Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
8Guideline Threshold Validation As summarized in Table 1, 64 WMA mixtures from 44 field projects with moisture susceptibility data available from mix design, construction, production, and field/forensic evalu- ation as well as technical reports and papers were identified from 11 web-survey respondents in addition to NCHRP Proj- ects 9-47A and 9-49A. Figure 10 and Figure 11 present the validation of moisture susceptibility thresholds for HWTT SIP and IDT strength test parameters, respectively. The validation for HWTT stripping slope and MR test parameters (i.e., MR ratio and wet MR stiff- ness) was not performed since no WMA mixtures had those test results available. In Figure 10 and Figure 11, green markers indicate good field pavement performance, and red markers indicate poor field pavement performance with moisture- related distresses (i.e., raveling, stripping, etc.) observed. Based on the comparisons of moisture susceptibility parameters against the corresponding flow chart thresholds for these labo- ratory tests in Figure 1, the WMA mixtures that fall in the green shaded zone are expected to have good performance, while those that fall in the red shaded zone are potentially susceptible to moisture damage. Figure 10 presents the HWTT SIP results of 20 out of 36 WMA mixtures identified from one state DOT and NCHRP Proj ect 9-47A. No stripping in the HWTT test was observed for the other 16 mixtures, and thus, they are not included in Fig- ure 10. As illustrated, only 4 WMA mixtures fell in the light gray zone, indicating SIP values lower than the minimum threshold of 3,500 load cycles. The two WMA mixtures with extremely low SIP values corresponded to a field project in Michigan, which used a soft PG 52-34 virgin binder. According to the web- survey responses, no moisture-related distress was observed on the WMA mixtures with HWTT SIP results available. Therefore, an approximately 89% (i.e., 32 out of 36) performance cor- relation was achieved for 36 WMA mixtures when comparing their HWTT SIP results against the proposed threshold. Figure 11 presents the TSR and wet IDT strength results of 53 WMA mixtures identified from four state DOTs, one contractor, and NCHRP Projects 9-47A and 9-49A. Two different thresholds of 65 psi and 100 psi for the wet IDT strength were proposed for on-site and off-site PMLC speci- mens, respectively, to account for the stiffening effect of the reheating process (Epps Martin et al., 2014). As illustrated in Figure 11 (a) for on-site PMLC specimens, only 5 WMA mix- tures fell in the light gray zone, indicating wet IDT strength values lower than 65 psi or TSR values lower than 70%. Two out of those five mixtures corresponded to the field project in Michigan that employed a PG 52-34 virgin binder men- tioned previously. According to the field evaluation results, no moisture-related distress has been observed to date on any of the WMA mixtures with TSR and wet IDT strength results available. Therefore, 25 out of 30 (or approximately 83%) WMA mixtures showed adequate correlation between laboratory test results for on-site PMLC specimens and field pavement performance. The TSR and wet IDT strength results for off-site PMLC specimens represented in Figure 11 (b) show that only 5 WMA mixtures had wet IDT strength values lower than 100 psi, and therefore fell in the light gray zone. Among those five mix- tures, two exhibited raveling in the wheel paths, which cor- responded with the proposed thresholds for the IDT strength test parameters. In addition, all of the WMA mixtures that fell in the dark gray zone had good field pavement per- formance. In general, 20 out of 23 (or approximately 87%) WMA mixtures showed good correlation with respect to the proposed thresholds of 70% TSR and 100 psi wet IDT strength for the off-site PMLC specimens. To further validate the proposed thresholds for IDT strength test parameters, various TSR and wet IDT strength values were evaluated using the receiver operating characteristic (ROC) analysis by comparing against the moisture susceptibility data from 64 WMA mixtures. The ROC analysis was performed in accordance with the following rules and the TSR and wet IDT strength results were summarized in Table 2 and Table 3, respectively. C H A P T E R 4
9 0 3,000 6,000 9,000 12,000 15,000 H W TT S IP SIP = 3,500 Figure 10. HWTT SIP threshold validation. Survey Respondents Number of WMA Mixtures Moisture Susceptibility Parameters National Center for Asphalt Technology (NCHRP 9-47A) (West et al., 2014) 22 (on-site PMLC) HWTT stripping inflection point (SIP) 19 (on-site PMLC) TSR & Wet IDT Strength 1 (off-site PMLC) TSR & Wet IDT Strength Washington State University (NCHRP 9-49A) (Wen et al., 2013) 5 (on-site PMLC) HWTT SIP 4 (on-site PMLC) TSR & Wet IDT Strength COLAS SolutionsTM 2 (on-site PMLC) HWTT SIP 2 (on-site PMLC) TSR & Wet IDT Strength Alaska Department of Transportation (Saboundjian et al., 2011) 1 (off-site PMLC) TSR & Wet IDT Strength Colorado Department of Transportation 2 (on-site PMLC) TSR & Wet IDT Strength Georgia Department of Transportation 3 (on-site PMLC) TSR & Wet IDT Strength 2 (off-site PMLC) TSR & Wet IDT Strength Iowa Department of Transportation 2 (off-site PMLC) TSR & Wet IDT Strength Missouri Department of Transportation 17 (off-site PMLC) TSR & Wet IDT Strength Washington Department of Transportation 7 (off-site PMLC) HWTT SIP Table 1. Data collected to verify guideline thresholds.
10 (a) (b) 40% 60% 80% 100% 120% 0 50 100 150 200 TS R Wet IDT Strength (psi) NCAT WSU GDOT COLAS CDOT TSR = 70% IDT Strength = 65 psi 40% 60% 80% 100% 120% 0 50 100 150 200 TS R Wet IDT Strength (psi) NCAT MODOT GDOT IADOT AKDOT TSR = 70% IDT Strength = 100 psi Figure 11. TSR and wet IDT strength thresholds validation; (a) on-site PMLC specimens, (b) off-site PMLC specimens. TSR Threshold 65% 70% 75% 80% True Positive 2 2 2 2 False Negative 0 0 0 0 False Positive 8 8 10 15 True Negative 43 43 41 36 TPR 1.00 1.00 1.00 1.00 FPR 0.16 0.16 0.20 0.29 Accuracy 0.85 0.85 0.81 0.72 Table 2. ROC analysis results for TSR thresholds.
11 True Positive: lab results < thresholds & moisture damage identified in the field; False Negative: lab results > thresholds & moisture damage identified in the field; False Positive: lab results < thresholds & no moisture dam- age identified in the field; True Negative: lab results > thresholds & no moisture dam- age identified in the field; True Positive Rate (TPR) = True Positive / (True Positive + False Negative); False Positive Rate (FPR) = False Positive / (False Positive + True Negative); and Accuracy = (True Positive + True Negative) / Total Population. As illustrated in Table 2, TSR thresholds of 65%, 70%, and 75% showed higher accuracy values than that of 80%. Referring to the ROC space shown in Figure 12(a), the TSR thresholds of 80%, 70%, and 75% were closer to the âperfect classificationâ corner than that of 65%, indicating a better predictive power. Therefore, 70% was one of the most effec- tive TRB thresholds in discriminating moisture-resistant ver- sus moisture-susceptible WMA mixtures. The ROC analysis results shown in Table 3 indicate that desirable accuracy values (approximately 0.9) were achieved by all four wet IDT strength thresholds. Referring to the ROC space shown in Figure 12(b), the thresholds of 80, 90, and 100 psi were closer to the âperfect classificationâ corner than that of 70 psi, indi- cating a better predictive power. Though slight improvement could be obtained by reducing the wet IDT strength threshold from 100 psi to 80 psi, the previously proposed threshold for wet IDT strength of 100 psi for off-site PMLC specimens was adequate in delineating moisture-susceptible versus moisture- resistant WMA mixtures. Based on the results presented in Figures 10 through 12, the correlations between the proposed moisture susceptibility thresholds in the flow chart and field pavement performance are promising (i.e., 89% for HWTT SIP, and 83% and 87% for TSR and wet IDT strength for on-site and off-site PMLC specimens, respectively). Therefore, the proposed flow chart shown in Figure 1 could be considered for implementation by state DOTs and contractors in order to identify and minimize moisture susceptibility in WMA mixtures. Wet IDT Strength Threshold 70 psi 80 psi 90 psi 100 psi True Positive 1 2 2 2 False Negative 1 0 0 0 False Positive 0 1 2 3 True Negative 21 20 19 18 TPR 0.50 1.00 1.00 1.00 FPR 0.00 0.05 0.10 0.14 Accuracy 0.96 0.96 0.91 0.87 Table 3. ROC analysis results for wet IDT strength thresholds. (a) ROC Space 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 TP R FPR TSR=65% TSR=70% TSR=75% TSR=80% Perfect Classification Better Worse (b) FPR 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 TP R ROC Space IDT Strength=70 IDT Strength=80 IDT Strength=90 IDT Strength=100 Perfect Classification Better Worse Figure 12. ROC space; (a) TSR thresholds, (b) Wet IDT strength thresholds.
