Traffic input module for mechanistic-empirical pavement design with weigh-in-motion data

Abstract

• Weigh-in-motion data provides large quantity of traffic information for pavement design. • Reliability of pavement design highly depends on the quality of truck traffic data. • The weigh-in-motion recorded traffic data may not be sufficiently accurate for satisfactory pavement design. • The accuracy of weigh-in-motion traffic data can be improved through machine learning algorithms. The success of the mechanistic-empirical pavement design guide implementation depends largely on a high level of accuracy associated with the information supplied as design inputs. Truck axle load spectra play a critical role in all aspects of the pavement structure design. Inaccurate traffic information will yield an incorrect estimate of pavement thickness, which can either make the pavement fail prematurely in the case of under-designed thickness or increase construction cost in the case of over-designed thickness. The primary objective of this study was to create an accurate traffic design input module, and thus to improve the quality of pavement designs. The traffic input module was created with the most recent data to better reflect the axle load spectra for pavement design. The unclassified vehicles by weigh-in-motion devices were analyzed and a neural-network-model-based classification method was utilized to determine the appropriate allocations of unclassified vehicles to truck classes. The updated truck traffic information includes average annual daily truck traffic, truck volume monthly adjustment factors, truck volume lane distribution factors, truck volume directional distribution factors, truck volume class distributions, traffic volume hourly distribution factors, distributions of for single-axle, tandem-axle, tridem-axle, and quad-axle loads, average axle weight, average axle spacing, and average number of axle types.