Investigation of Weigh-in-Motion Measurement Accuracy on the Basis of Steering Axle Load Spectra

Abstract

Weigh-in-motion systems are installed in pavements or on bridges to identify and reduce the number of overloaded vehicles and minimise their adverse effect on road infrastructure. Moreover, the collected traffic data are used to obtain axle load characteristics, which are very useful in road infrastructure design. Practical application of data from weigh-in-motion has become more common recently, which calls for adequate attention to data quality. This issue is addressed in the presented paper. The aim of the article is to investigate the accuracy of 77 operative weigh-in-motion stations by analysing steering axle load spectra. The proposed methodology and analysis enabled the identification of scale and source of errors that occur in measurements delivered from weigh-in-motion systems. For this purpose, selected factors were investigated, including the type of axle load sensor, air temperature and vehicle speed. The results of the analysis indicated the obvious effect of the axle load sensor type on the measurement results. It was noted that systematic error increases during winter, causing underestimation of axle loads by 5% to 10% for quartz piezoelectric and bending beam load sensors, respectively. A deterioration of system accuracy is also visible when vehicle speed decreases to 30 km/h. For 25% to 35% of cases, depending on the type of sensor, random error increases for lower speeds, while it remains at a constant level at higher speeds. The analysis also delivered a standard steering axle load distribution, which can have practical meaning in the improvement of weigh-in-motion accuracy and traffic data quality.

Keywords: axle load sensors; axle load spectra; bending beam; heavy traffic; overloaded vehicles; overweight vehicles; piezoelectric; piezoquartz; steering axle; weigh-in-motion.

Figure 1

An example of the effect of (A) systematic and (B) random components of relative error on axle load spectrum (ALS).

Figure 2

Scheme of the steps performed in the analysis to investigate the scale and sources of errors on weigh-in-motion (WIM) stations and to determine the standard steering axle load spectrum.

Figure 3

The investigated technologies of WIM axle load sensors: (A) quartz piezoelectric (piezoquartz); (B) bending plate.

Figure 4

Example of steering axle load spectrum (SALS) determined for WIM station no. 58.

Figure 5

Distributions of (A) mean values of SALS and (B) standard deviations of SALS calculated on the basis of data from WIM systems equipped with bending beam or quartz piezoelectric axle load sensors.

Figure 6

Example of the effect of bias of SALS determined for data from WIM station no. 58 and caused by: (A) differences in air temperatures; (B) differences in vehicle speed.

Figure 7

Distributions of mean values of SALS in relation to air temperature based on data from WIM systems equipped with: (A) quartz piezoelectric; (B) bending beam axle load sensors.

Figure 8

Distributions of standard deviations of SALS in relation to air temperature based on data from WIM systems equipped with: (A) quartz piezoelectric; (B) bending beam axle load sensors.

Figure 9

Distributions of means of SALS in relation to vehicle speed based on data from WIM systems equipped with: (A) quartz piezoelectric; (B) bending beam axle load sensors.

Figure 10

Distributions of standard deviations of SALS in relation to vehicle speed based on data from WIM systems equipped with: (A) quartz piezoelectric; (B) bending beam axle load sensors.