FWD reliability

peak deflections

The box plot of the normalized center deflections for five out of the seven devices is presented. The average normalized center deflections at 24 kips (105 kN) are shown to range from 2.9 to 3.7 mils (75 to 95 microns). The plot also shows that device B produces the lowest deflections, while device A1 produces the highest deflections. It is also important to note that device B has the shortest load pulse duration and devices A (A1 and A2) and C have the longest durations, which could potentially explain some of the observed differences.

An ANOVA is conducted to determine if the differences in normalized center deflections among the devices are significant. A p-value, nearly equal to zero, was obtained indicating that there are statistically significant differences among the devices, confirming the variations observed in the box plot.

load and deflection time histories

The load and deflection time histories are plotted for the five devices. A review of the load time histories shows that device B exhibits the shortest load pulse, whereas devices A (A1 and A2) and C display the longest load pulses. Additionally, the load pulses generated by devices A (A1 and A2) and C closely resemble a haversine shape. Furthermore, device D shows double peaks, while device B displays noise in the measured load throughout the entire duration. It is important to note that the data from device C is filtered, whereas the data from device D is not. It remains unclear whether the data from the remaining devices are filtered or not.

A review of the deflection time histories reveals that deflections recorded by device B exhibit the shortest pulse, while those recorded by devices A (A1 and A2) and C display the longest pulses. This disparity is attributed to the varying durations of load pulses applied by each device. In addition, the peak deflection measured using device B is the lowest and those measured using devices A (A1 and A2) and C are the highest, especially for device A1. Finally, the deflections measured using devices A1 and A2 have relatively high negative deflections, when compared to the other devices, and remain negative after the 40 msec mark.

Numerous notable and minor issues related to noise, double peaks, and excessive negative deflections are identified, prompting efforts to implement improvements aimed at enhancing decades-old technologies. Among these issues, the excessive negative deflections emerge as particularly critical, potentially impacting the reliability of backcalculation results.

nonlinear dynamic backcalculation

Nonlinear dynamic backcalculation is performed using data from device D (ERDC Dynatest HWD) to determine the rigid pavement layer properties. The backcalculated layer properties are then used to calculate the deflection time histories for device B with the shortest load pulse and devices A1 and C with the longest load pulses, in addition to device D. The purpose of this exercise is to quantify the effect of the shapes and durations of the load pulses on the surface deflections and try to explain the variations between the various devices that were observed earlier.

A review of the calculated deflections in orange reveals an increase in the peak deflections with longer load pulse durations. In addition, the measured and calculated deflections match for device D, which is expected since the backcalculation utilized data from the same device. For device B, the calculated peak deflection is slightly higher than the measured center deflection. For device C, the calculated deflections closely align with the measured deflections in both shapes and magnitudes, while noting that the load pulse for this device is the longest. Lastly, notable differences are evident between the calculated and measured deflections for device A1, apparent in both peak and negative deflections.

need for next-gen devices and calibration procedures 

Drawing from the initial analysis of the roundup data, it is evident that there are opportunities for targeted enhancements to each device to bolster the reliability of data collection. These improvements encompass, among others:

• Increasing the reliability in measured deflections over the entire time duration, including the peak deflections and negative deflections.
• Producing smoother load pulses characterized by sinusoidal or haversine shapes with minimal or no noise.
• Recording smoother deflection pulses with minimal or no noise.
• Recording load and deflection data over longer time durations, where needed. 

In addition to the equipment-specific improvements, it is advisable to enhance the calibration procedures for both load cells and deflection sensors across the entire duration of data collection (time history), rather than focusing solely on peak responses. Furthermore, it is recommended that F/HWD manufacturers provide the raw data from geophones (velocity transducers) and accelerometers for transparency and public access, facilitating verification, future re-analysis, and advanced modeling.

roundup report

A comprehensive report on the initial evaluation of the rigid pavement section from the FAA roundup can be provided upon request via email.