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.
c h a p t e r 2 2.1 Continuous Compaction Control and Intelligent Compaction Roller-integrated CCC has been used in Europe since the late 1970s, while IC technology has been available since the late 1990s. This section summarizes the history of CCC and IC as well as key results of previous research studies on CCC and IC in both Europe and the United States. 2.1.1 history of Continuous Compaction Control and Intelligent Compaction The history of CCC and IC is summarized here; the reader is referred to Mooney & Adam (2007) for a more detailed account. The initial research on roller-integrated measure- ment dates to 1974, when Dr. Heinz Thurner of the Swedish Highway Administration performed field studies with a 5-ton tractor-drawn Dynapac vibratory roller instrumented with an accelerometer. The tests indicated that in the frequency domain the ratio between the amplitude of the first harmonic and the amplitude of the excitation frequency could be cor- related to the state of compaction and the stiffness of the soil as measured by the static plate load test. In 1975, Dr. Thurner founded the firm Geodynamik with partner Ã ke SandstrÃ¶m to continue development of the roller-mounted compaction meter. In cooperation with Dr. Lars Forssblad of Dynapac, Geodynamik developed and introduced the Compactom- eter and the compaction meter value (CMV) in 1978. The new method was introduced to the technical community at the First International Conference on Compaction held in Paris, France, in 1980 (Thurner & SandstrÃ¶m 1980, Forssblad 1980). Dynapac began offering the CMV-based Compactom- eter commercially in 1980. A number of roller manufactur- ers (e.g., Ammann, Caterpillar, Ingersoll Rand) subsequently began offering the Geodynamik CMV Compactometer mea- surement system. Sakai introduced the compaction control state of Practice value (CCV) in 2004 (Scherocman et al. 2007). The CCV fol- lows in the footsteps of the CMV by using harmonic con- tent from the measured drum vibration to estimate the com- pacted state. Bomag introduced the Omega value and corresponding Terrameter in 1982. The Omega value provided a continu- ous measure of compaction energy and at the time served as the only alternative to CMV. In the late 1990s, Bomag in- troduced a vibration modulus E vib , which provides a mea- sure of dynamic soil stiffness (e.g., KrÃ¶ber et al. 2001). The Omega value was thereafter discontinued for new machines. Ammann followed with the introduction of soil stiffness pa- rameter k s (also called k B ) in 1999 (Anderegg 1998, Anderegg & Kaufmann 2004). The introduction of E vib and k s signaled an important evolution toward the measurement of more mechanistic, performance-related soil properties (e.g., soil stiffness/modulus). The current commercially available roller-based measurement values (MVs) are summarized in Table 2.1. Specifications for quality assurance (QA) of earthwork compaction using roller-integrated CCC were first introduced in Austria (1990), Germany (1994), and Sweden (1994). Revi- sions to these original specifications have been made in each country. The International Society of Soil Mechanics and Geotechnical Engineering (ISSMGE) recently endorsed the Austrian specifications for CCC (Adam 2007). The Austrian/ ISSMGE and German specifications each permit multiple options for using CCC in earthwork compaction QA. The most common and simplest approach uses CCC to identify weak areas for evaluation via a static plate load test (PLT), a lightweight deflectometer (LWD), or density spot testing. Acceptance is based on these weak areas meeting prerequisite PLT modulus, LWD modulus, or density requirements. The most advanced CCC-based QA specifications involve corre- lating roller MVs to PLT modulus, LWD modulus, or density in a defined calibration area. If a suitable correlation is found,
â a target roller MV is determined from the MV versus spot test regression equation. Acceptance is based on comparison of roller MV data collected in a production area to the target roller MV. Based on a survey of European practices, the cali- bration approach is challenging to implement and requires a high level of on-site knowledge (G. BrÃ¤u, personal commu- nication, 2008; D. Adam, personal communication, 2008). A complete description of the European specifications is pro- vided in Section 2.2. When Geodynamik first introduced the compaction meter and CMV, vibratory drum technology was rudimentary by current standards. Vibration was implemented mechanically via a two-piece âclamshellâ eccentric mass assembly within the drum. If rotated in one direction with frequency â¦ (rad/ s), the two eccentric masses would join together and provide maximum eccentric mass moment m o e o and therefore maxi- mum time-varying centrifugal force F(t) per Equation 2.1. If operated in the reverse rotational direction, m o e o and F(t) would be a minimum. In the 1990s, vibratory roller technol- ogy became much more sophisticated. Bomag introduced the Variocontrol roller with counterrotating eccentric masses and servo-hydraulic control of the vertical component of F(t), referred to as F ev (see Section 18.104.22.168). Ammann introduced the ACE roller with servo-hydraulic two-piece eccentric mass moment and frequency control in 1999 (see Section 22.214.171.124). Dynapac followed suit with variable eccentric mass moment control in 2006 (see Section 126.96.36.199). The remaining roller manufacturers use the traditional two-piece and thus two- amplitude eccentric mass assembly. In the roller community the maximum vertical excitation force F ev is commonly re- ferred to as theoretical amplitude A (see Equation 2.2) and is equal to the peak displacement of the drum (with mass m d ) if suspended in air. (2.1) (2.2) The introduction of servo-controlled eccentric excitation has catalyzed the term intelligent compaction, where the vi- bratory force amplitude and/or frequency are automatically adjusted in an attempt to improve roller performance and compaction. Currently, the intelligence in IC is limited. The Ammann/Case, Bomag, and Dynapac IC rollers automati- cally decrease the vertical vibration force when undesirable operating conditions are detected (e.g., jump mode). Further, some rollers (e.g., Bomag, Ammann/Case) have the ability to automatically reduce the eccentric force amplitude when a user-defined threshold roller MV has been reached. In a broader sense, however, intelligent compaction is in its in- fancy. Considerable advances in truly intelligent compaction are anticipated over the next decade. 2.1.2 prior Investigations of roller vibration and roller-Integrated measurement Systems Roller-integrated measurement of soil properties was initi- ated within the roller manufacturer community, and there- fore early literature on the topic is limited. Issues pertaining to roller instrumentation and vibration behavior over a broad range of operating frequencies and amplitudes have only re- cently been addressed in the literature (Adam 1996, Adam & Kopf 2004, Brandl et al. 2005, Mooney et al. 2005, Rinehart & Mooney 2008). Recent experimental data collected with in- strumented roller compactors (Mooney et al. 2003, Anderegg & Kaufman 2004, Adam & Kopf 2004, Mooney & Rinehart 2007, van Susante & Mooney 2008) have revealed fairly com- plex nonlinear roller vibration behavior, including loss of contact between the drum and soil, drum and frame rocking, and chaotic behavior. Combined experimental and numerical investigations over the past 30 years have shed considerable light on roller- soil interaction and roller-integrated measurement systems. Early modeling efforts by Yoo & Selig (1979, 1980) employed a two degrees-of-freedom (DOF) model to represent steady state vertical drum and frame kinematics. These early studies were able to demonstrate the sensitivity of roller vibration table 2.1. Commercially available roller MVs. Roller MV Manufacturers Drum Vibration Parameters Used for Determining Roller MV Compaction meter value (CMV) Dynapac, Caterpillar, Hamm, Volvo In the frequency domain, ratio of vertical drum acceleration amplitudes at fundamental (operating) vibration frequency and its first harmonic Compaction control value (CCV) Sakai In the frequency domain, algebraic relationship of multiple vertical drum vibration amplitudes, including fundamental frequency, and multiple harmonics and subharmonics Stiffness k s Ammann, Case Vertical drum displacement, drum-soil contact force Vibration modulus E vib Bomag Vertical drum displacement, drum-soil contact force F t m e t F t( ) cos( ) cos( )= = o o ev â¦ â¦ â¦2 A m e m F m = =o o d ev d â¦2
to changes in soil stiffness and damping and were used to investigate compaction efficiency (e.g., maximizing transmit- ted force). Decoupling of the drum from the soil (i.e., loss of contact, partial uplift) was first experimentally shown and predicted through lumped parameter modeling by Quibel (1980), Machet (1980), and KrÃ¶ber (1988). Adam (1996) and Anderegg (1998) used lumped parameter modeling to characterize the various operational modes of roller vibra- tion, including nonlinear and chaotic vibration. In addition to drum/soil coupled behavior (i.e., full contact through- out), Adam (1996) characterized both partial loss of contact, where the drum decouples from the soil for a portion of each loading cycle, and âjumpâ mode (also referred to as double jump), wherein the drum loses contact for more than one cycle of vibration at a time (see Figure 2.1). The resulting nonlinear signal in jump mode includes a subharmonic at one-half the excitation frequency (Adam 1996, Adam & Kopf 2004). Anderegg (1998) and Anderegg & Kaufmann (2004) described jump mode and rocking mode vibration as cha- otic states. Employing chaos theory, Anderegg showed that rocking and jump mode vibration states occur above a cer- tain centrifugal force and soil stiffness combination (roller parameter specific). In current practice, IC roller compactors use automatic feedback control of the centrifugal force to prevent chaotic motion (e.g., Anderegg & Kaufmann 2004) because these motions are harmful to the machines and dan- gerous for the operator. van Susante & Mooney (2008) demonstrated through nu- merical model fitting to experimental data that soil behaves nonlinearly during roller vibration. The nonlinearity is at- tributed to partial loss of contact, curved drum surface, and stress-dependent soil modulus. They also demonstrated that rocking mode drum vibration occurs within the operating frequency range of most rollers and that this drum rocking can significantly alter vertical vibration response. In addition, a traveling roller interacting with underlying soil heterogene- ity and employing variable excitation frequencies and ampli- tudes results in transient behavior that in turn influences the measurement systems (van Susante & Mooney 2008). The coupling of modeling efforts with experimental studies has enabled numerous investigations of roller MVs. Mooney et al. (2003, 2005) presented the findings from a laboratory, field, and numerical modeling investigation into the relation- ship between roller MVs and soil compaction properties for clay, sand, and crushed rock materials. Here, the frequency content of the drum acceleration signalâakin to CMV and CCVâwas explored. Both laboratory and field studies showed that soil stiffness and harmonic content-based MVs are much more sensitive to the compaction process than is measured dry density. Dry density may increase 10% from loose deposition to full compaction, whereas laboratory- measured stiffness and field roller MVs may increase by more than 100% (Mooney et al. 2003). This contributes to the relative difficulty in correlating roller MVs to dry density compared to roller MV versus spot-test modulus. The study also showed that roller MVs exhibit much greater sensitivity to the compaction of a lift if the underlying (sublift) mate- rial is stiff. The results of roller-based measurement on soft soil indicated that low CMV is insensitive to changes in soil properties. Adam & Kopf (2004) investigated the relationship between roller MV, soil modulus, and eccentric force amplitude (see Figure 2.2). Their study revealed how the various operational modes are influenced by eccentric force amplitude (referred to as relative amplitude in Figure 2.4) and soil modulus. More importantly, their study revealed the sensitivity of roller MVs to soil modulus within each operational mode. Figure 2.3 il- lustrates that Ammann k s and Bomag E vib increase fairly lin- early with soil modulus throughout both continuous contact and partial uplift (the most commonly observed field behav- iors). CMV increases linearly with soil modulus during par- tial uplift but is insensitive to soil modulus during continuous contact (CMV below approximately 10). Figure 2.4 confirms the insensitivity of CMV to soil modulus for CMV <10. The Adam & Kopf numerical study also showed that roller MVs are amplitude dependent (see Figure 2.2). Experimental investigations by KrÃ¶ber et al. (2001), Hartmann (2002), and Mooney & Rinehart (2007, 2009) revealed various degrees- of-amplitude dependence on roller-measured soil stiffness. The study by Hartmann also found that E vib decreased with increasing roller speed. 2.1.3 Correlation Studies A number of studies have been performed over the past 20 years to relate roller MVs to spot-test measurements (e.g., density, PLT modulus, LWD modulus). Floss et al. (1991) Figure 2.1. Observed modes of vibratory roller op- eration (from Adam & Kopf 2004).
â Figure 2.2. Relative roller MVs depending on soil stiffness (from Adam & Kopf 2004). Figure 2.3. Sensitivity of roller MVs to soil modulus (from Adam & Kopf 2004); kB = ks.
reported dozens of correlations between CMV and PLT moduli E V1 and E V2 as well as between CMV and density (or % compaction) for coarse-grained, mixed, and fine-grained soils. The resulting regressions were found to be linear and sometimes nonlinear. Their results revealed higher correla- tion coefficients from CMV to PLT modulus correlation than from CMV to density correlation. The reason for this, they concluded, is that the measurement depth of the PLT is closer to that of CMV than is the measurement depth of density tests (sand cone, nuclear gauge). They found that roller MV is influenced by moisture content for fine-grained soils. Spe- cifically, at a constant density, MV was found to increase with decreasing moisture content. KrÃ¶ber et al. (2001) investigated the relationship between Bomag E vib and PLT moduli E V1 and E V2 during field testing on a silty gravel. Their results showed a strong linear correlation between E vib and both E V1 and E V2 (R2 > 0.9). Their results also showed that E vib was equivalent in magnitude to E V1 during early compaction passes and nearly equal to E V2 at full com- paction. Hartmann (2002) explored the correlation between E vib and PLT moduli E V1 and E V2 as well as the influence of eccentric amplitude on the correlations. Hartman found no E vib amplitude dependence on a soft silty soil but significant amplitude dependence for a gravelly sand soil. Preisig et al. (2006) explored the correlation of Ammann k s to E V1 and E V2 using more than a dozen data sets from sandy gravels. As shown in Figure 2.5, when the results from eight sandy gravel sites are combined, the resulting k s versus E V correlations are quite good. Visually, the correlations for individual soils are not as evident. The research found that if only data near the fully compacted state were used (as de- fined by M e2 /M e1 = E V1 /E V2 < 3.5), the correlations improved considerably. Based on these data, they called into question the European calibration approaches that use data from low-, medium-, and full-compacted states. Based on their gravel data sets, Preisig et al. argued that a correlation can be de- veloped by using measurements on fully compacted material and assuming that the linear regression line passes through the origin. BrÃ¤u et al. (2004) attempted to develop universal regres- sion relationships using roller MV and spot-test measure- ment data from dozens of sites. They divided results by soil type (granular, mixed grain, cohesive), by layered versus ho- mogeneous [i.e., >1.5-m (4.9-ft)-thick] conditions, and by roller vibration amplitude (high and low) used during mea- surement. Their results revealed significant scatter. BrÃ¤u et al. concluded that the approach is possible but that there was too much uncertainty in the variables of the archived data. Mooney et al. (2003, 2005) performed a study to cor- relate harmonics-based roller MVs (CMV, CCV) to spot-test measurements (dry density, dynamic cone penetrometer) for sand subgrade soil and crushed rock base material. Their work showed that strength of the correlation and sensitivity of the roller MV improved significantly if the sublift material was stiffer. Petersen (2005) found poor correlations between E vib and spot-test measurements and attributed the results to stress dependency of soil modulus and inherent soil hetero- geneity that affects roller MVs differently than in situ tests. White & Thompson (2008) performed a study to correlate CMV to spot test measurementsâdry unit weight, dynamic cone penetrometer (DCP) index, Clegg impact value (CIV), and LWD modulusâfor five cohesionless base materials Figure 2.4. Empirical relationship between CMV and EV2 illustrating the insensitivity of CMV to Ev2 for CMV<10 (results and figure courtesy of Dynapac).
â multiple linear regression analyses between MVs (CMV or MDP, both from a Caterpillar roller), moisture content, and spot test measurements (density, DCP index, CIV, or E LWD ). Testing was performed on an A-1-b soil. Spot testing was per- formed after passes 1, 2, 4, 8, and 12 on three lanes. Regres- sion relationships between both MVs and DCP index, CIV and E LWD each exhibited R2 between 0.85 and 0.95. The R2 for the regression relationship for density varied between the two different MVs, with CMV having R2 = 0.68 and MDP having R2 = 0.92. White et al. (2008a) conducted a field investigation to cor- relate Caterpillar CMV to spot test measurements. Acceptable correlations were found between CMV and DCP index and between CMV and dry unit weight. No correlation was found between CMV and E LWD . Thompson et al. (2008) performed correlation analysis between Ammann k s and measurements from various spot-test methods. R2 values were between 0.3 through linear regression analysis. Data were collected over the entire compaction range for each base material. By av- eraging the data over a uniform area for each compaction state, high correlation was observed (R2 > 0.90 for 20 of 28 correlations). The relationships between Ammann k s and spot-test mea- surements were investigated by White et al. (2007). The cor- relations observed for subgrade and subbase soils were found to depend heavily on the range of compaction states over which the soil was tested. For subgrade soil, k s was found to be linearly proportional to LWD modulus E LWD and CIV and showed a power-function relationship with DCP index. For one test area comprised of subgrade material, a test roller was used to evaluate k s output and in situ test results. k s was re- lated to rut depth through a linear relationship for k s ranging from 10 to 35 MN/m and rut depths ranging from 10 to 60 mm (0.4 to 2.4 in). Thompson & White (2007) performed Figure 2.5. Correlations between Ammann ks and PLT moduli. Note: Me1 = EV1, Me2 = EV2. Top plots show all data; bottom plots show data for Me2 /Me1 = EV1 /EV2 < 3.5 (from Preisig et al. 2006).
0 and 0.8, depending on the spot test being correlated (dry unit weight, E V1 , E LWD , CIV 4.5-kg , DCP index) on A-6(9) subgrade material. It was found that k s was dependent on the moisture content of the subgrade soil (R2 = 0.61). Correlations were also examined on A-1-b soil but were not found. This was attributed to the small range of k s values (30 to 40 MN/m) observed in this test area. Rahman et al. (2008) investigated the relationship between Bomag E vib and spot-test measurements. For two sandy soils, they concluded there was no clear correlation between E vib and modulus determined from LWD, FWD, Geogage, or DCP. The authors attributed the poor correlations to differences in measurement depth. Though not mentioned as a reason for the poor correlations, these investigators used variable control IC to gather E vib data. As described above and herein, roller MVs can be strongly amplitude dependent. Rahman et al. found that E vib was sensitive to moisture. 2.1.4 prior Investigations of measurement depth of IC/CCC rollers A review of the literature reveals limited results regarding the measurement depth of an IC or CCC roller (i.e., the depth to which an MV is representative). Two experimental stud- ies and one numerical study provide some insight. Several other general statements are made about measurement depth but are not substantiated with theory, experimental results, or references to other literature. Floss et al. (1991) present the results of a study in which soft mattresses were buried in granular soil at four depths to 1.2 m (3.9 ft). Once the excava- tions were refilled and fully compacted, several different roll- ers were operated along the track containing the mattresses. The study concluded that measurement depth for low- versus high-vibration amplitudes was difficult to interpret but that measurement depth increased with increasing roller static weight. The following guidelines were proposed as a result of this study and now appear in the ISSMGE-recommended CCC specifications (Adam 2007; see Section 2.3.2 below): â¢ For 2-ton rollers the approximate measurement depth is 0.4 to 0.6 m (1.3 to 2.0 ft), â¢ For 10-ton rollers the approximate measurement depth is 0.6 to 1.0 m (2.0 to 3.3 ft), â¢ For 17-ton rollers the measurement depth is greater than 1 m (3.3 ft). For reference, common highway construction smooth drum and pad foot vibratory rollersâthose used in this studyâare 11- to 15-ton rollers. In addition to the mattress study, Floss et al. (1991) describe a two-layer study in which a layer of sandy gravel material was compacted above a stepped em- bankment of sandy silt material, resulting in three discrete layers of thickness for the sandy gravel (stiff) above the sandy silt (soft). They conclude that the results agree with those from the mattress study. Brandl & Adam (2000) do not present any results of theo- retical or experimental investigations regarding measurement depth but do state that âmeasurement depth depends on static load of the drum, vibration amplitude and frequency and also the soil.â The following standardized values for mea- surement depth are recommended in Brandl & Adam: â¢ For 2-ton rollers the measurement depth is approximately 0.4 to 0.6 m (1.3 to 2.0 ft), â¢ For 10-ton rollers the measurement depth is approximately 0.6 to 0.8 m (2.0 to 2.6 ft), â¢ For 12-ton rollers the measurement depth is approximately 0.8 to 1.5 m (2.6 to 4.9 ft). Anderegg & Kaufmann (2004) state that it is commonly accepted that 0.1 mm (0.004 in) of vertical drum vibration amplitude equates to 0.1 m (0.33 ft) of measurement depth; however, no theoretical or experimental justification is given. Classical foundation settlement analysis forms the underpin- nings for this rule of thumb, but it has not been rigorously analyzed or validated experimentally (R. Anderegg, personal communication, 2008). Brandl et al. (2005) present the results of another study in- volving rolling over buried mattresses. Mattresses were placed on an existing grade, and layers of sandy gravel material were placed and compacted above the mattresses with a 13-ton vi- bratory roller, eventually to a height of 1.05 m (3.40 ft). The authors conclude that the measurement depth of the roller at medium amplitude was 2.1 m (6.9 ft) in the sandy gravel material. Kopf & Erdmann (2005) performed finite element analysis of a 13-ton vibratory roller on stratigraphies of soft over stiff soil and stiff over soft soil. Their results indicate that measurement depth increased from about 0.7 to 1.4 m (2.3 to 4.6 ft) for stiff soil over soft soil and from about 0.6 to 1.2 m (3.9 ft) for soft soil over stiff soil, as the vibration force ampli- tude increased from low to high, 95 to 365 kN (21 to 82 kip). 2.1.5 Studies Utilizing In Situ Stress-Strain Instrumentation A review of the literature regarding the use of in-ground instrumentation to monitor the response and behavior of soil due to vibratory compaction yielded limited results, possi- bly due to the difficulties and costs associated with installing and securing accurate results from in-ground instrumenta- tion. Brandl & Adam (2000) measured vertical stress and displacement induced by a 13-ton vibratory roller at three depthsâ15, 50, and 85 cm (6, 20, and 33 in)âin a granular material. No discussion of in-ground sensor calibration or
â Ping et al. (2002) present roller-induced vertical stress data at three depths in a fine sand material. The study used 230- mm (9-in)-diameter Geokon earth pressure cells (EPCs), but no discussion of the calibration or placement procedures is provided. The results show peak stresses during compaction by a typical vibratory soil roller operating at 1.7 mm (0.07 in), with a theoretical vibration amplitude of 330 and 170 kPa (47.9 and 24.7 psi) at a depth of 0.23 and 0.4 m (9 and 16 in), respectively. For 0.8-mm (0.03-in) theoretical vibra- tion amplitude by the same roller, the peak stresses were 190 and 110 kPa (27.6 and 16.0 psi) at a depth of 0.28 and 0.56 m (11 and 22 in), respectively. The study also showed a decrease in vertical stress with increasing roller forward velocity and little to no change in vertical stress with increasing number of roller passes. It should be noted that in-ground sensor data could be unreliable before the soil is fully compacted as the sensor calibration depends on the density of the surrounding material (see Chapter 4 and Appendix D). DâAppolonia et al. (1969) installed vertical stress cells at depths of 0.3, 0.45, 0.6, 1.2, and 1.8 m (12, 18, 24, 47, and 71 in) and horizontal stress cells at a depth of 0.6 m (24 in) in a poorly graded dune sand. Two towed (usually towed by a dozer) roller compactors were used, weighing 6.25 and 3.15 tons and operating at 27.5 and 19 to 30 Hz, respectively. Maximum dynamic vertical stresses (i.e., static component removed) under the 6.25-ton roller at depths of 0.3, 0.45 0.6, 1.2, and 1.8 m (12, 18, 24, 47, and 71 in) were observed to be 103, 76, 62, 28, and 21 kPa (14.9, 11.0, 9.0, 4.1 and 3.1 psi), re- spectively, and were observed to be independent of the num- ber of roller passes. Dynamic horizontal stress data are not presented; however, these data are said to be independent of the number of roller passes. Static horizontal stress (i.e., due to overburden and residual stresses due to compaction) data are presented in terms of the coefficient of horizontal earth pressure K o defined as the ratio of horizontal to vertical stress. K o values at a depth of 0.6 m (24 in) for a TB compacted with the 6.25-ton roller in the direction parallel to the roller path range from about 0.8 to 1.1 and tend to increase with the number of roller passes. K o values for a TB compacted with the 6.25-ton roller at a depth of 0.6 m (24 in) in the direction perpendicular to the roller path range from about 0.8 to 2.75 and also tend to increase with the number of roller passes. K o values are generally lower for TBs compacted with the 3.15- ton roller and tend to increase with both number of roller passes and increasing frequency of vibration. 2.1.6 geostatistical Studies Grabe (1994) performed spatial analysis of CMV maps using a spectral density approach. Aside from a dominant wavelength in the data at a distance equal to the circumference of the drum that he attributed to soil sticking to the drum, placement procedures is presented. Values for vertical stress are not given; rather the results are presented as a percent- age of the maximum measured value, which occurred at a depth of 15 cm (6 in) under high-vibration amplitude. For the high-amplitude pass, the roller operated in jump mode, and the peak stresses at 50 and 85 cm (20 and 33 in) were observed to be about 75% and 25% of the maximum, respec- tively. For the low-amplitude pass the roller operated in con- tact mode and the peak stresses at 15, 50, and 85 cm (6, 20, and 33 in) were observed to be about 70%, 30%, and 10% of the maximum, respectively. For the low-amplitude pass, the displacements at 15, 50, and 85 cm (6, 20, and 33 in) were observed to be about 8, 1.75, and 0.4 mm (0.31, 0.07, and 0.02 in), respectively. For the high-amplitude pass, the displace- ments at 15, 50, and 85 cm (6, 20, and 33 in) were 13, 2.5, and 0.7 mm (0.51, 0.10, and 0.03 in), respectively. A phenomenon called the bow-wave effect, in which the soil in front of the drum experiences vertical extension before being compressed as the drum traverses over it, is observed at all three depths during both high- and low-amplitude passes. It is also shown that some roller passes result in permanent compressive dis- placements (i.e., compaction), whereas other passes result in permanent extension (i.e., loosening). Brandl et al. (2005) present the results of a similar study de- signed to measure stress and displacement at several depths. Two different rollers were used. No discussion of in-ground sensor calibration or placement procedures is presented. Roller-induced vertical stress is presented at a depth of 0.4 m (16 in) in a sandy gravel and at a depth of 0.35 m (14 in) in a clayey silt. Vertical displacement is presented at 0.1 m (4 in) in both the sandy gravel and clayey silt. Peak vertical stresses at 0.4 m (16 in) in the sandy gravel varied from 225 to 650 kPa (32.6 to 94.3 psi) across the range of amplitudes tested. The roller operated in contact mode for low-amplitude set- tings, partial loss of contact mode for medium-amplitude settings, and jump and rocking modes for high-amplitude settings. The peak vertical displacements at a depth of 0.1 m (4 in) in the sandy gravel were 1.7 and 3.25 mm (0.07 and 0.13 in) for low- and high-amplitude settings, respectively. In the clayey silt material, peak stresses at a depth of 0.35 m (14 in) were observed to range between 120 and 200 kPa (17.4 to 29.0 psi) for the range of low to high amplitudes. The roller operated in contact mode for all amplitudes. The peak verti- cal displacements at a depth of 0.1 m (4 in) in the clayey silt were about 10.5 and 19 mm (0.41 and 0.75 in) for low- and high-amplitude settings, respectively. It is noteworthy that the clayey silt TB was built in one day and that all testing was done on the following day. Even though the TB was fully compacted the first day, the first roller pass of the second day resulted in a permanent displacement of nearly 5 mm (0.20 in). Permanent settlements were very small or nonexistent after this first pass.
Grabe concluded that there were no patterns in the roller MV data maps. Grabe found that CMV values at close distances were correlated, and he concluded that the data had an infinite correlation length (limited by the sample size of the data). Petersen et al. (2007) examined why and how geostatistics could be used with CCC rollers. They concluded that the var- iogram parameters could be very useful. Peterson et al. found the variogram of roller MV data to be directionally depen- dent (i.e., anisotropic) and only used data in the driving di- rection. They proposed that roller MVs that change by more than the nugget over short distances suggest possible problem areas. White et al. (2008b) further examined variogram prop- erties and how they could be used for earthwork compaction QA. They concluded that the range could be used as a win- dow size for QA analysis and that the sill could be used as a target for uniformity of the data. They concluded there was anisotropy but that it can be ignored due to the high density of the data. 2.2 State of Current and Emerging IC Equipment The primary manufacturers of CCC and/or IC soil rollers include Ammann (offered under the Case name in the United States and referred to hereafter as Ammann/Case), Bomag, Caterpillar, Dynapac, Volvo (formerly Ingersoll Rand), Sakai, and Hamm. Ammann/Case, Bomag, Caterpillar, Dynapac, and Sakai rollers were in this study and are described fur- ther here. As summarized in Table 2.2, all manufacturers offer roller-integrated measurement systems. The six roller MVs used in practice include (1) CMV, developed by Geodynamik and used by Dynapac, Caterpillar, and Volvo; (2) CCV, a de- rivative of CMV developed by Sakai; (3) stiffness E vib , devel- oped and used by Bomag; (4) stiffness k s , developed and used by Ammann/Case; and (5) machine drive power (MDP), developed and used by Caterpillar. CMV, CCV, k s , and E vib require vibration and thus are applicable only on vibratory rollers. MDP does not require vibration but can be employed on vibratory rollers. The roller-integrated measurement sys- tems, feedback control, and GPS-based documentation for each manufacturerâs IC rollers are described in the following sections. 2.2.1 ammann/Case 188.8.131.52â MeasurementâValue The Ammann ACE Plus system calculates soil stiffness k s once per cycle of vibration. The measurement system is thoroughly described in Anderegg (1998) and Anderegg & Kaufmann (2004) and is briefly described here. To best un- derstand k s the basic vibration of the drum/soil system must be considered. Figure 2.6 shows a schematic of a roller and a two DOF model representing the vertical kinematics of the drum-frame system (Figure 2.6b), where m d and m f are the drum and frame masses, respectively; z d and z d are the drum displacement and acceleration, respectively; m o e o is the eccentric mass moment; and â¦ is the excitation frequency. Here, the soil is represented with a spring-dashpot Kelvin- Voigt model. The resulting free body diagram (Figure 2.6c) shows the drum/soil contact force F s comprised of four elements: drum inertia, frame inertia, eccentric force, and machine weight. Ammann determines drum inertia and eccentric force via measurement of vertical drum acceleration and eccentric position (frame inertia is neglected). The resulting equation of motion is a second-order differ- ential equation. The vertical drum displacement amplitude z d is determined via spectral decomposition and integra- tion of the measured peak drum accelerations (Anderegg & Kaufmann 2004). Solving this equation for k s when the drum velocity is zero (i.e., down-most position) yields Equation table 2.2. Summary of CCC and IC equipment investigated (as of August 2008). Roller Manufacturer Intelligent Compaction Features Roller-Integrated Measurement Automatic Feedback Control of: GPS-Based Documentation Ammann/Case Stiffness k s Eccentric force, amplitude, and frequency Yes Bomag Stiffness E vib Vertical eccentric force amplitude Yes Caterpillar MDP, CMV C None Yes Dynapac US CMV D Eccentric force amplitude Yes Volvo CMV V None No Sakai America CCV None Yes
â 184.108.40.206ââ FeedbackâControl The Ammann ACE Plus eccentric assembly (shown in Fig- ure 2.8) is comprised of outer and inner masses. The angle between the two masses, Î¸, is computer controlled and can be adjusted through a differential gear to provide maximum eccentric force (Î¸ = 0Â°), zero eccentric force (Î¸ = 180Â°), and any eccentric force in between (0Â° < Î¸ < 180Â°). The maxi- mum eccentric mass moment m o e o and associated theoreti- cal drum displacement amplitude A for the Ammann model ASC 110/130 are 8.8 kg-m (63.7 lb-ft) and 2.2 mm (0.09 in), respectively. The ACE Plus system performs closed-loop feed- back control of drum/soil contact force F s . Three operator- selected levels of F s are possible: â¢ Low force: F s(max) = 14 kN (3.1 kip), leading to measured z d = 0.4 to 1.5 mm (0.02 to 0.06 in); â¢ Medium force: F s(max) = 20 kN (4.5 kip), leading to mea- sured z d = 1.0 to 2.0 mm (0.04 to 0.08 in); â¢ High force: F s(max) = unlimited, leading to measured z d = 2.0 to 3.0 mm (0.08 to 0.12 in). With a selected force level, the roller adjusts the eccen- tric mass moment to maintain the F s(max) . The excitation fre- quency is adjusted to maintain a phase lag Ï between 140Â° and 160Â°. For high-force levels, the frequency required to maintain the appropriate Ï is 23 to 25 Hz. As the amplitude decreases, the frequency required to maintain the appropriate Ï is higherâup to 35 Hz. The ACE Plus system can also use a user-specified k s value as the control parameter. In the so-called plate modulus mea- surement mode, a limit value for k s is selected. When the pre- 2.3, where Ï is the phase lag between the eccentric force and drum displacement. The Ammann k s is effectively the ratio of F s to maximum vertical drum displacement z d(max) and occurs when the velocity equals zero (see Figure 2.7). k m m e zs d o o d = + ï£® ï£° ï£¯ ï£¯ ï£¹ ï£» ï£º ï£º â¦2 cos( )Ï (2.3) Accordingly, k s can be determined from measured drum acceleration and phase lag. The accuracy of Ammann mea- surement data is as follows: âz d = 0.001 mm (0.00004 in); âÏ = 0.5Â°; ââ¦ = 0.31 rad/s (âf = 0.05 Hz) (R. Anderegg, personal communication, 2007). Figure 2.6. (a) vibratory compactor schematic; (b) two DOF model representation of vibra- tory compactor; (c) free body diagram of forces acting on the drum. Figure 2.7. Illustration of ks during (a) contact and (b) partial loss of contact.
determined value of k s is reached, the ACE Plus system auto- matically decreases the eccentric mass moment to 0.5 F s(max) for the range chosen. In addition, the ACE Plus system calcu- lates an optimal speed based on a desired impact spacing of 2 to 4 cm (0.8 to 1.6 in). The operator can use a gauge in the roller cab to perform open-loop control of speed. Finally, the ACE Plus system monitors for unstable rocking or jumping (primarily through analysis of subharmonics) and automati- cally decreases the eccentric mass moment until stable opera- tion is restored. 220.127.116.11â âGPS-BasedâMappingâandâDocumentationâ Software The ACE Plus software (see Figure 2.9) marries k s data with x, y, and z coordinates collected via onboard GPS equipment. Using differential GPS with real-time kinematic (RTK), Ammann indicates accuracies are Â±10 cm (3.9 in). Without a reference signal (e.g., base station), the system has an ac- curacy within a few meters. Though k s is determined for each vibration cycle (and with an x-direction resolution of 2 to 4 cm [0.8 to 1.6 in]), GPS coordinates are acquired once per second (once per 1 to 3 m [3.3 to 9.9 ft]). Currently, the ACE Plus system collects k s data each cycle and reports an average k s with GPS data at a frequency of 1 Hz. Ammann indicates that the time resolution in seconds is inversely proportional to the number of GPS receivers (e.g., three receivers will give a time resolution of 0.33 s). Roller MV reporting is discussed in more detail in Section 3.1. Housed within a tablet PC onboard computer, the ACE Plus software maps a number of roller parameters in graphi- cal view, as shown in Figure 2.9a. Data are downloaded from the onboard PC in text file format via USB memory stick. The data can be evaluated with any program. Ammann also provides PC-based software (Figure 2.9b). 2.2.2 Bomag variocontrol System 18.104.22.168â MeasurementâValue The Bomag Variocontrol system calculates a âvibration modulusâ E vib using lumped parameter vibration theory and cylinder on elastic half-space theory. The principle behind E vib is presented in KrÃ¶ber et al. (2001) and is briefly described here. To determine E vib the drum/soil assembly is modeled as shown in Figure 2.6. Bomag uses constant frequency com- paction, with â¦ = 176 rad/s (f = 28 Hz). Bomag uses two ac- celerometers with measurement axes arranged at Â±45Â° from vertical to measure vertical drum acceleration. Phase lag is calculated (specific method is confidential information), enabling determination of the contact force F s per equilib- rium of forces shown in Figure 2.6. The drum displacement is computed (confidential information). The combination of F s and z d data yield force-deflection curves from which soil stiffness can be extracted (Figure 2.10). Along the way to determining E vib , Bomag calculates a se- cant stiffness k from the compression portion of each F s ver- sus z d cycle (Figure 2.10). To relate the measured F s versus z d behavior and stiffness k to modulus E, Bomag utilizes a theo- retical solution for a rigid cylinder resting on a homogeneous, isotropic, elastic half-space (see Figure 2.11) developed by Lundberg (1939). Lundbergâs theory is a static solution and Figure 2.8. Two mass eccentric assembly (courtesy of Ammann).
â Figure 2.9. (a) Ammann ACE Plus onboard tablet PC and software; (b) office PC software (courtesy of Ammann). Figure 2.10. Contact forceâdrum displacement behavior, Fs = FB (adapted from KrÃ¶ber et al. 2001). Figure 2.11. Drum on elastic half-space and relationship between stiffness k and modulus E (adapted from van Susante & Mooney 2008).
26 relates z d , F s , drum length L, and radius R to Poissonâs ratio Î½ and Youngâs modulus E of the half-space as shown in Equa- tion 2.4: (2.4) where b is the contact width as given by: b R v E L F s = Ã Ã( ) Ã Ã Ã 16 1 2â Ï (2.5) Bomag uses Î½ = 0.25 for soil. The nonlinear relationship between stiffness k and E is shown in Figure 2.11. The Vario- control system determines the appropriate E (referred to as E vib ) via a fitting approach. 22.214.171.124â FeedbackâControl The Bomag Variocontrol system uses a counterrotating ec- centric mass assembly that is directionally vectored to vary the vertical excitation force (Figure 2.12). The counterrotat- ing masses each create a centrifugal force. When the masses are opposite each other in their rotation cycles, the centrifu- gal force and thus the eccentric force are zero. Conversely, when the counterrotating masses pass each other, the eccen- tric force is maximum. The Variocontrol system rotates the entire counterrotating mass assembly to control the vector angle Î± at which maximum and minimum eccentric forces occur. Figure 2.12 illustrates this concept. As a result, when the Variocontrol system provides maximum vertical excitation (maximum drum displacement amplitude), the horizontal excitation is zero. Conversely, when the vertical excitation is zero (minimum drum displacement), the horizontal excita- tion is maximum. Accordingly, when rewritten to account for vectoring, Equation 2.1 appears as Equation 2.6: F t m e t F t( ) sin( )cos( ) cos( )= = o o ev â¦ â¦ â¦2 Î± (2.6) The Variocontrol system allows the operator to preselect from six maximum theoretical drum displacement ampli- tude options (and corresponding vector angles): 0, 0.6, 1.2, 1.7, 2.1, and 2.5 mm (0, 0.024, 0.047, 0.067, 0.083, and 0.098 in). Within a maximum amplitude setting, the Variocontrol system begins operation at the vector angle Î± corresponding to maximum amplitude until the target E vib value is reached. Once reached, the vector angle is decreased in the areas where the target E vib has been reached [to a theoretical amplitude = 0.4 mm (0.016 in)]. Figure 2.13 illustrates this principle. The Variocontrol system also allows for manual mode operation Figure 2.12. Bomag counterrotating eccentric mass assembly and vectoring of assembly to vary vertical eccentric force (courtesy of Bomag). Figure 2.13. Principle of variable-amplitude system (courtesy of Bomag). z v E F L L bd s= Ã Ã Ã Ã +ï£« ï£ï£¬ ï£¶ ï£¸ï£· 2 1 18864 2( â ) . ln Ï
â 1980. CMV is defined as the ratio of the second harmonic of the vertical drum acceleration frequency domain amplitude A 2â¦ (operating frequency â¦) divided by the first harmonic of the vertical drum acceleration frequency domain amplitude Aâ¦ multiplied by a constant c (typically 300; see Equation 2.7). CMV 2= c A A â¦ â¦ (2.7) A sister parameter called the resonance meter value (RMV) is defined by the ratio of the 0.5â¦ subharmonic acceleration amplitude to the first harmonic. Subharmonic content occurs when the drum begins to jump. CMV is determined by first performing spectral analysis of the measured vertical drum acceleration over two cycles of vibration (Figure 2.15). The reported CMV is the average of a number of two-cycle calcu- lations. Geodynamik typically averages over 0.5 s; however, this is customized to meet the manufacturerâs needs. For ex- ample, the Dynapac Compaction Meter reports CMV every 1.0 s, implying that the reported CMV is an average of ap- proximately 15 two-cycle calculations values. CMV precision is governed by 1% distortion resolution of the accelerometer. Per Equation 2.7, 1% acceleration distortion equates to CMV = 3 or Â± 1.5. 126.96.36.199â FeedbackâControl The Dynapac Compaction Optimizer (DCO) performs feedback control of the eccentric excitation force (and thus theoretical amplitude) to prevent jumping. As shown in Fig- ure 2.16, a dual-mass eccentric configuration is used to pro- in one of the six maximum amplitude settings. In manual mode the vector angle corresponding to the selected maxi- mum amplitude remains constant throughout operation. Because there is theoretically no vertical vibration in setting 1 (amplitude = 0 mm), E vib is not measured. In Variocon- trol mode the minimum amplitude for which E vib is reported is 0.4 mm (0.016 in). Finally, the Variocontrol monitors for jumping and automatically decreases F ev until stable opera- tion is restored. 188.8.131.52â âGPS-BasedâMappingâandâDocumentationâ Software Bomagâs documentation system (ÎCM 05) includes a tablet PC, mobile software, and a USB memory stick for data trans- fer (see Figure 2.14). A screenshot from the Bomag software is also illustrated in Figure 2.14. Bomag can accept any GPS receiver capable of providing GGA or PJK data via RS232 in- terface. Bomag has also used a correction service (e.g., Omni- star, Starfire) with reported accuracies of Â±10 to 50 mm (0.4 to 2 in). Bomag has used 10-Hz GPS receivers. The E vib at the sampled GPS coordinates is reported and stored; additional data averaging is not performed. 2.2.3 dynapac Compaction analyzer and Compaction optimizer 184.108.40.206â MeasurementâValue The Dynapac Compaction Meter uses the CMV as a mea- sure of the level of compaction. The CMV was developed by Geodynamik in the 1970s and introduced commercially in Figure 2.14. Bomag documentation system (images courtesy of Bomag).
Figure 2.15. Method to determine CMV involves spectral analysis (b) of two cycles of vertical drum acceleration time history data (a). Figure 2.16. Overview of Dynapacâs Compaction Optimizer (DCO) (courtesy of Dynapac).
â Figure 2.16. Overview of Dynapacâs Compaction Optimizer (DCO) (courtesy of Dynapac). tor the current position of the machine and to position the measured values. Since the DCA produces a CMV every 1.0 s, the spatial resolution of recorded data depends on the roller speed used for measurement. Dynapac recommends a speed of approxi- mately 1.0 m/s (3.3 ft/s), which provides an x-direction spa- tial resolution of 1.0 m (3.3 ft). The y-direction resolution is the width of the drum [typically 2.13 m (7.0 ft)]. Data can be exported from the field computer as a text file to be imported for further analysis. Paper printouts can be created for further documentation. The DCA software is available in an office version, so all preparations can be made and the final data can be analyzed. 2.2.4 Caterpillar 220.127.116.11â MeasurementâValues Caterpillar uses the Geodynamic CMV measurement sys- tem. In addition, Caterpillar uses MDP. The use of MDP as a measure of soil compaction is a concept originating from study of vehicle-terrain interaction (see Bekker 1969). MDP uses the concepts of rolling resistance and sinkage to deter- mine the stresses acting on the drum and the energy nec- essary to overcome the resistance to motion (Figure 2.18). MDP is calculated as: MDP g= + ï£« ï£ï£¬ ï£¶ ï£¸ï£· +P WV a g mV bâ sin â( )Î¸ (2.8) where P g is the gross power needed to move the machine, W is roller weight, a is machine acceleration, g is acceleration of gravity, Î¸ is slope angle (roller pitch), V is roller velocity, and m and b are machine internal loss coefficients specific to a vide any theoretical amplitude between 0 and 2 mm (0 to 0.078 in). The DCO maintains a vibration frequency of 28 Hz. Jumping is prevented by monitoring the RMV. When a threshold RMV is approached, the amplitude is reduced ac- cordingly. The DCO allows operation in one of six automatic settings, with maximum amplitudes of 0.40, 0.65, 0.90, 1.40, 1.80, and 2.00 mm (0.016, 0.026, 0.035, 0.055, 0.071, and 0.079 in). During operation the DCO continuously compares the measured RMV with the threshold RMV. If the measured RMV is less than the threshold RMV, the roller is operated at its maximum amplitude (e.g., 0.40, 0.65 mm). Otherwise, the amplitude is reduced accordingly. Dynapac does not cur- rently use feedback control based on CMV. 18.104.22.168ââ âGPS-BasedâMappingâandâDocumentationâ Software The Dynapac Compaction Analyzer (DCA) and accompa- nying field computer registers all pertinent roller data (e.g., CMV, pass number, amplitude, frequency, GPS coordinates) with presentation in graphical format for the operator (Figure 2.17). The DCA is compatible with any GPS receiver brand as long as the correct National Marine Electronics Association (NMEA) messages are available. Dynapac has used differen- tial GPS (DGPS) receivers with submeter accuracy for x and y coordinates. RTK receivers have been used with better results. Dynapac has also used satellite correction via Omnistar. The DCA records the z coordinate; however, with DGPS the ac- curacy is normally less than the layer thickness and is thus not displayed. The DCA software records all compaction data in local coordinates via local transformation between world geodetic system (WGS 84) and the local grid. The DCA also utilizes the road alignment parameters to show the opera- Figure 2.17. Dynapac onboard software and PC (courtesy of Dynapac).
0 particular machine (White et al. 2006). The second and third terms of Equation 2.7 account for the machine power associ- ated with sloping grade and internal machine loss, respec- tively. Therefore, MDP represents only the machine power associated with material properties. Prior to its use, MDP is calibrated for Î¸, m, and b (see Equa- tion 2.7). First, the orientation of the roller pitch sensor is found by noting the pitch readings when the roller is parked on the same sloping surface facing uphill and downhill. The average of these two readings is the pitch offset applied to all later sensor readings. The internal loss coefficients m and b are then found by operating the roller on a relatively uniform, unchanging calibration surface. P g and slope compensation (i.e., the second term of Equation 2.7) are monitored while operating the roller in both forward and reverse directions at the range of roller speeds anticipated during construction operations, generally 3 to 8 km/h (1.9 to 5.0 mph). At each roller speed the difference between P g and slope compen- sation is taken as the internal machine loss. Plots of slope- compensated machine power versus roller speed provide linear relationships from which the internal loss coefficients m and b are calculated. By incorporating both slope compen- sation and internal machine loss into Equation 2.7, MDP for roller operation on the calibration surface is approximately 0 kJ/s. MDP is a relative value referencing the material proper- ties of the calibration surface, which is generally a stiff, fully compacted soil. Positive MDP values therefore indicate ma- terial that is less compact than the calibration surface, while negative MDP values would indicate material that is more compact (i.e., less roller drum sinkage). 22.214.171.124ââ âGPS-BasedâMappingâandâDocumentationâ Software Caterpillarâs Compaction Viewer software was used dur- ing testing. A Navigator 10.4cTS operator interface is located within the operatorâs field of view and performs data acquisi- tion functions as well as displaying real-time position and compaction values. The system uses RTK GPS with accuracy capabilities of Â±10 mm (0.4 in) in the horizontal plane and Â±20 mm (0.8 in) in the vertical plane. The roller width is di- vided into a series of 30-cm (12-in) divisions for mapping roller coverage. Postprocessing and visualization of data can be performed with Caterpillarâs Compaction Viewer soft- ware on a PC. Caterpillar has developed a production version mapping and documentation system for use with commer- cially available machines. This system is compatible with the AccuGrade software produced by Caterpillar for earthwork applications (Figure 2.19). 2.2.5 Sakai 126.96.36.199â MeasurementâValue Sakai CCC rollers employ a CCV index. The unitless CCV is an extension of the CMV. The Sakai measurement system involves one accelerometer to monitor vertical drum vibra- tion. Analog bandpass filters are used to capture acceleration at the excitation frequency â¦ 0 and at 0.5 â¦ 0 , 1.5 â¦ 0 , 2 â¦ 0 , 2.5 â¦ 0 , and 3 â¦ 0 (see Figure 2.20). The amplitudes at each of these frequency components are used to determine the CCV, as shown in Equation 2.9. CCV is provided at a rate of 5 Hz. CCV = + + + + + ï£® ï£° ï£¯ ï£¹ ï£» ï£º Ã A A A A A A A 1 3 4 5 6 1 2 100 (2.9) 188.8.131.52ââ âGPS-BasedâMappingâandâDocumentationâ Software Sakai has implemented GPS-based mapping and docu- mentation software for its Compaction Information System. The onboard PC and screen are illustrated in Figure 2.21. Of- fice PC software is also provided. 2.3 Existing CCC Specifications Specifications to use roller-integrated measurement sys- tems for CCC have been introduced in Austria (in 1990, with revisions in 1993 and 1999), Germany (1994, with re- vision in 1997), and Sweden (1994, with revision in 2004). The ISSMGE recently developed recommended construction specifications based primarily on the Austrian specifications Figure 2.18. Simplified two-dimensional free body diagram of stresses acting on a rigid compaction drum (MDP increases with increasing z2 ).
â Figure 2.19. CD700 Caterpillar documentation system illustrating completed passes (left) and CMV (right) (courtesy of Caterpillar). Figure 2.21. Example of Sakai display (courtesy of Sakai). Figure 2.20. Drum acceleration frequency domain components used in Sakai CCV (adapted from Scherocman et al. 2007).
(Adam 2007). In the United States, Minnesota implemented pilot specifications for CCC in 2007â2008 and has developed a revised 2009 specification. The principal components of the various specifications and planned revisions are described below. 2.3.1 german CCC Specifications German specifications for earthwork QC/QA using CCC were introduced in 1994 and updated in 1997. Further revi- sions are expected in 2009. Referred to as ZTVE-StB, the Ger- man CCC specifications apply to subgrade and embankment soils. The lack of CCC specifications for base and subbase lay- ers is predicated on the belief that roller MVs measure much deeper than the 20- to 30-cm-thick base course layers used in Germany. There are two ways in which CCC can be specified in Germany. First, CCC can be implemented through initial calibration of roller MVs to PLT modulus or density and sub- sequent use of the correlation during QA. A second approach uses CCC to identify weak areas for spot testing via PLT, LWD, or density methods. The key elements of each approach are described here, as are the proposed 2009 revisions. 184.108.40.206â âCalibrationâApproachâ(MethodâM2âinâ GermanâSpecifications) This CCC approach involves two principal steps: (1) on- site initial calibration to develop the correlation(s) between the roller MV to be used and soil density or PLT modulus (E V2 ); (2) identification of roller MV target value (MV-TV) consistent with required density or E V2 ; and (3) acceptance testing by comparing roller MV data against the MV-TV. Calibration is performed on an area equal to three 20-m (66-ft)-long (minimum) test strips (see Figure 2.22). Roller MV data are collected during roller operation on a low degree of compaction test strip (e.g., after one compaction pass), a medium degree of compaction test strip (e.g., three to five compaction passes), and a high degree of compaction test strip (multiple passes until no further compaction observed). Three to five static PLTs or density tests are performed on each test strip. Regression analysis is performed on the roller MV versus spot-test data (see Figure 2.22). When using the PLT, the German procedure uses the unload-reload secant modulus (M E2 or E V2 ). The correlation coefficient r must be â¥ 0.7 (R2 â¥ 0.5). Additional spot tests may be performed to achieve R2 â¥ 0.5. If R2 â¥ 0.5 cannot be achieved, CCC is not permitted via Method M2. The regression equation and required minimum values for E V2 or density lead to determination of an MV-TV. In Ger- many, minimum E V2 values must be achieved for the top of subgrade. Specifically, E V2 must be 45 MPa for clay or silty soils and 80 to 100 for granular materials. Density requirements (typically 98% standard Proctor) exist for all layers below the top of subgrade. There are no moisture requirements. During acceptance testing, 90% of all roller MVs in an evaluation area must exceed the MV-TV value. There are cur- rently no additional criteria for acceptance using this method. The German specification does not permit variable frequency and amplitude control or jump-mode during calibration or acceptance testing. The project site must be homogeneous in soil type and in underlying stratigraphy; otherwise CCC- based QA is not recommended. The German group overseeing the CCC specifications is planning some moderate updates (G. BrÃ¤u, personal com- munication, 2008). The LWD or dynamic PLT may be used in place of the static PLT. Target values of LWD modulus (E LWD ) will be published but were not available at press time. In addi- tion, the issue of uniformity criteria was addressed. The Ger- man specifications will require that the 10% of MVs that fall below MV-TV be reasonably distributed around the evalua- tion area. Acceptance of this issue will be subjective and left to the on-site engineer. The German specifications will include Figure 2.22. Illustration of calibration approach for German specifications.
â performed, involves compaction with roller-integrated mea- surement until the mean roller MV increases by no more than 5%. Acceptance is then based on static PLT or LWD (dynamic PLT) modulus at the weakest area. In the Austrian/ISSMGE specifications, roller MVs must be dynamic (i.e., based in part on measurement of drum acceleration). The specification is applicable to all subgrade, subbase, and base materials and recycled materials that can be compacted dynamically and statically. For soils compacted dynamically, measurement occurs during compaction. For soils compacted statically, dynamic measurement occurs after static compaction. If the fine-grained portion [< 0.06 mm (0.002 in)] exceeds 15%, moisture content must be given special attention; however, moisture content criteria are not specified. 220.127.116.11ââ âMethodâ1:âAcceptanceâBasedâonâ Calibration The more recently developed Austrian/ISSMGE roller- integrated CCC method involves the development of a re- lationship between roller MV and the initial PLT modulus E V1 or E LWD . Density spot testing is allowed as an alternative, although it is not recommended. Calibration is required over the entire width of the construction site and for a length of at least 100 m (328 ft) for each material (subgrade, subbase, and base). Roller-integrated measurement must be carried out with constant roller parameters (frequency, amplitude, and forward velocity) throughout calibration. Roller MV data are captured during each measurement run, and subsequent PLT or LWD testing is performed at values of low, medium, and high roller MV (see Figure 2.23). PLT is required at a mini- mum of nine locations. If LWD testing is used, the average of language permitting the use of automatic feedback control intelligent compaction rollers during compaction but will prohibit their use during calibration and acceptance testing. 18.104.22.168â âCCCâtoâIdentifyâWeakâAreasâforâSpotâTesting In this approach, CCC is used to map the compacted area. The weakest spots are identified from the roller-generated maps for spot testing (density methods or PLT). A mini- mum number of spot tests are specified (e.g., four per 5,000 m2). To meet acceptance, each density or E V2 value must be greater than or equal to the desired value. If acceptance is not achieved, the soil must be reworked until the criterion is met. Assuming roller operating parameters are held constant and the soil, moisture, and subsurface have not changed, then by inference all other areas of the map meet acceptance. There is no initial calibration required for this approach. This ap- proach is the more common of the two approaches used in Germany (G. BrÃ¤u, personal communication, 2008). 2.3.2 austrian/ISSmge Specifications Austria first introduced roller-integrated CCC specifica- tions in 1990 with revisions in 1993 and 1999. Further revi- sions are not currently being considered. The ISSMGE re- cently developed recommended CCC specifications (ISSMGE 2005), largely based on Austrian standards (Adam 2007). The Austrian/ISSMGE specifications allow two different ap- proaches for roller-integrated CCC. The first approach in- volves acceptance testing using a regression-based correlation developed during on-site calibration. An alternative approach, recommended for small sites or where calibration cannot be Figure 2.23. Illustration of calibration approach for Austrian/ISSMGE specifications.
four E LWD values at a minimum of nine locations is reported (hence 36 LWD tests are required). The engineer of record is given the authority to design the rolling and measurement pattern used during calibration. Linear regression analysis is performed on the resulting roller MV versus E V1 or E LWD values (see Figure 2.24). The regression coefficient R2 must be â¥ 0.5; additional PLT or LWD tests may be performed to achieve R2 â¥ 0.5. The engi- neer of record may remove outliers if good cause exists. This approach may not be carried forward to production QA if R2 < 0.5. Using the regression equation and a specified E V1 or E LWD (see Table 2.3 for Austrian values) leads to determi- nation of a minimum roller MV (MIN) and a mean roller MV (ME). As illustrated in Figure 2.24, the MIN corresponds to 0.95 E V1 or E LWD , and the ME corresponds to 1.05 E V1 or E LWD . The MAX value is defined to be 1.5 MIN. The Austrian/ ISSMGE acceptance criteria are summarized as follows: â¢ The mean roller MV must be â¥ ME; â¢ 100% of roller MVs must be â¥ 0.8 MIN; â¢ 90% of roller MVs must be â¥ MIN. In addition to these requirements, compaction must be con- tinued until the mean roller MV is less than 5% greater than the mean value from the previous pass. The Austrian/ISSMGE specification also requires the following uniformity criteria: â¢ If 100% of roller MVs â¥ MIN, then the roller MV coef- ficient of variation (COV) for the entire area must be â¤ 20%. â¢ If 0.8 MIN â¤ minimum roller MV â¤ MIN, then 100% of roller MVs must be â¤ MAX = 1.5 MIN. The recommended control area over which acceptance should be performed has traditionally been 100 m (328 ft) long by the width of the roadway. However, recent experience with 200- to 500-m (656- to 1,640-ft)-long control areas has shown effective results (D. Adam, personal communication, 2008). These ISSMGE/Austrian correlations and acceptance criteria are valid for roller/soil contact and partial loss of con- tact roller operation. The Austrian/ISSMGE specifications permit measurement during double jump mode; however, a separate calibration is required for such operation. 22.214.171.124â âMethodâ2:âAcceptanceâBasedâonâPercentageâ ChangeâofâMVs For small construction sites and areas where calibration cannot be reasonably performed, the Austrian/ISSMGE rec- ommends the following method. Compaction should be con- tinued until the mean roller MV is less than 5% greater than the mean roller MV from the previous pass. Subsequently, PLT or LWD testing is conducted at the weakest area as de- termined by the roller MV output. The E V1 or E LWD must be greater than or equal to the required value (e.g., Table 2.3 for Austria). A minimum of three PLT or nine LWD tests must be performed in the weakest area. 2.3.3 Swedish CCC Specifications Specifications for the use of CCC on unbound materials in Sweden were first introduced in 1994; current use of roller- integrated CCC is governed by 2005 specifications (ATB Vag 2005). The QA of unbound material is mandated at two sur- face levels: (1) top of the base course and (2) a layer 300 to 750 mm (1 to 2.5 ft) below the top of the base layer. Typi- cally, Swedish construction includes a 300- to 700-mm (1- to 2.3-ft)-thick base layer and a 300- to 500-mm (1- to 1.6-ft) subbase or frost protection layer. Therefore, QA is typically performed on the surface of the base and subbase layers. QA is not required for the subgrade due primarily to the consid- Figure 2.24. Roller MV vs. EV1 /ELWD regression and key parameters in Austrian/ISSMGE specifications. table 2.3. EV1 and ELWD values required (Austria). Level E V1 (MN/m2) 1 m below subgradea 15 (cohesive); 20 (cohesionless) Top of subgrade 25 (cohesive); 35 (cohesionless) Top of subbase 60 (rounded); 72 (angular) Top of base 75 (rounded); 90 (angular) Level E LWD (MN/m2) 1 m below subgradea 18 (cohesive); 24 (cohesionless) Top of subgrade 30 (cohesive); 38 (cohesionless) Top of subbase 58 (rounded); 68 (angular) Top of base 70 (rounded); 82 (angular) aIf fill section is to be constructed.
â erable thickness of base and subbase layers used. The maxi- mum percentage of particles less than 0.06 mm (0.002 in) permitted in base and subbase layers is 7%; therefore, by de- fault CCC is only performed on material with predominantly cohesionless soil. Swedish specifications permit the use of roller-integrated CCC to identify weak spots for PLT. First, it is useful to ex- plain the general QA specification. In Sweden, conventional QA of base and subbase layers is based solely on PLTs per- formed at a minimum of eight randomly selected locations within each 5,000 m2 (1,993 yd2) control area. Density and moisture QA are not prescribed. The number of tests can be reduced to five if no previous control area has failed or if the standard deviation Ï is small. The unload-reload PLT deformation modulus E V2 and the ratio E V1 /E V2 are used. All measured E V2 values must exceed a layer-dependent mini- mum value for acceptance (see Table 2.4). The average E V2 should also meet the criteria summarized in Table 2.4. If criterion 2 is violated, an alternative can be used for the top 500 mm (1.6 ft) only. table 2.4. Unbound material acceptance criteria [per 5,000-m2 (1,993-yd2) control areaï¿½. Depth Below Base Course Surface (mm) N Asphalt Pavement Concrete Pavement (1) E V2(min) (MPa) (2) E V2(ave) (MPa) (3) E V2 /E V1 Alternative if (2) not met (1) E V2(min) (MPa) (2) E V2(ave) (MPa) (3) E V2 /E V1 Alternative if (2) not met 0â250 8 125 â¥ 140 + 0.96Ï â¤ 2.8 105 â¥ 120 + 0.96Ï â¤ 2.8 0â250 5 125 â¥ 140 + 0.83Ï 1 + 0.013 E V2 105 â¥ 120 + 0.83Ï 1 + 0.015 E V2 251â500 8 32 â¥ 40 + 0.96Ï â¤ 3.5 45 â¥ 55 + 0.96Ï â¤ 3.5 251â500 5 32 â¥ 40 + 0.83Ï â¤ 1 + 0.063 E V2 45 â¥ 55 + 0.83Ï â¤ + 0.046 E V2 500â550 8 32 â¥ 40 + 0.96Ï NA 45 â¥ 55 + 0.96Ï NA 500â550 5 32 â¥ 40 + 0.83Ï NA 45 â¥ 55 + 0.83Ï NA 551â650 8 20 â¥ 30 + 0.96Ï NA 30 â¥ 35 + 0.96Ï NA 551â650 5 20 â¥ 30 + 0.83Ï NA 30 â¥ 35 + 0.83Ï NA 651â750 8 15 â¥ 20 + 0.96Ï NA 20 â¥ 25 + 0.96Ï NA 651â750 5 15 â¥ 20 + 0.83Ï NA 20 â¥ 25 + 0.83Ï NA NA= not applicable When using roller-integrated CCC, the number of PLTs can be reduced to two. The PLTs are conducted at the two weakest areas as indicated by the roller MV data map. The number of PLTs can be reduced from two to one if no control area has failed the test or the previous control areas show small varia- tions. The criteria for acceptance are summarized in Table 2.5. The E V2 value in these points must not be lower than the threshold value, which is different for different levels below the top of the base layer and for flexible and rigid pavements. For granular base materials an additional criterion based on the E V2 /E V1 ratio must also be fulfilled. LWD testing cannot currently be used in place of static PLT for QA of base and subbase materials. Sweden does recom- mend QC/QA during subgrade compaction via either PLT or LWD testing. LWD testing can be used instead of PLT âif similar results can be shown.â The Swedish specifications pro- vide recommended E V2 and E LWD for depths of 800 mm (2.6 ft) or greater (see Table 2.6). A note within the specifications states that the average of five LWD tests can be used over a 2,500-m2 (997-yd2) area. table 2.5. Unbound material acceptance criteria when CCC used [per 5,000-m2 (1,993-yd2) control areaï¿½. Depth Below Base Course Surface (mm) N Asphalt Pavement Concrete Pavement (1) E V2(min) (MPa) (2) E V2 /E V1 Alternative if (2) not met (1) E V2(min) (MPa) (2) E V2 /E V1 Alternative if (2) not met 0â250 1â2 125 â¤ 1 + 0.0136 E V2 105 â¤ 1 + 0.0162 E V2 251â500 1â2 32 â¤ 1 + 0.078 E V2 45 â¤ 1 + 0.056 E V2 500â550 1â2 32 NA 45 NA 551â650 1â2 20 NA 30 NA 651â750 1â2 15 NA 20 NA
2.3.4 minnesota doT pilot Specifications In 2007 the Minnesota Department of Transportation (Mn/DOT) developed pilot specifications for QC/QA of granular and nongranular embankment soil compaction using CCC and/or LWD. At the time of this writing, Mn/ DOT was in the process of revising these specifications; the most recent version is available online at http://www.dot. state.mn.us/materials/gbintellc.html. The 2007 specification required QC by the contractor and QA by the engineer on designated proof layers to ensure compliance with control- strip-determined roller-integrated measurement target val- ues and LWD target values. Proof layers are designated at the finished subgrade level (directly beneath the base) and at certain additional levels depending on the height of the constructed embankment. Additional proof layers are re- quired for every 600 mm (2 ft) of placed granular soil thick- ness and every 300 mm (1 ft) of placed nongranular soil thickness. The engineer has the authority to modify proof layer designations. The Mn/DOT specification requires construction of control strips to determine the intelligent compaction tar- get value (IC-TV) for each type and/or source of soil. Note that Mn/DOTâs use of the term âintelligent compactionâ is equivalent to CCC as defined in this study in that automatic feedback control of roller operating parameters is not per- mitted during measurement. Additional control strips are required if variations in material properties that affect the IC-TV are observed by the engineer. Each control strip must be at least 100 m (328 ft) long and at least 10 m (33 ft) wide, or as determined by the engineer. Lift thickness must be equal to planned thickness during production. To determine the moisture sensitivity correction for the IC-TV, a control strip is constructed at or near each extreme of 65% and 95% of standard Proctor optimum moistureâthe moisture content limits specified in Mn/DOT earthwork construction. The resulting data are utilized to produce a moisture correction trend line showing a linear relationship of the IC-TV with moisture content. The control strip construction procedure is as follows: table 2.6. Recommended pLt and LWD QA values at depth. Depth Below Base Course Surface (mm) Construction with Only Base and Subbase Material Above Crushed Rock Construction with Only Base and Subbase Material Above Sand Subgrade E V2 (MPa) E LWD (MPa) E V2 (MPa) E LWD (MPa) 800 12 10â15 16 12â18 900 9 8â12 11 10â14 1,000 6 5â8 8 7â11 1,100 4 4â5 5 5â8 1,200 3 3 4 3â5 1,300 2 2 3 3 â¢ The bottom of the excavation is mapped with the roller to create a base map. This map is reviewed by the engi- neer to ensure that the control strip subsurface is uniform and to identify areas that must be corrected prior to fill placement. â¢ The embankment soil is placed in lifts; each lift is com- pacted with repeated passes and with roller compaction measurement. The optimum value is reached when addi- tional roller passes do not result in a significant increase in the roller MV as determined by the engineer. â¢ Moisture content testing is required at a minimum of 1 per 3,000 m3 of earthwork; the moisture must be maintained within 65% to 95% of standard Proctor optimum. â¢ Lift placement and CCC is repeated for additional lifts until the level-of-proof layer has been reached. â¢ The control strip IC-TV is defined to be the optimum MV obtained from the roller measurements during construc- tion of the control strip. The optimum value is reached when additional passes do not result in a significant in- crease in MV, as determined by the engineer of record. The IC-TV is defined such that 90% of the MVs are greater than 90% of the IC-TV. The IC-TV values for control strips at moistures near 65% and 95% of optimum are used to create a moisture correction trend line. It should be noted that the Mn/DOT pilot specifi- cations also require LWD testing (three per proof layer) and the establishment of LWD target value (LWD-TV) for each proof layer. The LWD-TV is corrected for moisture in a way similar to that for IC-TV. During QA the engineer observes the final compaction re- cording pass of the roller on each proof layer. For acceptance at each proof layer during general production operations, all segments shall be compacted so that at least 90% of the MVs are at least 90% of the moisture-corrected IC-TV prior to placing the next lift. All of the MVs must be at least 80% of the moisture-corrected IC-TV. The contractor must recom- pact (and dry or add moisture as needed) until all areas meet these acceptance criteria.
â If a significant portion of the grade is more than 20% in ex- cess of the selected corrected IC-TV, the engineer shall reeval- uate the selection of the applicable control-strip-corrected IC-TV. If an applicable corrected IC-TV is not available, the contractor shall construct an additional control strip to reflect the potential changes in compaction characteristics. Control section size criteria are currently under development. The engineer will also perform an LWD test and a mois- ture content test at the minimum rate of one LWD test measurement per proof layer, per 300 m (1,000 ft) for the entire width of embankment being constructed during each operation. The engineer may perform additional LWD tests and moisture content tests in areas that visually ap- pear to be noncompliant or as determined by the engineer. Each LWD test measurement taken shall be at least 90% but not more than 120% of the moisture-corrected LWD-TV obtained on the applicable control strip prior to placing the next lift. The contractor shall recompact (and dry or add moisture as needed) to all areas that do not meet these requirements.