A-Priori Travel Time Predictor for Long Term Roadworks on Motorways

Abstract

Road users have never had as much travel information as is available today. However the extent of congestion on major roads has also never been as critical as it is now. For this reason road authorities, including Rijkswaterstaat1, aim to inform road users as best they can in an effort to allow the road user to make a more educated decision on travel and to increase the confidence they have in travel times. In a bid to improve traffic flow on motorways, many roadworks are carried out yearly with a large number planned for the coming years. This contributes to congestion and delays in the short term however and leads to a greater uncertainty in travel times. Many techniques and models already exist to predict travel times under ‘irregular’ traffic conditions. For situations where roadworks are due to be carried out in the future however, no models or methods explicitly exist which allow travel times to be predicted in advance. It is this problem that this research project attempts to tackle. The main objective for this research is to develop a methodology incorporated in a model, which is capable of predicting travel times on motorway corridors for situations during roadworks that are to be carried out in the future. To achieve this objective the research question is posed: How can a-priori travel times be predicted on motorway corridors for situations during roadworks, prior to the commencement of the roadworks? The objective is achieved by firstly consulting external research on the topics of travel time estimation with models and the influence of roadworks on travel times. Using the acquired knowledge a modelling approach is developed which makes use of the basic principles of traffic flow based on the conservation of vehicles and first order traffic flow theory. The developed model makes use of traffic flow profiles and capacity profiles, which are processed by an LWR-model using a Godunov scheme. Traffic is numerically fed through the model and where it exceeds capacity, congestion occurs and propagates backwards in space according to first order traffic flow theory and in keeping with the general characteristics of real traffic flow. From the modelled data, speeds are derived for each iterated section. This allows for travel times per section and total travel times along a certain trajectory starting at a specific time of day to be calculated. These travel times form the prediction for the corresponding motorway corridor. The effects of roadworks are incorporated in the model through a reduction of the road capacity in the capacity profile. This is performed by applying a capacity reduction factor to the available capacity. This reduction factor is determined using characteristics of the roadworks which correspond to certain reduction values taken from extensive research preformed externally. The traffic flow profile is also adjusted for the effects of mobility management, which is commonly applied during roadworks in the Netherlands. Mobility management is an organised attempt to reduce the level of traffic demand on routes where road capacity is not expected to be able to cope with traffic demand, such as during roadworks. A mobility management factor is therefore applied to the traffic demand profile to reduce demand as a consequence of this. The model is evaluated using a roadworks study case on the A12 between The Hague and Gouda. The results of the model, in which a base capacity2 of 2100 veh/hr/ln is applied, show a good likeness to the recorded travel times during the performed roadworks. An absolute relative error of less than 5% is recorded for the travel times during the main peak periods. These results are produced with the application of a mobility management factor of 6-7%, which corresponds to the expected values for this specific case. The performance requirement for the error of travel times during the entire day is also achieved in the case study. The research shows that predicting travel times for future roadworks is possible and moreover can be performed in a relatively accurate fashion without the necessity of an overcomplicated model. Producing traffic flow demands is achievable, however estimating the extent of mobility management and the indirect reduction of traffic demand is more complicated. Road capacity during roadworks is affected and estimates are made of the reduced workzone capacity. The capacities found show a good likeness to recorded data, however small adjustments in the capacity reduction have the potential for large travel times variations. For this reason the application of confidence bandwidths, as applied, is valuable. Further difficulties in determining capacities stem from the inability to produce operational capacity estimations where no congestion occurs. The application of a base capacity solves this, however the applied value cannot be generically validated with great ease. The application of the model is most suited to implementation for road user information through a website or incorporated in a route planner. The use of the model in roadwork planning is also possible, but will require alterations to model. The case study results are encouraging, however the model requires further validation over a wider range of roadworks as varying locations and roadwork characteristics may lead to differing results. Further research is recommended into a simple capacity reduction method for roadworks. Research on the effect of mobility management and an effective method to estimate the effect of it is also recommended. The implementation of these as well a generic manner of determining a base capacity in the model are further recommended as possible adjustments to improve the model.