Knowledge CentreTechnical Resources SearchConference PapersCASE STUDIES OF REGRESSION AND MARKOV CHAIN MODELS

CASE STUDIES OF REGRESSION AND MARKOV CHAIN MODELS

Abstract

There are two main types of models available for modelling pavement performance behaviour. They are regression type models and models based on Markov Chains. Regression type models are the easiest to use and allow for the analysis of various factors. The advantages of Markov Models are that they can be calculated with a minimum of two years of data unlike regression models which require data over a period of years to predict trends. In addition, Markov Models permit the use of expert opinion or a Bayesian approach in the development of performance curves. This paper looks at three case studies of performance models for pavements and Bituminous Surface Treatments (BSTs) in a remote area where there were no available performance models. The initial models were based on two years of data from 1990. These models are compared with the actual performance of the pavements and BSTs. In addition, the Markov models have been recalculated with 6, 9, 13, 19 and 23 years of data and the models are compared to the original model and actual performance over time. The analysis of the comparative data indicates that in most instances the Markov models underestimated pavement performance. 

Conference Paper Details

Session title:
PAVEMENT PERFORMANCE CASE STUDIES
Author(s):
Chris Uchwat
Donaldson MacLeod
Topics:
Pavements
Year:
2012