||Congestion in the greater Copenhagen area is extensive and a very large part of this takes place in urban networks. Here, the operation of traffic signals are crucial for traffic performance, distributing capacity for both private and freight traffic, public transport, cyclists and pedestrians. Therefore the necessity of setting traffic signals according to the traffic volumes on the roads is indisputable. Many traffic signals however are not up to date and work in a way that is far from desirable. This means that road users often experience unnecessary stopping due to red lights. It increases the travel time and imposes an additional cost to the road users due to extra fuel consumption. Also the immediate environment is inflicted by emissions and noise which are a direct consequence of the poor settings of traffic lights.
The aim of this project is to study the possibilities within the area of optimising co-ordinated traffic signals. Many things indicate that optimising traffic signals is cost effective and an important part of the project is to examine what impacts signal optimising can bring for the users and for the society as a whole.
Two distinct areas form the foundation of the investigations. The one area is the road stretch of Folehaven – the continuation of The Holbæk freeway towards Copenhagen, consisting of 10 traffic signals. The other area is located in Vanløse an area including the continuation of Jyllingevej towards Copenhagen and a part of Ring Road 2. Here 12 traffic signals have been investigated. These areas represent two different road structures. Folehaven is one long stretch making it obvious to investigate the possibility of setting up the best conditions for road users – a so called green wave. Vanløse is more characterised as a network containing different co-ordinations of traffic signals and the challenge is mainly to define a strategy to traffic signal optimisation in this area. These areas each feature some additional obstacles for instance various ways of dynamic control which also is to be dealt with. All together the two areas are bringing good preconditions to analyse various aspects of co-ordinated signals. Also the investigation of two areas increases the sturdiness of the results.
In both areas a detailed analysis of the traffic related circumstances has been carried out based on gathered data material. Primarily these data are based in traffic volumes and the travel pattern. Throughout the analysis especially Folehaven indicated that the co-ordination could be improved.
The study of what impacts signal optimisation will bring is performed by modelling optimised signal settings in both areas. Through micro simulation the effects of this is estimated. To perform the simulations the software program VISSIM has been chosen because the use of this program is widespread in Denmark. Also the program is well-founded through decades of research. Based on primarily traffic counting in intersections and on roads, along with gathering of GPS data from both areas, both traffic models are calibrated to represent the actual traffic conditions – especially with regard to traffic volumes and travel times.
Optimisation of co-ordinated traffic signals is a complicated process with many regards to consider in both traffic and safety measures. This call for using computer based software which to a much greater extend can pay regard to all signal settings within a co-ordination. For optimising the signal settings, primarily concerning offsets, the optimisation program TRANSYT is chosen. TRANSYT is the most applied signal optimisation program in this country. Through optimisation TRANSYT seeks to minimise a performance index composed by the number of stops and the total delay in the network considered. For this purpose a hill climb algorithm is used which, despite its simple structure, has proven highly efficient in numerous studies, Danish as well as foreign. With the latest version of TRANSYT it is possible to make the optimisation process based on simulation in the VISSIM traffic model, rather than the mesoscopic traffic model used in TRANSYT. By simulating in VISSIM the dynamic aspects of traffic can be reproduced to a much greater extend. The explicit use of VISSIM in the optimisation process underpins the necessity of
having models that highly reproduce the actual traffic in the real network making calibration of the models very important.
The signal optimisation in the two areas mostly concerns the change in the mutual displacement of the traffic signal change times (offsets). In Folehaven this resulted in some distinct gains in the three periods modelled, the morning and afternoon peak hours respectively plus two hours around midday (7.00-9.00, 11.00-13.00 and 15.00-17.00). For instance the overall number of stops in the system was reduced with 25 percent in the morning hours, the period that generally gave the best results. In this period the overall delay is similarly reduced by 28 percent. Also this brought reductions in fuel consumption of five percent at most in the three analysed periods. Considering sections of the road stretch of Folehaven the travel times in particular was brought down with more than one minute which is a noticeable change to the road users travelling on this section. Moreover the vehicles are stopping less in all periods of the day – on average avoiding more than one stop. The overall optimisation of signal settings in TRANSYT has brought a significant improvement of the co-ordination of traffic signals in all periods of the day. Especially in the peak hours stops and delay were reduced, and consistently the optimisation comparatively resulted in greatest reductions when the potential was greatest i.e. when the amount of stops and delays were at the most.
Corresponding to this theoretically based study the optimised offsets have been implemented in the actual traffic signals on the road stretch of Folehaven in a result of collaboration with the municipality of Copenhagen. Thereby the actual change in travel times could be investigated. Because severely congestion on the inner part of the section, coming from outside the traffic model, couldn’t be modelled in the morning hours the results could therefore not be imitated completely. In order to compare travel times before and after the signal optimisation based on GPS data from 3x34 vehicles, it is verified that traffic before and after the signal optimisation is directly comparative. The number of trips fit for use was found to be quite extensive and the results in Folehaven were evident in all periods of the day. The travel times on the entire section was reduced significantly due to the new offsets, however the reductions in the morning peak hours was not as extensive as shown by simulation. On the other hand the results in the other two periods corresponded well to the simulation results and were to some degree a little better.
In this analysis of the GPS data it was possible to assess to what extend VISSIM could predict the consequences of changing the signal settings. With the exception of a few sections affected by the congestion not modelled, VISSIM estimated the changes in travel times with great accuracy. On most sections the divergence between model and reality was less than five percent, often 0-3 percent. This underlines the quality and sturdiness of using VISSIM as a tool in traffic planning.
The optimisation process in Vanløse is far more complicated than the one in Folehaven. As mentioned Vanløse consists of several parts of co-ordinated traffic signals each operating with different cycle times. This is a problem in relation to the optimisation process because TRANSYT can only operate with one cycle time at a time, and thereby only optimises the signals operating with the particular cycle time. At first the aim is to examine how the signals can be optimised overall in the network. The various co-ordinations are studied by optimising them stepwise and by harmonising the cycle times, thereby reducing the number of co-ordinations in the overall network. This gives TRANSYT the opportunity to optimise the signal settings in more traffic signals at a time. The signal plans have been changed and all green times adjusted to the new cycle times.
These adjustments of the signal settings lead to studies of the interaction between the various co-ordinations. The best performance was obtained in the scenario with all signals operating with the same cycle time. In general the results improved with fewer co-ordinations in the network and thereby fewer stepwise optimisations of the various co-ordinations. TRANSYT thus derive the best signal settings when
able to adjust all traffic signals at the same time. This is in spite the fact that some of the adjusted signal settings are not optimal to some of the individual traffic signals.
In general the improvements of the co-ordinations have shown good results through the simulations. In the morning and midday periods stops and delay were reduced by around 10 percent. Regarding different sections in the network the results varied somewhat but on the main road stretches there were generally improvements and stops and delay were reduced with up to 40 percent. In the afternoon period the results were also positive but not to the same extend. The reductions are not of the same extend as for Folehaven but the difference is to been seen in the light of the preconditions at hand. Above all the co-ordination in Folehaven has not been inspected in a period of years while the co-ordinated traffic signals in Vanløse were inspected in 2006. Thereby the potential for improvement in Folehaven is greater. Moreover the signal structure in Vanløse is more complicated and occasionally congestion is great in Vanløse compared to Folehaven. This congestion is evident regardless of the signal settings and therefore matters when comparing the potential for improvement.
The potential of signal optimisation is studied in regard to the socio-economical gains. For this purpose a modified version of the Ministry of Transport’s TERESA model is used. This model calculates the socio-economy of a given road project based on fuel consumption and travel delay. The variations of traffic during the day is approximated based on the three investigated periods. In this way the reductions of fuel consumption and travel delay are estimated for an entire workday.
The socio-economical gains in both Folehaven and Vanløse are equivalent with the outputs from the optimisation. In Folehaven the payback time equals three days with the given traffic. The present value is almost 45 million DKK during a period of five years. These results speak for themselves. And the profitability of the optimisation is indisputable. Based on the GPS data from Folehaven the actual gains were assessed by comparing the changes in travel times in the model and in reality. The result was a bit lower resulting in a payback time of four days and thus emphasises the cost-effectiveness in this project. In Vanløse the optimising of traffic signals will require greater changes in the signal settings which increase the cost of investment. The time of repayment thus amount to 46 days. The present value is 21 million DKK during the five year period, which is half as much compared to Folehaven. In such the project is still highly cost-effective.
If the optimisation of traffic signals is viewed upon from an environmentally point of view it is one of the most cost-effective instruments to improve air quality in rural areas. This is primarily a consequence from the reduction of stopping and idle running. In the optimisation of Folehaven the cost per saved ton CO2 is 343 DKK. Compared to other environmental projects the effects are much greater considering the investment costs.
Finally this socio-economic analysis was taken to a greater societal perspective. By optimising every co-ordination of traffic signals within the municipality of Copenhagen the potential gains were estimated. The uncertainty of this analysis is obvious but it still gives an overall picture of the effects of signal optimisation. The basis is an average of the economical gains obtained in the two areas. This suggests a yearly gain of 200 million DKK which should be compared to the investment costs with an estimated cost of 10 million DKK. This assessment demonstrates a great socio-economic potential by optimising the co-ordinated traffic signals.