PAVEMENT MATERIALS SELECTION: A DECISION THEORY APPROACH

By

Author

Presented To

Department of Engineering

ABSTRACT
In highway pavement design, the main factors considered are the expected traffic, subgrade strength, and the constituent materials for the pavement structure. Associated with each of these factors, however, are elements of uncertainty. Ignoring these uncertainties, means that either high factors of safety are applied in design, with its cost implications, or a high risk of failure must be accepted. A model is developed, using the Bayesian approach to statistical decision making, which accounts for the uncertainties in the pavement materials selection process. The main elements of the pavement materials selection model include the alternative materials for a given pavement layer; the states of nature associated with each alternative (material strength states); an estimation of the likelihood associated with each state of nature; an assessment (through utility functions) of the consequences of each combination of alternative material and state of nature; and a systematic evaluation of all alternative materials to determine the optimal material selection. The model is illustrated by its application to a case study involving the selection of a base course material from among alternative materials for use in a flexible pavement structure. With the uncertainties regarding the possible range of material strength states being expressed as probabilities, the materials selection process, proceeds by a prior analysis with existing information, which, after the acquisition of additional information, is updated by a posterior analysis. The optimal material selection is made on the basis of maximum expected utility. The case study shows that, the magnitudes of the expected utilities depend on the magnitudes of the prior, and, or posterior probabilities associated with the respective strength states; while any change in the magnitudes of these probabilities, depend on the sample likelihoods estimated from experimental results. It is recommended, as improvements to the model,that the utility functions be further refined, such that the effects of costs on materials selection are explicitly reflected in the decision analysis; and that, the possibility of incorporating procedures for the selection of surfacing materials be invest igated. The design approach proposed, when fully developed,is expected to bring about not only monetary savings, which otherwise would not have been made due to less than optimal selection of materials; but also result in more efficient use of scarce resources, to be reflected through a satisfactory return on the investment of highway funds.