2. Experimental Details

3.1.1 BRAKE FORCES –INDIVIDUAL BRAKE FORCE EVALUATION -ACCURACY

On the 3-S2 vehicle, brake torque data were collected during Tests 1 and 3 by a torque wheel installed on wheel number 5. The BFs achieved during the test were calculated by dividing the measured torque by the tire radius.

Over the duration of a PBBT test, the BF at a wheel varies with time. The BF value reported by the PBBT depends not only on the proper calibration and accuracy of the PBBT force sensor, but also on the processing of the data collected by the sensor as a function of time. Since the details of data processing used by each PBBT vendor were not known to the report authors, three distinct methods were used to calculate reference BFs from the torque wheel data. The best match of the three methods was used in the accuracy analysis. In summary, Method 1 reported the maximum BF measured at any time during the test. Method 2 computed and reported the average of the data falling within a given percentage of the maximum. Method 3 reported the BF at the time of test termination. Details are included in Appendix B.

The results for each replicate test of the laden and unladen conditions are tabulated in Appendix B (Table B2). Also, the brake forces measured by the torque wheel are plotted as a function of time in parallel with BFs (where available) measured by PBBTs (Appendix D). These plots are referred to as “time history” plots.

Figure 3 illustrates the percent deviation of PBBT reported BF from the BF computed with the torque wheel data (using the best match from the three methods) for the laden and unladen conditions, respectively. These data are the average from three repeat tests on a single wheel. The proposed FMCSA functional specifications for PBBTs call for

±2.5 percent accuracy of BFs for the PBBTs. The total accuracy range incorporates the torque wheel transducer accuracy, the tire radius measurement accuracy, and the error induced on the radius measurement by the varying contact geometries (dependent on the PBBT type), as detailed in Appendix B. The total acceptable range varies from ± 4.3 to ± 6.9 percent.

As can be seen in Figure 3, all PBBTs except the flat plates had less than a 3 percent deviation from the torque wheel results, and therefore their accuracy was considered acceptable without further consideration.

Figure 3. Average deviation of PBBT-reported BF from BF computed from the torque wheel data from wheel 5 of the 3-S2 in the laden and unladen conditions. The algorithm to compute BF from the torque wheel data which gave best fit was used to plot these data. The solid horizontal lines indicate the acceptable ranges for accuracy as listed in Table 5 for indirect standards.

For the FPs, low BFs were reported in the laden condition and high BFs were reported in the unladen condition. It should be noted that low reported brake forces do not necessarily lead to a safety concern. The deviations in the FP data must result from the effects of dynamic loading, data manipulation and/or algorithm for reporting BFs. Since the algorithm used by Hunter was unknown, but their FP demonstrated good prediction of decelEQ (Section 3.1.3), the significance of the 10 percent deviation from the reference BF values is not considered to be a safety issue or of critical significance. No time history data was available from the HEKA FP unit. As such, since deviations up to 30% were observed, this unit would not be considered acceptable for use in enforcement and further evaluation is warranted after appropriate modifications are made by the vendor.

The shape of the time history plots reported by the PBBT vendors appeared to match the data obtained from the torque wheel (Appendix D). The slight variations between the PBBT-reported brake forces and those calculated from the torque wheel data must result from differing algorithms. As such, it is recommended that for each type of PBBT, a common procedure be developed, adopted and documented. The algorithm for reporting the BF should include filtering to avoid any problems resulting from anomalous spikes. In doing so, assuming the unit is correctly calibrated, the reported BF should be PBBT-vendor independent.

3.1.2 BRAKE FORCES –OVERALL VEHICLE BRAKE FORCE EVALUATION -APPLICABILITY

In this section, we examine the applicability of the PBBT-reported BFs through comparison with the total BF produced by the vehicle during a 32.2 km/hr (20 mph) on-road stop. For the weakly braked vehicles, it was assumed that no wheels were skidding¹³ during the stops and thus that the maximum available brake forces were transmitted to the ground during the stop. As such, the total brake forces were computed using the equation:

F= Ma, where F is the overall vehicle brake force, M is the vehicle mass (in this case, GVW was used), and a is the average deceleration over the course of the stop. The GVW was measured with certified scales prior to the test and the average deceleration was computed using a linear regression of the slope of the velocity versus time data from the 5^th wheel data. (See Appendix B for details.) These values were considered the reference for applicability. The BF_TOT for the PBBT was simply the sum of the individual BFs measured on an axle-by-axle basis.

Figure 4 shows the PBBT-reported BF_TOT for each of the replicate tests for the weakly braked vehicles. The total vehicle BF deduced from the 32.2 km/hr (20 mph) stops is shown, along with the acceptable range of applicability (Table 5).

There may have been some skidding of the most strongly braked wheel (number 5 on the 3-S2 and number 2 on the 2-axle), but this could not be confirmed. Individual wheel speed data were not available from the tests.

Figure 4. BFTOT for weakly braked vehicles (Tests 1 -4). Data for replicate 1, replicate 2 and replicate 3 are plotted. The upper and lower dashed lines in the plots show the acceptable range of applicability (± 7.5 %, as listed in Table 5). The middle line represents the reference BF computed from the reference GVW (measured with certified scales) and the average deceleration from 32.2 km/hr (20 mph) on-road stops. If PBBT transducers have acceptable accuracy, PBBT or procedural modifications will be required to account for deviations beyond the acceptable range of applicability.

17

The breakaway torque brake tester (BTT) had acceptable applicability in reporting _BFTOT for weakly braked vehicles for all tests except Test 4, the empty 2-axle, in which two of the three measurements were higher than those computed from on-road stop. As shown in Appendix Table C4, the measured BFs for the strongly braked wheels (numbers 2 and 3) were high. These wheels may have been locked during the road stop, thus limiting the ratio of BF to WL to the road COF. In contrast, the BTT does not have a COF limit since the wheel can not slip in the grips. Torque wheel or wheel speed data were not available to confirm whether or not wheel lock-up occurred. A procedural modification, in which the maximum BF/WL ratio for strong brakes is limited to an assumed maximum road COF value, may have to be invoked in some cases for use in enforcement, so that the BTT does not report BFs higher than those which can be achieved on the road.

The flat plate brake testers (FPs) were split in their applicability in reporting BF_TOT for weakly braked vehicles. The applicability of the Hunter FP was acceptable in all cases. The HEKA exhibited erratic behavior, with only one repeat from each test near the acceptable range. The other repeats showed wide scatter, mostly due to low reported BF_TOT. Since the deviations were not systematic, it was not possible to isolate the cause, nor make recommendations for correction. It was most likely due to the handling of the dynamic data and the algorithms used to compute BFs.

Four of the five roller dynamometers (RDs) showed either acceptable applicability for reporting BF_TOT, or slightly higher values than those measured during road stops. This was clearly seen in Test 3 (unladen 3-S2). Since none of the roller surfaces had a COF higher than that expected for the road, the discrepancy was likely due to either geometric effects from the tire/roller contact patch, or to low speed of the rollers (<2 km/hr or <1.2 mph, for portable RDs) compared to 32.2 km/hr (20 mph) for the vehicle stops. The brake force generated can be higher at low speeds. The development of a scaling factor to account for the speed or geometry dependence may be required for use in enforcement. In contrast, the VIS RD showed somewhat lower BF_TOT than the other RDs. Since the individual torque wheel calibration check did not indicate this systematic difference (Fig. 3), it is suspected that a possible early test termination caused by the stronger brake, or a lower, and thus limiting, roller COF may have been the cause. Meeting the functional performance specifications and use of common test termination and data reduction procedures should adequately address these issues in the future.

3.1.3 WHEEL LOADS –INDIVIDUAL WHEEL LOAD EVALUATION -ACCURACY

As shown in Figure 5, acceptable accuracy of the wheel load measurements was observed for all PBBTs for which data were available. Data are included in Appendices B and C. The calibration was performed for the HEKA FP, but the data were not provided for publication. However, it is recollected that the weight calibration was acceptable. Weight calibration of the BTT was not performed due to minor damage to the hydraulic system as the PBBT was moved to get access to the concrete blocks. Similarly, the concrete blocks could not be transported to the off-site RAI in-ground RD, so an electronic shunt-calibration was performed instead, and results were accurate within 2.5%. Acceptable calibrations and documentation of the ability to meet the functional specifications for weighing accuracy will be required as part of compliance testing for use of all PBBTs for enforcement.