National Academies Press: OpenBook

Intelligent Soil Compaction Systems (2010)

Chapter: Chapter 8 - Case Study Evaluation of Specification Options

« Previous: Chapter 7 - Quality Assurance of Pavement Earthwork Using Roller-Integrated Continuous Compaction Control (Recommended Specification Options)
Page 119
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 119
Page 120
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 120
Page 121
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 121
Page 122
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 122
Page 123
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 123
Page 124
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 124
Page 125
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 125
Page 126
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 126
Page 127
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 127
Page 128
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 128
Page 129
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 129
Page 130
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 130
Page 131
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 131
Page 132
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 132
Page 133
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 133
Page 134
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 134
Page 135
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 135
Page 136
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 136
Page 137
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 137
Page 138
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 138
Page 139
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 139
Page 140
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 140
Page 141
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 141
Page 142
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 142
Page 143
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 143
Page 144
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 144
Page 145
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 145
Page 146
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 146
Page 147
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 147
Page 148
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 148
Page 149
Suggested Citation:"Chapter 8 - Case Study Evaluation of Specification Options." National Academies of Sciences, Engineering, and Medicine. 2010. Intelligent Soil Compaction Systems. Washington, DC: The National Academies Press. doi: 10.17226/22922.
×
Page 149

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.

  This chapter presents six case studies performed to illustrate implementation of the recommended quality assurance (QA) specification options detailed in Chapter 7 (see Table 8.1). These case studies were derived from field tests in Colorado, Florida, North Carolina, and Minnesota. The case studies pre- sented here include granular subgrade (FL15, FL23, NC20), nongranular subgrade (MN10), granular subbase (CO34), and aggregate base materials (FL19). Additional case studies are presented in Appendix D. Multiple specification options were investigated as part of each case study to enable direct comparison of option strengths and limitations. Both success- ful and unsuccessful implementations are presented and dis- cussed. Throughout this chapter, standardized terms from the Chapter 7 specification options are italicized and capitalized. As summarized in Table 8.1, compaction QA acceptance testing was conducted by the project QA agents using cri- teria based on dry unit weight requirements (γ d -TV), mois- ture requirements (w-TV), and/or static proof rolling (PR). For some case studies, comparisons with existing options are made. Table 8.1 summarizes whether each test bed passed (P) or failed (F) the existing contract specifications, if applicable. c h a p t e r 8 Case study Evaluation of specification options When appropriate, MV-TVs were established based on these same contract requirements. 8.1 Case Study I—Granular Subbase (TB CO34) QA Specification Options 1, 2a, 2b, and 3a were evaluated on a 12-m (40-ft)-wide by approximately 300-m (1,000-ft)- long by 30-cm (12-in)-thick granular subbase Evaluation Sec- tion in Colorado (see Figure 8.1). The existing soil [A-4(3)] in this cut section was excavated. Four 30-cm (12-in)-thick lifts of granular subbase material (A-1-a) were placed and com- pacted jointly by the research team using the Dynapac intelli- gent compaction (IC) roller and by the contractor using typi- cal vibratory and static equipment. This case study pertains to the fourth subbase layer. Contract QA specifications included γ d -TV = 100% standard Proctor maximum dry unit weight [21.6 kN/m3 (137.5 pcf) uncorrected for rock content] and w-TV = 3 ± 1% if the rock content (material retained on a #4 sieve) was less than 50%. Rock content exceeded 50% in this area; therefore, project QA was based solely on a static proof table 8.1. Summary of case studies presented. Test Bed Soil (AASHTO) CCC-QA Options Evaluateda Contract QA Pass/Failb 1 2a 2b 3a 3b 3c γ d -TV w-TV PR CO34 A-1-a ¸ ¸ ¸ ¸ — — NRc NRc Pd FL15 A-3 ¸ ¸ ¸ — — — F NS NR FL19 A-1-b — — — ¸ — — P NS NS FL23 A-1-b ¸ ¸ ¸ ¸ ¸ — P NS NS NC20 A-1-b ¸ — — ¸ — — P NS Pd MN10 A-6(5) — — — — — ¸ F NS NS a ¸evaluated; — not evaluated. b Based on existing contract requirements for γ d , w, and static proof roll (when required) and assessed by project QA agents; NS = not specified, NR= not required. c Soil had >50% rock content; therefore, γ d -TV and w-TV were requirements not enforced. d Section passed, but remediation was required in small area(s).

0 roll. For this case study, the specification options were evalu- ated using the contract γ d -TV. The roadway alignment was transected by pipe crossings at approximately 300-m (1,000-ft) intervals and formed natural production earthwork sections. The contractor and QA agents informally treated these 300-m sections as inde- pendent QA units; therefore, the research team chose to use these natural sections as Evaluation Sections. The Calibration Area at the southern end of the Evaluation Section was be- lieved to be representative of the Evaluation Section and was convenient given the south to north flow of haul trucks and earthwork material through the site (see Figure 8.1a). Spot- test measurements were performed with the nuclear gauge (γ d , w), using a probe depth of 200 mm (8 in) and with the 300-mm (12-in)-diameter Zorn light weight deflectometer (LWD; E LWD-Z3 ). Three to four LWD and two nuclear gauge tests were performed across the width of the drum lane and averaged to represent a single measurement for acceptance analysis. Each reported roller MV was determined by averag- ing over 1 m (3 ft) in the direction of roller travel. The subbase material for the Calibration Area and Eval- uation Section subbase layer was placed in one 30-cm (12- in)-thick lift. Logistically, the contractor placed material via side-dump trucks from south to north, beginning first in the Calibration Area and then the Evaluation Section. An area about 2 m (6 ft) wide on the east side was not filled to allow for some shoulder work. While the research team performed the Option 3a correlation analysis in the Calibration Area, the contractor placed and began compaction work in the Evalu- ation Section. After calibration was completed (about three hours), the research team compacted the remainder of the Evaluation Section according to the pass sequence summa- rized in Figure 8.1. Due to the contractor’s schedule, further compaction was not possible, and the QA agent performed a Figure 8.1. Overview of TB CO34 summarizing construction and compaction of the test bed and showing (a) Evaluation Section and Calibration Area, (b) spot testing with the nuclear density gauge in the Calibration Area, and (c) the A-1-a subbase material.

  static proof roll test after pass 8. The Evaluation Section passed the static proof roll test despite very small areas with high clay content being identified for remediation, which involved scraping the clay material out with a motor grader. Evalua- tion Section roller MV data from pass 8 (MV 8 ) and spot-test measurement data obtained immediately after pass 8 are used to evaluate some of the options. 8.1.1 acceptance Using Specification option 1 Acceptance for Option 1 is based on achieving specified dry unit weight or relative compaction target values (γ d -TV, RC-TV) in the weakest area(s) identified by the instrumented roller. The spot-test-based acceptance criteria should be de- termined by the appropriate agency and should be based on and modified as necessary per agency practice. Figure 8.2 presents the pass 8 MV data (MV 8 ) together with spot-test measurements performed immediately after pass 8. Three weakest areas were identified for nuclear gauge and LWD testing, as illustrated in Figure 8.2. All six spot-test measure- ments performed in these weakest areas failed to meet the γ d -TV = 21.6 kN/m3 (137.5 pcf) and RC-TV = 100% specified for this project (see Table 8.2). It can be argued, however, that requiring 100% relative compaction in the weakest areas is more stringent than the existing Colorado approach where test locations are randomly selected. For example, four of the six weakest areas would meet an RC-TV = 95% criteria. For this reason, individual agencies must carefully consider exist- ing QA criteria and spot-testing practice when establishing acceptance criteria for Option 1. An important premise behind Option 1 is that a positive correlation exists between roller MVs and spot-test measure- ments (i.e., low roller MVs correspond to low γ d and high roller MVs correspond to high γ d ). Nine spot-test measure- ments were performed over a range of roller MVs to assess the relationship. Figure 8.2 shows that a reasonable correla- tion (R2 = 0.52) exists between γ d and the roller MV in the Evaluation Section and suggests that Option 1 is valid for this Evaluation Section. Conversely, an acceptable correlation was not evident from the roller MV versus E LWD-Z3 data and would therefore call into question the validity of its use in Option 1. The difficulty obtaining a suitable roller MV-E LWD-Z3 relation- ship is discussed later. 8.1.2 acceptance Using Specification option 2a Acceptance for Option 2a is based on the percentage change in the mean MV %∆µ MVi from consecutive passes over an Evaluation Section. Acceptance is met when %∆µ MVi is less than the target percentage change (%∆-TV = 5% for the case study presented here, similar to Austrian CCC spec- ifications). Table 8.3 provides %∆µ MVi for several passes over the Evaluation Section. The evaluation of %∆µ MVi was per- formed on pass pairs 2-4, 4-6, and 6-8 because the interme- diate passes (3, 5, 7) used higher-amplitude vibration (see pass sequence in Figure 8.1) or were static passes. Option 2a requires constant roller operating parameters for measure- ment passes. Per the %∆-TV = 5% criterion, acceptance was met after pass 6. 8.1.3 acceptance Using Specification option 2b Acceptance for Option 2b is based on the spatial percent- age change in MV i between consecutive passes %∆MV i rather than the percentage change in mean MV of the entire Evalu- ation Section per Option 2a. Acceptance is met when %∆MV i ≤ %∆-TV for a specified percentage of the Evaluation Section (%Area-TV). Per Section 7.5.3.1, the recommended %∆-TV is two times the σ%∆MV determined from repeatability testing (as described in Section 7.1.5.1) with a maximum %∆-TV = 20%. However, higher %∆-TV may be used at the discre- tion of the engineer of record. The σ%∆MV = 12.5% for the CMV D data, but the roller MV data were qualitatively repeat- able. Therefore, %∆-TV = 25% was set for this case study. The %Area-TV was set at 80%. Figure 8.3 presents %∆MV i data for passes 4, 5, and 6 (%∆MV 6 ) and 6, 7, and 8 (%∆MV 8 ), the last two Measure- ment Pass pairs over the Evaluation Section. The cumulative distributions of the %∆MV i data are also shown. To com- pute %∆MV i , nearest neighbor interpolation was used to transform the MV i data onto a fixed grid—the method cur- rently used by most roller manufacturers. The process of transforming GPS-based roller MV data to a fixed grid is nontrivial and an area of ongoing research (e.g., see Facas et al. 2010). The cumulative distributions illustrate that roller MVs decreased by as much as 50% in some areas and in- creased by as much as 75% in others. These large changes are likely a manifestation of deviation in roller path and highlight one of the challenges with the gridding of MV data records. Figure 8.3 shows that the %∆-TV = 25% was met for 81% of the Evaluation Section after pass 6, and there- fore acceptance is met according to Option 2b. It should be noted that although the theory behind Option 2b is sound, it is limited practically by the difficulties associated with ac- curately transforming the MV data onto a fixed grid often with a y-scale resolution less than the 2-m-long drum length (see Chapter 3).

 Figure 8.2. Roller MV maps of the Calibration Area and Evaluation Section, MV spot-test measurement correla- tions, and cumulative distribution of MV data after pass 8.

  table 8.2. option 1: Spot-test Measurements from weakest areas. Weakest Area Location γ d (kN/m3) (pcf) Relative Compactiona 6-1b 21.0 (134) 97.2% 6-2 21.0 (134) 97.2% 7-1 21.0 (134) 97.2% 7-2 20.9 (134) 96.7% 9-1 20.0 (127) 92.6% 9-2 19.8 (126) 91.7% a Based on percentage of γ d -TV = 21.6 kN/m3 (138 pcf). b Reflects test 1 at position 6 in Figure 8.2. table 8.3. Summary of µMVi and %∆µMVi from tB Co34. Pass µ MVi %∆µ MVi 2 24.8 — 3 NAa — 4 26.9 6.9% 5 NAa — 6 28.1 4.1% 7 NAb — 8 28.6 1.8% a Pass was with high-amplitude vibration in reverse driving direction; MV data are not comparable to MV data from low- amplitude passes. Pass was static; no MV data are available. Figure 8.3. Percentage change in MV maps for passes 4, 5, and 6 (%∆MV6 ) and passes 6, 7, and 8 (%∆MV8 ) and cumulative distributions of percentage change showing percent of area meeting acceptance.

 8.1.4 acceptance Using Specification option 3a 8.1.4.1  Initial Calibration Option 3a requires calibration of roller MVs to spot-test measurements to determine an MV-TV. A Calibration Area was selected at the southern end of the Evaluation Section (Figure 8.1) and was compacted to provide low, medium, and full compaction states prior to measurement. Figure 8.2 shows MV data from a Measurement Pass over the Calibra- tion Area. Spot testing was performed at five or six locations per compaction state. The resulting regression relationships are shown in Figure 8.2. A suitable correlation was found be- tween CMV D and γ d (R2 = 0.52) but not between CMV D and E LWD-Z3 (R2 = 0.39). Based on the γ d -TV = 21.6 kN/m3 (138 pcf), the MV-TV = 48 (see Figure 8.2). Beyond the influence of measurement depth mismatch between roller MVs and spot-test measurements, the poor MV-E LWD-Z3 correlation is attributed to the high degree of local heterogeneity and the limitation of the LWD to reliably measure the stiffness of dry granular materials with rounded particles. The high degree of heterogeneity is illustrated by the spot-test measurements in Figure 8.4. The multiple data at each test location reflect the two moisture/density tests and three or four LWD tests performed across the drum lane as well as the range of roller MVs encountered in the 1 m (3 ft) of data averaged to represent a single point. Although γ d exhibits little variation across the drum lane [<0.2 kN/m3 (1.27 pcf)], the variation in E LWD-Z3 measured across the drum width (3 to 20 MPa) approaches the entire range from low to high compaction states shown in Figure 8.2. The limitation of the LWD to measure the stiffness of dry granular materials with rounded particles is illustrated in Figure 8.2. Spot-test measurements 10 and 13 to 16 were from areas of high roller MV and high γ d . However, the E LWD-Z3 values are often less than the results from the medium MV areas and in the case of point 9 similar to the results from the low-MV areas. 8.1.4.2  Assessment of Evaluation Section Acceptance for Option 3a is based on achieving the MV-TV over a specified percentage of the Evaluation Section (%Area- TV). In this case study, %Area-TV = 90% (similar to current German practice). Figure 8.2 shows MV 8 for the Evaluation Section. Pass 8 served as the final compaction pass and final Measurement Pass [A = 0.9 mm (0.035 in), v = 3.6 km/h (2.2 mph), f = 30 Hz]. The cumulative distribution of MV 8 data shows that only 3% of the Evaluation Section met the MV-TV; therefore, the Evaluation Section does not meet acceptance according to Option 3a. One critical requirement of Option 3a is that the Calibra- tion Area must be representative of the Evaluation Section. In Figure 8.4. Variation in Spot-Test Measurements across the drum width and in roller MV along 1 m (3 ft) in the driving direction.

  retrospect this is an inappropriate Calibration Area for the Evaluation Section. Figure 8.5 shows a comparison of CMV- γ d correlations from the Calibration Area and Evaluation Section, determined from the data presented in Figure 8.2. These correlations are clearly different. The Calibration Area is stiffer than the Evaluation Section (i.e., for a similar γ d the roller MVs in the Evaluation Section were on average 50% of those measured in the Calibration Area). The higher roller MVs in the Calibration Area are likely due to the stiffer sublift material. Test bed CO34 represented the fourth 30-cm (12- in) lift of subbase material placed in this Evaluation Section. The previous three lifts were also compacted by the research team using the same general boundaries for the Calibration Area and Evaluation Section. This previous work resulted in the Calibration Area receiving additional roller passes (i.e., for research purposes), resulting in a stiffer subsurface in the Calibration Area than the rest of the Evaluation Section. Acceptance of the Evaluation Section was reevaluated using the correlation from the Evaluation Section. Per the γ d -TV = 21.6 kN/m3 (137.5 pcf), the adjusted MV target value MV- TV adj = 30. The cumulative distribution of MV 8 data shows that 43% of the Evaluation Section met the adjusted MV-TV. Per Option 3a, this Evaluation Section would still not meet acceptance. 8.1.5 discussion The Evaluation Section met project QA acceptance based on a static proof roll after pass 8 and provides some insight into the CCC-based QA Specification Options. Acceptance per Option 1 was not met after pass 8 (the last pass) if the standard spot-test-based criteria are used (i.e., RC-TV = 100%). Using this criterion, Option 1 imposes more stringent requirements than current randomized spot-testing criteria. In the case study presented here, acceptance was met when the RC-TV was relaxed to 92%. The existence of a positive CMV D -γ d correlation indicates that the principle of Option 1 is valid. Acceptance per Options 2a and 2b was met after pass 6. While it is not known for sure if the Evaluation Sec- tion would have passed the static proof roll test after pass 6, these results indicate that Options 2a and 2b, as evaluated here, would have resulted in a similar level of quality as cur- rent practice. Implementation of Option 3a was unsuccess- ful because the Calibration Area was not representative of the Evaluation Section. An acceptable CMV-γ d correlation was found in the Calibration Area; however, this correlation was considerably different than that observed in the Evaluation Section. Implementation of specification options presented a number of challenges. Construction traffic, particularly haul trucks moving through the earthwork area, often forced less than ideal Rolling Patterns. Truck traffic made it difficult to create uninterrupted and repeatable path Evaluation Area roller MV maps. Developing the required correlations in the Calibration Area required that the haul trucks (delivering material to the Evaluation Section) drive around the active earthwork. Contractors commonly utilize haul truck traf- fic to compact soil, and mandating that they remain off the freshly placed soil represents a significant change to current practice. Further, the pace of the production earthwork place- ment and compaction—deemed typical—clearly limits the time the research team was able to spend in the Calibration Figure 8.5. Comparison of Calibration Area and Evaluation Section MV-γd regression relationships.

 Area. Including the time needed to construct the Calibration Area, the correlations were developed in approximately three to four hours, although a time frame of one to two hours would be more consistent with production schedules. 8.2 Case Study II—Stabilized Granular Subgrade (TB FL15) Specification Options 1, 2a, and 2b were evaluated on a 12-m (40-ft)-wide by 60-m (200-ft)-long by 30-cm (12- in)-thick ash-stabilized sand subgrade Evaluation Section in Florida (see Figure 8.6). Due to site limitations, this Evalua- tion Section is smaller than typical. The stabilized subgrade consisted of 23 cm (9 in) of granular subgrade material (A-3) mixed with 7 cm (3 in) of bed ash (Figure 8.6c). Compac- tion was performed solely by the research team with the Sakai CCC roller and the pass sequence depicted in Figure 8.6. The case study presented here utilizes the k s-CSM MV computed from independent roller instrumentation (see Section 3.4). Nuclear gauge testing was conducted with a 200-mm (8-in) probe depth (γ d , w), and LWD testing was conducted with a 300-mm (12-in)-plate-diameter Prima device (E LWD-P3 ). Five LWD and three nuclear gauge measurements were performed across the width of the drum lane and averaged to represent γ d and E LWD-P3 . Each reported roller MV was determined by averaging over 1 m (3 ft) in the direction of roller travel. Contract QA specifications included γ d -TV = 98% modi- fied Proctor maximum dry unit weight [γ d -TV = 15.3 kN/m3 (97.4 pcf), RC-TV = 98%). There were no QA requirements for moisture. The specification options were evaluated using the contractual γ d -TV. 8.2.1 acceptance Using Specification option 1 Figure 8.7 shows the MV 11 data map and the results of spot-test measurements performed in the roller-identified weakest areas immediately after the pass 11 Measurement Pass [A = 0.9 mm (0.035 in), v = 4.0 km/h (2.5 mph), f = 25 Hz)]. Two roller-identified weakest areas were selected for Figure 8.6. Overview of TB FL15 showing (a) a Measurement Pass, (b) spot testing with the nuclear den- sity gauge, and (c) mixing the ash and soil.

  spot testing. The third area was not tested due to time con- straints. Table 8.4 shows that none of the six spots tested met the γ d -TV or RC-TV; therefore, the Evaluation Section does not meet acceptance according to Option 1. Further compac- tion would have been required to meet acceptance based on current practice. Since the test bed was being constructed for research purposes, further compaction was not immediately pursued. As with the results from TB CO34, it can be argued that requiring 98% relative compaction in the weakest areas is more stringent than the existing Florida approach in which test locations are randomly selected. For example, five of the six weakest areas would meet an RC-TV = 94% criteria. For this reason, individual agencies must carefully consider exist- ing QA criteria and spot-testing practice when establishing acceptance criteria for Option 1. In addition, the fact that this area received 11 roller passes and was still not fully com- pacted according to Option 1 calls into question this particu- lar γ d -TV. It is important to note that there is not a positive correla- tion between the roller MV and γ d (Figure 8.7). This implies that the areas with low roller MVs do not necessarily cor- respond to areas with low density. This result implies that Option 1 should be used with caution, and a higher spot-test measurement frequency may be desired. 8.2.2 acceptance Using Specification option 2a Table 8.5 provides %∆µ MVi for several consecutive passes (with constant operating parameters as listed above). Based on a %∆-TV = 5%, the Evaluation Section meets acceptance after pass 9. Recall that the Evaluation Section had not met ac- ceptance based on current practice after pass 11. This implies that Option 2a is less stringent than current practice for this Evaluation Section. Figure 8.7. Spatial MV data obtained during the final Measurement Pass over TB FL15 and regression relationships between roller MV and Spot-Test Measurements. table 8.4. option 1: Spot-test Measurements from weakest areas. Location γ d (kN/m3) Relative Compactiona 1-1 15.2 (96.7) 97.4% 1-2 15.1 (96.1) 96.8% 1-3 14.7 (93.6) 94.2% 2-1 14.8 (94.2) 94.9% 2-2 14.8 (94.2) 94.9% 2-3 14.0 (89.1) 89.7% aBased on maximum modified Proctor γ d = 15.6 kN/m3 (99.4 pcf).

 8.2.3 acceptance Using Specification option 2b The repeatability of k s-CSM at the Florida work site was de- termined to be 5% based on the procedure in Section 7.5.3.1; therefore, the Option 2b %∆-TV was set equal to 10% for this case study. As with Case Study I, a %Area-TV = 80% was used. Figure 8.8 shows %∆MV i for passes 9 and 11. The cu- mulative distributions of the %∆MV i are also shown. Nearest neighbor interpolation was used to transform the data onto a fixed grid. As illustrated, 92% of the Evaluation Section met the %∆-TV after pass 9. This percentage is greater than the %Area-TV, and therefore acceptance was met according to Option 2b after pass 9. 8.2.4 discussion As shown in Table 8.1, TB FL15 did not meet contract QA requirements for γ d at the end of compaction. The test bed was constructed for research purposes, and further compac- tion (beyond 12 passes) was not immediately pursued. The Evaluation Section also failed to meet acceptance based on Option 1; however, it did meet acceptance based on Options 2a and 2b. These results imply that Options 2a and 2b may be less stringent than the other options and current practice. To table 8.5. Summary of µMVi and %∆µMVi from tB FL15. Pass µ MVi %∆µ MVi 5 50.8 — 6 NAa — 7 56.8 11.7% 8 NA — 9 58.5 3.0% 10 NA — 11 59.9 2.4% a Pass was static; no MV data are available. Figure 8.8. Percentage change in MV maps for passes 7, 8, and 9 (%∆MV9 ) and passes 9, 10, and 11 (%∆MV11 ) and cumulative distributions of percentage change showing percentage of area.

  improve reliability, it may be desirable to implement Option 2a or 2b in conjunction with Option 1. This case study also illustrated the importance of assessing whether or not a positive correlation exists between the roller MV and the spot-test measurements used for QA. Option 1 should be used with caution, and more spot-test measure- ments may be needed when an acceptable (R2 > 0.50) positive correlation does not exist. 8.3 Case Study III—TB FL19 Aggregate Base Construction of TB FL19 involved placing and compact- ing a nominal 0.15-m (0.50-ft)-thick aggregate base layer (AASHTO: A-1-b) over a compacted stabilized subgrade layer using the Dynapac smooth drum IC roller (with the CMV D measurement system). Maximum dry unit weight (γ dmax ) and corresponding optimum moisture content (w opt ) per modi- fied Proctor method (AASHTO T-180) were 18.2 kN/m3 (116.0 pcf) and 11.5%, respectively. Project specifications required that the base material be compacted to an RC-TV = 98% [γ d -TV = 17.86 kN/m3 (113.70 pcf)]. No moisture con- trol was required by the project specifications. To evaluate the influence of underlying support conditions on compaction layer roller CMV D , the stabilized subgrade layer was mapped at nominal A = 1.1 mm (0.043 in), f = 30 Hz, and v = 4.5 km/h (2.8 mph) settings, prior to placing the base layer. The Evaluation Section had plan dimensions of about 9.1 m × 275 m (30 ft × 917 ft). An approximate area of 2.4 m × 30 m (8 ft × 100 ft) at the southern end of the Evaluation Section was selected as the Calibration Area for Option 3a. The compaction layer fill material was placed using dump trucks, spread using a dozer, and leveled using a motor grader to the desired elevation. Compaction of the Calibration Area is described below. The Evaluation Section was compacted in manual mode using constant operation settings at nominal A = 0.9 mm (0.035 in), f = 30 Hz, and v = 4.0 to 4.5 km/h (2.5 to 2.8 mph). The Evaluation Section received uneven compac- tion from construction traffic during placing, spreading, and leveling operations. With the aid of the CMV D map on the onboard computer, compaction efforts were focused in areas with low CMV D . Compaction passes over the Evaluation Sec- tion varied from 6 to 15 roller passes. The compaction process was terminated when field observations indicated that ad- ditional compaction was not improving the CMV D values. More details on the construction and testing procedures fol- lowed on this test bed are provided in Appendix D. 8.3.1 acceptance Using Specification option 3a 8.3.1.1   Calibration Area Testing and Analysis to  Establish MV-TVs After the material was placed and leveled, the Calibration Area was scarified to the bottom of the compaction layer using a motor grader. The test area was compacted with 13 roller passes at nominal A = 0.9 mm (0.035 in), f = 30 Hz, and v = 4.0 to 4.5 km/h (2.5 to 2.8 mph) settings. In situ moisture- density (using nuclear gauge), CBR (from DCP), and E LWD-Z3 point measurements were obtained at five locations across the test strip with three measurements across the drum width at each test location (Figure 8.9). Moisture-density tests were conducted using a probe penetration depth of 150 mm (6 in). Point measurements were obtained after 1, 2, 3, 4, 8, and 12 roller passes. Figure 8.10 shows spot-test measurements in compari- son with the CMV D . CMV D data are shown as solid lines and point measurement as discrete points. Results show that both the CMV D and in situ measurement values increased with in- creasing passes. Compaction growth for CMV D and spot-test measurements are presented in Figure 8.11, with a hyperbolic fit to the average data. The curves indicate that, on average, the CMV D , E LWD-Z3 , and γ d measurements generally increased up to 12 roller passes. At 12 passes the required γ d -TV (98% of T-180 γ dmax ) could not be reached as the material was, on average, 3% dry of w opt (see Figure 8.12). Figure 8.9. Calibration Area and Evaluation Section (TB FL19).

0 Figure 8.10. Roller MV (CMVD ) and in situ compaction measurement comparison for select roller passes from the calibration area. Linear regression relationships between spot-test measure- ments and CMV D were developed based on spatially near- est data (see Figure 8.13). Because the required γ d -TV could not be achieved, the target values were selected from the field compaction curves for illustration purposes. As indicated in Figure 8.11, E LWD-Z3 and γ d corresponding to 98% of the hyper- bolic curve asymptote were considered as QA-TVs [E LWD -TV = 90 MPa and γ d -TV = 16.80 kN/m3 (107 pcf)]. Using the in- verse regression approach and an 80% upper prediction limit, MV-TV = 49 was established, corresponding to the E LWD -TV and γ d -TV. As described in Chapter 7, use of prediction limits in selecting the MV-TV accounts for the uncertainty in the relationships and can be used to increase the confidence in the estimated target value. 8.3.1.2  Assessment of Evaluation Section Following compaction passes, the Evaluation Section was mapped using constant operation settings at nominal A = 0.9 mm (0.035 in), f = 30 Hz, and v = 4.0 to 4.5 km/h (2.5 to 2.8 mph). The CMV D map and the frequency distribution of CMV D from the Evaluation Section are presented in Figures 8.14 and 8.15, respectively. Using Specification Option 3a, approximately 32% of the area did not achieve the MV-TV, and therefore this Evaluation Section would not meet the ac- ceptance requirements. An important finding is that the cor- relations between CMV D and spot-test measurements in the Evaluation Section were not as defined as in the Calibration Area (Figure 8.14). As discussed below, the reason for scatter

  Figure 8.11. Compaction growth curves for roller MV (CMVD ) and spot-test measurements from the calibration strip (point measurements represent average of three measurements across the drum width). in the Evaluation Section relationships is attributed to hetero- geneity in the underlying layers. In situ spot tests (γ d and E LWD-Z3 ) were performed at 23 test locations across the Evaluation Area, of which 12 loca- tions consisted of areas with CMV D < MV-TV. In situ spot tests and corresponding CMV D comparison measurements (based on nearest point data) are compared with calibration regression relationships in Figure 8.14. Out of 23 spot tests performed in the Evaluation Section, eight E LWD-Z3 and one γ d test measurements failed to meet their respective QA-TVs. To further investigate the areas that met the γ d -TV but not the E LWD -TV requirement, DCP tests were conducted at five select test locations (8, 13, 19, 23, 25), extending into the sta- bilized subgrade layer to a depth of about 380 to 500 mm below the surface. CBR values were calculated from the DCP index values. To help interpret the CBR values, a target CBR = 21 was determined using a regression relationship obtained from calibration, corresponding to the MV-TV = 49 and 80% prediction interval (Figure 8.13). CBR profiles showed a zone (about 70 to 250 mm in thickness) of low CBR (< 21) in the stabilized subgrade layer at depths of about 190 to 300 mm (7.6 to 12.0 in) below the surface at test locations 8, 13, 15, 17, Figure 8.12. Comparison of laboratory Proctor com- paction curves and in situ moisture and dry unit weight measurements.

 Figure 8.13. Simple regression relationships between roller MV (CMVD ) and Spot-Test Measurements with 80% prediction limits to establish MV-TVs (each test point represents an average of three spot-test measurements across the drum width). 19, 23, and 25 (see Figure 8.14). Low CMV D and E LWD-Z3 values at these test locations were likely influenced by these “weak” zones, while the compaction layer γ d values were not, with the exception of test location 8. CBR profiles at two test locations (15 and 17) with CMV D > MV-TV are also shown in Figure 8.14 for reference. These tests were terminated in the stabi- lized subgrade due to low penetration values [< 5 mm (0.198 in)/blow, which is about CBR > 48]. This example demon- strates how soft underlying layers affect CMV D and how DCP testing can be used to identify these conditions and illustrates the importance of ensuring that the Calibration Area is repre- sentative of the Evaluation Section. 8.3.2   Implementation of Spatial Statistics for  Nonuniformity Analysis To demonstrate the approach of establishing nonunifor- mity criteria using spatial statistics, an experimental semi- variogram of CMV D from the Calibration Area was devel- oped (Figure 8.15). A Gaussian theoretical model was fit to the experimental semivariogram with sill = 90, range = 7 m, and nugget = 0. The theoretical model was selected based on comparatively better least squares fit than other models (e.g., spherical, exponential), modified Cressie goodness of fit (Clark & Harper 2002), and cross-validation process (see Vennapusa et al. 2009 for full description of model fitting process). For acceptance using the sill-based nonuniformity criteria, an Evaluation Section with length equivalent to the Calibration Area should have sill values lower than the target sill (i.e., sill from the Calibration Area). The CMV D data from the Evaluation Section were evalu- ated for nonuniformity by plotting semivariograms for four select areas of length equal to the length of the calibration test area (~30 m). Only two of the four selected areas showed sill values < 90. It is clear that the Evaluation Section showed significantly greater spatial nonuniformity compared to the Calibration Area. This greater nonuniformity is attributed primarily to nonuniform conditions in the underlying sta- bilized subgrade layer, as shown in the stabilized subgrade CMV D map (Figure 8.16) and previously verified with full- depth DCP index tests (Figure 8.14). In short, the nonuniformity criteria established from the Calibration Area could not be met in the Evaluation Section due to the influence of heterogeneous underlying layer prop- erties. Similar to the MV spot-test calibration analysis for Op- tion 3a, it should be noted that this nonuniformity criterion is limited to conditions with Evaluation Section having similar spatial heterogeneity in underlying support conditions with respect to the Calibration Area. In this case study it was not. However, more research is needed in relating nonuniformity to performance for a better understanding of what level of uniformity is desired and how field operations can be im- proved to control nonuniformity. 8.3.3   Comparison to Existing German, ISSMGE, and  Mn/DOT Specifications Acceptance Criteria Comparison of acceptance using Specification Option 3a to the German, ISSMGE, and Mn/DOT acceptance criteria is presented in Figure 8.17. The German specifications (see Sec- tion 2.3.1) are very similar to Specification Option 3a, except the use of prediction limits is not specified. Acceptance re- quires that 90% of roller MVs in the Evaluation Section must exceed the MV-TV. As shown in Figure 8.17, the MV-TV from the linear regression relationship is 38. A frequency distribu- tion plot indicates that 94% of the CMV D is greater than the

  Figure 8.14. CMVD map of the Evaluation Section and CBR profiles at select locations.

 MV-TV; therefore, the Evaluation Section does meet the Ger- man acceptance requirements. ISSMGE specifications are described in Section 2.3.2. In brief, the mean TV in the Evaluation Section ME should be greater than ME-TV, all MVs within 0.8 MIN-TV, and MAX- TV with coefficients of variation (COVs) < 20%. Procedures for establishing ME-TV (~44), MIN-TV (~33), and MAX-TV (~50) are described in Figure 8.17. The frequency distribution plot of CMV D data from the Evaluation Area indicates that ME = 56 and is greater than the ME-TV, but only about 63% of the data falls within 0.8 MIN-TV and MAX-TV. Therefore, the Evaluation Area does not meet acceptance. Further, the CMV D COV was 27%, which exceeds the acceptance limit of 20%. The nonuniformity criteria were thus not achieved fol- lowing the ISSMGE criteria, and reworking the Evaluation Section would be required. Mn/DOT specifications are described in Section 2.3.4. In brief, the acceptance criterion calls for 90% of the roller MVs in the Evaluation Section to be within 90% to 130% of the MV-TV determined from the Calibration Area. The MV-TV (~53) determined using Mn/DOT procedure is shown in Fig- ure 8.17. The CMV D frequency distribution plot for the Eval- uation Section showed that only 56% of the MVs are within the 90% to 130% range. Therefore, the Mn/DOT acceptance criterion was not achieved. 8.4 Case Study IV—TB FL23 Granular Subgrade A Calibration Area (see test bed FL23 in Appendix D) was constructed prior to compaction in the Evaluation Section. Specification Options 1 and 2b do not require calibration, whereas calibration is required for Specification Option 3. QA for this project was based on achieving a target γ d . For illustration purposes, an E LWD target value was also selected for assessment of the Evaluation Section for Specification Op- tion 3a. Compaction operations in the Calibration Area and Evaluation Section were performed with a Case IC roller with the k s measurement system at constant settings, with nominal A = 1.1 mm (0.043 in), f = 30 Hz, and v = 4.0 km/h (2.5 mph). Moisture and dry unit weight tests (using nuclear gauge) were conducted using a probe penetration depth of about 150 mm (6 in), and LWD tests were conducted by excavating the loose surface sand to a depth of 50 to 75 mm (2 to 3 in) below the surface (see White et al. 2007). Construction of TB FL23 involved scarifying the in-place embankment subgrade material (AASHTO: A-1-b) to a depth of 0.2 to 0.3 m (0.6 to 1.0 ft) below grade and compacting using the smooth drum Case IC roller. Maximum dry unit weight (γ dmax ) and corresponding optimum moisture content (w opt ) as determined by the standard Proctor method (AAS- HTO T-99) for the subgrade material were 15.9 kN/m3 (101.2 pcf) and 8.3%, respectively. Project specifications required that the subgrade material be compacted to ≥ 95% standard Proctor γ dmax [γ d -TV = 15.11 kN/m3 (96.2 pcf)]. Moisture control was not required by the project specifications. The Evaluation Section (FL23B) consisted of plan dimensions of about 12 m × 275 m (36 ft × 825 ft; see Figure 8.18). A lane located adjacent to the production test bed with material in a relatively loose and noncompacted state was arbitrarily se- lected for the Calibration Area (FL23A). The calibration test area was about 2.4 m × 60 m (7.2 ft × 180 ft). 8.4.1 acceptance Using Specification option 1 Acceptance for this option is based on achieving the γ d -TV in roller-identified weakest areas, identified from the k s data map of the Evaluation Section. In situ γ d point measurements were conducted after pass 6 and focused in the weakest areas, where k s < 23 MN/m. The tests results are summarized in a histogram plot shown in Figure 8.19. The relative compac- tion at these test measurements was between 97% and 109% of standard Proctor γ dmax . Therefore, the Evaluation Section met the acceptance requirements of Specification Option 1 as well as current practice where testing is performed at discrete random locations. 8.4.2 acceptance Using Specification option 2a This specification option requires evaluating the change in mean k s (∆µk s ) with successive passes over the Evalua- tion Section. The Evaluation Section was compacted with six roller passes with constant roller operation settings as de- Figure 8.15. Semivarigoram, univariate statistics, and distribution of CMVD from the Calibration Area.

  Figure 8.16. Uniformity-based acceptance criterion for Evaluation Section using spatial and univariate statistics.

 Figure 8.17. Evaluation of the Evaluation Section using Specification Option 3a in comparison to German, ISSMGE, and Mn/DOT acceptance criteria.

  scribed above. As shown in Table 8.6, the target ∆µk s of 5% was achieved after pass 3. Therefore, the Evaluation Section met the acceptance requirements of Specification Option 2a. No QA point measurements were obtained immediately after pass 3. QA measurements obtained after pass 6 are described in Figure 8.19, which indicated that all the measurements met the project’s γ d -TV per current practice. 8.4.3 acceptance Using Specification option 2b This specification option requires evaluating the spatial percentage change in k s (∆k s ) with successive passes over the Evaluation Section. To allow pass-to-pass comparison of the spatial k s data, the data were placed on a fixed grid using kriging (see Figure 8.20). To perform the kriging, an omni- directional exponential variogram model was used and the data were assumed to be stationary. Fitting weights were not employed, and anisotropy was not considered. Average ∆k s and percent area with ≤ 5% ∆k s between successive passes is presented in Table 8.6. The target requirement of ∆k s ≤ 5% over 90% of the area was met at pass 5. It is interesting to note that while the acceptance is met after pass 5, acceptance would not have been met after pass 6. This is possibly due to decompaction during pass 5 and subsequent recompaction during pass 6. Further, the kriged maps show two distinctly different compaction curves for areas 1 and 2, as highlighted in Figure 8.20, while the ∆k s curve was similar for the two areas. As discussed in the following section, these two areas contained different underlying layer conditions. 8.4.4 acceptance Using Specification option 3a 8.4.4.1  Initial Calibration This specification option requires calibration of k s to spot- test measurements from a Calibration Area. The selected Calibration Area was compacted with eight roller passes at constant machine operation settings. In situ γ d and E LWD point measurements were obtained at 10 locations across the Cali- bration Area. Tests were conducted after 1, 2, 4, and 8 roller passes. Linear regression relationships between γ d , E LWD , and k s were developed based on spatially nearest data, as presented in Figure 8.21. Using the inverse regression approach and an 80% prediction interval, MV-TV = 23 MN/m was established, corresponding to the γ d -TV. Moisture content was not a sig- nificant parameter in the regression analysis. To demonstrate the use of E LWD as an alternate QA tool, E LWD -TV = 28 MPa corresponding to the MV-TV and 80% prediction interval was established, as shown in Figure 8.21. Figure 8.18. Calibration Area and Evaluation Section for TB FL23. Figure 8.19. Histogram of QA test measurements in the Evaluation Section. table 8.6. Summary of ∆ks and ∆µks successive passes in the evaluation Section. Parameter 1 2 3 4 5 6 µk s 16 20 20 21 20 20 ∆µk s — 20 3 2 -2 2 Percent area ∆k s ≤ 5% — 4 69 80 94 80

 Figure 8.20. Kriged spatial maps of ks from pass 1 to pass 6 and corresponding percent increase in roller ks for each consecutive pass. 8.4.4.2  Assessment of Evaluation Section After compaction operations in the Evaluation Section, the area was mapped using the same constant roller operation settings as in the Calibration Area. The k s map of the Evalu- ation Section is shown in Figure 8.22. Acceptance using this option is based on achieving k s ≥ MV-TV for a set percentage of the Evaluation Section, in this case 90% (similar to German practice). Analysis of the Evaluation Section showed only 20% of roller k s > MV-TV; therefore, the Evaluation Section does not meet acceptance per Option 3a. In situ γ d and E LWD point measurements were obtained at 40 test locations, with 33 test locations in areas where k s < MV- TV. Spot-test measurements and corresponding k s (based on

  nearest point data) are compared with calibration relation- ships in Figure 8.22. All of the γ d measurements met the re- quirement with 97% to 109% of standard Proctor γ dmax (see Figure 8.19). E LWD at 12 of 40 locations failed to meet the E LWD - TV. Inspection of Figure 8.22 reveals that the E LWD data from the Evaluation Section generally follows the same trend as that from the Calibration Area and therefore for the selected Cali- bration Area appears appropriate for QA based on correlation to E LWD -TV. However, the γ d data from the Evaluation Section do not exhibit the same trend as that from the Calibration Area, and therefore the selected Calibration Area is not suit- able for QA based on correlation to γ d -TV. This illustrates a limitation of this calibration approach. Because of the lack of a suitable Calibration Area, it is not possible to comment here on how Option 3a compares to existing spot-test-based QA for this case. To further investigate the locations that did not meet the E LWD -TV, DCP tests were conducted at five select locations (10, 18, 24, 18, and 37) extending to about 500 mm below sur- face. DCP tests were also conducted at two locations where k s > MV-TV (2 and 5) for comparison. CBR values were calcu- lated from DCP index values following ASTM D6951. CBR profiles at the test locations are shown in Figure 8.22. At lo- cations where E LWD tests failed, CBR profiles showed lower underlying support conditions. CBR values at those locations were in the range of 5 to 12 from 100 to 400 mm (4 to 16 in) depth, which is low compared to locations 2 and 5, where the average CBR was > 25 from 100 to 400 mm (4 to 16 in) depth. 8.4.5 acceptance Using Specification option 3b This specification option requires evaluating the change in k s (∆k s ) between successive passes in Calibration Area to es- tablish an MV-TV. Results obtained from the Calibration Area are presented in Table 8.7, with average ∆k s and percent area with ≤ 5% ∆k s between successive passes. The compaction growth curve for k s , average ∆k s , and γ d point measurements with a hyperbolic curve fit to the average data is presented in Figure 8.23. Five passes were required to achieve the re- quirement of ∆k s ≤ 5% over 90% of the Calibration Area. The γ d -TV was met at 9 of 10 test locations after four passes. The average k s corresponding to pass 5 was MV-TV = 23 MN/m. Acceptance using this option is based on Evaluation Sec- tion k s ≥ MV-TV. A cumulative frequency distribution plot of Evaluation Section k s is presented in Figure 8.24 and shows about 80% of k s < MV-TV. Note that this option mandates that the Calibration Area be representative of the larger Eval- uation Section, which as shown earlier was not the case for this case study. Therefore, comparisons to existing practice should be treated with caution. Figure 8.21. Linear regression relationships between spot-test measurements and ks from the Calibration Area (TB FL23A).

0 Figure 8.22. ks map of the Evaluation Section, CBR profiles at select locations, and comparison of production test measurements with the established QA criteria.

  8.4.6 Comparison to german, ISSmge, and mn/doT Specifications acceptance Criteria Acceptance criteria using the existing German, ISSMGE (2005), and Mn/DOT (2007) specifications are presented in Figure 8.25. Correlation between k s and γ d measurements is not recommended in the German specifications and therefore was not evaluated. E LWD values were considered. As shown in Figure 8.25, 35% of the Evaluation Section met this require- ment of MV-TV ≥ 21; therefore, the Evaluation Section does not meet acceptance. Correlation between k s and γ d measurements is not rec- ommended in ISSMGE specifications; therefore, E LWD values were again considered. The Evaluation Section k s shows that the mean k s (ME = 20) is less than the ME-TV = 21, only 46% of the MVs are above the MIN-MV (k s = 20), and ap- proximately 15% of the data fall below the 0.8 MIN-TV (k s = 16). Therefore, none of the acceptance criteria are met, and further compaction/reworking would be required. The coef- ficient of variation (COV = 19%) of k s is within the accep- tance limit of 20%. table 8.7. ∆ks from successive passes in the Calibration Area. ∆k s Pass 1 to 2 Pass 2 to 3 Pass 3 to 4 Pass 4 to 5 Pass 5 to 6 Pass 6 to 7 Pass 7 to 8 Average 20 5 4 -2 1 2 5 Percent area with ≤5% 0 52 75 94 87 89 59 Figure 8.23. Compaction curves for ks , ∆ks , and spot-test measurements. Figure 8.24. Cumulative frequency distribution plot of Evaluation Section ks.

 Figure 8.25. Acceptance using German, ISSMGE, and Mn/DOT specifications.

  Using the Mn/DOT specification, MV-TV = 25 is estab- lished from the calibration data. The Evaluation Section k s showed that only about 29% of the area is within 90% to 130% of MV-TV range. Therefore, the Mn/DOT criterion was not satisfied. 8.5 Case Study V—Granular Subgrade (TB nC20) Specification Options 1 and 3a were evaluated on an 18- m (60-ft)-wide by 500-m (1,640-ft)-long granular subgrade Evaluation Section in North Carolina (see Figure 8.26). The silty sand subgrade material (A-2-4) was compacted by the contractor using typical vibratory equipment. The Calibra- tion Area was full width [18 m (60 ft)] and occupied the northernmost 100 m (300 ft) of the Evaluation Section. A single Measurement Pass (i.e., a proof roll pass, i = pr) was performed with the Sakai CCC roller. Spot-Test Measurements were collected with the balloon density tester to a depth of 200 m (8 in) in the Evaluation Section and with the nuclear density gauge to a probe depth of 200 mm (8 in) in the Cali- bration Area and a Keros LWD with a 300-mm (12-in)-plate diameter (E LWD-K3 ). The case study presented here utilizes the k s-CSM MV computed from independent roller instrumenta- tion (see Section 3.4). Seven LWD and three nuclear gauge measurements were performed across the width of the drum and averaged to represent single γ d and E LWD-K3 measurements for regression analysis (i.e., in the Calibration Area). Each re- ported roller MV was determined by averaging over 1 m (3 ft) in the direction of roller travel. The contract QA specifications included a γ d -TV = 17.7 kN/m3 (113 pcf) based on 95% standard Proctor maximum dry unit weight (RC-TV = 95%) and a static proof roll (see Figure 8.26b). There were no QA requirements for moisture. The project QA agent performed a static proof roll test over the Evaluation Section after option evaluation was completed. One isolated area with soft underlying clay layers did not pass proof roll and required remediation. The specification op- tions were evaluated using the contractual RC-TV and an E LWD -TV for Option 3a derived from the γ d -TV (for illustra- tion purposes). 8.5.1 acceptance Using Specification option 1 Figure 8.27 presents MV pr data from the Evaluation Section. Spot-test measurements (γ d ) were performed in the roller- identified weakest areas. Results indicated relative compac- tion values of 100% and 102%, both exceeding the RC-TV = 95%. Therefore, the Evaluation Section met acceptance ac- cording to Option 1. The detailed calibration performed to evaluate Option 3a discussed below indicated that a positive Figure 8.26. Overview of TB NC20 summarizing construction and compaction of the test bed and showing (a) a Measurement Pass and (b) static proof roll test.

 Figure 8.27. Roller MV maps of the Calibration Area and Evaluation Section, MV and spot-test measurement cor- relations and cumulative distribution of MV data.

  correlation did exist between the roller MV and γ d . However, this correlation was not acceptable (R2 = 0.21), and in such cases Option 1 should be used with caution, and a higher number of spot-test measurements may be desirable. A small area of the Evaluation Section that failed the static proof roll test (and therefore required remediation) is also shown in Figure 8.27. Due to time constraints, spot tests were not performed in this area. Conversely, the roller-identified weakest area passed the static proof roll test. The weak MV-γ d correlation is one possible reason the roller-identified weak- est area passed the static proof roll test while an area with slightly higher MVs failed. 8.5.2 acceptance Using Specification option 3a 8.5.2.1  Initial Calibration Calibration of roller MVs to spot-test measurements was based on areas of low, medium, and high stiffness found within the Measurement Pass [A = 0.9 mm (0.035 in), v = 4.0 km/h (2.5 mph), f = 30 Hz] of the previously compacted Calibration Area at the northern end of the Evaluation Sec- tion. This is an alternative approach to creating discrete low, medium, and high compaction states. Four to five spot-test measurements were performed in low, medium, and high MV areas (see Figure 8.27). The resulting regression relationships are presented in Figure 8.27. A suitable correlation was not found between the roller MV and γ d (R2 = 0.21), and therefore Option 3a cannot be implemented. A good correlation was evident between the roller MV and E LWD-K3 (R2 = 0.69). For il- lustration, an E LWD-K3 -TV = 22 MPa was established based on the relationship observed between γ d and E LWD-K3 (R2 = 0.32, graph not shown) and the γ d -TV = 17.7 kN/m3 (112.7 pcf). Based on E LWD-K3 -TV = 22 MPa, the resulting MV-TV = 40. The difficulty in obtaining a good correlation between the roller MV and γ d is likely due to sublift variability and mea- surement depth. For example, locations 14 and 15 exhibit significantly different roller MVs, yet their γ d are the same. The roller MV at location 14 reflects stiffer sublift material compared to soft sublift conditions at location 15. The E LWD-K3 from the 300-mm (12-in)-diameter LWD that measures to 30 to 60 cm (12 to 24 in) deep supports this (i.e., the E LWD-K3 at location 14 is twice the E LWD-K3 at location 15. 8.5.2.2  Assessment of Evaluation Section Acceptance for Option 3a is based on achieving the MV- TV over a specified percentage of the Evaluation Section (%Area-TV). In this case study, %Area-TV = 90%, similar to German practice, and as determined above, MV-TV = 40. The MV map and cumulative distribution of MV data are shown in Figure 8.27. The cumulative distribution shows that 93% of the Evaluation Section met the MV-TV. Accordingly, the Evaluation Section met acceptance according to Option 3a based on the E LWD-K3 -TV. 8.5.3 discussion TB NC20 met contract QA requirements for γ d and passed the static proof roll test (except in one small area). The Evalu- ation Section also met acceptance based on Option 1. How- ever, the relationship between roller MVs and γ d , while posi- tive, was shown to be weak. Therefore, Option 1 should be employed with caution. Option 3a implementation was not valid, as a suitable correlation (R2 = 0.5) was not found be- tween the roller MV and γ d . In such cases, Option 3a could not be used for QA. Other options such as 1 or 2a/2b or a combination of options would need to be pursued. As in Case Study I, implementation of specification op- tions presented a number of challenges. At the North Caro- lina site, haul trucks entered the Evaluation Section in reverse, deposited material, and then drove forward out of the area (see Figure 8.28). They did not drive through the project due to a geomembrane placed beneath the material. This often forced less than ideal roller pass patterns and created haz- ardous conditions for personnel performing spot-test mea- surements. Performing correlation studies in a designated full-width Calibration Area required a change in how the earthwork contractor placed material. To perform repeatable Measurement Passes in the Evaluation Section, the research team had to wait for the earthwork contractor to completely Figure 8.28. Placing material (top) and performing spot-test measurements (bottom) in an active Evalua- tion Section at the North Carolina work site.

 finish hauling and placing a section. In typical production compaction practice, roller compactors are used throughout the hauling, placing, and grading operation. Careful planning and cooperation between the contractor and QA agents are critical for successful implementation of CCC-based QA. 8.6 Case Study VI—Mn10 nongranular Subgrade A test bed constructed as part of the MnROAD field testing program is presented herein as an approach to establish MV- TV relating to laboratory M r values for Specification Option 3c and an approach to adjust the MV-TV for moisture con- tent. Roller MV (E vib ) and spot-test measurements obtained from this test bed are described in Chapter 6 (Section 6.4.1). In brief, the test bed was constructed with a layer of non- granular subgrade material [AASHTO: A-6(5)] underlain by a relatively stiff and homogeneous subgrade layer. The sub- grade layer was moisture conditioned w opt by dividing the test bed into three sections to approximately -3%, 0%, and +3% of standard Proctor [maximum dry unit weight and optimum moisture content as determined by the standard Proctor method were 16.95 kN/m3 (107.90 pcf) and 16.4%, respectively]. Compaction passes were performed using a pad foot Bomag IC roller, and in situ w-γ d point measurements were obtained in parallel with the compaction process. The analysis and results presented below are for the purpose of illustrating the calibration approach for Option 3c. As such, detailed comparisons to current practice (i.e., γ d spot-test- based QA) and other existing CCC specifications are not made and the assessment of an Evaluation Section is not pur- sued. Laboratory M r tests were conducted on reconstituted laboratory-compacted specimens to develop w-γ d -M r rela- tionships for the subgrade material. Test results are described in Section 6.4.1. 8.6.1 relationship Between roller mv and Spot-Test measurements Linear regression relationships between γ d and E vib were de- veloped based on spatially nearest data, as presented in Figure 8.29(a). The regression relationship produced R2 = 0.37. Mul- tiple regression analysis was performed to incorporate the in- fluence of moisture content in the relationship. The multiple regression relationship improved the correlations with R2 adj = 0.54 [Figure 8.29(b)]. 8.6.2 establishing mv-Tvs Based on laboratory-determined Mr values for Specification option 3c The multiple regression relationship developed from labo- ratory testing was used to predict M r values for the in situ w-γ d point measurements. Since the laboratory relationship is valid only for the range of w-γ d of the laboratory samples, the spot-test measurements close to the laboratory sample Figure 8.29. (a) Simple linear regression relationship between Evib and γd and (b) multiple regression relationship between predicted and measured Evib.

  values were selected. Comparison of predicted M r and E vib for the selected data is presented in Figure 8.30(a), which showed a good correlation with R2 = 0.52. For illustration purposes, a target M r (M r -TV) = 50 MPa was selected for establishing a roller MV-TV. Using the inverse regression approach and an 80% prediction interval, a roller MV-TV = 51 MPa was estab- lished for the M r -TV. Using the multiple regression relationship developed for roller E vib , E vib contours are plotted over a moisture–dry unit weight plot as presented in Figure 8.30(b) (see Section 6.4.1). Because the prediction equations were comprised of only lin- ear terms, the E vib contours are linear and parallel lines that decrease with increasing moisture. To select a target zone of acceptance, an area bounded by ±2 percent of w opt , 90% satu- ration curve, and the MV-TV is highlighted in Figure 8.30b. For acceptance using Specification Option 3c, the production E vib must be above the MV-TV and the QA test locations must have moisture contents within the zone of acceptance. 8.6.3 adjustment of mv-Tvs for moisture Content Data obtained from the above-described test bed is used to demonstrate the approach of adjusting the MV-TVs for moisture content. Using the multiple regression relation- ship developed for roller E vib , E vib contours are plotted over a moisture–dry unit weight plot as shown in Figure 8.31. The highlighted target zone is selected based on w opt ±2% and 95% standard Proctor γ dmax . High (MV-TV 1 ) and low (MV- TV 2 ) E vib corresponding to low (w 1 ) and high (w 2 ) moisture contents, respectively, are selected. The equation to calculate MV-TV adj is presented in Figure 8.31. This MV-TV adjust- ment can be used for Specification Options 3a, 3b, and 3c. Acceptance is based on QA test locations meeting the MV- TV adj . Areas with E vib > MV-TV 1 are considered as passed and those with E vib < MV-TV 2 are considered as failed and need to be reworked for further QA testing. 8.7 Conclusions The following conclusions can be drawn from the case studies presented in this chapter: • Specification Option 1 requires minimal changes to typical current QA practices. Rather than selecting random points for spot-test measurements, QA inspectors use the roller MV data map to identify the weakest area(s). Depending on the number of weakest areas identified, the frequency of spot testing may increase compared to current practice. According to Option 1, if the roller-identified weakest area(s) meet acceptance, the rest of the Evaluation Section passes by association. However, given that testing locations are informed rather than random, requiring the weakest zones to meet 100% of the preexisting QA-TV may be Figure 8.30. Relationship between laboratory (a) Mr and Evib and (b) target zone of acceptance for Option 3c.

 more stringent than current random selection spot test- ing. Reducing the QA-TV may be more appropriate. In addition, an assumption underlying Option 1 is that the roller MV and spot-test measurement have an acceptable, positive correlation. • Specification Options 2(a, b) and 3(a, b, c) require modi- fication to current QA practices in that spot-test measure- ments do not form the basis for QA. Rather, acceptance is granted based on roller MVs. In Option 2a, acceptance is granted when the percentage change in the mean roller MV from pass to pass falls below a preset threshold. In the case studies presented here, Option 2a appeared to be less stringent than current practice, and it may be desirable to implement it in conjunction with Option 1 to improve reliability. • Specification Option 2b uses the percentage change in spa- tial roller MV data as the basis for QA. One challenge asso- ciated with Option 2b is that the method of transforming roller MV data onto a fixed grid to allow spatial compari- son is not trivial and reliable; proven methods do not yet exist. Based on the case studies here, Option 2b appears to be more stringent that current QA practices. • One major challenge to successfully implementing Speci- fication Option 3(a, b, c) is ensuring that the Calibration Area is representative of the Evaluation Section. Although using a roller MV data map of the Evaluation Section can aid in selecting an appropriate Calibration Area, this can be logistically challenging on a busy job site. • Option 3(a, b, c) requires a significant initial investment of time and careful, detailed analysis. This analysis is more complex than that currently required for QA purposes, and it is easy to make errors. Accordingly, QA inspectors will need careful training to ensure they are familiar with both the roller MV systems and the analysis required for the various options. • Construction traffic poses a challenge to implanting CCC- based QA. All of the options require careful, repeatable roll- ing patterns. However, construction traffic—particularly haul trucks moving through the earthwork area—often forces less than ideal roller pass patterns. Truck traffic often makes it difficult to create uninterrupted and repeatable Evaluation Area roller MV maps. • Developing the required correlations (e.g., for Option 3a) necessitates that haul trucks remain outside the Cali- bration Area once material has been placed and spot- test measurements are being performed. However, it is common for contractors to utilize haul truck traffic to compact soil, and therefore truck drivers are accustomed to driving through an earthwork area. Similarly, it is not uncommon for haul trucks to enter the Evaluation Section in reverse, deposit their material, and then drive forward out of the area (see Figure 8.28). This often forces less Figure 8.31. Moisture correction for Evib target values.

  than ideal Rolling Patterns and creates hazardous condi- tions for personnel performing spot-test measurements. Performing correlation studies in a designated full-width Calibration Area requires a change in how the earthwork contractor places material. To perform repeatable Mea- surement Passes in the Evaluation Section, the research team had to wait for the earthwork contractor to com- pletely finish hauling and placing a section. In typical production compaction practice, roller compactors are used throughout the hauling, placing, and grading op- eration. Careful planning and cooperation between the contractor and QA agents is critical for successful imple- mentation of CCC-based QA. • The pace of the production earthwork placement and com- paction frequently limited the time the research team was able to spend in the Calibration Area. Including the time needed to construct the Calibration Area, the correlations were developed in approximately 3 to 4 hours, although a time frame of 1 to 2 hours or less would be more consistent with production schedules.

Next: Chapter 9 - Conclusions »
Intelligent Soil Compaction Systems Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s National Cooperative Highway Research Program (NCHRP) Report 676: Intelligent Soil Compaction Systems explores intelligent compaction, a new method of achieving and documenting compaction requirements. Intelligent compaction uses continuous compaction-roller vibration monitoring to assess mechanistic soil properties, continuous modification/adaptation of roller vibration amplitude and frequency to ensure optimum compaction, and full-time monitoring by an integrated global positioning system to provide a complete GPS-based record of the compacted area.

Appendixes A through D of NCHRP 676, which provide supplemental information, are only available online; links are provided below.

Appendix A: Supplement to Chapter 1

Appendix B: Supplement to Chapter 3

Appendix C: Supplement to Chapter 6

Appendix D: Supplement to Chapter 8

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!