedures within the MUSIC project has its origins in the P0 policy developed by Smith in the late 70s as a mathematically sound optimisation procedure for setting traffic signals. Since then these optimisation procedures have evolved over many years in collaboration between the University of York, the Institute for Transport Studies at the University of Leeds and Hague Consulting Group. The procedures continue to be developed in the light of evolving:
- National and European policy objectives;
- Computational technology;
- Implementation technology, and
- Mathematical optimisation technology.
Methodology adopted for the MUSIC project
Within the MUSIC project the above six steps were applied to each of three demonstration sites and 'Before' and 'After' studies were conducted at each site to judge the success of the implementations. The demonstration sites were York (UK), Porto (Portugal) and Thessaloniki (Greece).
Alongside this programme of work a substantial programme of theoretical work was undertaken so as to improve the design of the crucial off-line optimisation procedures.
York Demonstration
In York the aim was to improve one of the city's Park and Ride routes. The traffic signal timing plans generated by the MUSIC approach were designed to enhance the benefits deriving from a new bus-lane; general traffic was held back, relocating a traffic queue during the morning peak hour to a position which allowed buses to bypass the queuing traffic.
The objectives of the MUSIC project in York were to:
- Decrease travel times experienced by Park and Ride buses;
- Limit rerouting along certain sensitive alternative routes;
- Limit the increase in car travel time which might be expected; and
- Encourage greater utilisation of Park and Ride.
The implementation in York was extremely successful and all the targets were met. The MUSIC timings have been in place since early 1998 and are still in place at the time of writing (October 1999). As a result of the MUSIC measures, bus travel times decreased and bus reliability increased and during the period of the MUSIC measurements a 25% increase in peak-hour patronage was recorded on the Park and Ride route studied.
Porto Demonstration
In Porto, timing plans were created for twenty eight traffic signals in the city centre and implemented for one month. The aims of the MUSIC project in Porto were to:
- Reduce public transport travel time on selected routes;
- Reduce vehicle travel time on selected routes;
- Reduce vehicle flow along selected routes with pedestrian crossings; and
- Reduce pedestrian red-light violations.
The project was not considered a great success in Porto with only just over half of the measurable objectives being met or improved. Delays to vehicles on certain routes were reduced and pedestrian red-light violations decreased during the period.
There appeared to be a number of reasons for the limited success in Porto:
- The city network model was the least 'mature' of all the models used in the MUSIC project;
- Physical difficulties with implementing the timings meant that the traffic signal plans put in place on street were not exactly those which the MUSIC optimisation procedure had designed; and
- The objectives set were extremely ambitious, seeking benefits in all areas with no compensatory losses allowed.
Thessaloniki Demonstration
Thessaloniki was the largest demonstration site in the MUSIC project. Timing plans for one hundred and twenty nine traffic signals were designed and implemented for two months. The objectives of the Thessaloniki demonstration were to:
- Reduce travel times along key corridors in the city (especially along major public transport routes);
- Reduce delays at key junctions; and
- Limit the increase in traffic flow within the city.
The results of the Thessaloniki demonstration were very good. Eleven of the eighteen targets were met and two were improved but not met. In five cases the measurable moved in a direction contrary to that desired; however in three of these cases it had, for institutional reasons, proved impossible to change the most relevant signal timings. Traffic congestion on-street appeared to have been reduced as a result of the MUSIC signal timings. At the time of writing (October 1999) Thessaloniki is running a variant of the MUSIC signal timings.
Conclusions
The MUSIC project has demonstrated that traffic signal timings can be developed to support wide ranging transport policy objectives.
The main recommendation of the MUSIC project is that city authorities should consider signal timing plans as an important tool for traffic and transport management. The MUSIC approach offers a low-cost, transferable methodology that can be used in any town or city with an existing network model.
The results of the on-street demonstrations in MUSIC strongly suggest that drivers do change their routes in response to traffic signal timings. It is vital for tools which design and assess signal timing plans to account for this driver response, so as to accurately assess the effects of putting plans into practice on street. However, the MUSIC project also highlighted the fact that this rerouting can take a long time to achieve and it was considered that, even after more than a month, the drivers had not completed the rerouting process. A research need was identified for a model that could accurately predict such intermediate states in a network.
The current optimisation procedures were utilised to produce the traffic signal timing plans used in the demonstrations. While these plans showed benefits in modelling and, generally, in real-life, the current optimisation procedures are heuristic rather than truly correct mathematical optimisation procedures. There is therefore an urgent research need for a rigorous optimisation method that can create traffic signal timing plans to support a variety of transport goals.
Finding More Information about the MUSIC approach
In order to maximise the benefits of the MUSIC approach and to aid transfer to other cities, a hand book and a publicity leaflet have been prepared. The MUSIC Handbook describes how any town or city could adopt the MUSIC approach. Details are available from York Network Control Group or the MUSIC web site at http://gridlock.york.ac.uk/music/. For copies of the leaflet or handbook or for any other enquiries about the music project please write to:
Mike Smith,
York Network Control Group,
Department of Mathematics,
University of York,
Heslington,
York.
ENGLAND
YO10 5DD
