Task Order 5328
Traffic Operations Research

A Low-Cost Wireless Mems System For Measuring Dynamic Pavement Loads

Pravin Varaiya
Department of Electrical Engineering and Computer Sciences
University of California, Berkeley

Summary

It is proposed to design, build, and conduct a 'a proof of concept' test of an experimental WIM system based on a radically different approach than current WIM systems. Figure 1 shows how the proposed system might be deployed at a freeway location or on a bridge.

Figure 1 Deployment of proposed WIM system on a multi-lane freeway or bridge location. The sensor nodes are only 3" in diameter; the access point is a 5" cube. Data from sensors nodes are sent to the access point via radio. The sensor nodes and access points are drawn at an exaggerated scale relative to lane width.

The proposed system must have two key properties:

1. Its single- or double-axis MeMS accelerometers directly measures the vibration (acceleration) of the pavement where the accelerometer is located. (In the scenario of figure 1, there are 12 locations.) The vibration is related to the instantaneous, dynamic load on the pavement by the passage of vehicles in the vicinity of the accelerometer.

2. The measurements can be processed to filter out noise and to extract important features of the dynamic load.

The goal of the proposed research is to determine if we can design and build a system with these properties.

Methodology

We explain our approach by comparing it with current practice. Current WIM systems are based on three types of sensors: bending plate, piezoelectric, and single load cells. All three sensors must be located in the path of the vehicle. They measure the vertical forces applied by axles to sensors over which the vehicle travels. Two sets of calculations must be performed on the WIM measurements to determine pavement loading under dynamic traffic conditions:

1. The measurements are used first to estimate the static weight of the axle. Accurate estimates of the axle load require continuing calibration of the WIM station and a controlled environment. (1)

2. The axle load estimates are then used to obtain the actual dynamic load that the pavement experiences. Calculations of the dynamic load depend on complex vehicle-pavement interaction models, which are never calibrated. Typically, these models are used to simulate dynamic loads.

The proposed approach differs both in the sensing modality and the computation of the dynamic pavement loads. Tiny MeMS accelerometers are placed directly on the pavement to measure the vibration or acceleration of the pavement (where the sensor is located) rather than to measure the force of the axle load on the sensor as in current WIM stations. Second, the measurements are processed in accordance with the following statistical relationship:

x = F(w, v, d, p),

in which
x = peak-to-peak variation of the sensor acceleration measurement
w = axle weight
v = speed
d = distance between wheel and sensor
p = parameter vector that characterizes the vehicle-pavement interaction
F = functional relationship that must be estimated

(1)Long Term Pavement Performance Program's Protocol for Calibrating Traffic Data Collection Equipment (1998) contains 16 pages of recommendations for checking equipment calibration and quality control steps to be taken in the field and in subsequent data processing.