Strength and Stress Estimation in Soil Bentonite Slurry Trench Cutoff Walls using Cone Penetration Test Data

Transcription

1 Strength and Stress Estimation in Soil Bentonite Slurry Trenh Cutoff Walls using Cone Penetration Test Data D.G. Ruffing 1, M.ASCE, P.E., J.C. Evans 2, F.ASCE, P.E., and C.R. Ryan 3, M.ASCE, P.E. 1 Geo-Solutions, th Avenue, ew Kensington, PA 15068; (724) ; druffing@geosolutions.om 2 Department of Civil and Environmental Engineering, Buknell University, Lewisburg, PA 17837; (570) ; evans@buknell.edu 3 Geo-Solutions, th Avenue, ew Kensington, PA 15068; (724) ; ryan@geosolutions.om ABSTRACT The Cone Penetration Test (CPT) is widely used for lassifying soils and assigning soil properties to the subsurfae beause it is robust and an be used to quikly ollet ontinuous data with depth. During and subsequent to the onstrution of a very deep soil-bentonite (SB) slurry trenh ut-off wall in Mayfield SW, Australia, the CPT was used to hek the ontinuity and quality of the SB bakfill material. The numerous CPT profiles onduted in the Mayfield wall provided a unique researh opportunity for examining how shear strength and effetive stress vary with depth. A method is proposed to estimate the undrained shear strength of the Mayfield Wall from the CPT ombined with some vane shear data. The results support previous studies that show that the shear strength and the effetive stress distributions in SB slurry trenh utoff walls are less than would be expeted from a geostati stress distribution. While there is some inrease in shear strength with depth, these data show the inrease is modest and onsistent with the less than geostati stresses. In addition, the authors reommend a general proedure for shear strength and horizontal stress estimation using CPT data in any SB slurry trenh appliation. ITRODUCTIO Continuous readings with depth, a robust testing system, and ease of use make the Cone Penetration Test (CPT) an important tool for lassifying soils and assigning engineering properties to the subsurfae. Although not ommonly speified for quality ontrol (QC) assessment of vertial utoff walls due to measurement auray limitations and ost onstraints, the CPT has been used on a selet number of soilbentonite (SB) slurry trenh utoff wall projets to gauge bakfill homogeneity and to assess in-situ strength development over time. The CPT with pore pressure readings (CPTu) was used as part of the QC program on the deepest SB wall ever built. The wall was onstruted in 2006 in Mayfield, Australia as part of a brownfield redevelopment projet. Two papers written about the projet, Jones et al. (2007) and Ryan and Spaulding (2008), address the wall onstrution and shear strength, respetively. The Mayfield wall was onstruted in a broad alluvial plain bordering the Hunter River near ewastle in ew South Wales, Page 1 of 10

2 Australia. The wall was installed through a medium sand alluvial layer that varied from 30 to 50 meters thik. Sandstone bedrok was enountered aross approximately half of the wall and a fine-grained aqualude over the remainder. At its deepest point, the wall was installed to a depth of 49 m and it has an overall length of 1500 m. The QC testing program for this projet inluded 24 CPTu profiles. Although the state-of-stress in SB utoff walls has been shown to be less than geostati, the stress distribution is not fully understood. Earlier works (Evans et al. 1985, Evans et al. 1995, Filz 1996, Ruffing et al. 2010) explore how the state-ofstress varies with depth and the impat of stress state on the hydrauli ondutivity of SB bakfill material. The main objetive of the Mayfield data analysis was to improve understanding of the magnitude and variability of the undrained shear strength (Su) in SB bakfill and to use these results to estimate the horizontal effetive stress. SHEAR STREGTH ESTIMATIO FROM CPTu DATA Five different relationships between CPTu data and Su, presented in Powell and Lunne (2005), were used to investigate the orrelation between shear strength and depth for the flowable SB fill. Undrained shear strength is highly dependent on the initial stress state, diretion of loading, rate of loading, and stress history of a soil (Mayne 2001). Methods used to predit Su from CPTu data must therefore rely on empirial orrelations developed from data olleted in soils with similar stress and depositional histories. The five shear strength estimation methods (Powell and Lunne 2005) that were evaluated in this study are shown on Table 1. TABLE 1. Investigated Shear Strength Estimation Methods for CPTu Data Method Equation Equation Where: ame umber Exess Pore Pressure Total Cone Resistane vo Theory Theory Effetive Cone Resistane S u S u Δu Δu qt kt vo q S u vo q mean S u mean q t u 2 Su ke Δu = a theoretial one fator that generally falls between 7 and 10, Δu = u 2 - u 0, and u 2 = pore pressure measured just behind one tip shoulder, u 0 = stati pore pressure kt = a theoretial one fator that generally falls between 10 and 20, q t = orreted tip resistane, σ vo = vertial effetive stress = a theoretial one fator, q = unorreted tip resistane, σ vo = vertial effetive stress = a theoretial one fator, q = unorreted tip resistane, σ mean = mean total stress ke = a theoretial one fator that generally falls between 1 and 13, q t = orreted tip resistane, u 2 = pore pressure measured just behind one tip shoulder (1) (2) (3) (4) (5) Page 2 of 10

3 For very soft lays, Powell and Lunne suggest using the method based on exess pore pressures. Sine the bakfill material in a SB wall an be desribed as very soft layey material, this "exess pore pressure" method was the first examined. However, using the exess pore pressure method and data from a representative CPTu log resulted in both unreasonable and infeasible shear strength values. For example, many of the shear strength values resulting from the appliation of this method to the Mayfield CPTu test results were negative. egative shear strength values are learly not possible and therefore this method was deemed inappropriate for SB utoff wall bakfill. Three of the methods, the total one resistane, σvo theory, and σmean theory require knowledge of the state of stress in at least one dimension. As indiated in the introdution, the state of stress in SB walls is not known with suffiient ertitude and thus methods that required input of stress state are onsidered inappropriate for prediting Su from CPTu data olleted in SB utoff walls. Due to the inappliability of these three methods requiring knowledge of the stress state and the infeasible results obtained from the analysis using the exess pore pressure method, the effetive one resistane method remained as the sole potential method for prediting Su vs. depth for the Mayfield Projet, and more broadly for any SB wall. Unlike the four methods just disussed, the effetive one resistane method requires no prior knowledge of the stress state or stati pore pressure, but, like the other methods, does require the user to selet a one fator, ke. Previous studies, as desribed in Powell and Lunne (2005), indiate that ke orrelates well with the pore pressure parameter, Bq. The pore pressure parameter data from the Mayfield data set and the presentation in Powell and Lunne (2005) indiate that a value of 12 for ke would be appropriate. Alternatively, if Su is measured independently by other means on a projet, then the ke fator an be diretly alulated using this additional oupled CPTu-Su data. On the Mayfield projet, the QC program inluded a limited number of vane shear tests to measure Su. In order to assess the one fator on this projet, the oupled CPTu and vane shear data was used to alulate one fators. This analysis also provided an opportunity to investigate whether or not the ke fator varies with depth. The ke fators as a funtion of depth obtained from omparison of the vane shear and CPTu data in the Mayfield wall are shown in Fig. 1 along with a best-fit, power funtion trend line. Page 3 of 10

4 Depth (m) 0 Cone Fator, Power Trendline Corretion Fators FIG 1. Cone Fator, ke, vs. Depth The data presented in Fig. 1 shows onsiderable satter. This satter is probably due to small sand and gravel or other non-representative inlusions to whih both the CPT and the vane shear are known to be sensitive. Small sand or gravel inlusions in this wall ould be expeted from the soils at this site. Despite the satter, the oneptual relationship between one fator and depth indiates that the one fator inreases with depth to an approximate maximum of 15. The arithmeti average of the data in Fig. 1 results in a one fator of The authors note that the average one fator alulated using the vane shear data, 11.5, ompares well to the one fator, ke = 12, predited from the pore pressure parameter disussed earlier. The effetive one resistane method and orretion fator, ke, were then used with the CPTu data to determine the undrained shear strength. The first step in determining Su from CPTu data was to orret the measured tip resistane was for pore pressure effets using the following equation: q t q (1 a) u (6) 2 Where, qt = orreted tip resistane, q = raw tip resistane, a = area ratio (0.73 for this projet), and u2 = pore pressure measured at the shoulder of the one After tip resistane orretion, exessively large values of one resistane found near the bottom of the CPT profiles were removed. These values were likely the result of the one enountering the trenh bottom making them unrepresentative of the strength of the bakfill. In addition to editing out values near the bottom, other Page 4 of 10

5 Depth (m) values at all depths were deleted if they were signifiantly larger than the surrounding data points. The authors believe that the unharateristially high readings observed in some of the profiles may have been aused by piees of gravel suspended in the bakfill and are therefore not representative of the overall strength of the bakfill. The elimination of this anomalous data was performed using the authors judgment, but the deleted data primarily onsisted of values of tip resistanes above kpa. Finally, after the tip resistane was orreted for pore pressure effets and anomalous values of tip resistane were removed, the shear strength was alulated using the effetive one resistane method presented on Table 1. Shear strength was alulated individually for eah data point. One Su was alulated for eah depth point for eah of the 24 data sets, all 24 data sets of Su vs. depth were averaged to reate one data set representing the average Su in the wall vs. depth. This average Su data set was then smoothed using a three point running average. The ombined, smoothed Su data set is presented in Fig Shear Strength (kpa) Cone Fator = 11.5 Best Fit Cone Fator from Fig 1 Vane Shear Data Point Very Soft Soft Medium FIG 2. Shear Strength vs. Depth for Mayfield Wall The plots of Su vs. depth, presented in Fig. 2, show the predited shear strength using both a onstant one fator of 11.5 and a varying one fator alulated from oupled CPT and vane shear data. For omparison, Fig. 2 also inludes the vane shear data and typial lay shear strength lassifiations. The presentation in Fig. 2 shows that regardless of the hoie of onstant or varying one fator, the shear strength alulated from the CPT data inreases only slightly with depth and is indiative of a very soft layey material. The shear strength alulated using the onstant one fator is used in subsequent analyses in this paper beause the results Page 5 of 10

6 are similar between the onstant and varying one fator and the one fator developed for this study, 11.5, ompares well with that suggested in the literature. EFFECTIVE STRESS FROM SHEAR STREGTH Shear strength is known to be diretly related to the effetive onfining stress (Lambe and Whitman 1969, Holtz and Kovas 1981,) and there should be a linear gain in shear strength with depth for effetive onfining stresses inreasing linearly with depth. In ontrast, if the effetive stress is onstant with depth, the shear strength should be onstant with depth. The relationship between shear strength and effetive stress has been the topi of numerous publiations. Several published relationships between shear strength and vertial effetive stress for a normally onsolidated soil, in the form of the Su/σ vo ratio, are presented on Table 2. TABLE 2. Relationships between Shear Strength and Vertial Effetive Stress Soure Details Su/σ vo Lambe and Whitman 1969 OCR = / 0.1 Mayne 2001 Degroot 2001 Theoretial Wroth-Prevost Model (Φ =26º, OCR = 1) ormalized Ratio from Critial State Model (Φ =26º, OCR = 1) SHASEP at OCR = 1 to 4: Su/σ vo = 0.19(OCR) 0.75 SHASEP at OCR = 1: Su/σ vo = 0.18(OCR) a 0.22 a Robertson and Cabal Φ =26º in diret simple shear a OCR = 1.3 for SB materials (Evans and Ryan 2005) For Su in the relationships shown on Table 2, the effetive vertial stress is generally the major prinipal stress. In normal soils the vertial effetive stress is the major prinipal stress and therefore the largest ontributor to the soil strength. For the Su derived from CPTu data olleted in SB bakfill, it is not lear whether the relationship between shear strength and effetive stress is one between the vertial, horizontal, or mean stress. While the state of stress in soil bentonite slurry trenh utoff walls is not fully understood and is expeted to vary in all three diretions, it an be reasonably assumed, at signifiant depths, that the horizontal effetive stress is the major prinipal stress and therefore ontrols the soil bakfill strength. With the assumption that the major prinipal stress is horizontal, the alulated bakfill shear strength from the CPT test oupled with the Su/σ relationship from the information in Table 2 an then be used to alulate the horizontal effetive stress in the utoff wall. This alulation of horizontal stress inherently assumes that the σ vo omponent of the published Su/σ vo ratios represents the major priniple stress for normal soils and thus the same relationship an be diretly applied to horizontal stress alulation in the SB bakfill where the major priniple stress is expeted to be the horizontal Page 6 of 10

7 Depth (m) Depth (m) stress. The alulated horizontal effetive stress from the CPTu data is plotted next to the stress values predited by geostatis, the arhing model (from Evans et al 1995) and the modified lateral squeezing (MLS) model (from Ruffing et al 2010) in Fig. 3. Equations and information used for Fig. 3 are presented on Table 3. TABLE 3. Effetive Stress Equations for Presentation in Figures 3a & 3b Stress Diretion Model Effetive Stress Equation Equation umber Geostatis σ h = γ b depth k ob (7) Horizontal Arhing σ h MLS σ h = σ v arhing = σ h MLS k ob (8) Where: the soil properties, σ v arhing, and, σ h MLS are from Ruffing et al (2010), γ b = unit weight of bakfill, σ v = vert. effetive stress, k ob = 0.5, σ h = horiz. effetive stress The presentation in Fig. 3 inludes two plots of the horizontal effetive stress alulated using the onsensus Su/σ ratio (0.22) for a normally onsolidated soil and using the upper bound of the Su/σ ratio (0.3), both from Table 2. (9) 0 Horizontal Effetive Stress (kpa) Horizontal Effetive Stress (kpa) Geostati Geostati Arhing Arhing MLS MLS Cone Fator = 11.5 Cone Fator = 11.5 FIG 3a. Horizontal Effetive Stress vs. Depth (S u /σ = 0.22) FIG 3b. Horizontal Effetive Stress vs. Depth (S u /σ = 0.3) Page 7 of 10

8 Fig. 3a shows that the horizontal effetive stresses predited from the shear strength alulated from the Mayfield CPTu data using the onsensus Su/σ ratio (0.22) are signifiantly higher than the horizontal effetive stresses predited by the MLS model in the depth range of 0-15 m. The predited stress from the MLS model mathes most losely after a depth of about 15 m. Above 15 m the alulated effetive stress from the CPTu data is higher than that predited by any of the models presented. Fig. 3b shows that the horizontal effetive stress alulated using the shear strength alulated from the Mayfield CPTu data and a Su/σ equal to the upper bound of the published ranges (0.3) mathes well with the horizontal effetive stress predited by the arhing model from approximately 0 10 m and with the MLS model from approximately m. DISCUSSIOS Ryan and Spaulding (2008) presented alulated the shear strength in the Mayfield wall using the total one resistane method. An assumption of the vertial effetive stress and the in situ pore pressure in the wall, both of whih are not well known, are required for this alulation method. The shear strength alulated in Ryan and Spaulding (2008) has the same shape as that presented in Fig. 2 using a onstant orretion fator, but the values presented there tend to be even less than those presented herein. This is due to inherent differenes between the total and effetive one resistane methods and the hoie of a larger one fator (Ryan and Spaulding used an kt of 15). Both the alulation method hoie and one fator seletion are ritial in determining the shear strength from CPTu data. Sine the CPTu data is, in reality, a funtion of mean effetive stress and that the ontrolling major prinipal stress is likely not the vertial stress for the entire depth of the wall makes alulating the shear strength of SB bakfill from CPTu data very ompliated. A prinipal stress rotation from vertial to horizontal in a diretion perpendiular to the axis of the wall, no doubt ours with depth. Also, the stress may be expeted to vary in all three dimensions with the longitudinal stress being the lowest. Therefore, as mentioned, the stress onditions in an SB wall are different than those in a normally onsolidated soil where the vertial stress is the major priniple stress and the minor prinipal stresses are equal. The effet of these differenes in stress onditions upon the CPTu test results is not known although the authors believe that the horizontal effetive stress is the major prinipal stress in SB utoff walls and therefore the largest ontributor to strength development in SB bakfill, partiularly at great depth where lateral squeezing ontrols rather than arhing. It is important to note that nearly all ommerially available CPTu systems were developed for use in a wide range of soil onditions and are therefore of limited preision when used in evaluating a normally onsolidated material of a very soft onsisteny. This limitation must be onsidered when seleting the CPTu test for use in an overall QC program on an SB utoff wall. Page 8 of 10

9 COCLUSIOS In summary, the effetive one resistane method using a onstant one fator is a reasonable method for alulating Su from CPTu data olleted in a SB wall beause it does not require any prior knowledge of the in-situ stress state or the initial pore pressure in the wall. In addition, this method produes shear strength values of a magnitude that are onsistent with those from other studies. Based on the analysis presented herein, the authors reommend the following equation for alulating soil bentonite bakfill shear strength from CPTu data: S u q (1 a) u ke 2 2 (10) u Where, q = raw tip resistane, a = area ratio, and u2 = pore pressure measured at the shoulder of the one. A one fator, ke, of 11.5 was seleted for this projet. A limitation of this method (and many others) is its dependene on the hoie of the one fator. Cone fator hoie in this appliation should be evaluated further. In this ase, there was suffiient data in the form of vane shear tests to support a one fator of The shear strength in the Mayfield Wall, alulated using Eq. 10, was generally kpa whih is 60% 100% higher than the estimated shear strengths presented in Ryan and Spaulding (2008). In both ases, however, the shear strength of soilbentonite bakfill results in a "very soft" lassifiation of the bakfill. The shear strength alulated using Eq. 10 an also be used to alulate the horizontal effetive stress in a utoff wall using a Su/σh ratio of The upper bound of this range, 0.3, is reommended beause this will onservatively result in the lowest alulated stress. The resulting equation for alulating horizontal effetive stress from CPTu data is: σ ' h q (1 a) u ke 2 2 (11) 0.3 u The horizontal effetive stress alulated using the CPTu data olleted in the Mayfield wall and a Su/σh ratio of 0.3 losely mathed the horizontal effetive stress predited by the arhing model from 0 10 m and the MLS model from m. As disussed above, it is important to reognize the general limitations of the CPTu test in the testing of SB utoff walls. These inlude the auray of the load ells in the one as well as the heterogeneity of the wall. Many of the ones urrently available are intended for use in a wide range of soils, inluding heavily overonsolidated deposits with high strength. For this reason, the load ells may not be aurate in the low stress onditions found in SB walls. Also, despite efforts by Page 9 of 10

10 designers and ontrators to ensure homogeneity in SB bakfill, there an be signifiant differenes in the shear strength of the bakfill due to the presene of large lay lumps, roks, and sand. These anomalies result in measured values of shear strength that may be higher than that of the bulk utoff wall. Engineering judgment will always be required to aount for these inherent issues when using CPTu data to alulate the in-situ shear strength or effetive stress of SB bakfill. REFERECES Evans, J.C., Fang, H-Y, and Kugelman, I.J., (1985). "Containment of Hazardous Materials with Soil-Bentonite Slurry Walls." Proeedings of the 6th ational Conferene on Management of Unontrolled Hazardous Waste Sites Washington, DC ov. 1985: Evans, J.C., Costa, M.J., and Cooley, B., (1995), The State of Stress in Soil Bentonite slurry Trenh Cutoff Walls, Geoenvironment 2000, Geotehnial Speial Publiation o. 46, Y. B. Aar and D.E. Daniel, Eds., Amerian Soiety of Civil Engineers, pp Filz, G.M. (1996). Consolidation Stresses in Soil-Bentonite Bak-filled Trenhes, Proeedings, 2 nd International Congress on Environmental Geotehnis, M. Kamon, Ed., Osaka, Japan, pp Holtz, R.D., and Kovas, W.D. (1981). An Introdution to Geotehnial Engineering. Upper Saddle River, ew Jersey: Prentie-Hall, In. Lambe, T.W., and Whitman, R.V., (1969). Soil Mehanis, SI Version. ew York, ew York: John Wiley and Sons, In. Mayne, P.W., (2001). "Stress-Strain-Strength-Flow Parameters from Enhaned In- Situ Tests," International Conferene on In-Situ Measurement of Soil Properties and Case Histories Bali, Indonesia, May 21st 24th 2001, pp Powell, J., and Lunne, T., (2005). Use of CPT Data in Clays/Fine Grained Soils. Studia Geotehnia et Mehania, Vol. XXVII, o. 3-4, Ruffing, D. G., Evans, J. C., and Malusis, M. A., (2010). Predition of Earth Pressures in Soil-Bentonite Cutoff Walls, ASCE GeoFlorida 2010 Advanes in Analysis, Modeling and Design GSP 199, pp Ryan, C.R., and Spaulding, C.A., (2008). "Strength and Permeability of a Deep Soil Bentonite Cutoff Wall." The Challenge of Sustainability in the Geo-Environment Proeedings of the Geo-Congress 2008, ASCE ew Orleans, LA. Jones, S., Spaulding, C., and Symth, P. (2007). "Design and Constrution of a Deep Soil-Bentonite Groundwater Barrier Wall at ewastle, Australia." 10th Australian ew Zealand Conferene on Geomehanis Common Ground, Ot Page 10 of 10