Weigh-in-Motion (WIM) systems provide detailed traffic information that includes traffic volume, speed of each vehicle, axle spacing, individual axle and gross vehicle weights, vehicle classification, Equivalent Single Axle Load (ESAL), estimated freight load data, and more. Recently, WIM systems began to integrate digital pictures of vehicles as part of the data, providing visual information on the vehicle, which would lead to even more usages and applications. Such detailed traffic information would provide benefits for making better decisions on transportation systems, but one caveat is that the cost of installation and maintenance of a WIM system is very high. Therefore, the Minnesota Department of Transportation (MnDOT) has been looking into ways to reduce the overall WIM cost, and this research project is one of those efforts. This project consists of two parts. The first part is the development of a low-cost WIM system in which the cost is reduced as much as possible with minimal loss of data quality. The second part is evaluation of the low-cost WIM system by installing one on a highway and then comparing the collected data with another type of WIM system. In building a low-cost system, the cost savings were done in three areas. First, low-cost piezoelectric polymer film sensors (will be simply called BL sensors, which is a short name for the product name, Roadtrax BL Piezoelectric Axle Sensor) were used instead of the expensive crystalline-quartz piezoelectric sensors (will be simply called Lineas sensors, which is also a short name for the product name, Lineas Quartz Sensor). Second, a customized PC-based controller was built based on off-the-shelf parts and used. This was much cheaper than proprietary controllers available on the current market. Third, loop detectors were not used and were replaced with an algorithm. Thus, the cost of installation and maintenance of loop detectors was saved. With these multi-facet efforts, the total material cost of a four-lane highway WIM system excluding installation was $8,885. The same four-lane system using the conventional Lineas sensors and a commercial controller was estimated at $133,072. Consequently, the research team was able to cut the WIM system cost by 93%. To evaluate the low-cost WIM system, the site chosen by MnDOT was the Automatic Traffic Recorder (ATR) #213 site (a retired ATR site) which is located at mile point (MP) 18.1 on Minnesota Highway 61 (MN 61). This site has an instrumentation cabinet, power, a phone line for modem communication, and other wiring infrastructure for convenience, reducing the overall project cost. BL sensors were installed on all four lanes, and an in-house custom WIM controller (a PC-based controller) was installed in the instrumentation cabinet. A major advantage of selecting this site was that the WIM #30 site is available for comparison at about 1.8 miles south from this site. The WIM #30 site uses Lineas WIM sensors and is equipped with a same type of a PC-based, in-house built controller. Data was collected for over 10 months and analyzed. Since one of the issues was how the known thermal sensitivity of BL sensors affects accuracy of measurements, both sites collected pavement temperature data and recorded it as part of each vehicle record. To collect WIM data under a wide range of temperature conditions, data were collected from the coldest months of winter to the warmest months of summer. The analysis was focused on the original goal of finding ways of increasing the accuracy of the low-cost WIM system through calibration of weights based on temperature. Accuracy of any WIM system depends on three factors: the accuracy of axle sensors themselves, quality of installation, and calibration accuracy. It was assumed that quality of installation and weight calibration based on a known vehicle was the same, although there are some differences in installation methods that may affect accuracy. All weight analyses were done using class-9 steer axles, since they are known to have a tight predictable distribution of weights. Data showed the following results. Root Mean Square Error (RMSE) of daily volumes computed over 10 months (from 12/23/2014 to 10/31/2015) was 94 vehicles per day, which is about 1.47% of the WIM #30 average volume. Given that there are three exits between the two WIM sites of the highway, these daily volume differences are considered minimal, implying that both BL and Lineas sensors made no differences in volume counts. According to class-9 steer axle data, weights measured by BL sensors changed from an average of 5 kips to an average of 20.7 kips when pavement temperature was increased from —22.50F to 97.50F. This is a 314% weight increase over the temperature increase of 1200F, which clearly indicates that the errors are too high and weights must be calibrated based on temperature. On the other hand, Lineas sensors increased only a small amount over the same temperature change, i.e., it only increased from 9.0 to 11.2 kips on average, which is a 24% increase. This result clearly indicates that BL sensor data will not be useful unless weights are calibrated based on temperature, while Lineas sensors maintain reasonably stable weights over the same temperature range. After temperature based calibration, the average of class-9 steer axle weights measured by both BL and Lineas sensors were approximately the same, but BL sensors produced a 46% higher standard deviation. In summary, a low-cost WIM system was successfully built and installed by cutting cost by 93%. The data analysis showed that the average of the weight measurements by the low-cost WIM system could be as accurate as the conventional WIM system built using Lineas sensors, as long as the weights are calibrated based on pavement temperature. However, standard deviation of the low-cost WIM system was much higher than that of the Lineas sensor based system. (Author/publisher)
Samenvatting