You are on page 1of 14
Towards systematic CPT interpretation K. BEEN and M..G. JEFFERIES, Golder Associates This paper describes a systematic interpretation of the cone penetration test with pore pressure measurement (CPTu) based on all three independent measurements from the CPTu, which is quasi-independent of site specific factors. The first step in the approach is to recognize the similarities of sands and clays within a critical state framework for soil behaviour, in particular the description of the initial state of the material. The CPTu can be used to determine the initial state, which largely determines subsequent material behaviour, in any type of soil. However, drainage conditions and the type of soil being penetrated by the cone affect the interpretation of initial state. The second important step in the interpretation is therefore to use the CPTu measurements to estimate the drainage conditions and mechanical behaviour (or type) of the soil being penetrated and to fold this understanding back into the assessment of initial state. Introduction The cone penetration test (CPT) is probably the most widely used in situ test in soil mechanics, with the possible exception of the SPT. Modern electronic instruments usually incorporate a pore pressure transducer, resulting in the so-called CPTu. There appear to be three levels at which CPT data can be viewed and utilized. On the simplest level, the CPT is used by many engineers merely as a stratigraphic logging tool. It does an excellent job of this, especially with the additional pore pressure measurement facility of the CPTu, and although the CPT is not a downhole geologist it has demonstrated success over the past 30 years in providing a reliable method of soil classification. At the second level, some engineers use the instrument as an indexing measurement for the in situ state of the soil; methods of interpretation which result in OCR of clays and the relative density or state parameter of sands fall into this category. On the third level the instrument is used to determine strength parameters, such as Cu, or @; this step requires calibration of the CPT results to independent strength measurements for any particular soil. Each aspect of the CPTu interpretation (soil classification, in situ state or strength) is essentially carried out independently. There is only a loose link between the soil classification interpretation and the state or Predictive soil mechanics. Thomas Telford, London, 1993 121 BEEN AND JEFFERIES strength interpretation. Different methods of interpretation are also used for drained and undrained response of the soils to cone penetration. The first step in the interpretation is thus for the engineer to decide whether the soil is a ‘sand’ (drained) or a ‘clay’ (undrained). Partially drained soils, such as silts, are not adequately covered by existing interpretation methods. ‘A modern CPTu gives three separate readings, qc, f, and u, which are seldom used together to enhance the interpretation, except for those soil classification charts which use a normalized penetration resistance and a friction ratio on the axes. This paper provides a unified conceptual background, using all three CPTu measurements, that is applicable to all soil types including silts. The framework is quasi-empirical, but relies on critical state soil mechanics for its fundamentals. Wroth (1988) pointed out that an empirical method should be based on a physical appreciation of why the properties should be related; it should be set against a background of theory and expressed in terms of dimensionless vari- ables. The method described satisfies these basic requirements. It should be noted that the method applies only to right cylindrical electronic piezocones and procedures which conform to the proposed International Reference Test Procedure (de Beer et al., 1988). Standar- dization of the equipment and procedures is required for any such empirically calibrated methodology. In addition, the friction ratio is important and the penetrometers should therefore have an independent transducer to measure sleeve friction (as opposed to the so called ‘subtraction cones’). Normalized CPT parameters The normalized, or dimensionless variables, used for the CPTu inter- pretation in this paper are defined as follows: Q= Ge-ewloo or — Qp = (Ge- Po)/ Po qa) F = £./(qe— ovo) Q) Bg = (u— uo)M(qe~ v0) @) where: qcis the cone tip resistance corrected for end area effects (sometimes q,), and therefore requires use of the CPTu, rather than the CPT eo, Oo are the vertical total and effective stresses before penetration Po, Ps are the mean normal total and effective stresses before penetration f, is the cone sleeve friction (corrected for end area effects) uis the measured CPTu pore pressure, in this case 5mm behind the conical tip up is the initial pore pressure in situ, 122 SYSTEMATIC CPT INTERPRETATION These dimensionless variable groups follow Wroth (1988). In this paper, the mean normal stress is sometimes preferred over the vertical stress. Mean normal stress allows more flexibility in the approach as soils with different Ko values do not have to be treated separately. If measurements of Ko are not available, values should be estimated (but preferably not from Ko = 1—sin¢') and a simple linear relationship between p’ and 0% exists. Many authors use oy in the above dimension- less groups and for the purposes of discussion in this paper the two can be treated as equivalent. However, when quantitative analyses are made, it is necessary to distinguish between the two definitions of Q (see eqn. (1)). The parameter Q is normally used alone for CPT interpretation to determine in situ density, state or strength parameters. Q and F combined result in a good soil classification chart (see, for example, Robertson (1990) for a recent update). B, has been used less frequently, but can be correlated successfully to OCR for a particular soil stratum (e.g. Wroth, 1988) although no universal relationship has been found. In addition, Houlsby (1988) suggests that (q-—u)/olo = Q(1—Bg) +1 should be a simple function of OCR. Relationships between parameter groups incorporated in the systematic approach The approach described in this paper utilizes several relationships between the dimensionless parameter groupings noted above. These relationships have all been published previously and are briefly des- cribed below. ‘A systematic interpretation of the CPT should include the influence of drainage on penetration resistance. Piezocones (i.e. the CPTu) provide an indirect measurement of the influence of drainage by measuring the pore water pressure induced during penetration as well as the conven- tional tip resistance and sleeve friction. The additional pore pressure information can be included in the soil classification interpretation through the parameter grouping Q(1 - Bg) +1 plotted against F, rather than simply Q against F (Jefferies and Davies, 1991). Figure 1 shows the usual CPT classification chart after Robertson (1990) and the extended version of Jefferies and Davies (1991). For drained tests in sands, Been et al. (1987) use the relationship between Q, and y (the state parameter, ot offset from the critical state line on an e—log p’ diagram, Been and Jefferies, 1985) shown on Fig. 2 to interpret the CPT. They point out that the correlations for different sand types appear to be related to the slope of the critical state line A for each sand, but the relationships proposed are less than perfect. It is noted that the parameter k is the intercept of the Q versus y lines with 123 1000 Ty 7 1000 rn ui 100 10 Normalized cone resistance(4; ~ OvcVo've meneasnnr wert Eovanuee L 4 bituil il = a E oa 1 10 (0 04 08 42 [ 2] Normalize treton rato t/q ~ Ga) x 100% Pore pressure ratio By [ q 1. Sensitive, fine rained 6. Sands — clean sand to sity sand pL 2. Organic sols — peats 7. Gravely sand to sand o ' ° 3. Clays — clays to sity cay 8. Very si sand to clayey" sand Foo 4, Sit mites — clayey sit 9. Very sti fine grained © tosity day 5. Sand mixtures — sity sand * Heavily overconsoidated or cemented to sandy sit fa) Fig. 1. CPTu soil classification systems (a) after Robertson (1990) and (b) proposed classification chart using Q(1 — B,) and F (after Jefferies and Davies, 1991) zt a 600, 400} (st a STATE PARAMETER, KuNTERCEPT AT Y=0 (@ 254 LEGEND werene ba a Seesde as eae tees ‘Bur ae swoch v 2 34 56 8 0 OCR * ey’ / v0 ) Fig. 2. Interpretation of in situ state from the CPT. (a) CPT interpretation for sands in terms of and d (after Been et al., 19874), and (b) CPT interpretation for clays in terms of log OCR (after Crooks et al., 1988) BEEN AND JEFFERIES STEADY STATE LINE void RATIO VIRGIN CONSOLIDATION UNE! 109 9" Fig. 3. Definition of state for sands (1) and clays (log OCR) the = 0 axis, corresponding to penetration of the cone into the material at its critical state. Penetration at the critical state is expected to be a strong function of the critical state friction angle (expressed as M in this paper), as well as the critical state volumetric hardening parameter A. The relationship between k and critical state parameters can therefore be improved by incorporation of M into the interpretation. The parameter m, which is the slope of the Q versus line, is also a function of soil type. Q can readily be related to OCR in clays (Wroth, 1988; Sills et al., 1988; Crooks et al., 1988) and there are numerous relationships between OCR and other parameter groupings such as (u — up)/o%o in the literature (e.g. Mayne, 1988). Been et al. (1988) show how OCR is converted to a state parameter y value comparable to that for sands using the mod-Cam clay critical state model (Roscoe and Burland, 1968): log 2+ W(x — A) @) where A = 1—x/A and R is OCR expressed in terms of pinax/po rather than the conventional vertical stress definition. Figure 3, from Been et al. (1988), shows the geometric relationship between these two parameters for clays. The mod-Cam clay model is necessary only to define the spacing ratio between the critical state line and the normal consolidation line. The critical state model therefore provides a basis for extending Q versus OCR relationships for clays into Q versus ¢ relationships which are comparable to those for sands. For sands y/ and OCR are independent quantities and such a transformation is not possible, although in clays it appears that the transformation is reasonable as the influence of OCR clearly dominates that of ‘A difference between the Q-/ relationships for sands and clays is drainage, or rather, pore water pressure. CPT penetration in sands is drained, while in clays it is essentially undrained and in silts it is logR 126 LL 500 $s VALUE ao 100 50] Los apne? LeseNo (09< \<0.20 F920 4<0.30 e030 -0.2 Or Fig. 4. Unified relationship of Q,(1 — B em 8 0 = cws bo a 4) to state parameter and critical state parameters M and font = 34088 a » (b) oF oa NOLLVIAYAAALNI Ld OLLVWALSAS BEEN AND JEFFERIES probably partially drained. Figure 4, adapted from Been et al. (1988), shows a unified relationship of Q,/(1 — By) versus , giving: gq = Qp/(1 — Bg) = k* exp (—m*y) 6) The parameters k* and m* are given as functions of the critical state parameters M and A. In algebraic form these are: KIM = 3+ 0.85/A ©) m* = 11.9- 13.34 ” It is noted that the relationship in eqn. (5) covers a range of material behaviours from fully drained sands to undrained clays. It is reasonable to expect that partially drained silts are encompassed by this relationship although this has not yet been demonstrated Basis for proposed interpretation The proposed interpretation is intended to be a systematic, critical state based, interpretation of the CPTu. In this context, critical state soil mechanics is used in the sense of Wood (1990). The important factor is that a critical state model for soil behaviour must consider volume changes. The model itself should predict how strength (c, or ) changes with state (OCR or density and stress level). This feature distinguishes critical state behaviour from other models and is crucial for a unified framework, as it is amply evident that soil properties remain the same while a soil behaviour (such as dilation, c, or 6) may change when stress or density changes occur. It is not sufficient to describe the same sand at two different states with different sets of parameters (e.g. 4). Within the context of critical state models, the most important parameters describing the soil properties are: M, A, « and the yield function (for example, the mod-Cam clay function used to define the relationship between y and OCR in the previous section). In addition to model parameters for the soil such as M, A and x, it is necessary to know the initial state of the soil before a prediction of soil performance under loading can be made. The initial state is described by pj, Ko, and OCR or y. Permeability should also be known for problems which are neither perfectly drained nor undrained. In a critical state framework, there are not many parameters which influence soil response during cone penetration, but they may be divided into model parameters and initial state parameters. The model parameters M, A and « describe the soil type and can readily be measured in the laboratory. Initial estimates of these parameters can also be made with reasonable accuracy by experienced engineers if the soil type is known. The initial state parameters OCR and have a much larger influence on cone penetration and cannot be estimated a priori by 128 SYSTEMATIC CPT INTERPRETATION engineers. The CPT should therefore be used to derive OCR and ¥. Permeability estimates can be made independently on the basis of dissipation tests during the CPTu. The proposed systematic approach to CPT interpretation is thus developed as follows: (a) Estimates or measurements of in situ stress conditions (0%, Ko) are made, (b) CPT is used in ‘classification’ mode to determine soil type, from which model parameters M and A are estimated. (0) CPT results are used to determine in situ state of soil. The above is nevertheless inadequate for engineering design. The following are also required: (a) CPTu dis pressures, (e) Laboratory tests to determine soil model properties. In some cases laboratory tests will require the initial CPT data to estimate the in situ state to ensure that the boundary conditions in the laboratory test are correct. (A) Reinterpretation of CPT results with the revised soil properties. (g) Derivation of design parameters (such as undrained strength or ) from the constitutive model as well as the CPT and laboratory data. sipation tests to determine in situ permeability and pore In practice, it is desirable that steps (b) and (c) are carried out automatically in the field during the CPT testing process. In addition, estimates of undrained strength or density can be made during step (). However, step (c) requires knowledge of the soil type, as defined by the critical state parameter A, and to a lesser extent M. These parameters may vary significantly from one soil layer to the next, so that it is not practical to use an engineer’s estimate for the initial field interpretation. While it is reasonable to associate a single A value with each soil (which is done in this paper), A varies continuously during the soil profile as different soils are encountered. The only way to deal with variation in properties adequately is if the value of A is estimated on a continuous basis from the penetration test data itself. What is needed is a basis for an algorithm to transfer the soil classification data from the ‘classification’ part of the CPT interpretation directly into the second level, or state, part of the interpretation. As noted above, this must be done on a continuous basis through the profile. between A and soil type index The CPT has proved itself a good tool for soil classification through several decades of application. It is a mechanical device that measures a 129 BEEN AND JEFFERIES (CENTRE OF CRCLES FOR ZONAL CLASSFICATION 1008 R=18 24 100: 6 2.76 x ms S 5 g 7S Z o ms 4 10: Bon 2 on Yo 70 F(%) Fig. 5. Revised soil classification chart with concentric zonation (after Jefferies and Davies, 1992). Refer to Fig. 1 for description of zones complex soil response to loading, and does not see the soil or estimate the grain size as an engineer or geologist might do to classify the soil. Given the CPT’s success at soil classification, there must be a link between the measured mechanical behaviour of the soil and the soil type. This link should be expressed in a soil type index calculated from the CPTu measurements. Figure 1 shows the existing CPT classification charts. Jefferies and Davies (1992) have shown that the boundaries on Fig. 1(b) can be approximated as concentric circles provided the vertical and horizontal scales are appropriately chosen (and unequal). Figure 5 shows the revised chart. With the concentric circle approximation, the soil type is indicated by the circle radius, which is thus used as a soil behaviour type index. A soil index I, is defined as follows: Ie = V{3—log[Q(1 — B,) + 1]}? + (1.5 + 1.3(log F)}? @) In eqn. (8), the factor 1.3 is the mapping used to obtain a plot with concentric circles and the centre of the circles is at log [Q(1 — B,) + 1] = 3, log[F] = -1.5/1.3. The logarithms are to the base 10. I. is a material behaviour index, ranging from gravels to clays. Table 1, from Jefferies and Davies (1992), summarizes the classification index. 130 SYSTEMATIC CPT INTERPRETATION, Table 1. Soil behaviour type from the classification index Ic CPT index I. | Zone Soil classification 1.<1.80 6 Sands—clean sand and gravel to silty sand 18

You might also like