Intelligent Soil Compaction Systems (2010)

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 5 Intelligent systems sense their environment and adapt to improve performance. In the context of current roller tech- nology, IC involves sensing via vibration-based measurement and adapting via feedback control of roller parameters. Auto- matic feedback control (AFC) of the applied excitation force has been enabled by the recent introduction of variable exci- tation force amplitude and, in some cases, variable excitation forcing frequency. Here, the roller adapts per the roller MV. Based on the unpredictable dependence of roller MV on ex- citation amplitude and frequency (characterized in Chapters 2, 3, and 4), specifications for QA using current CCC tech- nology should require roller operation with constant opera- tional parameters (amplitude, frequency, speed, direction). Therefore, CCC-based QA should not be performed during AFC operation (in its current form). AFC may be used during the compaction process (i.e., during non-QA roller passes). To this end, this chapter explores current AFC used by IC rollers and characterizes the benefits of using AFC for soil compaction. At the time of this study, three manufacturers offered com- mercially available AFC of excitation force: Bomag, Case/ Ammann, and Dynapac. At a minimum level, each manufac- turer controls the vertical excitation force amplitude to pre- vent unstable jump mode vibration of the roller (see Section 2.1.2). When the measurement system senses jump mode, the vertical excitation amplitude is decreased until the measure- ment system indicates stable vibration. This level of AFC is aimed at protecting the roller from accelerated wear and the operator from chaotic response of the roller. The philosophy behind additional AFC varies across manufacturers. Dynapac does not perform additional AFC. Its philosophy regarding optimal compaction is to maximize excitation force during compaction (I. Nordfelt of Dynapac, 2009, personal commu- nication). At the end of compaction, Dynapac recommends low-amplitude excitation as a finishing pass; however, this is not automatically performed. As described in Section 2.2.2, Bomag controls vertical excitation amplitude based on the relationship of the current E vib to a limit E vib . The limit E vib is entered into the onboard computer by the operator prior to compaction. The vertical excitation is maximized within one of five levels chosen by the operator and decreased if E vib exceeds the limit. As described in Section 2.2.1, Ammann and Case/Ammann control the excitation amplitude and fre- quency to maintain one of three levels of force transmitted to the soil. The operational approach and responsiveness of the Ammann/Case and Bomag AFC systems is presented in Section 5.1. AFC-based IC aims to provide improved compaction ef- ficiency (e.g., fewer passes) as well as more uniform compac- tion (e.g., Adam & Kopf 2004). Due to the amplitude and frequency dependence of roller MVs, it is difficult to assess the benefits of AFC using roller MVs. The benefits of AFC- based IC must, therefore, be determined via independent assessment of compaction (i.e., density testing). Section 5.2 presents an investigation of compaction efficiency and uni- formity from IC based on spot-test measurements. 5.1 Operational Evaluation Of AFC-Based IC 5.1.1 Bomag variocontrol Bomag AFC (termed Variocontrol) was used over an area of compacted material where a wide range of soil stiffness was present (TB MN42). The data in Figure 5.1 illustrate the op- erational principle of Bomag AFC. The roller was set to operate at maximum theoretical amplitude, A max = 1.5 mm (0.059 in), and a preset limit of E vib = 45 MPa. Recall that A is a surrogate for vertical excitation force (Equation 2.2). The operator must choose one of five maximum eccentric force levels [A max = 0.6, 1.1, 1.5, 2.1, 2.5 mm (0.024, 0.043, 0.059, 0.083, 0.098 in)] and must prescribe a limit E vib . Here, the choices of A max = 1.5 mm analysis of intelligent soil Compaction

 (0.059 in) and limit E vib = 45 MPa imply medium compactive effort and medium soil stiffness. As shown in Figures 5.1a and b, if E vib > limit E vib , A is reduced to 0.6 mm (0.024 in). When E vib < limit E vib , A is increased to A max . The E vib record during AFC suggests that a very soft zone exists from x = 7 to 30 m. The implication within the Vario- control system is that the soil in this area requires further compaction, and A max is therefore prescribed. In actuality the material may be a different soil, it may be too moist, or the sublift soil may be soft, and additional passes would not improve conditions. For x < 7 m and x > 30 m, the implica- tion is that the soil has reached near-final compaction and the lowest-possible A should be prescribed to prevent over- compaction or loosening. Here, A = 0.6 mm (0.024 in) is pre- scribed because it is the lowest A for which E vib can be reliably measured. The dependence of E vib (and all roller MVs) on A can result in a misleading record of soil stiffness while operating in AFC mode. This is evidenced by the comparison of E vib records collected during AFC and constant A roller passes in Figure 5.1a. Constant A = 0.6 mm (0.024 in) and A = 1.5 mm (0.059 in) roller passes were performed prior to the AFC pass. Com- parison of E vib records during constant A operation revealed quite a different stiffness profile. The nature and degree of roller MV dependence on A (termed MV-A dependence) are a complex function of material modulus functions and layered structure (see Chapter 4). A roller MV may increase with in- creasing A (termed positive MV-A dependence), as shown for Figure 5.1. Bomag Variocontrol with f = 28 Hz, limit Evib = 45 MPa, and Amax = 1.5 mm (0.059 in; TB MN42).

  x > 30 m in Figure 5.1a. Conversely, a roller MV may decrease with increasing A (termed negative MV-A dependence), as shown for 7 < x < 30 m. The result of MV-A dependence can be a different and more variable roller MV record while op- erating in AFC mode than in constant A mode. In Figure 5.1a, constant A operation and measurement reveal a much more uniform subsurface stiffness than that measured during AFC. Due to the E vib -A dependence, mea- surement during AFC reveals an artificial and misleading level of variability in soil stiffness. The influence of E vib -A dependence can be observed from x = 6 to 7 m, as high- lighted in Figures 5.1c and d. Here, E vib decreases below the limit E vib , triggering an increase in A from 0.6 to 1.5 mm (0.024 to 0.059 in). The A = 0.6 mm (0.024 in) pass sug- gests a subtle decrease in soil stiffness; however, due to the strong negative MV-A dependence, E vib decreases from 40 to 15 MPa. While a legitimate decrease in soil stiffness had to be present to trigger the increase in A, the soft area sug- gested by the E vib record is primarily a result of strong MV-A dependence. The MV-A dependence described above can also trigger AFC changes in A. Figure 5.2 shows Bomag AFC operation with a limit E vib = 100 MPa and A max = 2.5 mm (0.098 in). At x = 31 m, E vib eclipses 100 MPa. The resulting decrease in A coupled with the strong positive E vib -A dependence causes E vib to dip below 100 MPa and trigger an increase in A. This increase in A and the positive E vib -A dependence again causes E vib to eclipse 100 MPa. The sequential decrease, increase, and decrease in A is driven by E vib -A dependence and not by a legitimate soft area. A second aspect of interest is the responsiveness of AFC (i.e., how quickly the roller adapts to threshold changes in soil stiffness). A closer look at two zones highlighted in Figures 5.1c, d, and f illustrates the responsiveness. The limit E vib was crossed six times during the AFC pass. The AFC reacted (i.e., A began to change) within 0.2 to 0.3 m (0.7 to 1 ft) of the po- sition where the limit E vib was crossed. The response distance, defined as the distance required for the roller to completely modify A, was found to be approximately 1 m (3.3 ft). These reaction and response distances speak to the ability of AFC to address very localized areas of poor compaction. The re- action and response distances are a function of roller speed; here, the roller was traveling at a typical operating speed of 1 m/s (1 ft/s). Bomag AFC was used on a second area where variable stiffness was present (TB MN43). Constant A passes reveal a strong positive E vib -A dependence. In one area (x = 55 to 70 m), the surface gravel was underlain by old asphalt pavement, as reflected by the very high E vib values in Figure 5.3a. During a constant A = 2.5 mm (0.098 in) pass, the Bomag roller ex- hibited jump mode vibration from x = 59 to 67 m as a result of the high underlying stiffness (Figure 5.3). During an A = 0.6 mm (0.024 in) pass, the roller did not experience jump- ing. During AFC operation with a limit E vib = 100 MPa, A was reduced for much of the stiff zone (see Figure 5.4), preventing the roller from entering jump mode. Similar to the behavior observed in Figure 5.2, Bomag AFC is artificially engaged by the positive E vib -A dependence. At numerous locations de- picted in Figure 5.3c, E vib eclipses 100 MPa and triggers a de- crease in A. What follows are several cycles of E vib increase and Figure 5.2. Bomag Variocontrol with f = 28 Hz, limit Evib = 100 MPa, and Amax = 2.5 mm (0.098 in; TB MN42).

0 Figure 5.3. Bomag operation with and without feedback control, with Amax = 2.5 mm (0.098 in) and Evib limit = 100 MPa (TB MN43). Figure 5.4. Comparison of Ammann constant-amplitude mode with AFC mode (TB MN43); A = theoretical amplitude and zd is measured amplitude.

  decrease due to the AFC of A and the positive E vib -A depen- dence. The resulting variable E vib record is artificial. 5.1.2 ammann aCe As described in Section 2.2.1, the Ammann AFC system (termed Ammann Compaction Expert, ACE) maintains one of three contact force settings F s(max) . The F s(max) is controlled by varying the eccentric mass positions (and thus A) and the frequency. Frequency is controlled to maintain the phase angle between vertical drum displacement and eccentric force between 140º and 160º. The level of A is then varied to maintain low [14 kN (3.1 kip)], medium [20 kN (4.5 kip)], or high (unlimited) F s(max) (Anderegg & Kaufmann 2004). The Case/Ammann roller was used in constant A and AFC mode on a variable stiffness test bed (TB MN43). Two con- stant A passes were performed [A = 0.4 mm (0.016 in) and 1.7 mm (0.067 in)], and two AFC mode passes were performed [F s(max) = 20 kN (4.5 kip) and F s(max) = unlimited]. Figure 5.4 presents the measured soil stiffness k s , theoretical amplitude A, actual amplitude z ˆ , and vibration frequency f. The F s(max) and phase angle data are not stored by the roller measurement sys- tem. Values of k s varied from 15 to 80 kN/m and represented a range of soft to stiff conditions. During constant A opera- tion, z d remained reasonably consistent with A in soft areas. In stiffer areas, z d can be significantly greater than A (e.g., x = 45 to 70 m). The Ammann roller exhibited jump behavior from x = 55 to 60 m during constant A = 1.7 mm operation, wherein the measurement of k s and z d is unreliable (values go to zero). To maintain constant F s(max) while soil stiffness increases during AFC passes, the controller decreases A and increases excitation frequency. This occurs during both AFC passes. From x = 55 to 63 m, the soil stiffness remained fairly con- stant and yields a constant F s(max) . In this case the controller does nothing since the F s(max) is at the desired value. From x = 63 to 80 m, the stiffness decreases. To maintain F s(max) , the controller increases A. Vibration amplitude and frequency seemed to have an influence on k s values. The k s data from each pass trend similarly; however, the k s at discrete locations could be highly variable between passes. For instance, at x = 47 m, k s ranges between 25 and 50 MN/m. 5.2 Influence of AFC on Compaction Test beds in Colorado and Maryland were selected to in- vestigate the influence of AFC on compaction efficiency and uniformity according to measurement values (MVs) and spot- test measurements. Operating the rollers in AFC and constant- amplitude modes for alternate roller passes provided the op- portunity to further evaluate AFC response distance and the effectiveness of AFC in preventing roller jumping. The test beds were selected primarily because they represented dif- ferent underlying layer conditionsΔrelatively uniform versus highly variable. 5.2.1 Uniform Underlying layer Conditions—Bomag Two granular base layers (TBs CO16 and CO17) were prepared side by side and compacted in AFC and constant- amplitude modes, respectively, to provide similar conditions for comparing compaction efficiency and resulting unifor- mity for different machine operations. E vib measurements were recorded for the same test bed in constant-amplitude mode and in AFC mode with two different limit E vib values. During test strip construction, spot-test measurements were obtained at several intermediate compaction passes. The test bed conditions and material types are presented in Figure 5.5. The moisture content of the compaction layer material was relatively uniform between 4.0% and 5.6%. The underlying compacted subbase layers (TBs CO11 and CO12) were rela- tively uniform and similar in terms of coefficient of variation (COV) according to spot-test measurements (see Table 5.1). TB CO16 was compacted in AFC mode for passes 1 through 12 and in manual mode for passes 13 and 14. AFC with limit E vib = 100 MPa was used for passes 1 through 8 with A max = 2.50 mm (0.098 in). For passes 9 through 12, limit E vib = 120 MPa with A max = 1.90 mm (0.075 in) [v = 4 km/h (3.6 ft/s) constant throughout]. Spot-test measurements (γ d , E LWD-Z2 , and California Bearing Ratio [CBR]) were obtained after 4, 8, 12, and 14 passes. Figure 5.6 shows roller MVs (solid lines) in comparison with spot-test measurements. For reference, final pass spot-test and roller measurements for the underly- ing subbase layer (TB CO11) are also presented. Figure 5.7 shows the roller MVs in comparison with amplitude mea- surements during compaction in AFC mode for passes 4, 8, and 12 and in manual mode for pass 14. TB CO17 was compacted in manual mode with A = 0.70 mm (0.028 in) for passes 1 through 8, A = 1.90 mm (0.075 in) for passes 9 and 10, and A = 0.70 mm (0.028 in) for passes 11 and 12 [v = 4 km/h (3.6 ft/s) constant throughout]. Spot-test measurements (γ d , E LWD-Z2 , and CBR) were obtained after 4, 8, 10, and 12 roller passes. Roller MVs in comparison with spot- test measurements are presented in Figure 5.8. For reference, final pass measurements for the underlying subbase layer (TB CO12) are provided in Figure 5.8. Roller MV and spot-test measurement compaction curves for TBs CO16 and CO17 are provided in Figure 5.9. The aver- age amplitude during AFC mode compaction for TB CO16 decreased from pass 1 to pass 8, consistent with increasing average E vib values. The COV of roller MV and spot-test mea- surements after pass 8 for the two test beds are summarized

 in Table 5.1 for purposes of comparing the final compaction pass uniformity. Compaction curves from spot-test mea- surement average values from the two test beds show similar trends with no noticeable difference. The COV of spot-test measurements is similar after pass 8 for the two test beds. The COV of roller MVs for TB CO17 (21%) is higher than the COV of MVs for TB CO16 (12%). Given the significant fluctuations in A (see Figure 5.8) and the roller MV-A de- pendence described in Figures 5.1 and 5.2 and visible in Fig- ure 5.9, lower COV of roller MVs does not necessarily reflect increased uniformity. Final-pass roller MVs from constant- amplitude operation (pass 12 or pass 14) for the two test beds produced similar COV (21%). During constant-amplitude operation, the COV of roller MVs provides a good measure of uniformity. The synopsis of results from this test bed study is that AFC operations did not produce improved compac- tion or uniformity compared to constant-amplitude mode compaction for these test bed conditions [i.e., 0.15 m (0.5 ft) thick layer of base atop uniform subbase]. 5.2.2 nonuniform Underlying layer Conditions—Bomag and dynapac Results from granular base layer (TBs MD11, 13, 14) placed over compacted granular base layer (TBs MD6, 8, 9), and un- derlain by subgrade (TBs MD2, 4, 5) are presented in this sec- tion. The influence of heterogeneous underlying layer con- ditions on MVs obtained in AFC mode is evaluated here in comparison with MVs obtained in constant-amplitude mode for two different roller MVs (CMV D and E vib ). These results are particularly interesting in that they demonstrate the influ- ence of underlying layer conditions on AFC operations and resulting MVs. For CMV D measurements, AFC compaction was performed using a preselected A max setting. For Dynapac IC, the vibration amplitude in AFC mode is controlled to pre- vent jump mode only. The degree of jump is measured by the bouncing value (BV). Reportedly, the vibration amplitude was reduced when BV approached 14. CMV D was arbitrarily reported in the output as 250 for BV > 25. Figure 5.5. Plan and profile views of granular base (TB CO16/17), subbase (TB CO11/12), and subgrade layer (TB CO1/2) construction. table 5.1. Comparison of CoV of roller MVs and spot-test measurements. Parameter COV (%) CO16 CO11a CO17 CO12 Pass 8 (AFC) Pass 14 (Manual) Pass 12 (Manual) Pass 8 (Manual) Pass 12 (Manual) Pass 12 (Manual) E vib 12 21 12 21 21 18 E LWD-Z2 12 10 14 10 10 15 γ d 1 — 2 1 2 2 CBR 22 — 21 29 25 21 a Underlying subbase layer.

  Figure 5.6. Comparison of roller MVs with spot-test measurements from TB CO16 granular base layer (MVs ob- tained in AFC and constant-amplitude modes) and underlying TB CO11 granular subbase layer (MVs obtained in constant-amplitude mode). Figure 5.7. Roller MV and A measurements on TB CO16 granular base layer.

 Figure 5.8. Comparison of roller MV with spot-test measurements from TB CO17 base layer (AASHTO: A-1-a) un- derlain by CO12 subbase layer (AASHTO: A-1-a) (MVs obtained in constant-amplitude mode). The geometry of the soil layers and associated soil classi- fications are presented in Figure 5.10. The granular base lay- ers were placed at relatively consistent moisture content (w = 3.5% to 5.0%). The subgrade TB MD2 was relatively soft and homogeneous, and TBs MD4 and 5 were relatively stiff and heterogeneous. TBs MD4 and 5 consisted of an isolated zone of fractured rock mixed with the subgrade soil. For TB MD2 the average E LWD-Z2 = 5.9 MPa, with COV = 10%; for TB MD4 the average E LWD-Z2 = 19.8 MPa, with COV = 45%; and for TB MD5 the average E LWD-Z2 = 30 MPa, with COV = 48%. After compaction passes, the test beds were mapped with the rollers using constant-amplitude and AFC settings. E vib measurements were obtained in constant-amplitude mode with nominal A = 0.70 mm (0.028 in) and A = 1.90 mm (0.075 in), and in AFC mode with limit E vib = 40 MPa and 80 MPa [f = 30 Hz and v = 4 km/h (3.6 ft/s) were constant] with A max = 2.50 mm (0.098 in). CMV D measurements were obtained in constant-amplitude mode with nominal A = 0.80 mm (0.031 in) and A = 2.40 mm (0.094 in) and in AFC mode with maximum A = 2.50 mm (0.098 in) [f = 28 Hz and v = 4 km/h (3.6 ft/s) were constant]. Figure 5.11 presents E vib values obtained in constant- amplitude and AFC modes for TBs MD11, 13, and 14. E FWD measurements are shown in comparison with E vib values dur- ing constant A = 0.70 mm (0.028 in) operation. Amplitude measurements (shown as gray lines) are also presented in comparison with E vib measurements (shown as black lines) obtained in AFC mode. Similarly, Figure 5.12 presents CMV D

  Figure 5.9. Comparison of roller MV and spot-test measurement compaction curves from TB CO16 (operated in constant-amplitude mode) and TB CO17 (operated in AFC mode).

 Figure 5.10. Plan and profile views of granular base (TBs MD6, 8, 9, 11, 13, 14) and subgrade layer (TBs MD2, 4, 5) construction (layer thicknesses are nominal). measurements obtained in constant-amplitude and AFC modes for TBs MD11, 13, and 14, along with E FWD measure- ments in comparison with CMV D measurements with con- stant A = 0.80 mm (0.031 in) operation. The underlying layer roller MVs are also shown in Figures 5.11 and 5.12 for reference. Both roller MVs and E FWD measurements showed similar variation along the test beds, with relatively soft and homogeneous conditions on TB MD11 and relatively stiff and heterogeneous conditions on TBs MD13 and 14. E vib measurements in constant-amplitude mode with A = 1.90 mm (0.075 in) showed roller jumping (jump values > 0) on TBs MD13 and 14, generally at locations with E FWD-D3 > 120 MPa and E vib > 100 MPa. During AFC mode compac- tion with maximum E vib = 80 MPa, no roller jumping was observed across the test beds and the amplitude was effec- tively reduced to A = 0.6 mm (0.024 in) at locations where E vib was > 80 MPa. For AFC operation with maximum E vib = 120 MPa, the amplitude was reduced to A = 0.6 mm (0.024 in) where E vib ≥ 120 MPa, but roller jumping was not prevented. Response distance for this test bed study was about 0.5 to 1 m (1.6 to 3.3 ft). CMV D measurements in constant-amplitude mode with A = 2.40 mm (0.094 in) showed roller jumping at several loca- tions across TBs MD13 and 14 with E FWD > 120 MPa. In AFC mode the roller generally maintained A > 2.00 mm (0.079 in) for the three test beds. No jumping was observed on TB MD13. Jumping was noticed on TB MD14 from about 0 to 15 m, where the amplitude increased despite the increase in BVs. Response distance for this test bed study was about 0.5 to 1.0 m (1.6 to 3.3 ft). 5.3 Conclusions The following conclusions can be drawn from the results presented in this chapter: • The dependence of roller MVs on A can provide a misleading record of soil stiffness when operating in AFC mode. Both positive and negative MV-A dependence were observed dur- ing testing and resulted in an artificial and misleading level of variability/uniformity in soil stiffness. For this reason, AFC that involves changing A is not recommended during measurement passes in QA (see Chapter 7). • The roller MV-A dependence can trigger AFC changes in A. This is problematic when roller MVs hover around a target or limit MV. • The response distance of AFC-based IC evaluated here was found to be approximately 1 m when operating at typical roller speeds, indicating that rollers in AFC mode can respond to relatively localized changes in soil conditions. AFC-based IC rollers would struggle to react to very localized areas of soft soil (e.g., above a buried pipe or narrow backfilled trench). • An investigation of the influence of AFC-based IC on com- paction efficiency and uniformity revealed no measurable benefits of AFC mode over constant-amplitude mode. Spot- test measurements obtained from side-by-side test beds (e.g., one compacted using AFC and one compacted with constant low-amplitude vibration) did not show any significant dif- ferences in soil compaction or uniformity of soil properties. Final-pass constant-amplitude roller MVs recorded on both test beds revealed no difference in uniformity.

  Figure 5.11. Roller MV, amplitude, and jump measurements from AFC and manual high- and low-amplitude set- tings with comparison to EFWD measurements and underlying layer roller MVs.

 Figure 5.12. Roller MV, amplitude, and BV measurements from AFC and manual high- and low-amplitude set- tings with comparison to EFWD measurements and underlying layer roller MVs.