Resumen
The improvement of travel time reliability is valuable to travellers, since it helps them to better organise their activities or synchronize their schedules with other peoples'. Reliability benefits can be in the same order of magnitude as the travel time savings. Since 2014, reliability benefits have been included as non-mandatory in French appraisal guidelines. But although several valuation methods for reliability are available, various applicability and consistency issues remain and a comprehensive, consistent among modes and simple feasible method is needed. Thus the aim of this paper is to provide a comparison of methods with a practical perspective for a short term implementation, illustrated by an application on dedicated bus lane projects. This comparison will be derived from existing valuation methods and values from international literature, and will consider both the valuation method and the travel time reliability forecasting method in order to provide a comprehensive method. This paper analyses the empirical robustness and consistency of the reliability benefits resulting from various methods for several modifications of the travel time distributions with a practical perspective. Two categories of methods for the valuation of travel time reliability, mean-dispersion models and mean-lateness models, are tested on simplified case studies of dedicated bus lane projects: mean-dispersion methods and mean-lateness methods. Then the forecast of reliability indicators for such projects is discussed. Statistical models are tested to estimate 90th percentile and finally some tests are run to analyse the sensitivity of reliability benefits to the perimeter over which they are calculated (link or itinerary).