Cone Penetrating Testing (2007)

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.

29 Soils are very complex materials because they can be com- prised of a wide and diverse assemblage of different particle sizes, mineralogies, packing arrangements, and fabric. More- over, they can be created from various geologic origins (marine, lacustrine, glacial, residual, aeolian, deltaic, alluvial, estuarine, fluvial, biochemical, etc.) that have undergone long periods of environmental, seasonal, hydrological, and thermal processes. These facets have imparted complexities of soil behavior that relate to their initial geostatic stress state, nat- ural prestressing, nonlinear stress–strain–strength response, and drainage and flow characteristics, as well as rheological and time–rate effects. As such, a rather large number of different geotechnical parameters have been identified to quantify soil behavior in engineering terms. These include state parameters such as void ratio (e0), unit weight (), poros- ity (n), relative density (DR), overconsolidation ratio (OCR), strength parameters (c, , cu  su), stiffness (E, Eu, Gmax, G, D, K), compressibility (p, Cr, Cc, Cs), consolidation coefficient (cvh), permeability (k), creep (Cae), subgrade reac- tion coefficient (ks), spring constants (kz), lateral stress pa- rameters (KA, K0, KP), Poisson’s ratio (, u), dilatancy angle (), strain rate parameters ( ), and more. In this section, the evaluation of select geotechnical param- eters from CPT data is addressed, including various post- processing approaches based on theoretical, numerical, analytical, and empirical methods. In the survey results, DOT geotechnical engineers have indicated that CPT results are currently being used to assess several soil parameters that relate to highway design and construction. Selected relationships utilized in the data reduction of the cone, piezocone, and seismic cone tests are presented in the subsequent subsections. As with conventional practice, soils are grouped into either clays or sands, in particular referring to “vanilla” clays and “hourglass” sands. That is, the corre- lations can be expected to apply to “well-behaved” soils of common mineralogies (i.e., kaolin, quartz, feldspar) and typical geologic origins (e.g., marine and alluvial). It can be noted that alternative evaluations of soil properties and pa- rameters are available and that a spreadsheet format best allows for “tuning” and site-specific correlations for partic- ular geologic settings and soil materials. The procedures chosen herein represent a selection of methods based on the author’s understanding and experiences in United States and Canadian practices. A number of nontextbook geomaterials can be found throughout North America (e.g., loess, cemented soils, carbonate sands, sensitive structured clays, residual and tropical soils, glacial till, dispersive clays, and collapsible soils) that will undoubtedly not fall within the domain and applicability of these relationships. For those materials, it is suggested that site-specific calibration, test- ing, and validation be performed by a research institution working with the state DOT. Some guidelines and methods in assessing nontextbook geomaterials are given by Lunne et al. (1996), Coutinho et al. (2004), Schnaid et al. (2004), and Schnaid (2005). It may be noted that no uniform and consistent methodology currently exists to interpret all necessary soil engineering para- meters within a common framework. For specific concerns in the interpretation of CPT data, various parameters have been derived from analyses based in limit equilibrium, plasticity, elasticity, cavity expansion, strain path, stress path, finite elements, discrete elements, finite differences, and dislocation-based theories. At this time, the subsequently noted procedures are based largely on mixed theories tempered with experience and available calibrations with laboratory test results and/or backcalculated values from full-scale load tests and performance monitoring. SHEAR WAVE VELOCITY Shear wave velocity (Vs) is a fundamental measurement in all civil engineering solids (steel, concrete, wood, fiberglass, soils, and rocks). Vs can be obtained for all types of geoma- terials, including clays, silts, sands, gravels, and fractured and intact rocks, as well as mine tailings and fills. The val- ues of Vs can be readily determined by laboratory tests, including resonant column, ultrasonics, bender elements, torsional shear, and special triaxial apparatuses (Woods 1978) and by a variety of different field geophysical tests, including crosshole, downhole, suspension logging, spectral analysis of surface waves, refraction, and reflection (Campanella 1994). As noted earlier, the incorporation of one or more geo- phones within the penetrometer facilitates the conduct of SCPT. This is a version of the downhole geophysics test and may be conducted either by pseudo-interval or true-interval methods, depending on the equipment available, care taken in execution of the test, and degree of reliability needed in the assessed Vs profile. It is best practice to measure the Vs by CHAPTER SIX SOIL PARAMETER EVALUATIONS

direct methods such as the downhole geophysics test and SCPT. However, in some instances, it may be necessary to estimate the Vs profile by means of an empirical correlation if a seismic penetrometer is not available. Also, the correla- tive relationships may be employed to check on the reason- ableness of Vs readings obtained by SCPT and/or identify unusual geomaterials that may fall into the category of unusual or nontextbook type soils (Lunne et al. 1997; Schnaid 2005). For uncemented, unaged quartzitic sands, Baldi et al. (1989) suggested that Vs may be evaluated from the following relationship: Sands: Vs  277 (qt) 0.13 (vo) 0.27 (8) where Vs  shear wave velocity (m/s), qt  corrected cone tip resistance (MPa), and vo  effective overburden stress (MPa), as shown in Figure 28 (upper). For clay soils, Figure 28 (lower) shows a generalized interrelationship between shear wave and cone tip resistance for soft to firm to stiff intact clays to fissured clay materials (Mayne and Rix 1995) that determined: Clays: Vs  1.75 (qt) 0.627 (9) In addition to measured tip resistance (qt in kPa), the correl- ative relationship was significantly improved for intact clays if the in-place void ratio (e0) was also known. Of particular interest are interpretative methods that accommodate all types of soils. In one approach, an estimate of the in situ Vs can be made from (Hegazy and Mayne 1995): All Soils: Vs (m/s)  [10.1 log qt11.4]1.67 [fs/qt 100]0.3 (10) where qt  tip resistance and fs  sleeve resistance are input in units of kPa. The relationship was derived from a database that included sands, silts, and clays, as well as mixed soil types, and thus is interesting in that it attempts to be global and not a soil-dependent relationship. Another database from well-documented experimental sites in sat- urated clays, silts, and sands showed that Vs relates directly to the sleeve friction fs, reported in units of kPa (Mayne 2006b): Vs  118.8 log (fs)  18.5 (11) UNIT WEIGHT The saturated unit weight of each of the soil layers is needed in the calculation of overburden stress and in the other calculations. The unit weight is best achieved by obtaining undisturbed, thin-walled tube samples from borings. How- ever, in many soils, undisturbed samples are difficult to 30 obtain, particularly clean sands, cohesionless silts, and grav- els. Moreover, during CPT, samples are not routinely obtained; therefore, indirect methods for assessing unit weight are desirable. Based on the survey, nearly 40% of DOTs assume the unit weight (Appendix A, Question 35). Another 15% use an estimate based on the 12-part SBT clas- sification, as discussed by Lunne et al. (1997). An alternative approach uses results from large-scale calibration chamber tests to evaluate the dry unit weight (d) of sands from normalized cone tip resistance (qt1) given by Eq. 7 (chapter 5). The trend is presented in Figure 29. A regression line is given for uncemented unaged quartz to siliceous sands that has only a rather modest coefficient of determination (r2  0.488). Also shown are calibration chamber test data for four different carbonate sands (calcareous type), clearly showing that the relationship FIGURE 28 Shear wave velocity estimate from CPT data in (upper) clean quartz sands (after Baldi et al. 1989), and (lower) clay soils (after Mayne and Rix 1995).

31 should be used with caution in sands and that mineralogy and cementation can be important facets of geomaterials. For saturated soils, the correlation in Figure 30 is based on a large data set of soils, including soft to stiff clays and silts, loose to dense sands and gravels, as well as mixed geo- materials (n  727; r2  0.808). For these, the saturated total unit weight depends on both Vs (m/s) and depth z (meters). Also shown for comparative purposes (but not included in the regression) are data from intact rocks whereby a maxi- mum unit weight (rock  26 kN/m3) and maximum shear wave velocity (Vs  3300 m/s) can be taken as limiting val- ues. A set of alternate expressions for the dry and saturated unit weights is available in terms of Vs and vo (Mayne 2006b). By adopting a characteristic specific gravity of solids (Gs), the total unit weight of saturated soils can be directly estimated from CPT fs (kPa), as presented in Figure 31. In general, the evaluation appears good for soft to stiff clays of marine origin, fissured clays, silts, and a variety of clean quartz sands. Note the effect of specific gravity in affecting the relationship for higher unit weights in the soft fresh- water glacial lake clays at the Northwestern University site, as well as the lower unit weights for soft lacustrine clay of Mexico City. POISSON’S RATIO The value of Poisson’s ratio () is normally taken for an isotropic elastic material. Based on recent local strain mea- surements on samples with special internal high-resolution instrumentation (e.g., Burland 1989; Tatsuoka and Shibuya 1992; Lehane and Cosgrove 2000), the value of drained  ranges from 0.1 to 0.2 for all types of geomate- rials at working load levels, increasing to larger values as failure states are approached. The value for undrained loading is u  0.5. SMALL-STRAIN SHEAR MODULUS Soils are commonly associated with shearing during loading modes, deformations, and failure, and thus are best repre- sented in terms of their stress–strain–strength behavior in terms of simple shear. The slope of a shear stress ( ) versus shear strain (s) curve is the shear modulus (G). The small- strain shear modulus (termed G0, or Gmax), also known as the initial tangent dynamic shear modulus (Gdyn), is a fundamen- tal stiffness that relates to the initial state of the soil. This stiffness applies to the initial loading for all stress–strain– strength curves, including static, cyclic, and dynamic types of loading, as well as undrained and drained conditions (Bur- land 1989; Mayne 2001; Leroueil and Hight 2003). The small-strain shear modulus is calculated from the total soil mass density (T  T/g) and shear wave velocity (Vs), where g  9.8 m/s2  gravitational acceleration constant: Gmax  T Vs2 (12) FIGURE 29 Dry unit weight relationship with shear wave velocity and depth. FIGURE 30 Saturated soil unit weight evaluation from shear wave velocity and depth.

In lieu of shear modulus, the stiffness can be expressed in terms of an equivalent Young’s modulus of soil through elas- tic theory: Emax  2Gmax (1  ) (13) where   0.2 applies for drained and u  0.5 for undrained conditions. SOIL STIFFNESS The value of small-strain shear modulus, Gmax (and corre- sponding Emax), applies strictly to the nondestructive range of strains, where s  104 as a decimal (or s  106%). For loading levels at strains higher than these, modulus reduction curves (G/Gmax  G/G0) must be implemented. For cyclic loading and dynamic problems in geotechnical engineering, Vucetic and Dobry (1991) present G/Gmax curves in terms of soil plasticity and shear strain (s). The appropriate value of shear modulus is then obtained from: G  Gmax (G/Gmax) (14) The G/Gmax curves can be presented in terms of a logarithm of shear strain (s), as discussed by Jardine et al. (1986, 2005a) and Atkinson (2000), or alternatively in terms of mobilized shear stress ( / max), as discussed by Tatsuoka and Shibuya (1992), Fahey and Carter (1993), and LoPresti et al. (1998). The mobilized shear stress is analogous to the reciprocal of the factor of safety ( / max  1/FS). In terms of fitting stress-strain data, G/Gmax versus mobilized stress level ( / max), plots are visually biased toward the intermediate- to large-strain regions of the soil response. In contrast, G/Gmax versus log s curves tend to accentuate the small- to intermediate-strain range. The ratio (G/Gmax) is a reduction factor to apply to the maximum shear modulus, depending on current loading conditions. 32 A selection of modulus reduction curves, represented by the ratio (G/Gmax), has been collected from monotonic lab- oratory shear tests performed on an assorted mix of clayey and sandy materials (Mayne 2006b). The results are pre- sented in Figure 32 (upper), where G  s  secant shear modulus. These laboratory tests include static torsional shear and special triaxial tests with internal local-strain measurements. An assumed constant value of  has been applied with the conversion E  2G(1  ) to permit plot- ting of E/Emax versus q/qmax, where q  (1  3)  devia- tor stress. Undrained tests are shown by solid dots and drained tests are indicated by open symbols. In general, the clays were tested under undrained loading (except Pisa), and the sands were tested under drained shearing conditions (except Kentucky clayey sand). Similar trends for the vari- ous curves are noted for both undrained and drained tests on both clays and sands. The nonlinear representation of the stiffness has been a major focus of the recent series of conferences on the com- mon theme: Deformation Characteristics of Geomaterials (e.g., Jardine et al. 2005a). A number of different mathe- matical expressions can be adopted to produce closed- form stress–strain–strength curves (e.g., LoPresti et al. 1998). One rather simple algorithm involves a modified hyperbola (Fahey and Carter 1993; Fahey 1998) with pre- sented results for modulus reduction (G/Gmax) versus mobi- lized stress ( / max  1/FS) shown in Figure 32 (lower). It can be seen that a limited range of the exponent (0.2  g  0.4) tends to encompass many of the laboratory tor- sional shear and triaxial compression data. The modulus reduction can be given by: G/Gmax  1  ( / max)g (15) with g  0.3  0.1 for “well-behaved” soils (uncemented, insensitive, not highly structured). FIGURE 31 Saturated unit weight evaluation from CPT sleeve friction reading and specific gravity of solids.

33 An equivalent stiffness of soils is also afforded by means of the constrained modulus (D) obtained from one-dimensional consolidation tests. In lieu of e-logv graphs developed from consolidation tests, the data may be plotted in terms of verti- cal stress versus vertical strain and the tangent slope is defined as the constrained modulus D  v/v, where v  e/(1  e0). From elastic theory, the constrained modulus relates to the equivalent elastic Young’s modulus (E) and shear modulus (G) for drained loading conditions: (16) For foundation settlement analyses, a representative constrained modulus of the supporting soil medium is usu- ally sought. In practice, it has been usual to correlate the modulus D to a penetration resistance (e.g., Mitchell and Gardner 1975; Schmertmann 1978b; Jamiolkowski et al. D E G       = ⋅ − + − = − − ( ) ( )( ) ( ) ( ) 1 1 1 2 2 1 1 2      1985). From a collection of diverse geomaterials ranging from sands, silts, intact organic and inorganic clays, and fis- sured soils (Mayne 2006), Figure 33 (upper) shows that a relationship for “well-behaved” soils might take the form: D  C (qt  vo) (17) with an overall representative value of C  5 for soft to firm vanilla clays and normally consolidated (NC) hourglass sands. However, for organic plastic clays of Sweden, a considerably lower C  1 to 2 may be appropriate. For cemented (Fucino) clay, a value C  10 to 20 may be more appropriate. With SCPT data, an alternate correlation can be sought between D and small-strain shear modulus (Gmax), as pre- sented in Figure 33 (lower). In this case, a similar adopted format could be (Burns and Mayne 2002): D  G Gmax (18) FIGURE 32 Monotonic modulus reduction curves from (upper) static torsional and triaxial shear data on clays and sands, and (lower) using modified hyperbolic expression proposed by Fahey and Carter (1993). FIGURE 33 Trends between constrained modulus and (upper) net cone resistance, and (lower) small- strain shear modulus of various and diverse soils.

with assigned values of G ranging from 0.02 for the organic plastic clays up to 2 for overconsolidated quartz sands. In the future, additional studies with multiple regres- sion, artificial neural networks, and numerical modeling may help guide the development of more universally applied global relationships. STRESS HISTORY Clays The stress history of clay soils is classically determined from one-dimensional oedometer tests on high-quality undis- turbed samples. The yield point in one-dimensional loading (i.e., consolidation test) denotes the preconsolidation stress (p), formerly designated vmax or Pc. In normalized form, the degree of preconsolidation is termed the overconsolida- tion ratio, OCR  (p/vo). For intact clays, a first-order estimate of the preconsolidation stress can be obtained from the net cone tip resistance (Mayne 1995; Demers and Leroueil 2002), as shown in Figure 34: p  0.33 (qt  vo) (19) It can be seen that this expression underestimates val- ues for fissured clays. This is because the macrofabric of cracks and fractures affect the field measurements of the CPT as the blocks of clay are forced away from the axis of penetration. In contrast, any fissures or cracks within the small laboratory oedometric specimens are closed up during constrained compression in one-dimensional loading. An example of the profiling of preconsolidation stress with depth by cone penetrometer data is illustrated in Figure 35 using results from the national experimental test site at Both- 34 kennar in the United Kingdom (Nash et al. 1992). Extensive geological, laboratory, and in situ field tests have been con- ducted in the soft clays having thicknesses up to 30 m and a shallow groundwater table of 0.5 to 1.0 m below grade. Using a variety of different sampling techniques, a reference profile of p has been established from consolidation tests using three laboratory devices at different universities: (1) incremental loading oedometers, (2) constant rate of strain consolidometers, and (3) restricted flow tests. The first-order evaluation from net cone resistance is shown to be in good agreement with the lab- oratory results. With piezocone testing, a separate and independent assessment of p in intact clays can be made from the pore- water pressure measurements, as shown in Figure 36. Notably, data for fissured clays lie above the trends. The first-order relationships for intact clays can be expressed as (Chen and Mayne 1996): Midface Filter Element: p  0.40 (u1  u0) (20) Shoulder Filter Element: p  0.53 (u2  u0) (21) As indicated by Figure 36, a slight additional trend with plasticity index (PI, or IP) was determined from the database using multiple regression analyses. For Type 1 piezocones, the penetration porewater pressures are positive for all clay consistencies, ranging from soft to hard intact clays to fis- sured deposits. For Type 2 piezocones, the trend is similar for soft to firm to stiff intact clays; however, for over- consolidated fissured clays the porewater pressures can be negative, thus providing a nonunique relationship. From a theoretical perspective, the value of preconsolida- tion stress can also be ascertained from the effective cone tip resistance (Mayne 2005): FIGURE 34 First-order relationship for preconsolidation stress from net cone resistance in clays.

35 Midface Filter Element: σp′ = 0.75 (qt − u1) (22) Shoulder Filter Element: σp′ = 0.60 (qt − u2) (23) The previous relationships give redundancy to the interpre- tation of yield stress in clays by means of CPTu data; however, this is interesting as it lends support to the values obtained should they agree. That is, multiple methods give an opportu- nity to confirm and corroborate the interpreted soil parameters Larsson and Mulabdic (1991). A noted discrepancy offers a reason to investigate why the conflict exists, as well as a cau- tion that additional testing (i.e., oedometer, CRS consolidome- ter) may be warranted, particularly in unusual soil formations. For the Bothkennar site, the additional evaluations of stress history using excess porewater pressures (u2) and effective cone resistance (qt  u2) are presented in terms of the OCR in Figure 37. Again, as with the earlier profiles in Figure 35, good agreement between the laboratory consolidation data and field methods is evident. Sands The evaluation of stress history for clean, uncemented, unaged quartz sands is a more challenging assignment for two primary reasons: (1) oedometric e-logv curves for sands are very flat, thus making detection of a yield stress problematic; and (2) undisturbed sampling of clean quartz to siliceous sands is quite difficult, and though now attainable by new freezing methods, remains very expensive. There- fore, a relationship for obtaining OCR in clean quartz sands has been empirically derived from statistical evaluations on 26 different series of CPT calibration chamber tests (Kul- FIGURE 35 Comparison of CPT-based evaluation of preconsolidation stress with laboratory consolidation tests in Bothkennar soft clay (data from Nash et al. 1992). CRS = constant rate of strain; IL Oed = incremental loading oedometer; RF = restrained flow, Svo (vo) = effective vertical (overburden) stress. hawy and Mayne 1990; Lunne et al. 1997; Mayne 2001). Chamber tests are very large diameter triaxial specimens having diameters and heights on the order of 0.9 to 1.5 m. Cone penetration is conducted after preparation of the sand sample (dry, moist, saturated) at the desired relative density, effective confining stress levels, and stress history (Jami- olkowski et al. 2001). For purposes herein, the sands are primarily siliceous (quartz and feldspar) with applied stress histories ranging from NC to overconsolidated states (1  OCR  15). Multi- ple regression analyses of the chamber test data (n  636) from anisotropically consolidated sands indicate that the induced OCR is a function of the applied effective vertical stress (vo), effective horizontal stress [ho  (K0 · vo)], and measured cone tip resistance (qt), as indicated by Figure 38. Here, the OCR is shown normalized by Q  (qt  vo)/vo. The results can be presented by the following closed-form expression (Mayne 2005): (24) where   effective stress friction angle of the sand, vo  effective overburden stress, and atm  a reference stress equal to one atmosphere  1 bar  100 kPa  1 tsf. From the OCR, the apparent preconsolidation stress of the sand can be calculated from: p  OCR vo (25) OCR atm atm = ⋅( ) − ⋅ ( 0 192 1 0 22 . / ( sin ) / . qt vo      ) ⎡ ⎣ ⎢⎢ ⎤ ⎦ ⎥⎥ − ⎛ ⎝⎜ ⎞ ⎠⎟ 0 31 1 0 27 . sin ' .

An example of the procedure for evaluating stress history from cone tip stress measurements in clean sands is afforded from a quarry site near Stockholm investigated by Dahlberg (1974). The site was comprised of a Holocene deposit of clean, glacial, medium-coarse sand having an initial 24 m thickness overlying bedrock. After the upper 16 m was removed by quarrying operations, a series of in situ testing [SPT, CPT, pressuremeter test (PMT), and screw plate load test] were per- formed in the remaining 8 m of sand, in addition to special bal- loon density tests in trenches. The groundwater table was located at the base of the sand just above bedrock. Index param- eters of the sand included mean grain size (0.7  D50  1.1 mm); uniformity coefficient (2.2  UC  3), mean density T  1.67 g/cc, and average DR  60%. Using results from four Borros-type electric CPTs at the site, Figure 39 shows the measured qt readings and interpreted profiles of OCR and p in the Stockholm sand. Results of the screw plate load tests were used by Dahlberg (1974) to interpret the preconsolida- tion stresses in the sand, which are observed to be comparable to the known values from mechanical overburden removal, where the stress history can be determined by calculating the OCR  (v  vo)/vo, using the prestress: v  (16 m)(16.4 kN/m3)  262 kPa (2.72 tsf) Mixed Soil Types If seismic cone data are obtained, then the small-strain stiff- ness may be used together with the overburden stress level to evaluate the effective preconsolidation stress in all soil types (clays, silts, and sands). An original database com- piled by Mayne et al. (1998) on a variety of 26 intact clays worldwide has been supplemented with recent data on two cemented clays (Fucino, Italy and Cooper Marl from Charleston, South Carolina) in Figure 40. In addition, data 36 FIGURE 37 Comparison of CPTu-based evaluations of preconsolidation stress with laboratory consolidation tests in Bothkennar soft clay using excess porewater pressures and effective cone resistance (data from Nash et al. 1992). CRS = constant rate of strain, u2  porewater pressure, qt  u2  effective cone resistance, IL Oed  incremental loading oedometer, RF  restricted flow. FIGURE 36 First-order trends of preconsolidation stress in clays with excess porewater pressures measured by (upper) Type 1 piezocones, and (lower) Type 2 piezocones.

37 all soil types may be considered in a consistent manner, whereas the separation of soil layers into “clay-like” and “sand-like” often result in mismatched profiles of preconsoli- dation stress with depth. EFFECTIVE STRESS STRENGTH Sands The strength of soils is controlled by the effective stress frictional envelope, often represented in terms of the Mohr– Coulomb parameters:   effective friction angle and c  effective cohesion intercept. For clean sands, a com- monly used CPT interpretation is based on considerations FIGURE 38 Chamber test data showing trend of OCR/Q for clean quartz and siliceous sands. FIGURE 39 Results from Stockholm quarry sand site showing (left) cone tip resistances, (center) OCR profiles from excavation, and (right) preconsolidation stresses (data from Dahlberg 1974). from Po River sand (Ghionna et al. 1995) and Holmen Sand (Lunne et al. 2003), where the stress histories of the granu- lar deposits are well-documented, are also included. Finally, results from Piedmont residual fine sandy silts at the National Geotechnical Test Site (NGES) at Opelika, Alabama, are also considered (Mayne and Brown 2003). The overall relationship for intact geomaterials is shown in Figure 40 and expressed by: p  0.101 atm 0.102 G0 0.478 vo 0.420 (26) with a statistical coefficient of determination r2  0.919 for intact soils. The approach is evidently not valid for fissured geomaterials. The advantage of this particular approach is that

of an inverted bearing capacity (BC) theory supplemented with CPT calibration chamber data from five sands (Robertson and Campanella 1983). However, the flexible- walled chamber test results were not corrected for bound- ary size effects. In that approach, the expression for peak friction angle of clean quartz sands is given by the approx- imation (c  0):   arctan [0.1  0.38 log (qt/vo)] (27) An alternate expression derived from a much larger compi- lation of a calibration chamber database from 24 sands, where the cone tip stresses were adjusted accordingly for relative size of chamber and cone diameter (D/d ratio), was proposed by Kulhawy and Mayne (1990):   17.6º  11.0º log (qt1) (28) where qt1  (qt/atm)/(vo/atm)0.5 is a more appropriate form for stress normalization of CPT results in sands (e.g., Jami- olkowski et al. 2001). The relationship for  with qt1 is shown in Figure 41. Recently, a database was developed on the basis of undis- turbed (primarily frozen) samples of 13 sands. These sands were located in Canada (Wride and Robertson 1999, 2000), Japan (Mimura 2003), Norway (Lunne et al. 2003), China (Lee et al. 1999), and Italy (Ghionna and Porcino 2006). In general, the sands can be considered as clean to slightly dirty sands of quartz, feldspar, and/or other rock mineralogy, excepting two of the Canadian sands derived from mining operations that had more unusual constituents of clay and other mineralogies. In terms of grain size distributions, these granular geomaterials include ten fine sands, four medium sands, and one coarse sand (Italy). The sands from Canada 38 were slightly dirty, having fines contents (FC) between 5%  FC  15%, whereas the other sands were all relatively clean with FC  4%. Mean values of index parameters (with plus and minus one standard deviation) of these sands indi- cated: specific gravity (Gs  2.66  0.03), fines content (FC  4.36  4.49), particle size (D50  0.35  0.23 mm), and uniformity coefficient (UC  D60/D10  2.80  1.19). At all sites, results from electric SCPTu were available, except the China site where only CPTu was reported. Each undisturbed sand was tested using a series of either isotropi- cally and/or anisotropically consolidated triaxial shear tests. Additional details are discussed by Mayne (2006a). The sand database was used to check the validity of the friction angle determinations from in situ CPT tests. The relationship between the triaxial-measured  of undis- turbed (frozen) sands and normalized cone tip resistance is presented in Figure 41. Here, the CPT proves to be an excellent predictor in evaluating the drained strength of the sands. The two outliers from LL and Highmont Dams are mine tailings sands from Logan Lake, British Columbia, that contained high percentages of clay minerals (as noted) and are both underpredicted by the CPT expression. Mixed Soil Types An interesting approach by the Norwegian University of Sci- ence and Technology (NTNU) is an effective stress limit plasticity solution to obtain the effective stress friction angle for all soil types (Senneset et al. 1988, 1989). In the fully developed version, the NTNU theory allows for the determi- nation of both the effective friction angle () and effective cohesion intercept (c) from CPTU data in soils. For the simple case of Terzaghi-type deep BC (angle of plastification P  0), and adopting an effective cohesion intercept c  0, the effective friction angle can be determined from normalized CPT readings Q  (qt  vo)/vo and Bq  (u2  u0)/(qt  vo) using the chart shown in Figure 42. FIGURE 41 Peak triaxial friction angle from undisturbed sands with normalized cone tip resistance. FIGURE 40 Preconsolidation stress evaluation from small-strain shear modulus in soils.

39 An approximate form for a deterministic line-by-line evaluation of f for the NTNU method is given by (Mayne and Campanella 2005): (degrees)  29.5º Bq0.121 [0.256  0.336 Bq  log Q] (29) that is applicable for 0.1  Bq  1.0 and range: 20º    45º. For Bq  0.1 corresponding to granular soils, the previ- ous expression for clean sands would apply. UNDRAINED SHEAR STRENGTH OF CLAYS For geotechnical applications involving short-term loading of clays and clayey silts, the undrained shear strength (su  cu) of the soil (formerly termed c  cohesion) is commonly sought for stability and BC analyses. The classical approach to eval- uating su from CPT readings is through the net cone resistance: su  (qt  vo)/Nkt (30) where Nkt is a bearing factor. More papers and research pro- grams have focused on the assessment of relevant value of NkT for an interpretation of su than for any other single parameter (e.g., Keaveny and Mitchell 1986; Konrad and Law 1987; Yu and Mitchell 1998), without any consensus reached. This is because, in part, the value of su is not unique, but depends on the direction of loading, strain rate, boundary conditions, stress level, sample disturbance effects, and other factors (Ladd 1991). Indeed, a suite of different undrained shear strengths are available for a given clay soil. For the basic laboratory shear modes, there are many available apparatuses, including CIUC, PSC, CK0UC, direct shear simple (DSS), DS, PSE, CK0UE, UU, UC, as well as hollow cylinder, true triaxial, and torsional shear (Jamiolkowski et al. 1985; Kulhawy and Mayne 1990). Depending on the particular agency, firm, or institution given responsibility for assessing the appropriate Nkt, different test modes will be chosen to benchmark the su for the CPT. In lieu of the classical approach, an alternate and rational approach can be presented that focuses on the assessment of p from the CPT. The magnitude of preconsolidation stress (σp′) is uniquely defined as the yield point from the e-logv plot obtained from a consolidation test. The influence of OCR in governing the undrained shear strength of clays is very well established (e.g., Trak et al. 1980; Leroueil and Hight 2003). Therefore, the OCR profile already evaluated by the CPT results can be used to generate the variation of undrained shear strength with depth in a consistent and rational manner. A three- tiered approach can be recommended based on: (1) critical-state soil mechanics, (2) empirical normalized strength ratio approach, and (3) empirical method at low OCRs, as discussed later. For all cases, a representative mode for general problems of embankment stability, foundation-BC, and slopes and exca- vations in clays and clayey silts can be taken as that for DSS. From considerations of critical state soil mechanics (CSSM), this simple shear mode can be expressed in nor- malized form (Wroth 1984): su/voDDS  1⁄2 sinOCR (31) where  1  Cs/Cc  plastic volumetric strain potential, Cs  swelling index, and Cc  virgin compression index of the material. For many clays of low to medium sensitivity, 0.7    0.8, whereas for sensitive and structured clays, a higher range between 0.9    1.0 can be observed. If the compression indices and  are not known with con- fidence, a recommended default form based on three decades of experimental laboratory work at the Massachusetts Insti- tute of Technology has been proposed (Jamiolkowski et al. 1985; Ladd 1991; Ladd and DeGroot 2003): su/voDSS  0.22 OCR0.80 (32) which is clearly a subset of the CSSM equation for the case where   26º and   0.80. Finally, at low OCRs  2, the back analyses of failure case records involving corrected vane strengths for embankments, FIGURE 42 Effective stress friction angle for sands, silts, and clays from NTNU method.

footings, and excavations, it has been shown that the mobi- lized undrained shear strength may be taken simply as (Trak et al. 1980; Terzaghi et al. 1996): su  0.22 p (33) which is a subset of both the CSSM and the Massachusetts Institute of Technology approaches. Available experimental data support the CSSM approach, as shown by Figure 43. For NC clays, the normalized undrained shear strength to effective overburden stress ratio (su/vo)NC increases with effective friction angle. In Figure 44, the larger influence of stress history is shown to dominate the ratio (su/vo)OC for overconsolidated soils. Notably, the CSSM adequately expressed the increase with OCR in terms of a power function. It is important here to note the exception for fissured clay materials (specifically, London clay from Brent Cross) that have a macrofabric of cracking and preexisting slip surfaces. Fissured soils can exhibit strengths on the order of one-half of the values associated with intact clays. For these cases, fis- sured clays occurring below the groundwater table can be identified by zero to negative porewater pressures taken at the shoulder position (Type 2); thus, u2  0 (Lunne et al. 1997). For Type 1 piezocones, zones of fissured clays can be demarcated by a low ratio u1/qt  0.4, in comparison with intact clays that exhibit characteristic ratios on the order of u1/qt  0.7 (Mayne et al. 1990). An illustrative example of post-processing CPTs in clays to determine the undrained shear strength variation with depth is shown in Figure 45. This is the national geo- technical experimentation site in soft varved clay at the University of Massachusetts–Amherst (DeGroot and Lutenegger 2003). A series of five CPTs produced the total (corrected) cone tip resistances presented in Figure 45 40 (left) showing a subsurface profile with 1 m clay fill over a desiccated clay crust to an approximate 4 m depth overlying soft silty clay. The groundwater lies 0.5 to 1 m below grade. The net cone resistances were processed to evaluate p values as noted in Eq. 19 and produce the overconsolidation ratios shown in Figure 45 (center). These were used in turn with an effective   21º from Eq. 31 to obtain the profile of undrained shear strengths, as seen in Figure 45 (right). The results are in good agreement with the laboratory reference oedometer tests and corre- sponding DSS strength tests at the site. On particularly critical projects, it is warranted to perform additional strength testing to confirm and support the CPT interpretations, rather than rely solely on one test method. For instance, reference benchmarking of su values can be estab- lished using field vane tests with appropriate corrections (e.g., Leroueil and Jamiolkowski 1991) or by laboratory strength testing on high-quality samples (e.g., Ladd and DeGroot 2003). FIGURE 43 Normalized DSS undrained shear strength versus effective friction angle in normally consolidated clays. FIGURE 44 Relationship for DSS undrained strength with , OCR, and degree of fissuring in clays.

41 SENSITIVITY In soft clays and silts, the sensitivity (St) is considered as an index to problematic construction and field perfor- mance difficulties. The reference test for determining St is the field vane shear (Chandler 1988), although laboratory testing methods can include the unconfined compression test, miniature vane, and fall cone. With the CPT, the fric- tion sleeve reading can be considered indicative of a remolded undrained shear strength: fs  sur (e.g., Gorman et al. 1975). Thus, an indicator as to the sensitivity (St) of the deposit may be obtained by taking the ratio of peak shear strength to remolded value. Mostly, the value of St is sought for soft clays; therefore, using the aforementioned relationship for peak strength at low OCRs (i.e., su  0.22p) combined with the evaluation of preconsolidation stress from net cone resistance [i.e., p  0.33(qt  vo)] suggests that (OCRs  2): St  0.073(qt  vo)/fs (34) If a direct and accurate measure of in-place sensitivity is necessary, follow-up testing with the vane shear is prudent. RELATIVE DENSITY OF CLEAN SANDS In clean sands with less than 15% fines content, it is common practice to assess the relative density (DR) by in situ tests. For the CPT, a number of different expressions have been devel- oped from large-scale chamber tests (e.g., Schmertmann 1978a; Robertson and Campanella 1983; Jamiolkowski et al. 1985); however, those correlations did not consider the boundary effects that cause reduced values of qt measured in flexible walled chambers (e.g., Salgado et al. 1998). A recent FIGURE 45 Results from Amherst soft clay site showing: (left) corrected cone tip resistances, (center) over- consolidation ratios, and (right) undrained shear strengths (laboratory data from DeGroot and Lutenegger 2003). FIGURE 46 Relative density relationship with normalized tip stress and sand compressibility from corrected chamber test results (after Jamiolkowski et al. 2001). reexamination of a large CCT data set by Jamiolkowski et al. (2001), which incorporates a correction factor, has found that a mean relationship in terms of normalized cone tip stress can be expressed by: (35) and the effects of relative sand compressibility can be con- sidered by reference to Figure 46. The aforementioned database on undisturbed (frozen) sands also lends an opportunity to assess this revised expres- sion. As seen in Figure 47, the corresponding CPT data on D qR t vo = ⋅ ⋅ ⎛ ⎝⎜ ⎞ ⎠⎟ −100 0 268 0 675. ln / / .    atm atm ⎡ ⎣ ⎢⎢ ⎤ ⎦ ⎥⎥

42 FIGURE 47 Relative density of undisturbed (frozen) quartz sands versus normalized cone tip resistance. FIGURE 48 Relative density of carbonate sands in terms of normalized cone tip resistances. these 15 sands fall generally within the bounds established from the CCT results, with the Canadian sands indicative of high-compressibility materials and the Japanese sands trend- ing on the low-compressibility side. In sands of carbonate and calcareous composition, the expected trend would follow that for high-compressibility bounds because of particle crushing (Coop and Airey 2003). For this case, a newly created data set from CPT chamber testing on carbonate sands has been compiled, including Quiou Sand (Fioravante et al. 1998), Dogs Bay Sand (Nutt and Houlsby 1991), Ewa Sand (Morioka and Nicholson 2000), and Kingfish Platform (Parkin 1991). These data con- firm that carbonate sands would fall at the higher side of trends reported for quartzitic sands because of their higher compressibility, as shown by Figure 48. GEOSTATIC LATERAL STRESS STATE The geostatic horizontal stress is represented by the K0 coef- ficient, where K0  ho/vo. In general, laboratory data on small triaxial specimens and instrumented oedometer tests FIGURE 49 Lateral stress coefficient K0 from total stress cell field measurements versus OCR in clays. FIGURE 50 Lateral stress coefficient K0 versus OCR from laboratory tests on sands. indicate that the following relationship can be adopted in uncemented sands and well-behaved clays of low to medium sensitivity: K0  (1  sin) OCR sin (36) For structured and cemented soils, higher values of K0 can be realized, somewhat related to the clay sensitivity (Hamouche et al. 1995). Figure 49 shows field K0 data from total stress cell (TSC) measurements (or spade cells) in clays that generally agree with the above relationship. Results from self-boring PMTs also give a similar trend between K0 and OCR for a variety of clay soils (Kulhawy and Mayne 1990). For clean sands, data from large calibration chamber tests and small laboratory triaxial and oedometer test series show the K0  OCR trends in Figure 50. Related to the previous OCR Eq. 24, the derived formulation for the lateral stress

43 The Kp limit is shown in Figures 49 and 50 for the K0 − OCR relationships for clays and sands, respectively. Illustration of the approach for K0 profiling in sands by CPT is afforded from the previous case study of quarried glacial sand near Stockholm (Dahlberg 1974). To utilize Eq. 37 for evaluation of K0, an a priori relationship between K0 and OCR must be made; that is, Eq. 36. Samples of the sand were reconstituted in the laboratory at the measured in-place densities and subjected to consolidated drained triaxial shear testing to determine   40º (Mitchell and Lunne 1978). Using Eq. 28 provides a comparable evaluation of  from the four CPTs, as seen in Figure 52. The CPT data together with  are used in Eq. 24 to obtain the OCR (Figure 39) in either Eqs. 36 or 37 to produce the profiles of K0. As seen in Figure 52, these are in agreement with the reported field K0 values determined from lift-off pressures in PMTs performed at the site (Dahlberg 1974). EFFECTIVE COHESION INTERCEPT For long-term stability analyses, the effective cohesion inter- cept (c) is conservatively taken to be zero. The intercept is actually a projection caused by the forced fitting of a straight line (form, y  mx  b) to a strength envelope that is actually curved (Singh et al. 1973). Several difficulties are associated with assessing a reputable value of c to a particular soil, including its dependency on the magnitude of preconsolida- tion stress (p), strain rate of loading (d/dt), and age of the deposit (t). The projected c is actually a manifestation of the three-dimensional yield surface that extends above the fric- tional envelope, as discussed by Hight and Leroueil (2003). For short-term loading conditions, an apparent value of c FIGURE 51 Lateral stress evaluation of quartz sands from CPT results in chamber tests. FIGURE 52 CPT post-processing for peak  and coefficient K0 in Stockholm sand. coefficient from chamber tests is shown in Figure 51 and expressed by: (37) A maximum value for K0 can be set by the passive stress coefficient (KP), which for a simple Rankine case is given by: (38)KP = +( ) = + − tan º / sin sin 2 45 2 1 1       K qt vo 0 0 22 0 31 0 192= . . . ⋅ ⎛⎝ ⎞⎠ ⋅ ′⎛⎝ ⎞⎠ ⋅σ σ σatm atm OCR( ) 0 27.

may be assessed from the stress history (Mayne and Stewart 1988; Mesri and Abdel-Ghaffar 1993): c  0.02 p (39) COEFFICIENT OF CONSOLIDATION Porewater pressures generated during cone penetration in fine-grained soils are transient. Once the penetration process is halted, the excess pressures will decay with time and the transducer reading will eventually reach equilibrium corre- sponding to the hydrostatic value (u0). The rate of dissipation is governed by the coefficient of consolidation (cvh): (40) where k  coefficient of permeability, D  constrained modulus, and w  unit weight of water. For most natural soft marine clays, the horizontal per- meability is only around 10% to 20% higher than the verti- cal value (Mesri 1994; Leroueil and Hight 2003). A summary of the laboratory series of permeability tests on different natural soft clays is given in Figure 53, whereby both standard vertical measurements of hydraulic conduc- tivity (kv) are compared with horizontal values (kh) using radial permeameter devices. For varved clays and highly stratified deposits, the ratio of horizontal to vertical perme- abilities may range from 3 to 5, and very rarely approaches 10. Guidelines to permeability anisotropy are given in Table 3. The most popular CPTù method to evaluate cvh in soils at present is the solution from the strain path method (SPM) reported by Houlsby and Teh (1988), although other available c k D vh w = ⋅   44 procedures are discussed by Jamiolkowski et al. (1985), Gupta and Davidson (1986), Senneset et al. (1988, 1989), Jamiolkowski (1995), Danzinger et al. (1997), Burns and Mayne (1998, 2002a), and (Abu-Farsakh and Nazzal 2005). For the SPM solution, Teh and Houlsby (1991) provided time factors for a range of porewater pressure dissipations. The degree of excess porewater pressure dissipation can be defined by U*  u/ui, where ui  initial value during penetration. The modified time factor T* for any particular degree of consolidation is defined by: (41) where t  corresponding measured time during dissipation and a  probe radius. The SPM solutions relating U* and T* for midface u1 and shoulder u2 piezo-elements are shown in Figures 54 and 55, respectively. These can be conve- niently represented using approximate algorithms as shown, thus offering a means to implement matching data on a spreadsheet. In terms of calibrating the approach, a fairly comprehen- sive study between laboratory cv values and piezocone ch values in clays and silts was reported by Robertson et al. (1992). Assumptions were made between the ratio of T c t a I vh R * = ⋅ ⋅ 2 oitaR yalC eht fo erutaN kh/kv 5.1 ot 1 syalC suoenegomoH Sedimentary Clays with Discontinuous Lenses and Layers, Well-Developed Macrofabric 2 to 4 Varved Clays and Silts with Continuous Permeable Layers 1.5 to 15 Adapted after Leroueil and Jamiolkowski (1991). Note: kh = horizontal hydraulic conductivity; kv = vertical hydraulic conductivity. TABLE 3 PERMEABILITY ANISOTROPY IN NATURAL CLAYS FIGURE 53 Comparison of horizontal and vertical permeabilities on natural clays (after Leroueil et al. 1990).

45 horizontal to vertical permeability to address possible issues of anisotropy during interpretation. The study compared laboratory-determined results with the SPM solution (Teh and Houlsby 1991) using data from Type 1 piezocones (22 sites) and Type 2 piezocones (23 sites), as well as eight sites where backcalculated field values of cvh were obtained from full-scale loadings. With the SPM approach in practice, it is common to use only the measured time to reach 50% consolidation, desig- nated t50. An illustrative example of determining t50 for a 15-cm2 Type 2 piezocone dissipation in soft varved clay at the Amherst NGES is shown in Figure 56. At a dissipation test depth of 12.2 m and groundwater located at zw  1 m, the measured t50  9.5 min. Using a standard adopted reference at 50% dissipation, the modified time factors are T50*  0.118 for Type 1 midface elements and T50*  0.245 for Type 2 shoulder elements. Then, the calculated coefficient of consolidation is deter- mined from: (42) where ac  probe radius and IR  G/su  rigidity index of the soil. For a 15-cm2 penetrometer, ac  2.2 cm. Using a value IR  40 and the measured t50  9.5 min gives cvh  0.79 cm2/min. The SPM may also be used to fit the entire pore- water decay curve, as shown on Figure 56. If dissipation tests are carried out at select depth intervals during field testing, a fairly optimized data collection is achieved by the SCPTù, because five measurements of soil behavior are captured in that single sounding: qt, fs, ub, t50, and Vs. The results of a (composite) SCPTù in the soft Amherst clays are depicted in Figure 57. Here the results of a seismic cone sounding are augmented with data from a c T a I tvh c R = ⋅ ⋅50 2 50 * FIGURE 54 Strain path solution for CPTu1 dissipation tests (after Teh and Houlsby 1991). FIGURE 55 Strain path solution for CPTu2 dissipation tests (after Teh and Houlsby 1991). FIGURE 56 Measured dissipation at Amherst NGES and definition of t50 at 50% consolidation.

46 For sands, the operational rigidity index can be evaluated from the Vs measurement using calibrations based on undis- turbed frozen sand specimens (Mayne 2006b). PERMEABILITY The permeability may be evaluated by means of the interre- lationship with the coefficient of consolidation and con- strained modulus (D) such that: (44)k c D vh w = ⋅  FIGURE 57 Seismic piezocone test with dissipations (termed SCPTù) at the Amherst soft clay test site. FIGURE 58 Evaluation of rigidity index from plasticity index and OCR (after Keaveny and Mitchell 1986). separate series of dissipations conducted by DeGroot and Lutenegger (1994). RIGIDITY INDEX The rigidity index (IR) of soil is defined as the ratio of shear modulus (G) to shear strength ( max). From considerations of cavity expansion theory and critical-state soil mechan- ics, the undrained value of rigidity index (IR  G/su) in clay and silts can be evaluated directly from the CPTu data (Mayne 2001): (43) where M  6sin/(3  sin). As this is an exponential function, the derived values are particularly sensitive to accu- rate CPT measurements and therefore require proper satura- tions for the filter and cone assembly to obtain u2 readings and correction of measured qc to total qt. If undisturbed samples of the material are available, the rigidity index can be measured in laboratory DSS or triaxial compression tests on undisturbed samples or, alternatively, estimated from expressions based in critical state soil mechanics (Kulhawy and Mayne 1990). An empirical corre- lation for IR developed from triaxial test data has been related to clay plasticity index and OCR (Keaveny and Mitchell 1986), as presented in Figure 58. I M q q uR t vo t = + ⎛⎝⎜ ⎞⎠⎟ − − ⎛ ⎝⎜ ⎞ ⎠⎟ −exp . . . 1 5 2 925 2 9 2  25 ⎡ ⎣⎢ ⎤ ⎦⎥

47 For this approach, results from piezo-dissipation testing are used together with an appropriate rigidity index to eval- uate cvh, and an estimate of D is obtained from either of the relationships with net cone resistance or small-strain shear modulus (or both), as discussed previously. Alternatively, a direct empirical method has been pro- vided by Parez and Fauriel (1988) based on the measured t50 value from the dissipation curves, as presented in Figure 59. An approximate expression for the overall mean trend (dashed line) is also shown. Some additional considerations in the evaluation of piezo- dissipation tests include (1) stress release of rods, and (2) dilatory responses. During the hydraulic push, pressure is placed on the cone rods in the advancing penetration. If a dissipation test is to be performed, then the rod pressure should likely be maintained during the time decay readings, because the release of the rod pressure may cause a stress drop in the initial readings. This is especially evident in Type 1 piezocone dissipation (Campanella and Robertson 1988); however, this can also occur in Type 2 readings conducted in stiff clays and silts. For Type 1 piezocones, porewater decay with time is always monotonic (decreases with time). For Type 2 piezo- cone filters in soft to firm soils, a similar monotonic decay is observed. However, during Type 2 dissipation tests in stiff clays and silts, a dilatory response can occur, whereby the measured porewater pressures initally increase after the halt of penetration, climb to a peak value, then decrease with time. Interpretations of piezocone data for dilatory response are discussed by Sully and Campanella (1994) using an empirical approach and by Burns and Mayne (1998, 2002b) within a model based on cavity expansion and critical-state soil mechanics framework. For the latter, a simplified method to address this is presented in Mayne (2001). OTHER SOIL PARAMETERS A number of additional soil parameters may be deter- mined from cone penetration results, yet are beyond the scope covered herein. Some guidance toward reference sources that address selected parameter topics is given in Table 4. ADDITIONAL CONSIDERATIONS: LAYERED SOIL PROFILES When pushing a cone penetrometer in layered soils, the advancing probe will sense portions of a deeper layer before that stratum is physically reached. For instance, the tip resis- tance in a uniform clay underlain by sand will register an increase in qt before the sand layer is actually penetrated. Similarly, a cone advancing through a sand layer underlain by softer clay will start to “feel” the presence of the clay before actually leaving the sand; therefore, the qt will reduce as the lower clay is approached. The result is that there will be an apparent false sensing of soil interfaces when CPTs are conducted in layered stratig- raphies having large contrasts between different soil types. Efforts to investigate these relative effects have been made using numerical simulations by finite-element analyses FIGURE 59 Direct evaluation of soil permeability from t50 measured in piezo-dissipation tests (after Parez and Fauriel 1988; Leroueil and Jamiolkowski 1991). skrameR ecnerefeR retemaraP lioS Attraction, a' = c' cotφ' Senneset et al. (1989) Defined as intercept from plot of net resistance (qt − σvo) vs. effective overburden (σvo'). Related to c' (below) California Bearing Ratio Pamukcu and Fang (1989); Amini (2003) Relates to pavement design Effective Cohesion Intercept, c' Senneset et al. (1988) Mohr–Coulomb strength parameter. Relates to attraction term above. Modulus of Subgrade Reaction, ks Newcomb and Birgisson (1999) NCHRP Synthesis 278 Resilient Modulus, MR Mohammad et al. (2002) Used in the design of highway pavement sections State Parameter of Sands, Ψ Been et al. (1986, 1987, 1988) Critical state approach for sands Strain Rate and Partial Saturation Randolph (2004) Conduct CPTu ìtwitch testin g” at variable rates of penetration TABLE 4 ADDITIONAL GEOTECHNICAL PARAMETERS DETERMINED BY CPT

(e.g., Vreugdenhil et al. 1994) and experimentally using miniature CPTs in chamber tests with alternating deposited layers of sand and clay. In the case of a sand layer that is sandwiched between upper and lower clay layers, Ahmadi and Robertson (2005) discussed the means to correct the apparent measured qtA in the sand to an equivalent qtA* for full thickness layer. The problem is depicted in Figure 60, as per Robertson and Wride (1998). The apparent measured value of cone tip resistance in the middle sandy layer (qtA) is influenced by the value of apparent cone resistance in the clay layers (qtB), the thickness of the sand (Hs), and the diameter of the penetrometer (dc). Results from numerical simulations and limited field data are 48 presented in Figure 61. A recommended conservative cor- rection is given by the lower bound that can be expressed as: qtA*  qtA {1  0.25 [0.059(Hs/dc)1.77]2} (45) Of course, it should also be realized that modern elec- tronic piezocone testing involves three or more continuous recordings with depth. Thus, the interface layering can be best ascertained by cross referencing the qt, fs, and u2 read- ings of the CPTu next to a carefully controlled log from an adjacent soil test boring and evaluated within the con- text of the available engineering geology understanding of the region. FIGURE 60 Situation of thin-layer effect on measured cone tip resistance (after Robertson and Wride 1998). FIGURE 61 Thin-layer correction factor based on numerical (Vreugdenhil et al. 1994) and field CPT data (after Ahmadi and Robertson 2005).