Estimating Travel Time Distribution for Reliability Analysis

Abstract

With increasing mobility, users are more sensitive to the variability of travel time. Therefore, travel time reliability has become an important performance measure for transportation facilities. A variety of performance measures have been proposed to quantify reliability. Many of these indicators are based on percentiles of travel time. A good knowledge of the distribution of travel time is needed to properly estimate these values. If there were no congestion, travel time would be estimated by a Normal distribution. Congestion, however, distorts the distribution and another statistical distribution is needed to model travel time reliability. Different distributions are presented in the literature as the best way to model the travel time distribution. The impact of the use of distributions of individual or aggregated travel times on a period is discussed. This paper presents a comparison of six statistical distributions used to model travel time thus its reliability. These six distributions are the Lognormal, Gamma, Burr, Weibull, Normal Mixture (a mixture of two Normal distributions) and Gamma Mixture (a mixture of two Gamma distributions). Tests were conducted to identify the parameters of these different statistical distributions which fit best with the travel time empirical distribution. Tests are conducted on the basis of real time data collected on a three kilometer weaving section of the A4-A86 French motorway. Analyses are conducted with regards to the percentiles of travel time which are the most used for the travel time reliability assessment.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 15 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.