An accurate deterioration prediction model should include, as completely as possible, all the factors that affect deterioration and the effects of past maintenance. Many of the deterioration models that currently exist in the literature are simple models that include only a few explanatory variables, and most do not incorporate the effects of maintenance. The literature has reported that the performance of many of these models has been poor. Even in cases where more complex models have been formulated, researchers have reported that the estimates of important explanatory variables have counterintuitive signs. The usefulness of these models in the planning of future maintenance is therefore questionable. The reason for the unexpected signs of the parameters is that models use data collected from in-service pavements. Maintenance on in-service pavements is not carried out by highway agencies exogenously; instead, it is carried out in response to factors such as condition and traffic. This implies that the condition of a highway pavement at any point in time is determined by the point of intersection of two simultaneous processes: (a) the process of deterioration by which the pavement quality degrades and (b) the behavior of the agency performing maintenance in response to this deterioration. An appropriate model for predicting conditionover time is therefore a simultaneous equation specification that reflects both the processes described above. In this paper, a simultaneous-equation deterioration model is formulated for highway pavements and estimated from data collected in the field. On coparison with a traditional single-equation model, it is seen that the simultaneous-equation approach produces realistic parameter estimates. This paper appears in transportation research record no. 1272, Pavement management and rehabilitation 1990.
Abstract