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.
60 C h a p t e r 6 6.1 Introduction This chapter presents examples to illustrate use of the PRS guidelines developed in Chapter 5. The examples use field data to develop AQC and performance relationships for chip seal and diamond grinding. Examples also were developed for thin overlay, microsurfacing, joint reseal- ing, and dowel-bar retrofit using simulated data because field data were not available. The latter examples are presented in Appendix A. 6.2 Characterization of Pre-Existing Conditions The selection of a preservation treatment and the resulting design depends on how the existing pavement conditions are characterized. Sufficient pavement condition data are needed to assess the structural deficiency in an existing pavement and the appropriate overlay thickness. For application of preservation treatments, not only must the existing pavement be structur- ally sound, but each treatment has a âwindow of opportunityâ to be effective. Therefore, treat- ment selection, timing, and location of a preservation treatment are the key components for its effectiveness. The selection of preservation treatments for flexible and rigid pavements was discussed in Chapter 3. Most pavements with load-related distresses of low severity and extent are candidates for preservation treatments. Many state DOTs use decision trees or tables from past experience to evaluate the candidacy of pavement preservation treatments based on extent and severity of existing surface distresses. Generally, highway agencies make such decisions prior to award of construction contracts; thus, contractors are not responsible for treatment selection on a given project. In this research, it was assumed that the preservation treatments are selected based on the pre-existing condition and are applied at optimum times. 6.3 Chip Seal The process for developing guidelines for the PRS is described in Chapter 2; step-by-step guidelines developed specifically for chip seal construction are described in Chapter 5. In this section, a detailed example that uses both laboratory and field data is described to illustrate the use of the proposed PRS guidelines. 1. Select a Preservation Treatment Chip seals are typically used as preservation treatments for flexible pavements. Chip sealing involves the application of asphalt (typically an emulsion) to the pavement surface, followed Examples
examples 61 by the application of rolled aggregate chips. Generally, chip seals are applied to seal longitudi- nal, transverse, and block cracking; inhibit and retard raveling/weathering; improve friction; improve ride quality; and inhibit moisture infiltration. Typical performance measures for chip seals are aggregate loss (raveling), stripping, bleeding, and flushing. 2. Select Candidate Material and Construction Characteristics and Performance Measures Table 6-1 lists the AQCs and associated performance measures identified for the chip seal portion of the PRS. The table also lists the parameters measured in the field or laboratory for the various AQCs, and related test methods. The relationships between the AQCs and functional performance measures are described below. Emulsion-Aggregate Adhesive Bond Strength Aggregate loss is the primary form of distress in chip seals at intermediate temperatures. One of the main causes of aggregate loss is a lack of adhesive bond strength between the aggregate and emulsion such that significant cover aggregate is lost upon loading. The adhesive bond between the aggregate and emulsion is a function of the construction practices used during the seal con- struction. Construction-related factors such as the time between application of the aggregate layer onto the emulsion and the first rolling pass, the type of compaction effort applied, the number of roller passes, and the curing time allowed prior to traffic opening can affect the adhesive bond formed between the aggregate and emulsion and thus the aggregate loss observed (Lee and Kim 2008). The Vialit testing of extracted field samples proposed in this PRS directly measures the strength of the adhesive bond formed during the construction of the chip seal treatment. Gradation The performance-uniformity coefficient (PUC) is a performance indicator of aggregate gra- dation and its uniformity. In chip seal surface treatments, gradations that are more uniform perform better than those that are less uniform in terms of the aggregate loss and bleeding failure criteria. The PUC of the aggregate source will affect the bleeding and aggregate loss per- formance of the chip seal surface treatment being constructed. The concept of the PUC is based on McLeodâs chip seal failure criterion that 70% is the ideal aggregate embedment for chip seal surface treatments. The PUC is the ratio of the percentage passing at a given embedment depth (PEM) to the percentage passing at twice the embedment depth (P2EM) in a sieve analysis curve (Lee 2007, McLeod 1971). Table 6-1. Proposed AQCs for the preliminary PRS. AQCs Related Performance Measure Test Parameter Proposed Test Method Emulsion-Aggregate Adhesive Strength Aggregate Loss % Aggregate Loss Vialit Test (Lab) Gradation Aggregate Loss Performance-Uniformity Coefficient Gradation Analysis of Vialit Samples (Lab) Mean Profile Depth (MPD) and Visual Inspection Bleeding and Skid Resistance MPD after 1 Week and % Bleeding Laser Profiler (Field) Emulsion Application Rate (EAR) Aggregate Loss and Bleeding Percentage of Optimum EAR for PUC Analysis Ignition Oven: Vialit Samples (Lab) Aggregate Application Rate Aggregate Loss and Bleeding Percentage of Optimum EAR for PUC Analysis Ignition Oven: Vialit Samples (Lab)
62 performance-related Specifications for pavement preservation treatments The PEM and P2EM values represent the bleeding and aggregate loss failure criteria, respectively, with regard to the gradation. The PEM value is defined as the percentage passing that corresponds to 70% of the median particle size on the gradation curve. The P2EM value is defined as the percent- age passing that corresponds to 1.4 times the median particle size, with the median particle size defined as the particle size of which 50% of the gradation passes through the sieve. For chip seal, a low PEM value is desired because a low percentage of the gradation passing at the bleeding failure criterion indicates that the aggregate particles in that range of the gradation are larger and less susceptible to bleeding. However, a high P2EM value indicates that a low percentage of aggregate particles do not meet the aggregate loss criterion, and therefore less aggregate loss is expected. The aggregates are assumed to be embedded in emulsion up to 70% of its median (M) particle size. Particles smaller than 0.7M will be submerged completely in the emulsion and experience bleeding. Therefore, smaller particles should be larger than 0.7M to avoid bleeding. In contrast, particles bigger than 1.4M will be less than 50% embedded in emulsion and therefore are likely to be lost when trafficked. A PUC value close to zero indicates a more uniformly graded aggregate. Thus as the PEM value approaches 0% and the P2EM value approaches 100%, gradation uniformity increases, resulting in less bleeding and aggregate loss. Mean Profile Depth The mean profile depth (MPD) is a measure of the exposed texture depth of a chip seal surface treatment (Transit New Zealand 2005). As the emulsion-aggregate rate (EAR) or embedment depth increases, the MPD will decrease, and where the EAR (or embedment depth) is decreased for a given aggregate structure, the MPD will increase. The MPD is defined as the average of the mean segment depths of all of the segments of the profile, with the mean segment depth being the average value of the profile depth of the two halves of a segment having a given baseline, as shown in Figure 6-1 (ASTM E1845). MPD is an indicator of surface roughness (i.e., macrosurface texture) and aggregate exposure depth of the chip seal. Roughness is important because it provides the skid resistance and fric- tion needed for vehicles to brake adequately. The aggregate exposure depth is a function of the Figure 6-1. Schematic diagram of MPD definition (ASTM E1845).
examples 63 aggregate embedment depth, which is directly related to bleeding performance of chip seals. Lower MPD values increase the likelihood of bleeding and skid-resistance problems. Emulsion and Aggregate Application Rates The EAR and aggregate application rate (AAR) are both critical to the performance of chip seal surface treatments. Previous research has shown a considerable variability between the mea- sured and design EAR and AAR in chip seal construction (Adams and Kim 2011). 3. Establish AQC-Performance Relationships and Determine Performance Limits Table 6-1 summarizes the relationships established between the AQCs and their associated chip seal performance measures. The performance limits related to each AQC are described in this section. Adhesive Bond Strength Versus Aggregate Loss One of the most critical performance measures for a chip seal surface treatment is aggregate loss. The strength of the adhesive bond that forms between the emulsion and aggregate used in chip seal construction determines the ability of the chip seal to retain aggregate under traf- fic loading. Previous research measured the adhesive bond strength directly by using Vialit aggregate loss impact loading tests on specimens extracted directly from the constructed field sections (Im 2013, Adams and Kim 2011). The data were obtained from various modified and unmodified emulsion types at different test temperatures. The research showed a linear rela- tionship between bond strength (measured using a pneumatic adhesive tensile testing instru- ment) which tested emulsions in accordance with AASHTO TP91, âDetermining Asphalt Binder Bond Strength by Means of the Bitumen Bond Strength Test,â and percent aggregate lost from the Vialit test (see Figure 6-2). Therefore, Vialit test aggregate loss results can be used Figure 6-2. Bond strength versus Vialit test aggregate loss performance (Kim et al. 2017). 0 5 10 15 20 25 0 50 100 150 200 250 % A gg re ga te L os s BBS (psi) Unmodified at 15C Modified at 15C Unmodified at 25C Modified at 25C Poor Performing at 15C Linear Model y = â0.0765x + 24.641 RÂ² = 0.7493
64 performance-related Specifications for pavement preservation treatments as an AQC for the bond strength between the aggregate and emulsion used in a chip seal and as an indicator of the chip sealâs resistance to aggregate loss. This research also showed that the Vialit aggregate loss test differentiates between modified and unmodified emulsions at different temperatures. Vialit Test Aggregate Loss Threshold Limit Determination To develop the Vialit test aggregate loss threshold for the PRS, limits need to be derived based on the traffic demand expected for the constructed chip seal section. For example, roadways with higher traffic levels often have higher speeds, and vehicles are more susceptible to windshield damage due to aggregate loss than is the case for lower volume roads. Therefore, the acceptable aggregate loss threshold value will be lower (i.e., more restrictive) for higher traffic levels than lower traffic levels. In contrast, at lower traffic levels, the aggregate loss threshold should be less restrictive compared to the threshold at higher traffic levels. The PRS will establish threshold values for the three traffic (AADT) levels: 1. low traffic (0â500) 2. medium traffic (501â2500) 3. high traffic (2501â20,000) These traffic levels are those used in an NCHRP project dealing with PRS for emulsified asphal- tic binders used in preservation surface treatments (Kim et al. 2017). The research team recom- mends 20,000 vehicles as the upper AADT limit for high traffic, based on a study of high-traffic chip seal practices across the United States â although chip seals are constructed at AADT counts that exceed 20,000 vehicles in California, Colorado, and Montana (Gransberg and James 2005). Because the performance of chip seals constructed at high-traffic volumes is heavily dependent on local factors, such as climate, traffic speed, aggregate quality, contractorâs experience, and equipment, the high-traffic upper limit is conservatively set at 20,000 vehicles. To develop Vialit aggregate loss limits for the PRS, an aggregate loss limit that differentiates between acceptable and unacceptable mixture performance is needed. Two aggregate loss limits were adopted from existing research studies, based on laboratory and field chip seal experiments. The first limit is the maximum allowable aggregate loss limit for the lowest traffic level. The Alaska Department of Transportation (McHattie 2001) defines âacceptableâ field aggregate loss as 10% or less for any traffic situation where a chip seal is constructed. Previous research has also found that 10% aggregate loss limit characterizes acceptable aggregate loss performance in third-scale model mobile load simulator (MMLS3) testing (Lee 2007, Adams and Kim 2011, Lee and Kim 2009) and that if a chip seal exhibits 10% aggregate loss in the laboratory, it is likely to exhibit significant aggregate loss in the field. The relationship between the MMLS3 test results and Vialit test aggregate loss results was examined in research for the North Carolina DOT (NCDOT) (Kim and Im 2015). The research found that, for unmodified emulsions (used for low traffic), the Vialit test aggregate loss is double the aggregate loss caused by MMLS3 testing. For modified emulsions (used for high traffic), the Vialit test aggregate loss is about 50% more than the aggregate loss caused by MMLS3 testing. Therefore, aggregate loss threshold limits for the Vialit test were established as 20% for low traf- fic and 15% for high traffic and the average of the low- and high-traffic limits for medium traffic (i.e., 17.5% aggregate loss). PUC (Gradation) Versus Aggregate Loss Relationship Gradation is an AQC that relates directly to the performance of chip seal treatments. The effect of gradation on performance is dependent on the PUC, which indicates the degree of uniformity of the aggregate source. For chip seal surface treatments, more uniform gradations
examples 65 perform better than less uniform gradations in terms of aggregate loss and bleeding failure cri- teria (Adams and Kim 2011, Lee and Kim 2009). Figure 6-3 shows the relationship between the PUC and aggregate loss performance (Adams and Kim 2011, Lee and Kim 2009). PUC Threshold Limit Determination Approach The approach used to determine the PUC threshold limit for the PRS is based on the concept that 100% of the optimum EAR (which is based on the performance-based mix design) yields the appropriate baseline for aggregate loss. The performance-based mix design has been shown to minimize both the potential for aggregate loss and bleeding problems in chip seal mixtures (Adams and Kim 2011). Previous research (Adams and Kim 2011) has shown that the sensitivity of the aggregate loss performance to the PUC parameter (i.e., gradation) is related to the EAR and, more specifically, to how close the measured EAR is to the design optimum EAR. Therefore, the approach is as follows: First determine the asymptotic percent aggregate loss corresponding to the optimum EAR; then for a given EAR, determine the PUC threshold value that corresponds to this percent aggregate loss from the respective curve (e.g., Figure 6-4). Mean Profile Depth Versus Bleeding The mean profile depth (MPD) of a chip seal can be measured using a three-dimensional laser profiler that tracks the MPD as a function of time. MPD is related directly to the roughness and bleeding performance of the seal (Adams and Kim 2011). For the PRS, MPD measured after 1 week in service is proposed, because sweeping typically is conducted within 1 week after con- struction and the MPD can be measured using the laser while traffic control is still set up. The MPD can also be used as an indicator of potential performance problems in a chip seal treatment (Kim and Im 2015). The PRS threshold limit for the MPD parameter was determined based on the performance data. The relationship between the MPD and bleeding was observed in field sections constructed by NCDOT pavement preservation construction personnel (Adams and Kim 2011). Locations were selected where no significant surface distresses existed before the treatment application on straight alignment and with no steep grades. Any cracks near the sections were sealed before construction. Figures 6-5 and 6-6 show the MPDs measured after 1 week and the performance observed after 1 year in service for granite and lightweight aggregate, respectively. The figures show that the sections with the lower MPD values (below 2.12 mm) after 1 week exhibited Figure 6-3. Percent aggregate loss versus PUC (Adams and Kim 2011). % A gg re ga te L os s PUC
66 performance-related Specifications for pavement preservation treatments Figure 6-4. Approach for developing threshold values for the PUC based on design rates. 0 2 4 6 8 10 12 14 16 0 10 20 30 40 50 60 % A gg . L os s PUC 55% of Optimum EAR 70% of Optimum EAR 85% of Optimum EAR 100% of Optimum EAR 115% of Optimum EAR Figure 6-5. Bleeding (1 year) and MPD (1 week) of granite aggregate sections. Section 1 Section 2 Section 3
examples 67 bleeding after 1 year in service. For chip seal sections with AADT between 1,000 and 4,500 vehi- cles, sections with a 1-week MPD below 2.12 mm exhibited bleeding within 1 year of service but sections with a 1-week MPD above 2.12 mm did not exhibit bleeding. Therefore, 2.12 mm was selected as the threshold value for the 1-week MPD AQC for PRS, based on the results for granite aggregates. However, a different MPD value might be selected for higher AADT levels. Emulsion and Aggregate Application Rates Variance from the design EAR and AAR can influence the performance of the chip seal (e.g., aggregate loss and bleeding). For example, applying an AAR higher than the design value on a seal could lead to aggregate loss. The combined effects of these two parameters will ultimately determine the performance. Therefore, these two parameters cannot be used independently to assess penalties for chip seals. Although the PRS will be established using Vialit test aggregate loss percentages, the PUC and the MPD as AQCs, and the EAR and AAR need to be measured to determine the percent- age of the optimum EAR and AAR for the PUC analysis. The EAR and AAR can be obtained by conducting ignition oven tests (as specified in ASTM D6307) using chip seal specimens. For the purposes of the chip seal PRS, the specimens extracted for Vialit testing will be used to determine the percentage of aggregate loss AQC. This approach allows the EAR and AAR to be captured without additional samples being required. 4. Determine Thresholds and Limits for AQC The steps for establishing limits for the AQCs and quality measures outlined in Chapter 5 were followed. These steps are summarized for chip seals as follows. 1. Determine AQC-performance relationships. The relationships between each AQC and per- formance have been established. The aggregate loss measured from the Vialit test, the MPD, Figure 6-6. Bleeding (1 year) and MPD (1 week) of lightweight aggregate sections. Section 7 Section 8 Section 9
68 performance-related Specifications for pavement preservation treatments and the PUC demonstrates the ability to predict the key performance measures associated with chip seal treatments. 2. Set specification limits. The specification limits for the AQCs were determined based on laboratory and field performance data, as detailed in the previous section. For the Vialit test aggregate loss AQC, the specification maximum limit is a function of the traffic level as defined above. The specification minimum limit for MPD was set at 2.12 mm, based on the results for granite aggregate. However, a different MPD value may be selected for higher AADT levels. Lastly, for the PUC, the threshold value was determined to be a function of the percentage of the optimum EAR and AAR measured from ignition oven tests following chip seal construction. 3. Decide on a quality measure. The recommended quality measure for chip seals was decided as the PWL. 4. Define AQL. The upper AQL for chip seal treatments is recommended to be a PWL of 90 for this chip seal demonstration example, based on typical AQL values, that is, 90% of samples from a lot must pass the AQC specification limit to receive 100% pay. 5. Define RQL. The RQL for this chip seal demonstration example was determined to be a PWL of 60, based on the typical range of RQL values, that is, 60% of samples from a lot must pass the specification limit to be eligible for reduced pay. Determining PWL for Chip Seal Field Demonstration Sections The PWLs were calculated for the chip seal field sections. Tables 6-2 and 6-3 present the PWL values for the Vialit test aggregate loss and MPD tests, respectively. These PWL values were used to determine whether the contractor for a sample lot would have received full pay, reduced/partial pay, or been rejected in the proposed PRS. Section ID Traffic Volume Lower Spec. Limit Avg. MPD Std. Dev. Q Sample Size PWL Pay Conclusion MD-A 4500 2.12 1.99 0.24 -0.54 3 34.5 <60 lot rejected MD-B 4500 2.12 2.2 0.06 1.33 3 100 Full pay MD-C 4500 2.12 2.12 0.08 0.00 3 50 <60 lot rejected MD-D 4500 2.12 2.07 0.16 -0.31 3 41.4 <60 lot rejected MD-E 4500 2.12 2.21 0.04 2.25 3 100 Full pay MD-F 4500 2.12 2.22 0.07 1.43 3 100 Full pay Table 6-3. MPD PWL values. Section ID Traffic Volume Upper Spec. Limit Avg. Loss Std. Dev. Q Sample Size PWL Pay Conclusion MD-1 4500 15 11 2.54 1.57 9 95.2 Full Pay MD-2 4500 15 8.4 1.46 4.52 9 100 Full Pay MD-3 4500 15 12.2 2.1 1.33 9 91.4 Full Pay MD-7 4500 15 2.8 0.84 14.52 9 100 Full Pay MD-8 4500 15 4.2 1.03 10.49 9 100 Full Pay MD-9 4500 15 1.8 1.4 9.43 9 100 Full Pay MD-10 1000 17.5 7.3 1.67 6.11 9 100 Full Pay MD-11 1000 17.5 9.2 3.1 2.68 9 100 Full Pay MD-12 1000 17.5 13.6 1.75 2.23 9 99.7 Full Pay MDV-1 2000 17.5 10.2 1.1 6.64 9 100 Full Pay MDV-2 2000 17.5 16.7 2.7 0.30 9 59.5 <60 Lot Rejected Table 6-2. Vialit test aggregate loss PWL values.
examples 69 5. Specify Test Methods to Measure AQC The following section describes the test methods to measure different AQCs. Vialit Aggregate Loss To obtain samples for conducting the Vialit aggregate loss tests, steel Vialit plates are placed onto the road surface prior to the start of construction. A chip seal treatment is then constructed on top of the Vialit plates (see Figure 6-7). After the compaction phase of the construction is complete, the sample is allowed to cure for a least 1 hour prior to starting the extraction process. The details of the test procedure are described in Adams and Kim (2011). Mean Profile Depth (MPD) Measurement and Visual Pavement Inspection A three-dimensional (3-D) laser surface texture profiler is shown in Figure 6-8. This portable laser profiler is used to obtain the MPD data used in a non-destructive manner. The test proce- dure is described in ASTM E 1845 specifications. Using the laser profiler, the MPD is measured 1 week after construction. Because MPD measurement requires traffic control, it is recom- mended that this task be coordinated with the sweeping procedure that typically occurs about 1 week after construction. Current practice among state highway agencies is to inspect (visually) a newly constructed chip seal on the day of construction and again during the first summer of construction. As bleeding typically will occur within the first year in service (Adams and Kim 2011), conducting a visual inspection for bleeding during the sealsâ second summer in service is desirable. Because the 1-week MPD is an indicator of bleeding potential within the first year of the life of the seal, measuring the 1-week MPD for each constructed section and checking against the threshold limit can identify sections that have high bleeding potential. Also, conducting a visual inspec- tion prior to the end of the warranty period agreed upon by the contractor and agency (typically 1 year) would determine if bleeding occurred in sections that have measured MPD values below the threshold value. If the section exhibits bleeding levels higher than 50%, the contractor should Figure 6-7. Vialit test apparatus.
70 performance-related Specifications for pavement preservation treatments repair or replace the chip seal at no cost to the agency. This requirement minimizes the risk to the agency by ensuring that bleeding does not occur within the agreed-upon warranty period while also avoiding penalizing the contractor unless significant bleeding can be validated instead of relying on a predictive indicator of bleeding. In summary, the 1-week MPD measure allows a state highway agency to identify and prioritize potentially problematic sections. This approach allows for an efficient allocation of inspection personnel and resources. In addition, as agencies build a database of 1-week MPD values versus bleeding performance, the 1-week MPD limit can be calibrated locally. The NCDOT Pavement Condition Survey Manual (NCDOT 2012) provides a visual inspec- tion method for identifying bleeding severity in a chip seal. The manual places bleeding severity in three categories: â¢ Light bleeding: condition is present on 10%â25% of the section â¢ Moderate bleeding: condition is present on 25%â50% of the section â¢ Severe bleeding: condition is present on greater than 50% of the section In this method, each wheel path of a two-lane roadway represents 25% of the section, as defined by the manual. Bleeding is evaluated across the entire length of the chip seal for any significantly bled areas that justify replacement (see Figures 6-9 and 6-10). Current practice is to replace a bled section when the bleeding is severe, or greater than 50% of the section, as exhibited in Figure 6-10. These PRS recommend that a section that exhibits severe bleeding within the warranty period should be rejected and replaced or repaired at the expense of the contractor. The rationale behind this recommendation is that most highway agencies would not repair or replace a chip seal with light or moderate bleeding as the seal would not be considered to have reached the end of its service life. Ignition Oven Test The measured material application rates can be determined according to the ignition oven test procedure in AASHTO T 308 using chip seal samples extracted from the field after construction. Also, through ignition testing, the EAR could be determined (Adams and Kim 2011). Figure 6-8. Photo of 3-D laser profiler prototype used in this study.
examples 71 6. Establish a Sampling and Measurement Plan As noted previously, the risks associated with incorrectly accepting or rejecting a lot are related to sample size and method. The procedure outlined in Chapter 5 was followed to develop guide- lines for a sampling and measurement plan for chip seal treatment: 1. Determine which party performs acceptance testing. The contractor and agency must agree on the testing party. 2. Determine the type of acceptance plan to be used. Stratified random sampling, a modified version of random sampling commonly used in pavement construction acceptance sampling, involves dividing a lot into several sublots of equal size and selecting random samples within each sublot. 3. Develop verification sampling and testing procedures. Verification sampling is used to verify the accuracy of acceptance test results. The decision to use split or independent sampling depends on the goals of the agency. For this example, it is assumed that the agency or indepen- dent third party will measure the Vialit aggregate loss and MPD at the recommended sampling frequency for each sublot for verification. In practice, the agencyâs verification test methods are used solely for verification, and acceptance methods proposed by the contractor are first compared to the results of the agency verification tests. Figure 6-9. Light bleeding example (NCDOT 2012). Figure 6-10. Severe bleeding needing replacement (NCDOT 2012).
72 performance-related Specifications for pavement preservation treatments 4. Select the appropriate verification sampling frequency. The verification sampling frequency by the agency should be approximately 10% of the acceptance sampling rate by the contractor. In practice, the verification testing frequency is decided for economic, rather than statistical, reasons and is agreed upon by the agency and contractor. For this example, it is assumed that the procedure is already established. 5. Determine lot size and sample size. The evaluation of the aggregate loss AQC involves the extraction of field samples for Vialit aggregate loss testing in a temperature-controlled labora- tory environment. Therefore, lots and sublots should be defined logically as segmented lengths of a project. For this example, the recommended lot size is 5,000 ft. long. Sublot lengths from which stratified random sampling should be performed are recommended to be 100 ft. The risks associated with sampling depend on sample size. Based on earlier research, it is recom- mended that a sample size of nine be used for each sublot for the Vialit aggregate loss AQC (Adams and Kim 2011). Because taking nine samples for each sublot throughout the entire lot is time-consuming and transportation of these samples to the laboratory would be difficult, it is recommended that three sublots be selected randomly from the lot for sampling. However, individual agencies may increase the number of sublots sampled to minimize risk if sufficient resources and personnel are available. The same lot and sublot plan described for aggregate loss should be used for measuring the MPD AQC. However, a sample size of three MPD laser scans could provide statistically significant and representative results (Adams and Kim 2011). Current practice requires gradation measure- ments taken at the aggregate quarry to ensure that the aggregate specified in the chip seal contract meets the gradation requirements. Therefore, no field sampling is required for the evaluation of the PUC AQC. Current quality control checks of gradation should be maintained, and the PUC can be checked using the gradation data. 7. Develop Pay Adjustment Factors Pay adjustment factors are necessary for acceptance plans in developing PRS. However, estab- lishing pay reduction factors to determine partial pay using typical approaches based on reduc- tion of service life concepts is not appropriate for performance measures such as aggregate loss, which is the most critical distress for chip seal treatments (Lee 2007). For instance, most aggre- gate loss occurs within the first days and weeks in service, and field observations have shown that aggregate loss early in the life of the seal does not necessarily lead to a reduction in the service life of the seal (Im 2013). Therefore, the opinions of State DOT pavement maintenance practitioners regarding pay adjustment factors were used as a starting point but should be validated prior to implementation. Vialit Aggregate Loss Key issues associated with aggregate loss are vehicular damage claims and the public perception of chip seal treatments as an effective treatment alternative. Another concern with aggregate loss is its contribution to bleeding (Lawson et al. 2007). Given the established maximum specification thresholds of 20%, 17.5%, and 15% aggregate loss for low, medium, and high traffic, respectively, a set of samples for a lot that fails to meet the AQL of 90, but exceeds the RQL of 60, should be considered for partial pay. A relationship between aggregate loss and pay factor could not be developed because of the inability to quantify the effect of aggregate loss on bleeding failure or public perception/ satisfaction with the quality of the sealing work. However, as a starting point for these PRS, the research team surveyed State DOT pavement maintenance practitioners to obtain recommen- dations on reasonable partial pay factors for PWL ranging from 60 to 90. The survey results were averaged and rounded to the nearest 5% (see Table 6-4).
examples 73 Also, the survey respondents unanimously recommended that the contracted party also should be responsible for addressing any vehicle damage claims at no cost to the state highway agency. Mean Profile Depth (MPD) Measurements and Visual Inspection for Bleeding The 1-week MPD and visual inspection method described for assessing bleeding potential in a surface treatment in the PRS provides a pass/fail criterion. If bleeding over 50% is identified visually as present, the lot is rejected and should be repaired or replaced at cost to the contractor. If minimal or no bleeding is present, the lot passes and the contractor receives full pay. The research team recommends that no partial pay factors should be established for bleeding because most bleeding is a result of aggregate loss issues (Lawson et al. 2007) and thus the contrac- tor would be penalized twice for the same problem as the proposed PRS assesses a pay reduction for the aggregate loss that led to bleeding. The other main cause of bleeding is the lack of binder resistance to non-recoverable strain at high temperatures (DâAngelo and Dongre 2007). This defi- ciency is a binder performance problem addressed by the performance-graded specifications for the binder selected for the job and is not appropriate for penalty in these PRS. Construction-related bleeding performance is a result of either aggregate loss, which when significantly present is already penalized in these PRS, or the over-application of emulsion (known as flushing), which could result in a rejection of the lot based on visual inspection. Performance-Uniformity Coefficient (PUC) Previous research has established the importance of aggregate gradation for the performance of chip seal surface treatments as it relates to aggregate loss and bleeding potential (Lee 2007). State highway agencies conduct regular quality control testing to ensure that gradation is within the specified limits. The PUC can be determined easily from the gradation analysis data to ensure that the aggregate selected will not contribute to aggregate loss or bleeding. These PRS recom- mend that the PUC of the aggregate should be assessed on a pass/fail basis for use in the con- struction of chip seal treatments at the time of quality control inspection based on the PUC limit derivation approach described above. Emulsion and Aggregate Application Rates Significant differences between the target and design application rates can significantly affect the performance of a chip seal treatment. However, because the effect of variance from the target EAR and AAR would be captured by the Vialit aggregate loss test and the bleeding assessment in these PRS, the research team decided that no separate pay factor adjustment is recommended to account for this variance. 8. Summary This example has demonstrated the following: â¢ Vialit testing of extracted field chip seal samples can be used to assess the raveling potential of chip seals for different aggregates, binder types, design rates, and traffic levels. â¢ A unique relationship exists between the MPD and the percentage of bleeding under both laboratory and field traffic loading conditions, such that threshold values can be established according to aggregate size to predict bleeding potential in chip seals. Table 6-4. Pay factors vs. PWL for aggregate loss. PWL Range (%) Pay Reduction (%) 90-100 Full Pay 75-90 25% Pay Reduction 60-75 50% Pay Reduction 0-60 Reject; No Pay
74 performance-related Specifications for pavement preservation treatments â¢ The PWLs calculated for each AQC can be used to determine if a lot will receive full pay (AQC>90), partial pay (60<AQC<90), or no pay (AQC<60) for chip seal treatments. â¢ The 1-week MPD measurement in combination with the visual inspection of sections below the 1-week MPD threshold could be used as an indicator of the potential for bleeding prob- lems within the first year in service. â¢ A relationship exists between the PUC measured and aggregate loss; the PUC can be used as a performance measure for the gradation of aggregate selected for chip sealing. â¢ Bleeding should be addressed on a pass/fail basis, with failure defined as bleeding present above 50%. Any repair or replacement costs should be the contractorâs responsibility. â¢ The PUC is an indicator of uniformity of gradation; it should be addressed on a pass/fail basis during regular quality control testing on quarry material. 6.4 Diamond Grinding The following AQCs and performance measures were identified and selected for diamond grinding treatment: 1. Performance measures: Faulting, slab curling/warping, friction, and roughness 2. AQCs: Surface smoothness, aggregate type and hardness, spacing of saw blades, depth of saw cuts 3. Final selected AQC and performance measures: Profile-based indices (IRI or DLI) as AQCs and expected surface roughness (IRI), percent slab cracked, and faulting as performance measures. This example demonstrates the use of a mechanistic-empirical approach for establishing such relationships between the AQCs and performance measures. The following is the summary for developing such relationships: 1. Determine profile-based indices from the measured longitudinal profiles before and after application of diamond grinding. 2. Evaluate the resulting change in dynamic axle load response before and after application of diamond grinding. 3. Use variation or changes in axle load spectra (before and after the treatment) to predict pave- ment performance by using the Pavement-ME analysis for the pavement sections. 4. Relate change in AQCs to variation in the expected performance. PRS guidelines assumed that the criteria for treatment selection with respect to existing pave- ment conditions and appropriate treatment timing were established and followed. The steps presented in the guidelines (Chapter 5) were followed to develop general PRS guidelines and this example for diamond grinding treatment on rigid pavements. 1. Select a Preservation Treatment Diamond grinding has been identified as one of the primary concrete pavement preservation treatments used to improve pavement surface smoothness. 2. Select Candidate Material and Construction Characteristics and Performance Measures Roughness is a key performance measure in evaluating the effectiveness of diamond grind- ing. Numerous profile-based indices can be used to quantify surface roughness, such as IRI and DLI. IRI and DLI decrease in magnitude as surface roughness decreases; they are appro- priate candidates as an AQC for PRS. Because IRI seems to give similar results to DLI, it is recommended for use as the AQC, because it is the most commonly used index.
examples 75 3. Establish AQC-Performance Relationships and Determine Performance Limits To validate relationships between the AQC and pavement performance, it was necessary to evaluate the performance of the pavement sections that were diamond ground. The mechanistic- analysis procedure outlined in Chapter 5 was repeated for several pavement sections to relate IRI and DLI to expected performance. Table 6-5 summarizes the characteristics of 14 JPCP sec- tions included in the LTPP database. These pavement sections were selected for the following reasons: â¢ Data before and after longitudinal profiles are available, â¢ Unique file formats for profile data are available in the LTPP database, â¢ Distribution in the four LTPP climatic zones, â¢ Treated with only diamond grinding (i.e., no other maintenance treatments), â¢ Significant change in roughness just after diamond grinding is evident. A few pavement sections that showed an increase in roughness after application of diamond grinding treatment were included to explore the reasons for an ineffective treatment. The analyses were conducted for all pavement sections to determine the effect of diamond grinding on pavement profiles and profile-based indices, dynamic axle loads, and predicted pavement performance in terms of faulting, cracking, and IRI for a 20-year design life using the Pavement-ME, and the expected SLE due to diamond grinding. These analyses revealed the following: â¢ The change in IRI and DLI were reasonably correlated with each other, as seen in Figure 6-11. â¢ Grinding treatment was generally effective for all but two pavement sections (8-3032 and 27-3009). The profiles and faulting of these sections showed that diamond grinding did not sig- nificantly reduce the magnitude of faulting, especially in Section 27-3009. Using PSD analy ses, a sharp decrease in amplitude of wavelengths in the profile signal response was observed for the smoothed pavement sections. The ineffective grinding in a few sections may be because grinding only removed roughness and not the contributor to faulting (poor drainage or in-adequate load transfer). The changes in profile indices shown in Figure 6-12 indicate that IRI and DLI follow the same trend (increase or decrease). Because IRI is the most prevalent and well-understood roughness index in current practice, IRI was selected as the AQC for developing general PRS guidelines and used in the example for diamond grinding treatment of rigid pavements. Table 6-5. LTPP concrete pavement sections selected for diamond grinding analysis. No. Section ID State Age (years) Climate Zone AAWD 2 FI3 AADTT4 5 Slab thickness (in) Subgrade typeBefore Application1 After 1 6-3010 CA 9.03 10.24 11.69 DNF 55 0 4088 8.8 A-6 2 13-3017 GA 10.04 10.99 11.47 WNF 118 15 2702 9.9 A-5 3 27-4050 MN 15.73 18.92 20.82 WF 95 1452 220 8 A-3 4 42-3044 PA 10.73 10.91 11.11 WF 186 263 3864 12.7 A-2-4 5 46-3010 SD 9.84 9.88 10.81 WF 98 1055 418 9.3 A-2-4 6 55-3009 WI 10.97 11.74 11.76 WF 128 609 356 8.2 A-6 7 4-7614 AZ 12.47 14.00 14.47 DNF 36 0 1743 10 A-2-4 8 16-3017 ID 18.17 18.95 19.92 DF 89 345 748 10 A-4 9 49-C431 UT 6.98 7.73 7.78 DF 96 396 1087 9.8 A-1-b 10 8-3032 CO 17.83 18.77 20.96 DF 95 346 289 8.6 A-1-a 11 27-3009 MN 13.07 14.10 15.04 WF 119 1022 2812 7.5 A-6 12 38-3006 ND 14.76 19.84 21.00 DF 195 1417 416 8.4 A-4 13 20-3015 KS 12.48 13.75 14.36 WF 73 261 932 9.2 A-6 14 39-9006 OH 13.43 13.58 14.55 WF 134 307 3073 9.4 A-1-b Note: 1Age at the time of treatment application, 2Average annual wet days, 3Freezing index, 4Average annual daily traffic, and 5AASHTO soil classification
Figure 6-11. Change in IRI versus change in DLI before and after grinding. y = 0.0488x - 0.0095 RÂ² = 0.8618 -8 -6 -4 -2 0 2 4 6 8 -150 -100 -50 0 50 100 150 Ch an ge in D LI (1 0- 2 in .) Change in IRI (in/mi) '+': reduction in IRI '-': increase in IRI Figure 6-12. IRI and DLI before and after grinding. (a) IRI (b) DLI 84.4 87.3 88.1 157.5 155.3 158.5 91.1 120.3 121.0 114.0 86.8 79.1 77.8 110.0 49.4 41.5 59.9 72.7 69.3 61.0 78.1 96.4 109.3 118.1 186.6 70.0 52.5 66.3 0 50 100 150 200 6-3010 13-3017 27-4050 42-3044 46-3010 55-3009 4-7614 16-3017 49-C431 8-3032 27-3009 38-3006 20-3015 39-9006 O ve ra ll IR I ( in/ mi ) Section ID Before grinding After grinding 5.2 6.0 5.5 11.8 9.6 11.7 4.4 5.8 8.5 5.4 8.1 4.6 7.2 7.0 3.8 3.7 3.7 6.2 5.9 7.4 3.7 4.8 7.9 7.7 12.5 3.9 4.2 6.0 0 2 4 6 8 10 12 14 6-3010 13-3017 27-4050 42-3044 46-3010 55-3009 4-7614 16-3017 49-C431 8-3032 27-3009 38-3006 20-3015 39-9006 O ve ra ll D LI (1 0-2 in .) Section ID Before grinding After grinding
examples 77 â¢ When diamond grinding reduced the surface roughness, it generally reduced the dynamic loads. The few sections that showed higher roughness after grinding exhibited a minimal change in dynamic loads or in some cases (Section 27-3009) higher dynamic loads. This indi- cates a positive correlation between an effective grinding treatment that reduces roughness and the resulting dynamic loads experienced by the pavement. Figure 6-13 is an example of a shift in dynamic load for an effectively treated section (42-3044) versus an ineffectively treated section (27-3009). â¢ The axle load spectra before and after grinding can be used as inputs in the Pavement-ME software to predict a change in the predicted performance. A shift in the dynamic loads after grinding should be considered to predict performance of the pavement in terms of cracking, faulting, and IRI. The predicted performance using Pavement-ME indicated that grinding may not always improve pavement performance and these variations are related to change in dynamic loads and surface profile. Examples of change in performance for Sections 42-3044 and 27-3009 are shown in Figures 6-14 and 6-15, respectively. Section 42-3044, which exhib- ited an immediate reduction in roughness due to grinding, showed an improvement in perfor- mance over time as compared to Section 27-3009 where grinding, treatment was ineffective. Figures 6-16 through 6-18 show the change in IRI and DLI versus the change in long-term faulting, cracking, and roughness, respectively. Such relationships support the mechanistic- empirical approach of relating AQCs to performance and that the AQC candidates (IRI and DLI) correlate well with expected performance. â¢ The relationships established between DIRI (IRI after grinding minus IRI before grinding) and predicted performance at 20 years (in terms of faulting, cracking, and IRI) were used Figure 6-13. Examples of dynamic axle load spectra before and after grinding. (a) Single axle â42-3044 (b) Tandem axle â42-3044 (c) Single axle â27-3009 (d) Tandem axle â27-3009 0 0.1 0.2 0.3 0.4 0.5 0.6 0 10000 20000 30000 40000 Lo ad d ist rib ut io n Single Axle Loads Before Grinding After Grinding 0 0.05 0.1 0.15 0.2 0 20000 40000 60000 80000 Lo ad d ist rib ut io n Tandem Axle Loads Before Grinding After Grinding 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0 10000 20000 30000 40000 Lo ad d ist rib ut io n Single Axle Loads Before Grinding After Grinding 0 0.02 0.04 0.06 0.08 0.1 0.12 0 20000 40000 60000 80000 Lo ad d ist rib ut io n Tandem Axle Loads Before Grinding After Grinding
78 performance-related Specifications for pavement preservation treatments Figure 6-14. Predicted performance for pavement Section 42-3044. (a) Percent slabs cracked (b) Faulting (c) IRI 0 20 40 60 80 100 0 5 10 15 20 Sl ab s C ra ck ed (% ) Pavement Age (years) Before Grinding After Grinding 0 0.05 0.1 0.15 0 5 10 15 20 Fa ul tin g (in ) Pavement Age (years) Before Grinding After Grinding 0 50 100 150 200 250 300 0 5 10 15 20 IR I ( in/ mi le) Pavement Age (years) Before Grinding After Grinding
examples 79 Figure 6-15. Predicted performance for pavement Section 27-3009. (a) Percent slabs cracked (b) Faulting (c) IRI 0 20 40 60 80 100 0 5 10 15 20 Sl ab s C ra ck ed (% ) Pavement Age (years) Before Grinding After Grinding 0 0.05 0.1 0.15 0 5 10 15 20 Fa ul tin g (in ) Pavement Age (years) Before Grinding After Grinding 0 50 100 150 200 250 300 0 5 10 15 20 IR I ( in/ mi le) Pavement Age (years) Before Grinding After Grinding
80 performance-related Specifications for pavement preservation treatments to determine the SLE for each grinding treatment. The SLE was estimated by subtracting the before grinding âtime-to-thresholdâ from after grinding âtime-to-thresholdâ in years. Threshold values for faulting (0.157 inches or 4 mm), cracking (15%), and IRI (200 inch/ mile) were selected for illustration. A negative SLE reflects declining performance after grind- ing treatment and suggests the pavement will reach a threshold sooner than the do-nothing alternative. The relationships between SLE and DIRI and DDLI are shown in Figures 6-19 and 6-20, respectively. The relationship between a candidate AQC and expected performance can be used to estimate rational pay adjustments based on the expected life extension. (a) IRI (b) DLI 0 1 2 3 4 5 6 7 8 9 10 Ch an ge in sl ab s c ra ck ed (% ) -6 -4 -2 0 2 4 6 8 Change in DLI (10-2 in.) '+': reduction in DLI '-': increase in DLI 0 1 2 3 4 5 6 7 8 9 10 -100 -80 -60 -40 -20 0 20 40 60 80 100 120 Ch an ge in sl ab s c ra ck ed (% ) Change in IRI (in/mi) Observed data Non-linear Model '+': reduction in IRI '-': increase in IRI Figure 6-17. Relationship between AQCs and 20-yr cracking performance. (a) IRI (b) DLI y = 8Ã10-5x - 0.0006 RÂ² = 0.8096 -0.012 -0.010 -0.008 -0.006 -0.004 -0.002 0.000 0.002 0.004 0.006 0.008 0.010 -100 -50 0 50 100 Fa ul tin g (in ) Change in IRI (in/mi) '+': reduction in IRI '-': increase in IRI y = 0.0014x + 7Ã10-5 RÂ² = 0.7029 -0.012 -0.010 -0.008 -0.006 -0.004 -0.002 0.000 0.002 0.004 0.006 0.008 0.010 -6 -4 -2 0 2 4 6 8 Fa ul tin g (in ) Change in DLI (10-2 in.) '+': reduction in DLI '-' : increase in DLI Figure 6-16. Relationship between AQCs and 20-yr faulting performance.
examples 81 (a) IRI (b) DLI y = 0.0784x - 0.3562 RÂ² = 0.8997 -8 -6 -4 -2 0 2 4 6 8 10 -100 -50 0 50 100 150 IR I (i n/ m i) Change in IRI (in/mi) '+': reduction in IRI '-': increase in IRI y = 1.3403x + 0.1813 RÂ² = 0.8295 -8 -6 -4 -2 0 2 4 6 8 10 -6 -4 -2 0 2 4 6 8 IR I (i n/ m i) Change in DLI (10 -2 in.) '+': reduction in DLI '-': increase in DLI Figure 6-18. Relationship between AQCs and 20-yr surface roughness performance. IRI (in/mi) SL E (ye ars ) -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 -150 -100 -50 0 50 100 150 Faulting SLE Cracking SLE IRI SLE '+': reduction in IRI '-': increase in IRI Figure 6-19. DIRI due to grinding versus predicted DSLE for different AQCs. 4. Determine Thresholds and Limits for AQC There is no single correct method for establishing specification limits. Furthermore, there is a distinct difference between the limits of AQC and quality measures. The following steps were used to establish limits for AQC and quality measures. 1. Determine AQC-performance relationships. These relationships have been discussed in the previous section. Both IRI and DLI can be used to predict the immediate effect of diamond grinding on surface roughness. Because IRI and DLI correlate well, IRI will be selected as the primary AQC for this example.
82 performance-related Specifications for pavement preservation treatments 2. Set specification limits. A synthesis of current practice (Merritt et al. 2015) surveyed 22 states regarding localized roughness provisions for IRI-based specifications. The survey found that 16 states use an IRI range of 80 to 200 inch/mile for determining pay adjustments (Merritt et al. 2015). An NCHRP study for determining quality adjustment pay factors for pavements suggested 65 to 95 inch/mile as an acceptable range of IRI for newly constructed pavements, with 65 inch/mile considered to be superior ride quality that should provide an incentive (National Academies of Sciences, Engineering, and Medicine 2012). This study also suggested 75 to 120 inch/mile as a level of roughness that requires corrective action. Based on these findings, the research team has determined that post-grinding IRI of 90 inch/mile could be required to provide adequate smoothness. Therefore, a one-sided upper specification limit of 90 inch/mile was adopted and used for evaluating quality measures, pay adjustments, and risks. 3. Decide on the quality measure. The recommended quality measure, which is often used in current statistical quality control in highway construction is PWL (Burati et al. 2003, National Academies of Sciences, Engineering, and Medicine 2012). An example for determining PWL is described later. 4. Define AQL material. PWL is used as a quality measure in pavement construction practices. However, in this case, a modified version of PWL can be used to reflect a change in pavement quality measure due to diamond grinding, i.e., DPWL (PWL after - PWL before treatment). The procedure to obtain this DPWL value is described below. 5. Define RQL material. The RQL is a subjective decision made by the agency or party setting the specification limits. The DPWL value that can be used as the RQL can be obtained from the example. A lot at RQL will receive a reduced pay factor corresponding to the level of quality; a lot may be rejected if DPWL is at or below the RQL. Summary: â¢ The AQC selected for developing diamond grinding PRS is IRI. â¢ Given that IRI is a measure of roughness, it theoretically should not have a negative con- sequence for being âtoo smooth.â Therefore, only an upper limit of IRI of 90 inch/mile is selected to represent adequate pavement smoothness. â¢ The DPWL is selected as a quality measure. Figure 6-20. DDLI due to grinding versus predicted DSLE for different AQCs. -4.0 -2.0 0.0 2.0 4.0 6.0 8.0 -6 -4 -2 0 2 4 6 8 SL E (ye ars ) DLI (10-2 in) Faulting SLE Cracking SLE IRI SLE '+': reduction in DLI '-': increase in DLI
examples 83 â¢ The DPWL value that can be used as the AQL can be obtained as shown in the example. â¢ The DPWL value that can be used as the RQL can be obtained as shown in the example. 5. Specify Test Methods to Measure AQC The existing well-established standards for measuring and evaluating surface roughness are rec- ommended. Most highway agencies use lightweight profilers for measuring profile-based specifi- cations (Merritt et al. 2015). In comparison to high-speed mounted profilers, lightweight profilers weigh significantly less and have a more manageable, lower operating speed which is ideal for operating in constrained conditions and along shorter sections of pavement. Lightweight profil- ers are limited in that most are set up to only measure a single wheel path and require at least two carefully coordinated runs to obtain a complete set of profile data for one lane. The FHWA Highway Performance Monitoring System (HPMS) field manual references AASHTO PP 37-04 (now RO 43-13) and ASTM E-950 for procedures to collect IRI data (FHWA 2014). AASHTO RO 43-13 (Standard Practice for Quantifying Roughness of Pavements) details the estimation of IRI with the use of a longitudinal profile index measured in accordance with ASTM E-950 (Standard Test Method for Measuring the Longitudinal Profile of Traveled Sur- faces with an Accelerometer Established Inertial Profiling Reference). In the HPMS field manual, IRI is reported in units of either m/km or inch/mile. These standards should be consulted by agencies and contractors to ensure appropriate procedures are followed when collecting rough- ness data in the field. 6. Establish a Sampling and Measurement Plan As previously mentioned, the risks associated with incorrectly accepting or rejecting a lot are related to the sample size. The procedure outlined in Chapter 5 was followed to develop guide- lines for a sampling and measurement plan for diamond grinding: 1. Determine which party performs acceptance testing. The parties (contractor and agency) involved in the project must agree upon the duties of performing acceptance testing. 2. Determine the type of acceptance plan to be used. A variable acceptance plan is best suited for measuring the magnitude of IRI. Construction and sampling variability can affect surface smoothness. The variable acceptance plan can measure the variability and determine a quality measure based on statistical parameters. 3. Develop verification sampling and testing procedures. Verification sampling is standard procedure and used to verify the accuracy of the acceptance test results. The decision to use split or independent sampling is unique to the goals of the agency. For this example, it is assumed that an agency or a third party will measure the surface profile for the entire project length. However, the speed and lateral and longitudinal reference points should match with the acceptance testing. In practice, it is appropriate that the agencyâs verification test methods are used solely for verification and that acceptance methods proposed by the contractor must first be compared to the results of agency verification testing. 4. Select the appropriate verification sampling frequency. As discussed before, the verification sampling frequency of the agency should be approximately 10% of the acceptance sampling rate of the contractor. In practice, verification testing frequency is selected based on economic, rather than statistical, reasons. This decision must be agreed upon by agency and contractor, and it is assumed that the procedure is already established for the purposes of this example. 5. Determine lot size and sample size. The evaluation of pavement surface roughness involves the measurement of longitudinal profiles. Therefore, lots and sublots should logically be defined as segmented lengths of a project. Based on a survey of highway practice, most agencies report pavement segment lengths in 0.1-mile (500-ft) increments for IRI-based specifica- tions (Merritt et al. 2015). For this example, an LTPP pavement section (500 feet long) can be defined as the lot. A sublot is defined as a pavement subsection with shorter segment length
84 performance-related Specifications for pavement preservation treatments within a lot. Although the risks associated with sampling will depend on sample size, the profile signal analysis showed poor sensitivity to the DLI (which was designed to capture the dynamic load of truck traffic) when profile segment lengths are less than 100 ft. Given that DLI is correlated with IRI, a sublot length of 100 feet was chosen. Thus, each of the pavement sections evaluated in Table 6-8 are considered as a single lot from a larger project. Each lot was subdivided into sublots of approximately 100-ft-long segments, and the IRI of each sublot was considered as a sample. This resulted in a sample size of five for each pavement section. The PWL estimation is based on these lot and sample sizes. 7. Select and Evaluate Quality Measurement Methods As discussed in Chapter 5, the quality measure will be DPWL. Using the procedure outlined above, the PWL before and after grinding was calculated for each lot. The quality measure DPWL was developed to represent the change in quality and is calculated using Equation 6-1. (6-1)After BeforePWL PWL IRI PWL IRI( ) ( )â = â This DPWL shows how much the construction quality has statistically demonstrated a shift toward or away from acceptable quality. A positive DPWL value indicates improvement in AQC due to grinding; a negative value indicates a decline in quality. An example of the PWL calcula- tion using the procedure outlined in Chapter 4 is presented for pavement Section 42-3044. Fig- ure 6-21 presents the sampled lot measurements and Table 6-6 shows the statistical parameters obtained for this lot. Figure 6-21. IRI before and after grinding for Section 42-3044. Table 6-6. Sublot statistical measurements for Section 42-3044. Sublot length (ft) IRI (in/mi) Before IRI (in/mi) After 0-100 105 55 100-200 133 71 200-300 190 86 300-400 227 87 400-500 133 64 Mean 157.5 72.7 Std. Dev. 49.6 14.1
examples 85 The PWL value of a lot is calculated using the quality index (Q value) of the specification limits. The Q-statistics are calculated using Equations 6-2 and 6-3. (6-2)Q x LSL s L = â (6-3)Q USL x s U = â where QL = quality index for the lower specification limit. QU = quality index for the upper specification limit. LSL = lower specification limit. USL = upper specification limit. xâ = the sample mean for the lot. s = the sample standard deviation for the lot. Given that IRI is used as the AQC in this example, only an upper specification limit of 90 inch/ mile is used. The quality indices before and after grinding are calculated as follows: 90 157 50 1.3 90 73 14 1.2 Q Q U before U after = â = â = â = Determine PWL from FHWA reference tables (Weed 1996) and interpolation: PD PWL PD PWL PD before before after after = = â = â = = = 93.2 100 100 93.2 6.8 89.9 10.1 Determine DPWL using Equation 6-1: 89.9 6.8 83.1after beforePWL PWL IRI PWL IRI( ) ( )â = â = â = This means that, approximately, an additional 83% of the Section 42-3044 shifted into an acceptable quality than before grinding based on the upper specification limit of IRI. This analy- sis was done for all the pavement sections and the results are summarized in Table 6-7. The PWL obtained from each âlotâ is used to develop pay factors unique to the IRI. Based on the AQC-performance relationships previously established, the PWL can be related to expected performance in terms of SLE that can be used to develop a pay equation that relates PWL levels to expected pay. 8. Develop Pay Adjustment Factors for Incentives and Disincentives Pay adjustment factors are necessary for variable acceptance plans in developing PRS. There- fore, a variable acceptance plan is selected for this example because samples exhibit a wide range of IRI, which cannot be rejected on a pass/fail basis. Given the established upper specification limit of 90 inch/mile, a set of samples that exhibit smoother (i.e., below 90 inch/mile) segments
86 performance-related Specifications for pavement preservation treatments is not only acceptable but exceeds the desired quality. Similarly, a measured roughness greater than 90 inch/mile should not be rejected if the roughness level does not substantially exceed the target quality, but it would not deserve full pay. Using the pay equation relationships, the rel- evant EP and OC curves were developed for assigning pay factors to different levels of acceptable and rejectable quality while minimizing the expected risks. 1. Predict pavement performance as a function of levels of quality. Existing PRS do not justify pay factors for a unique AQC-performance relationship, but engineering judgment and expe- rience are used to relate PWL to SLE. However, a relationship between PWL and pavement performance, which is substantiated by IRI-performance relationships, was established. Sub- sequently, these results were used to develop a relationship between SLE for diamond grind- ing performance measures and DPWL. Figure 6-22 shows the relationship between predicted SLE (based on predicted fatigue cracking, faulting, and IRI) and DPWL for all the pavement sections. The results show a general increase in the SLE with increasing DPWL. This suggests that, if a grinding treatment can smooth a large percentage of a lot (a pavement section) within the specification limits, the pavement should be expected to have an improved perfor- mance. These quality relationships were used to develop pay equations by using SLE based on cracking, faulting, and IRI. Equations 6-4 through 6-6 show the relationships between predicted SLE and varying levels of DPWL. SLE e Cracking PWL = + ( )â â + 15 1 (6-4) 0.0783 6.958 SLE PWLFaulting ( )= Ã â â0.034 0.4588 (6-5) SLE e IRI PWL = + ( )â â + 20 1 (6-6) 0.0714 4.3 where SLECracking, SLEFaulting, SLEIRI = service life extensions due to cracking, faulting, and IRI, respectively. DPWL = difference in PWL before and after treatment 2. Convert the expected performance into pay adjustment. The pay factors for varying life extensions can be calculated by using Equation 6-7, and they correspond to the estimated No Section ID PWLbefore PWLafter PWL 1 6-3010 54.3 100.0 45.7 2 13-3017 59.2 100.0 40.8 3 27-4050 56.2 100.0 43.8 4 42-3044 6.8 89.9 83.1 5 46-3010 0.0 100.0 100.0 6 55-3009 0.0 100.0 100.0 7 4-7614 47.1 100.0 52.9 8 16-3017 0.0 32.1 32.1 9 49-C431 6.4 15.2 8.7 10 8-3032 12.4 11.7 -0.7 11 27-3009 57.2 0.0 -57.2 12 38-3006 100.0 100.0 0.0 13 20-3015 0.0 6.1 6.1 14 39-9006 15.7 95.5 79.8 Table 6-7. DPWL due to grinding based on IRI.
examples 87 Figure 6-22. DPWL due to grinding versus predicted SLE for different performance measures. (a) Cracking (b) Faulting (c) IRI 0 2 4 6 8 10 12 14 16 -100 -50 0 50 100 SL E (ye ars ) PWL -4 -2 0 2 4 6 -100 -50 0 50 100 SL E (ye ars ) PWL 0 4 8 12 16 20 24 28 32 36 40 -100 -50 0 50 100 SL E (ye ars ) PWL
88 performance-related Specifications for pavement preservation treatments change in quality ranging from 0 to 100 DPWL because SLE is a function of DPWL [see Equa- tions 6-4 through 6-6]. PF C R R R D E O ( ) = â â1 (6-7) where PF = pay adjustment factor for treatment (same units as C) C = present total cost of treatment, use C=1 for PF D = design life of pavement or initial treatment E = expected life of pavement or treatment O = expected life of successive treatments R = (1 + INF) / (1+ INT) INF = long-term annual inflation rate in decimal form INT = long-term annual interest rate in decimal form Relationships were developed between pay factors and SLEs for cracking, faulting, and IRI, as shown in Figure 6-22. The curves in these figures are called expected pay (EP) curves and were used to (1) refine the levels of acceptable and rejectable quality, (2) develop OC curves to assess the associated a and b risk, and (3) ensure that payment factors are awarded in accordance with the level of quality achieved. Equations 6-8 through 6-10 are the pay adjustment plans for SLE due to diamond grinding for the predicted performance measures, i.e., fatigue cracking, faulting, and IRI, respectively. Tables 6-11 through 6-13 show the cor- responding pay factors for various ranges of quality for each performance measure affected by the diamond grinding treatment. PayAdj PWLcracking( ) ( )= â% 0.0000006 (6-8)4.21322 PayAdj PWLfaulting( ) ( )= â â% 1.225 8.1354 (6-9) % 200 1 exp 0.075 3.5 (6-10)PayAdj PWLIRI [ ]( ) ( )= + â â + 3. Adjust the AQL, RQL, and pay relationships to minimize risk. As discussed in the determi- nation of AQC limits, the AQL and RQL need to be established. For establishing AQL, the EP curves must be evaluated such that the payment plan awards 100% pay at AQL while an incentive can be given if the quality of work is above AQL. Tables 6-8 through 6-10 show the pay factors generated from the EP curves (see Figure 6-23). As seen in the tables, the AQL may be chosen around DPWL = 89.4 for cracking, 88 for fault- ing, and 47 for IRI, to ensure a contractor is not awarded bonus pay for AQL work. For establishing RQL, the EP curves can be used to determine the level of performance (in terms of life extension) that is deemed unacceptable and should result in reduced pay. This deci- sion is typically made to meet the needs of the agency to ensure that the pavement performs up to the established standards. For instance, in the cracking EP model shown in Table 6-11, an agency may decide that a life extension of less than 1 year is undesirable. Therefore, the RQL will be set at DPWL = 55, and any lot produced at a quality level below that will receive no pay. Simultane- ously, the agency is also deciding that any quality between AQL of DPWL = 89.4 and that RQL
examples 89 Table 6-8. Summary of pay factor for cracking. PWL SLE (years) PF (%) 0 0.014 0.000 5 0.021 0.001 10 0.031 0.010 15 0.046 0.054 20 0.068 0.182 25 0.101 0.466 30 0.148 1.005 35 0.218 1.924 40 0.321 3.376 45 0.469 5.545 50 0.684 8.643 55 (RQL) 0.991 12.912 60 1.421 18.629 65 2.010 26.099 70 2.795 35.662 75 3.795 47.691 80 5.007 62.591 85 6.386 80.803 89.4 (AQL) 7.7 100 90 7.846 102.802 95 9.278 129.041 100 10.586 160.171 Table 6-9. Summary of pay factor for faulting. PWL SLE (years) PF (%) 0 -0.2 -8.1 5 -0.080 -1.998 10 0.075 4.127 15 0.229 10.251 20 0.384 16.376 25 0.538 22.501 30 0.693 28.626 35 0.847 34.751 40 (RQL) 1.002 40.876 45 1.156 47.001 50 1.311 53.125 55 1.465 59.250 60 1.620 65.375 65 1.774 71.500 70 1.929 77.625 75 2.083 83.750 80 2.238 89.875 85 2.392 96.000 88 (AQL) 2.5 100 90 2.547 102.124 95 2.701 108.237 100 2.855 114.362
90 performance-related Specifications for pavement preservation treatments of DPWL = 55 will be accepted, but will receive reduced pay or a disincentive. This logic can be similarly applied to the EP models for faulting and IRI (see Table 6-11). Under the assumption that a life extension of less than 1 year is undesirable, the RQL for faulting and IRI performance are DPWL = 40 and 19, respectively. Table 6-12 summarizes the final AQL and RQL for diamond grinding based on cracking, faulting, and IRI. The OC curves were developed to assess the risk of receiving a payment that correctly corresponds to the level of quality sampled. These OC curves are shown in Figures 6-24 through 6-26. When evaluating the risks associated with receiving appropriate pay for predicted cracking, OC curves can be examined. The level of quality produced by a contractor as indicated on the x-axis can be matched with the OC curve with desired quality to evaluate the probability of receiving a pay factor which corresponds to the desired quality. In the case of predicted cracking, AQL is DPWL = 89.4. If a contractor produces AQL in the field, then the quality level of AQL must be matched with the OC curve at AQL. Figure 6-24 indicates that the pay adjustment plan that will award a pay factor greater than 1 has a probability of 50% of all lots sampled. This sug- gests that the contractor will receive a pay greater than 100% (pay for above AQL) half the time and receive a pay less than 100% (pay for below AQL) half the time. Given that several lots will be sampled for quality, this averages to 100% pay throughout the project, which is characteristic of an unbiased and fair adjustment plan to both the agency and the contractor. This also incentivizes the contractor to consistently aim for above AQL quality to offset the probability of performing below AQL and receive bonus pay. Also, the greater the sample size, the higher the probability of receiving pay greater than 100% if the produced quality is above AQL. Similarly, the prob- ability of receiving pay greater than 100% is lower if the produced quality is less than AQL. Table 6-10. Summary of pay factor for roughness. PWL SLE (years) PF (%) 0 0.27 5.87 5 0.38 8.42 10 0.54 12.03 15 0.76 17.03 18.9 (RQL) 1.0 22.29 20 1.07 23.86 25 1.50 32.92 30 2.07 44.57 35 2.84 58.87 40 3.82 75.54 45 5.04 93.80 46.7 (AQL) 5.5 100 50 6.50 112.47 55 8.16 130.31 60 9.92 146.24 65 11.69 159.66 70 13.36 170.41 75 14.83 178.68 80 16.08 184.84 85 17.09 189.33 90 17.87 192.54 95 18.46 194.81 100 18.90 196.40
examples 91 (a) Fatigue cracking (b) Faulting 0 20 40 60 80 100 120 0 20 40 60 80 100 Pa y ad jus tm en t (% ) PWL PF from SLE PF-PWL model -20 0 20 40 60 80 100 120 0 20 40 60 80 100 Pa y ad jus tm en t (% ) PWL PF from SLE PF-PWL model (c) IRI 0 50 100 150 200 250 300 0 20 40 60 80 100 Pa y ad jus tm en t (% ) PWL PF from SLE PF-PWL model Figure 6-23. Pay adjustment for grinding based on various performance measures. Quality characteristics1 Predicted cracking Predicted faulting Predicted IRI AQL 89.4 88 46.7 AQLSLE (years) 7.7 2.5 5.5 AQLPF (%) 100 100 100 RQL 55 40 18.9 RQLSLE (years) 0.9 1 1 RQLPF (%) 12.9 40.9 22.3 1Quality measure in units of âPWL Table 6-11. Summary of %PF determined from pay equation.
92 performance-related Specifications for pavement preservation treatments Figure 6-24. Predicted cracking OC curves. Figure 6-25. Predicted faulting OC curves. Figure 6-26. Predicted IRI OC curves.
examples 93 As mentioned before, an agency can set the sample size based on their resources and balancing the risk. Similar logic can be applied when evaluating the risks associated with receiving the appropri- ate pay factor for predicted faulting and roughness, shown in Figures 6-25 and 6-26, respectively. Table 6-12 summarizes the example PRS specifications for diamond grinding as evaluated for performance improvement due to predicted cracking, faulting, and IRI. Items Cracking Faulting Roughness AQC(s) IRI (in/mi) Lot size 0.1 mi Sample size 5 AQC threshold Upper specification limit = 90 in/mi Quality measure Quality thresholds AQL = 89.4 , RQL = 55 AQL = 88 , RQL = 40 AQL = 46.7 RQL = 18.9 Pay equation PF (%) = 6E- 7( )4.21322 PF (%) = 1.225( )- 8.1354 PF(%) = 200 / [1+exp(- 0.5( PWL PWL PWL PWL PWL PWL, PWL PWL PWL PWL)+3.5)] AQL pay factor 1.0 1.0 1.0 RQL pay factor 0.129 0.40 0.22 PF(A) at AQL 50% 50% 50% Table 6-12. Summary of parameters to develop diamond grinding PRS.