IMPACT OF ERROR IN PAVEMENT CONDITION DATA ON THE OUTPUT OF NETWORK3 LEVEL PAVEMENT MANAGEMENT SYSTEMS

摘要

The quality of pavement condition data is important not only in assessing the current condition of thenetwork but also in the prediction of future condition and the planning of future maintenance andrehabilitation (M&R) activities. This paper provides a quantitative assessment of the impact of errormagnitude and type (systematic and random) in pavement condition data on the accuracy of PMS outputs(i.e. forecasted needed budget and M&R activities in a multi-year planning period). The process developedto simulate the propagation of pavement condition errors to the output of PMS consists of fivecomponents: condition data generation, error perturbation, condition prediction, M&R prioritization, andoutput generation. This process was applied to the 2011 pavement condition dataset of the Bryan district,Texas. In 2011, this roadway network consisted of approximately 3,200 centerline miles. The study resultsshow that both systematic and random errors can highly distort some PMS output parameters even in errorranges that may be considered acceptable in practice. For example, the case study shows that, with 95%confidence, a ±10 standard error in a 0-100 condition index can result in 2-5.8% error in estimatedportions of the network needing maintenance, rehabilitation, or “do nothing.” Similarly, a constantadditive systematic error of -2 in a 0-100 condition index can result in 2-3% error in estimated portions ofthe network needing maintenance, rehabilitation, or “do nothing.” These effects tend to persist throughoutthe planning period. These findings can help highway agencies to optimize pavement condition datacollection processes by focusing on error levels and types that cause the greatest impact on PMS output.