Long-Term Pavement Performance Automated Faulting Measurement

CHAPTER 4. ANALYSIS RESULTS

In this chapter, the analysis and comparison results of the LTPP AFM, the ProVAL AFM, and the FDOT PaveSuite AFM are presented. To evaluate the LTPP AFM, six LTPP JPCP test sections were selected such that the manual distress survey dates matched or were close to the longitudinal profile survey dates. The same six LTPP test sections were also used to evaluate the ProVAL AFM.

Table 1 and table 2 show the evaluation results of the LTPP AFM. In table 1, the fourth column from the left represents the total number of transverse joints present on the 152.4-m test section. The transverse joint locations and faulting measurements were collected using manual distress surveys and were stored in LTPP PPDB table MON_DIS_JPCC_FAULT. Profile runs in ERD format from the MON_PROFILE_MASTER table for the LTPP test section that passed the ProQual quality check were used in the analysis. The column "TP" stands for "true positive," whereas column "FP" stands for "false positive." True positive means the true transverse joint was detected, whereas false positive means something other than the true transverse joint, such as a crack or pothole, was detected.

According to the data in table 1, for Georgia test section 133019 (STATE_CODE of 13 and SHRP_ID of 3019), the LTPP AFM detected all of the 25 true transverse joints with 0 false positives using the 5 ERD files collected from that section. For New York test section 364018, the true positive JDR was 95 percent, i.e., the LTPP AFM missed one true positive joint from ERD files 3 and 4. The JDR from the LTPP AFM for the six selected sections ranged from 95 to 100 percent.

Similarly, the same six LTPP test sections and five ERD files were used to evaluate the ProVAL AFM. According to the data in table 2, New York test section 364018 had a JDR of 67.5 percent, whereas the LTPP AFM JDR was 95 percent (table 1). The total number of false positives detected for the New York test section using the ProVAL AFM for the five profile ERD files (i.e., ERD1 to ERD5) was 33 (5 + 7 + 10 + 6 + 5). The total false positives detected using the LTPP AFM (table 1) for the same test section (364018) for five profile files was four. The JDR from ProVAL AFM for the six selected sections ranged from 58 to 99.4 percent.

Table 1 . LTPP AFM joint detection results using LTPP profiler data.

ERD = Engineering Research Division
FP = False Positive
JDR = Joint Detection Rate
SHRP = Strategic Highway Research Program
TP = True Positive

Table 2 . ProVAL AFM joint detection results using LTPP profiler data.

ERD = Engineering Research Division
FP = False Positive
JDR = Joint Detection Rate
SHRP = Strategic Highway Research Program
TP = True Positive

Table 3 shows the joint detection comparison results between the LTPP AFM and the FDOT PaveSuite AFM methods, using FDOT HSIP data collected on State Road 24 in Waldo, FL. FDOT's 609.6-m test section included a 152.4-m lead-in and lead-out, and a 304.8-m effective test length spanning 50 slab joints. The slabs were typically 6.1 m long by 3.66 m wide with a relatively smooth surface finish. Both AFM methods detected 48 true positives with a JDR of 96 percent. However, according to the data in table 3, the FDOT PaveSuite AFM detected eight false positives, whereas the LTPP AFM detected zero false positives.

Table 3 . FDOT and LTPP AFM joint detection results using FDOT HSIP.

AFM = Automated faulting measurement
FDOT = Florida Department of Transportation
HSIP = High-speed intertial profiler
JDR = Joint detection rate
LTPP = Long-Term Pavement Performance

Table 4 through table 7 show the analysis results for AFMs computed using both the ProVAL and the LTPP AFM methods. As discussed in the methodology, the LTPP AFM uses two methods, the slope method (in-house method) and the AASHTO R 36-12 Method-A, to compute faulting measurements. The ProVAL AFM uses AASHTO R 36-12 Method-A. Table 4 shows the average faulting for the entire test section (152.4 m) for all five profile runs computed using the LTPP AFM slope method. The average of absolute differences in faulting estimated by the LTPP AFM and faulting measured by the manual GFM on the six LTPP JPCP test sections are shown in the "Average section |Bias|" column. Test sections 133019, 370201, and 493011 have less than 1-mm bias/error as required by the AASHTO R 36-04 standard. However, test sections 313018, 364018, and 421606 have a bias/error greater than 1 mm.

Table 4 . LTPP AFM faulting results (slope method) using LTPP profiler data.

GFM = Georgia Faultmeter
SHRP = Strategic Highway Research Program

The average faultings for the entire test section (152.4 m) for all five runs computed using the ProVAL AFM are shown in table 5. Also, the average of the absolute difference in faulting estimated by the ProVAL AFM and faulting measured by the manual GFM on the six LTPP JPCP test sections are shown in the "Average section |Bias|" column of table 5. Sections 133019 and 313018 have less than 1-mm bias/error as required by the AASHTO R 36‑04 standard. However, test sections 364018, 370201, 421606, and 493011 have a bias/error greater than 1 mm.

Table 5 . ProVAL AFM faulting results using LTPP profiler data.

GFM = Georgia Faultmeter
SHRP = Strategic Highway Research Program

Table 6 shows the LTPP AFM joint faulting using AASHTO R 36-12 Method-A. Only four test sections were analyzed using AASHTO R 36-12 Method-A. Test sections 133019 and 370201 have less than a 2-mm bias/error. However, sections 364018 and 421606 have a bias/error of 13.81 and 3.68 mm, respectively.

Table 6 . LTPP AFM faulting results (AASHTO Method-A) using LTPP profiler data.

GFM = Georgia Faultmeter
SHRP = Strategic Highway Research Program

Comparison results of the FDOT PaveSuite AFM and the LTPP AFM using the FDOT HSIP data are shown in table 7. The average faulting measured from the FDOT PaveSuite AFM was 1.69 mm, while the LTPP AFM, using the slope method for the same data, was 1.62 mm. (For FDOT HSIP data, only the in-house method, i.e., the slope method, was used to compute joint faulting; AASHTO R 36-12 Method-A was not used.) Also, the average of absolute differences in faulting estimated by the FDOT PaveSuite AFM and the LTPP AFM slope method with the measured manual GFM were 1.05 and 1.14 mm, respectively. Both AFM results using the FDOT data were similar.

Table 7 . Joint faulting results using FDOT HSIP data.

AFM = Automated faulting measurement
FDOT = Florida Department of Transportation
GFM = Georgia Faultmeter
LTPP = Long-Term Pavement Performance