Evaluating the Service Life of Thermoplastic Pavement Markings: Stochastic Approach

Abstract

The study applied the Markov chain (MC) model that uses a transition matrix to transmit the probability of monitored pavement markings being in one service life state then changing into another service life state over a time interval. The service life prediction by MC models were then compared with those from linear models, testing if there were any clear advantages of using one model over the other in terms of predicting longevity of the marking retroreflectivity. The retroreflectivity data were collected by monitoring the coefficient of dry retroreflective luminance for 2 years using a handheld retroreflectometer. Using the MC model, the study found that the pavement marking retroreflectivity (PMR) degradation follows an exponential curve trend whereby the degradation rates decrease as the time increases. Significant differences were found in the deterioration of the markings based on the colors (white or yellow) and line type (center, lane line, or edge line). White thermoplastic edge lines on two-lane roadways were found to have a better performance (low deterioration rates) compared with the same lines on four-lane highways. Based on the transition probability matrix (TPM), it was observed that retroreflectivity is in an excellent or good state for a short period of time (54% probability) but is in a fair or poor state for a longer time (92% probability), suggesting the trend has a higher degradation rate at the beginning and a lower rate near the failure state. Keeping the minimum failure states at 15 and 1 mcd/m2/lx for white and yellow markings, respectively, the service life of white markings was found to be approximately 4 years (49.5 months) and it was found to be about 2.4 years (29 months) for yellow markings. The MC model findings were compared with those obtained through linear regression, which showed that white thermoplastic pavement markings take approximately 3.5 years (42 months) to deteriorate to failure state level, while yellow thermoplastics take about 2.1 years (25 months). The study concluded that there is a clear difference between the prediction using MC models compared with linear models, with MC models being more cost effective in terms of maintenance and replacement scheduling due to a longer life prediction.