Abstract:
Automated rutting data is condensed by means of statistical measures, mostly averages, over larger intervals to characterise pavement sections, thereby reducing the amount of data required to be analysed. Longer pavement sections, however, do not necessarily represent homogeneous performance conditions, and valuable information is lost as a result of the aggregation process.
Using this aggregated data as an input in pavement management systems (PMS) may result in inaccurate condition prediction, maintenance requirements, and budgetary forecasts. The level of aggregation, including the selected analysis pavement section lengths, influences the extent to which this information is inaccurately forecast.
This study investigates the potential effects, as influenced by varying aggregated section lengths, of using statistical aggregation measures, namely, averages and percentiles, to analyse high-density automated rutting data, on the distribution of rut depth over a pavement section, and the respective maintenance, funding, and pavement performance requirements.
When using the mean as the statistical aggregation measure, the study indicated that the dispersion of rut depth reduces with increasing averaged section length, resulting in inaccurate condition and maintenance requirement forecasts. Introducing percentiles as the aggregation measure revealed that percentiles offer more accuracy over averages. For annual financial planning, the 50th percentile is most suitable. For technical needs planning, the 75th percentile is better suited, considering higher percentiles (90th to 99th) where high priority roads and performance requirements are a concern.