24 EQUIPMENT AND PROCEDURES In terms of calibration then, if a device shows, for example, a repeatable but non-linear calibration, the use ofa linear approximation for the calibration would immediately result in a loss of accuracy, although the results may still be repeatable and precise, The loss of accuracy would be related fo the difference between the true and assumed calibration lines. Only by using the true non-linear calibra- tion could the results become accurate and precise, The use Of any incorrect calibration could result in repeatable (pre- cise) results which could be grossly in error and inaccurate. Precision or repeatability is not a guarantee of accuracy. ‘The most desirable situation is to have an instrument that is accurate and precise, This is a prerequisite to obtaining accurate and precise readings in the field, where it is important to record all information during the field testing that could influence the deduced readings (Chapter 3) 2.9 SUMMARY OF PERFORMANCE CHECKS AND MAINTENANCE REQUIREMENTS. ‘Table 2.1 summarizes the frequency with which many of the checks and calibrations related to CPT and CPTU testing should be done to ensure that the quality of the data collected is of an acceptable standard and comparable to data from other tests. Some aspects are also described in more detail in Chapter 3CHECKS, CORRECTIONS AND PRESENTATION OF DATA ‘The purpose of this chapter is to aid the reader in ensuring, that data obtained from a CPT/CPTU is of a known quality, and presented in a clear and systematic way. The chapter is, divided into three sections covering (a) factors that might affect the measurements and how to avoid or correct for them; (b) the presentation of data to ensure that it is of most, use and (¢) a summary guide to information required t0 enable checking of the data quality 3.1. FACTORS AFFECTING MEASUREMENTS AND CORRECTIONS The parpose of this section i to deseribe how various factors ay influence the results, how errors may be detected and hhow the results should be corrected to achieve the highest quality. Factors that affect measurements can be systematic ‘orrancom; the significant systematic effects are discussed in detail inthis chapter. It is generally accepted that the pore water pressures around @ penetrating cone influence the measured cone resistance and sleeve friction. The measured pore water pressure is also influenced by the location of the filter and sometimes by the axial load on the cone. Other factors such as temperature changes, inclination, calibration errors and wear ofthe cone can also influence the measurements. The factors discussed in Chapter 2 should also be faken into account. ‘The importance of the various corrections and errors will vary with the cone penetrometer being used, the deposit being investigated and the intended use of the data, How- ever, it is important that the corrections and errors be quantified in both magnitude and relative importance. 3.1.1 Pore water pressure effects on q. and f, Due to the “inner” geometry of a cone penetrometer the ambient pore water pressure will act on the shoulder area behind the cone and on the ends ofthe friction sleeve. This is. illustrated in Figure 3.1. This effect is often referred to as “the unequal area effect” and influences the total stress determined from the cone and friction sleeve. For the cone resistance the unequal area is represented by the cone area ratio a, which is approximately equal to the ratio of the cross-sectional area of the load cell or shaft, A,, divided by the projected area of the cone, Aq, as shown in Figure 3.1 ‘The corrected total cone resistance, g,, is given by the equation: a= ge ual ~0) Ga) ‘where up is the pore pressure acting behind the cone. ‘This effect was fist dentified when the CPT was used for deep water investigations, when it was observed that the cone resistance g. was not equal tothe water pressure. ‘To determine the cone area ratio, a, a simple calibration vessel of the type shown in Figure 2.18 is used. The calibration vesscl is designed to contain the penetrometer, suspended so that no dead loads are applied to the cone or fiction sleeve, and to apply an all-round air or water pressure, Readings are then taken of qc and , asthe pressure26 CHECKS, CORRECTIONS AND PRESENTATION OF DATA Cross sectional area (top) Ast Friction sleeve surface area As LJ Cross lie sectional area (bottom) Asp ue Ae Cross sectional area Figure 3.1 Pore water pressure effects on measured parameters, in the vessel is increased in increments. By plotting the cone resistance against the applied pressure, the cone area ratio ‘can be calculated as the slope of the line. An example is, shown in Figure 3.2, Many cone penetrometers have values of cone area ratios ranging from 0.9 to 0.55, but sometimes this ratio may be as, low as 0.38, as illustrated in Figure 3.2. A cone area ratio as, ow as 0.38 should be considered unacceptable when using, the CPT in very soft fine-grained soils, as the correction becomes the major contribution to g, with potentially increased loss of accuracy. Idcally it should be close to 1.0 and cones have been manufactured that attempt to achieve this. F 100 —— & a0 g 1a 5 so 3 y | 8 ao Se } @ o 2 0® Ee) pense 3 @ 2 ol a = Applied water pressure, u (MPa) Figure 3.2 Determination of “a” in a calibration vessel (after Baitaglio and Manisealco, 1983). The pore water pressure correction is especially important in sof fine-grained saturated soils, where pore pressures can, be large relative to the cone resistance, In Figure 3.3a, an example is given for four different piezocones at the Both- kkennar site with a values from 0.59 to very close to 1. The disagreement in qe values is sen to disappear once they have been corrected for pore water pressure effects into q,. In Figure 3.3b, a second example is shown from Aas et al (1986) ofthe effect of this correction for two piezocones with very different area ratios (0.7 and 0.38). It can be seen that coven with this big area ratio difference the results are in far better agreement once corrected (great care was taken with these tests) This correction is one of the main advantages of using piezocones since the pore pressure u can be measured and the unequal area effect correction made. To make this correction, the pore pressure measurement must be made behind the cone where the pressures are acting and in most cases this will be the u, location shown in Figure 1.1. “However, when the pore pressure is only measured on the cone (wu), rough estimate of the pore pressure sn can be ‘made based on an adjustment tou. One way to do this isto use the following formula (ta ~ Ma) = Kui = te) G2) ‘where uo isthe in situ pore pressure and K is constant ‘The value of K will vary with the stress history of the deposit and the sol type, This procedure should only be used a8 a last resort when no other data is available. The engineer ‘must have confidence that the soil type and stress history of the deposit have been suficiently well determined to allow a reliable selection of K. In Table 3.1, guidance is given on values for K based on available experience, but great care must be taken in its use. Since the fiction sleeve has “end areas” that will be exposed to pore water pressure, the measured sleeve friction will also be influenced by the pore water pressure effects (Figure 3.1). When excess pore pressures are generated, the pore pressures are normally different atthe upper (us) and lower (us) ends of the sleeve. Using the terminology in Figure 3.1, the corrected sleeve fiction, , can be given by: G3) ‘Table 3.1 Typical values oF adjustment factor Kin various soil types (modified after Sandven, 1990) Soil ype K lay CLAYS normally consolidated 06-08 = 2-3, CLAYS slightly OC, sensitive 05-07 6-9 CLAYS heavily OC, sit 0-03 © 10-12 SILT loose, compressible 0506 = 38 SILT dense, dilative 002 «35 SAND loose silly 0204 © 23 (0C= overcontoidted,FACTORS AFFECTING MEASUREMENTS AND CORRECTIONS 27 Cone resistance, 4,(MPa) 08. 0 02 04 06 10 Depth (m) Corrected cone resistance, 4,(MPa) 02 o4 06 08 10 12 go Figure 33a Example ofthe effect of carrecting q, in soft soils (Bothkennar). cand u (kPa) Corrected cone resistance, a; (kPa) es For cone B Gone B| : 3 - aks £ | £ i ; é 40 C1 2 us ‘Cone A Cone B n | | Figure 3.3b Example of the effect of correcting ge in soft sols (from Aas etal, 1984), The magnitude ofthe correction ean be reduced by having equal end areas (4,,=,) and making these end areas as small as possible, Normally it will be difficult to carry out this comection since 1 is seldom measured. However, experience has shown that for some commercially available cones the measured f can be influenced significantly by pore pressures, An example from an offshore profile at Gullfaks C (where up and ss were measured) is shown in Figure 3.4 Where the ratio of the corrected to the measured sleeve friction is plotted against depth. In this example, the errors are not that great since the end ateas of the sleeve were equal, for the cone penetrometer used, typically + 20%. The corrections can be more significant in soft fine-grained deposits where pore pressures are proportionally large com- ppared to measured sleeve frictions or when end areas are significantly different It is recommended to check the “end area” effect on the friction sleeve by measurement of 4,, and A,,and calibration,28 CHECKS, CORRECTIONS AND PRESENTATION OF DATA tests in the laboratory (similar to those mentioned above for, cone area correction). This will allow an evaluation of the relative importance of the effect. If wis not measured, itis, recommended that the correction should not be carried out. The pore pressure effect is one of the reasons why the sleeve friction readings are generally reported to be less reliable than the measured cone resistance (see, for example, de Ruiter, 1982; Lunne et al., 1986a). Corrected sleeve friction Measured sleeve friction (e002 04 06 08 10 12 14 Penetration below seafloor (m) 40 = Figure 3.4 Example ofthe effect of comecting f; (from Skomedal and Lunne, 1987), Us Taranto LL aay j os A hoe 2: 1 i+ London clay —__ below 1 8m J usu, Lightly Heavily ‘0c. ‘The corrections to both q. and f, will of course also have an effect on the assessed friction ratios (section 3.2.2). 3.1.2 Filter location ‘Theoretical studies and practical experience have shown that the measured pore pressure varies with the soil type and also the location of the filter clement. Figure 3.5 shows the predicted pore pressures along the : cone shaft from the recent theoretical efforts of Whittle and Aubeny (1991). They used the Modified Cam-Clay Model for Boston Blue Clay and the Strain Path Method. Similar 20 Pore pressures in soll 60° cone ' McC. BBC 15}. Au/dyo, Figure 35 Predicted pore pressure distributions (Whittle and Aubeny, 1991). Us| 15 uu, Figure 3.6 Measured pore prossure distributions (Robertson et al, 1986).studies have previously been done by Levadoux and Baligh (1986) and by Houlsby and Teh (1988), Figure 3.5 illustrates the rapid drop in pore water pressure that is normally experienced just behind the neck of the cone, This finding is confirmed by numerous field tests as summarized by Robert- son et al (1986) and shown in Figure 3.6. Generally, the largest pore pressures are measured in the zone beneath the cone where the compressive stresses are at a maximum. The cylindrical part immediately behind the cone will be in a 2one of normal stress relief, while both zones will be subjected to lange shear stresses. Henee, the large normal stresses will dominate the pore water pressure response beneath the cone, while large shear stresses dom- inate the response along the cone shaft In a saturated soil, an increase in normal stresses will yield positive pore pressures, whereas an increase in shear stresses may induce either positive or negative pore pressure changes, depending on the dilataney properties and the ‘mobilized shear stress level in the soil. Soft, normally consolidated to moderately OC fine-grained soils (undrained response) will, because of their contractive nature, yield positive pore pressures, both on the cone and along the cylindrical shaft, An example of this from the lightly over- consolidated Bothkennar clay i given in Figure 3.7. In dilative soils, like dense silts and sands and heavily overconsolidated clays, larger positive excess pore pressures ‘will stil be developed on the cone, whereas very low or even Pore water pressure (MPa) 8 go_02 a4 06 08 10 16| 18| 20 Figure 3.7 Fxamplo of pore water pressures at different filter locations ftom Bothkennar (from Powell and Quarterman, 1991). FACTORS AFFECTING MEASUREMENTS AND CORRECTIONS 29 Pore water pressure (MPa) PS_00_05 10 15 20 25 30 7 ee a i +t fi ot |) E ew | £10 | 3 & 1 1 t i I 1 i } i 5 \ Figure 3.8 Example of pore water pressure at different ‘iter Iecations from Gault clay. nogative pore pressures will be developed along the cone shaft. An example is given in Figure 3.8, which presents results from a dual piczocone test in the heavily over consolidated Gault clay. Lumne ef al. (1986a) found pore pressures on the very tip of the cone or midway on the conical part of the cone to be ‘very similar. Others have sometimes found a small tendency for the pore pressure nearer the cone tip to be smaller than that inthe middle of the cone face. This is reflected in some of the results presented in Figure 3.6, Robertson and Campanella (1988) also found that the filter size may influence the measured pore pressures, although little data is available to quantify the importance of this factor. AAt the moment the location of the filter element is not standardized. The ISSMFE Reference Test Procedures refer to the location behind the cone (us) as the preferred filter location. The advantages of this filter location can be surn- marized as: 1, The filter is much less prone to damage and wear. 2. Measurements are less influenced by element compressibility. 3, Pore pressures measured can be used directly to correct cone resistance (section 3.1.2). 4, Measured pore pressures during a dissipation test are less, influenced by procedure (locking rods or not), see section 544.30 For other locations, pore pressures measured on the cone ‘can, in some cases, provide a better definition of layers (especially thin ones or, when in heavily overconsolidated clays, the 1, pore pressure response is negative or very low, as shown in Figure 3.8). Pore pressures at both ends of the fiction sleeve are requited if proper correction of f, for pore pressure effects is to be made (section 3.1.1). ‘Hence, if only one filter location is to be included, the one immediately behind the cone (u.), is normally recommended. By having two or even three filter locations, even more information can be obtained from a piezocone test (for example, enhanced profiling capability and improved inter~ pretation in terms of OCR and even K,), Where practical circumstances allow dual or triple element piezocones to be used, considerable benefits may be achieved. An example of ‘triple element piezocone test in soft silty clay at Pentre is, shown in Figure 3.9. The increased definition of layers ean, be seen when the results of all three pote pressures are viewed together. ‘The importance of good saturation of the pore pressure measurement system should always be remembered if ‘meaningful results are to be obtained (section 2.3.7), Figure 3.10 shows examples of 1 profiles ina soft clay; the effects Of poor saturation are immediately obvious in Figure 3.10a where distinct steps are to be seen in the profile with litle definition, particularly at shallow depths. The profile in Pore water pressure (MPa) Pore water pressure (MPa) CHECKS, CORRECTIONS AND PRESENTATION OF DATA Pore water pressure (MPa) 00 os 10 15 20 Figure 3.9 Improved definition of layers from triple element piezovone tests Pore water pressure (MPa) povoz og op 08 po O2 04 08 a8 fo 02 G4 08 08 a 2 2 4 4 ‘ 8 8 8 e° 8 8 E10 0 ‘0 bx» 12 12 ul “ “" 8 ‘8 ‘8 8 8 8 2 20 2» Figure 3.10 Example showing effect of poor saturation on pore pressure measurements.FACTORS AFFECTING MEASUREMENTS AND CORRECTIONS 31 Figure 3.10b shows better response but stil poorer than tht of Figure 3.100 which shows by far tho best definition and response. All these profiles reach similar values of pore pressure at depth but with varying degrees of response, 3.1.3. Effect of axial load on pore water pressure readings During the development and calibration of some piezocone penetrometers, Bruzzi and Battaglio (1987) observed that the penetration pore pressure u could be influenced by the axial load applied to the cone. Laboratory and field inves- tigations were carried out with piezocones where the filter ‘material and filter position were varied. They found that the pore water pressure could be influenced as a result of the applied axial load causing deformations in the components of the probe, especially the pore pressure transducers and the filter elements. They found that by modifying the designs and using suitable filter materials, these effects could be eliminated. Special laboratory tests can be conducted to cheek for these effects. The test should involve loading the cone and filter element to see if a change in pressure is measured with the pressure transducer, An example of such a testis shown in Figure 3.11 from Bruzi and Battaglio (1987) where the pore pressure measuring system is seen to respond to the axial loading for one cone in Figure 3.11a, while the cone used in Figure 3.11b is seen to be free of this interaction. 10] q, (MPa) 0.5 Time (min) ux 102 (MPa) Most commercially available cones are now designed and checked to avoid these problems of axial load influence, but confirmation should be obtained that this is the case, espo- cially when using piezocones of a new design. 3.1.4 Temperature effects With any mechanical device containing load cells or sensors, temperature changes can affect the measurements. As part ofa laboratory and field testing study of a number of commercially available cones, Lunne et al. (1986a) showed that temperature changes may have significant effects on piezocone measurements. ‘The main reason that temperature effects can be significant is that a change in temperature can cause a shift in the load cell output at zero load, Modern high-quality piezocones have load cells that are to a large degree temperature compensated, although at small loads, such as in soft clays, this effect still needs to be carefully considered, In addition to careful temperature compensation of the load cells, there are two ways that temperature zero shifts may be avoided or corrected for: 1, Make sure that zero readings are taken at the beginning and end ofa test at the same temperature asin the ground. 2. Mount a temperature sensor in the cone and correct the ‘measured results based on laboratory calibrations. 25). 05 Time (in) -———_______. 0s Time (min) Migure 3.11 Examples of the effect of cone design and axial load on the measured pore water pressures, (a) poor design; (b) improved sesign (Bruzzi und Battaglio, 1987).32 CHECKS, CORRECTIONS AND PRESENTATION OF DATA ‘The first approach is normally the best to use, as outlined in section 2.3.3. One field example where temperature corrections proved important is given in Figure 3.12. Very similarly shaped profiles at a soft clay site were obtained on different visits, although they were displaced from each other. By applying a correction to the zero readings so that all profiles were ‘zeroed to the ground temperature, a much better agreement between the profiles can be seen, The results were taken from an early design of a McClelland cone; subsequent improvement in cone design significantly reduced tem- perature effects (Lunne et al, 1986a). If the temperature in the ground is not known or if itis ‘impractical to zet0 cone sensors at ground temperature, the second approach may be considered. However, this approach may be difficult to apply as the relationship between temperature and load cell output obtained in the laboratory under static conditions may be somewhat differ= ent from that of a heat flux through the penetrometer body during penetration, If reasonably steady temperature condi- tions are recorded in the body for prolonged periods, it may be possible to apply some form of correction for zero shift based on temperature ‘The use of a temperature sensor in the penetrometer body ‘may explain anomalies when penetrating mixed deposits, as, Cone resistance, q, (MPa) 0 os 1 OP (ONSOY CLAY) | |- Corrected data (to7°C) 4 _-Recorded data Test zeroed at 6°C Depth of penetration (m) 15}— a L Note eee All 4 tasts performed leervee with McClelland C atiac 46cm? cone go Ls _ Figure 3.12 Effect of temperature correction on field measure ments (from Lunne etal, 1986a). well as showing ifa reasonably constant temperature regime, thas been obtained. In sands, the temperature may inerease due to friction between the cone penetrometer and sand particles. Tem- perature increases of up to 30°C have been reported (see, for ‘example, Zuidberg, 1988). As sands usually have a high cone resistance, this may not be significant. However, with a relatively soft clay layer beneath a sand layer, the effect may become significant since the temperature compensation sys- tems take a finite time to respond to the temperature chan- ages. For example, Post and Nebbeling (1995) suggest that if ‘meaningful results are to be obtained in a soft clay layer which has a sand deposit above it, penetration should be stopped before the clay layer until the cone temperature has stabilized, The effects of temperature are not restricted to the cone penetrometer; changes in the temperature of the data acqui- sition system can also result in zero shifts in the recorded data. The temperature changes can result from both the ‘warming up of the data acquisition system and the ambient, temperature change. If the system is always switched on well in advance of the start of the test, it should only be ambient changes that are important within a test. Post and [Nebbeling (1995) suggest that, for the particular system they investigated, a S°C change in ambient temperature could result in a 10 kPa change in qc. While this may not appear {oo significant, the effect will vary with the data acquisition, system being used. The importance will also increase during Tong dissipation tests, especially ovemight, when tempera- ture changes could be large in certain climates. ‘The likely influence of all temperature effects should be quantified in some way. 3.1.5 Inclination In section 2.3.2 requirements for verticality of the thrust ‘machine and straightness of the rods were discussed with a view to ensuring that the penetrometer followed as vertical a ppath as possible, By including 2 simple slope sensor in the ‘penetrometer, the inclination can be monitored and recorded 50 that the path can be mapped. Figure 2.13 illustrated how this information could be used when necessary to allow corrections to be made to the depth of penetration. In this, particular example an error of 6 m in 37 m can be seen, 3.1.6 Callbration and resolution errors In sections 2.5 (Calibration of sensors) and 2.7 (Choice of capacity of load cells) many points were raised with regard, to ensuring that the output from the sensors can be converted ‘with known accuracy and confidence into the required CPT/ CPTU parameters q., f, and u. It is important that the potential accuracy or inaccuracy of the measurements as & result of errors attributable to calibration factors are quantified.FACTORS AFFECTING MEASUREMENTS AND CORRECTIONS 33 MPa MPa 9002 04 06 08 10 90_02_04 06 08 10 fatal 2 — 2 4 4 6 | 6 el rel 4, £ pi 10 + 8.4 [wel “4 16 - 16 18}-—+—+ 8 20 20 Figure 3.13 Effoct of resolution in cone resistance readings. If there is any indication of non lineatity in the calibra- tions, itis important to know what approximations have been ‘made in processing the results. This can be especially true when using high capacity load cells over the lower end of their range in soft deposits when intemal friction from ‘O* rings etc, can have marked effects on the initial responses. ‘The resolution of the measuring and recording system should also be known. In Figure 3.13 are examples of two sets of CPT profiles from a soft clay site (Powell and Quarterman, 1995). In the first set, significant scatter in the results is seen and has been attributed primarily to electrical noise superimposed on a small voltage output for the range of interest of a high capacity load cell used. The second set, shows what can be achieved when using a different cone with, load cells that give an output large enough to be unaffected, by the noise, The first set may well have been taken as implying a much more variable deposit than actually existed. 3.1.7. Effect of wear In section 2.6 reference was made to the importance of checking the cone and friction sleeve for damage and wear. If the allowable tolerances of the IRTP for cone diameter (Cable 4.1) are adhered to, then the maximum error in qe that can be obtained by assuming a 10 em’ cross-sectional area of the cone is 5%, simply from wear of the cone diameter. These, errors can be significantly greater if regular checks are not, made for wear ofthe cone and friction sleeve, The effect can of course be allowed for if the actual cone diameter is used. 3.1.8 Correction for CPTU zeroed at the bottom of aborehole When undertaking CPTUs for offshore soil investigations two different modes of testing are available, namely the downhole (or borehole) and seabed modes (see Section 2.2.2), Tae CPTU sensors are normally zeroed when the cone {s atthe seabed in the seabed mode andl atthe bottom of the borehole in the downhole mode, The result ofthe downhole tests need to be corrected in order to comrect for the error involved in zeroing atthe borehole base and also to allow the results from the two test modes to be compared. If the results of a downhole test, at a dopth z below the bottom ofthe borehole, are g* and uf respectively, then the values corresponding to a scabed test, g. and 2, can be computed by adding the effect ofthe water pressures in the boreliole, as tym ut + pwd G4) Ge= abt dea yy Bs) where d is the depth of the bottom of the borehole below the seabed, ‘a’ is the area ratio of the cone (see section 3.1.1) and 7, is the density of the sea water. Further, from equation 3.1, = gh + day + (1 — aua® + Yd) ft (laut + red 6) ‘and the excess pore pressure, Aup is Aun = uf + yy rd = yd +2) =ut— rez 3.7)34 CHECKS, CORRECTIONS AND PRESENTATION OF DATA | If the friction sleeve does not have equal end areas (see seation 3.1.1) then it is theoretically correct fo make a similac correction to the measured f;. However, in practice only q_ and u ae corrected forthe borehole effect. Zwvaag and Sunderland (1982) showed that there can be considerable discrepancy between seabed and downhole tests duc to the fact that zeroing of the sensors are done at two different levels 3.2. PRESENTATION OF RESULTS ‘The following section gives recommendations and require- ‘ments for the final presentations of the CPT and CPTU data, in the field report. The information that should be presented. falls naturally into the following three categories: i ‘¢ measured parameters ‘© corrected and derived parameters additional information. Each of these is discussed in turn below. 3.2.1. Measured parameters The measured parameters: qc, f, and u should be plotted on ‘one sheet, At any one site all profiles should be presented to a ‘common set of scales. The following are the scales recom ‘mended by ISSMFE Reference Test Procedure (1989) Depth sale: vertical axis, 1 unit length for 1 m Cone resistance qo: horizontal axis the same unit length for? MPa Sleeve fiction: horizontal axis the same unit length | for 50 kPa Pore water pressure: horizontal axis the same unit length 0020 kPa, Figure 3.14 shows an example where the measured values, have been plotted according to the above recommendations, The above scales may not always be the most useful or beneficial for a given site or purpose. Figure 3.15a shows, fan example of a soft clay site plotted to these scales. There isan obvious disparity between the lack of definition for the g. compared to the pore pressure, Figure 3.15b shows, the same data plotted to much more useful scales, Con sideration should always be given to the most appropriate scales to enable clear presentation of the information. It may be that if a profile contains layers with significantly different values of cone resistance and sleeve friction, the data should be presented with additional plots with the softer layers presented to an expanded scale, ‘After a stop in penetration when a dissipation test has been performed, the measured change in g. and u should be plotted versus time, It is practical to use either logarithmic ‘oF square root scale for time. An example is shown in, Figure 3.16. q.(MPa) ,(MPa) Pore pressure, u (MPa) o 4 8 12 16 2 po of 02 04 02 03 04 05 06 07 08 2 4 ; Riga 3.14 Data presented aocording to the Revommended Test Procedure.Cone resistance, q.(MPa) Sleeve friction, f,(kPa) PRESENTATION OF RESULTS 35 (MPa) 4,(MPa) Pore pressure, u (MPa) po_4o 80 po 01 po on o2 03, 04 08 os 2 2 2 4 4 4 3 6 6 3 8 8 = 219 10] 10 a, 2 12 14 14] 14 16] | 16 16 wo 18 19 } 20! 20 20 igure 3.158 Data presented according to the Recommended Test Procedure fora sft clay. Pore pressure, (MPa) Figure 3.1Sb Data presentation for maximum dotail fora sof clay, 002 04 06 08 10 40 60 190 150 200 go 02 0 06 08 10 A < AS 3 4 a 6| 6 eA 8 E a) 10 é 8.2 2 2 “4 4 1“ 16 1 ‘6 18 18 18 20 20 20| 36 (CHECKS, CORRECTIONS AND PRESENTATION OF DATA. t 3.2.2 Derived parameters | = The following parameters should be derived and plated 06 ‘when possible: | 3 + (a) /= cone resistance corrected for pore pressure effects So4 (section 3.1.1) 3 {i= sleeve fiction corrected for pore pressure effects $ al (ection 3.1.1) only valid when pore pressures have i 2 02) been measured at both ends of the friction sleeve. & R= friction ratio, usually in %. f 00 sums su Lun! a Oot 0410100 10.00 100.00 1000.00 Rak G8) ' Log time (min) on y 08; or ideally | ge fi Soe! ® 7 ce I s (b) , " é \ ~—f 4" ‘or more typically t § o4 ki i g Bek G1) 8 a 5 oa] Au } & 28, pore pressure ato = @.11) faa GRO PRG ae) oo aed) where Root time (min“®) Figure 3.16 Example of pore water pressure decay plotted on (a) Jog and (b) square root time seales. ‘excess pore pressure (u — i.) in situ equilibrium pore water pressure in situ total vertical stress. I q,(MPa) o 4 8 2 16 2 8 10 99 04 02 03 04 05 06 07 08 | 10} "a ah >= es eh) te. |PRESENTATION OF RESULTS 37 1.00, a 0.80 08 : j $080 sip! i 040) Ss o4 020! 02 0.00! 0.0) ‘0070.10.00 10.00'~ 100.60" 1000.00 0 5 10 15 20 25 30 Log time (min) Root time (min) ‘Figure 3.18 Example of normalized pore water pressure decay curves (a) log, (b) square root time scale. ‘Table 3.2 Checklist for information required when roceiving results from CPTICPTU to ensure and check data quality Question Answer Note | i 1. Type of cone penetrometer Manofatrer, capacity, spe. ISSMEE i standard i 2, Adhering ip interational standard ‘Compare with equtements in ISSMFE | "Ge 10 em’, 6% slove area = 130 mete) TRIP(ADDA) i 3. Ifanswer to 2 is no, what is difference? x-sect.aren TEA = 15 em’, a= 60 | ‘cone angle and Ay = 202 em’; ok i | sleve area = i 4 Location of ier) for measuring pp. and i ‘ype of fid 5 Area rat) ofcone ip ‘Nonmally in rangea = 059-088 “6 End areas offition eve st they are equal 7s ge corete for pre pressure ese? Compare with formula given inthis ehaper 8 Inf contd fer pore pressure fos? ‘Compare with formulas given inthis hapier 9. Whats bass for ois? Assumed? Base on measurement of yon samples? 10, When were sensors (gm) las calibrated? Compare with requirements given in specications or Table 21 11. ro readings Defoe an afer cach est Trportnto check if resus appear ported? “now-nomal” 12, Where were euings zeroed? Tempvtant for overwater testing (@.g. sea bottom or bottom of borehole) 1. Depth of ay pro-iling : plains any ising daa 14, Whats fequoncy of readings? The commercial rate is oe set every second, i. every 2m. Decided by pect, - requirements 15, For dsipation esting How wall waste initial par of the H (@) were the rods clamped or uslampod? disipatin curve dafned~ faster sampling (b frequency of readings rae to start with?38 CHECKS, CORRECTIONS AND PRESENTATION OF DATA Figure 3.17 gives one example of plotting ofthe derived parameters. As discussed in section 3.2.1, if the profile contains layers with low cone resistance and sleeve ftietion values, the horizontal scale should be expanded so that the measured values represent a larger part of the scale, In some cases with alternating layers with low and high qe and ‘fr values, he solution may be to give two sets of plots with two different horizontal scales ‘When a dissipation tet has been carried out, the normal- ized excess pore water pressure, @.12) where u, measured pore water pressure at time 1u;= measured pore water pressure at time = O tue = in situ equilibrium pore water pressure should be plotted against the log and/or square root of the time 1. The data from Figure 3.16 is replotted in this form in, Figure 3.18, Dissipation testing is discussed further in Chapter 5. 3.2.3 Additional information Information that must be provided ‘ach diagram with CPT or CPTU results shall include the following information: site name test no. date of performing test serial no, of cone penetrometer position of porous element(s) ‘ground water level (or water depth) name and signature of the operator and the company depth of predrilling if relevant. ‘The field report should also include the following, information: © Description of used and ‘manufacturer(s). © Cone geometry and dimensions if deviating from ISSMFE. International Reference Test Procedure. ‘© Calibration factors for all sonsors and the load range over which they apply. Capacity of each sensor, ‘© Zero readings of all sensors before and after cach test, and the temperature at which taken; alternatively change in zero readings expressed in kPa. ‘« Type of liquid used in the pore pressure measurement system, © Observed wear or damage on cone, friction sleeve or filter element, ‘¢ Any irregularities during testing relative to ISSMFE Ref cerence Procedure or other standard being used. equipment name of ‘¢ Area ratio of cone, a, and the friction sleeve end arcas. ‘© For dissipation tess it should be noted whether or not the rods were clamped or unclamped during dissipation In Table 3.2, a summary of the essential information that must be supplied with CPT/CPTU test results is given in the form of a checklist that can be copied and used when receiving CPTICPTU data. 3.3 CHECKS ON DATA QUALITY Provided the CPT equipment and procedures follow the standard reference described in Chapter 2, the results are normally of good quality, especially in dense or stiff soils, where the measured values are reasonably high compared to the capacities of the load cells. In soft soils there is a ‘greater chance that the measured values (notably q. and f,) may be less reliable due to the various factors discussed in, section 3.1 Before using the data for interpretation in terms of soil parameters for design, itis advisable to make the following, checks: 1. Check if the difference in zero readings before and after atest are within acceptable limits (in soft clays this may correspond to * 20 kPa for q,). Ifthe difference in zero readings is significant, the CPT results should be cor- rected or discarded, 2. If inclination of the cone penetrometer is measured during a test, this should be reviewed; for deep tests a significant inclination will tend to give false depth readings, 3. For piezocone data the pore pressure should be reviewed to see if the response is sufficiently good and adequately defining the stratigraphy. Filter location should be con- sidered and pore pressure data should be used to correct e data for pore pressure response (section 3.1.1). 4. If dissipation tests have been performed, the response time and equilibrium pore pressure should be evaluated to assess the level of saturation of the piezocone system, 5. Reference level for zeroing sensors should be noted especially for over-water or downhole tests; for exam- pile, inthe latter itis common to take zero readings at the bottom of the borehole. 6. A rough check to see if the measured qe are within the following acceptable lower limits is Ge? 0 (= Yor") where yoy is the average unit weight of soil and 2 is depth below ground level. Setting 7 equal to 15 KN/an” the check becomes: go> 152 (in kPa) : a‘The purpose of this chapter isto briefly describe some of the available standards, guidelines and specifications that can be referred to when planning a soil investigation programme that includes CPTSs, However, the fuctors that ean influence the results and thereby the specification and interpretation of CPT/CPTU tests and which were discussed in Chapters 2 and 3, must also be taken into account when devising a specification, It is not the purpose of this chapter to give a specification for CPT/CPTU. Several relevant documents are briefly discussed in the following, with emphasis on common features and identi- fication of special aspects. A summary is then made in tabular form for a quick and easy comparison between the basic features in the various documents (Table 4.1). Some additional national standards and guidelines are also included in the tabulated summary for comparison. 4.1 ISSMFE INTERNATIONAL REFERENCE TEST PROCEDURE FOR CONE PENETRATION TEST (CPT) The ISSMFE International Reference Test Procedure for Cone Penetration Test (referred to as IRTP in the following) was prepared over many years by the Technical committee on Penetration Testing appointed by the Intemational Soci- ety for Soil Mechanics and Foundation Engineering. The final version of the document was published by the Swedish Geotechnical Instituto in connection with the Intemational Conference on Soil Mechanics and Foundation Engineering STANDARDS AND SPECIFICATIONS in Rio de Janeiro, Brazil in 1989. Since this is the document ‘that is most referred to internationally and since it also forms the basis of most other CPT specification documents, it is included in Appendix A. The national standards or guide- lines described later in this chapter generally use the defini- tions given in the IRTP. ‘The IRTP is general in nature and covers electric, mechanical and hydraulic penetrometers. The possible ‘measurement of pore water pressure, and hence the piezo- cone, is included, but with very litte detail on specification, Since this book concentrates on the electric cone pen- etrometer and particularly the piezocone, Table 4.1 sum- marizes the basic requirements of the IRTP for these devices, only. Complete details can be found in Appendix A. 4.2. SWEDISH GEOTECHNICAL SOCIETY (SGF): RECOMMENDED STANDARD FOR CONE PENETRATION TESTS (1993) ‘The Swedish Geotechnical Society’s document Recommen- ded standard for Cone Penetration Tests was one of the first of the new generation of CPT standards that dealt in detail with CPTU as well. It contains information on the saturation and calibration of the pore pressure sensing component of the test ‘A very special feature of this document is thatthe test is divided ‘into three test classes according to the required accuracy of the results, Each class is described below and a summary of their required accuracies is given in Table 4.2srs ppeEE any anatey Lo sosv cies ene ee $50e 0 = = > = = enon aa Son ‘Sapent pe papers eon SOREN) FAL2 p40 wD 33 6 FEY EEL 28 ‘ues podnapisi eect 5} Ie 2 PENS (lames pouped easy ox “re See re pao sone vt 2 reaps ecm = pret ape: soma sep ne “ouput steanber ne) od = arutv @ et cies oss ote eva oe= ay ausy ows brs ee UES oss) anny sr EDT Tees papers nd aay ipsa 7 = = Sar ae ae = ag weeds ppoommony Smee wed tapes pos due ya pam wR esd joretwa sopearear airs42 ‘STANDARDS AND SPECIFICATIONS ‘Table 4.2 Generally accepted inaccuracies in different test classes; according to SGP/Swedish Geotechnical Society Recommendation Standard for Cone Penetration Tests (1993) ‘Test class Cone resistance Sleeve fiction Pore pressure xray (ea) (ePa) (ia) cern 100 10 0 or 40 4 5 ors 20 2 1 Class 1: designation CPT1 ‘This is the least accurate class, see Table 4.2, Normal equipment will be a S- to 20-tonne capacity friction cone or piezocone. ‘# Same tolerance for the cone geometry as IRTP. ‘@ Application: all types of soil. In soft fine-grained soils, the test class should only be used in those cases where the soil stratification prevents use of a higher test class. ‘© Possible interpretation: stratification and properties in sands and coarse silts, Stratification in stiff fine-grained soils can be made, provided that the pore pressure is ‘measured in the test (piezocone test). ‘© Limitations: if possible, « higher test class should be used in fine-grained soils. At great water depths, the test should ‘be made with a piozacone, Class 2: designation CPT2 This is the class of medium accuracy (Table 4.2) normally using a five-tonne capacity piezocone. ‘Narrower tolerances for the cone geometry than IRTP. ¢ Application: all types of soil. In dense sand, the penetra tion may be restricted. In soft fine-grained soils, the test should be made according to class CPT3 if the straifica- tion so permits. ‘¢ Possible interpretation: stratification and properties in all, types of soils. The accuracy, however, is limited in soft and medium stiff fine-grained soils. ‘¢ Limitations: the penetration is restricted in dense sands. Evaluation of properties in soft and medium stiff fine- ‘grained soils should be made with caution, Class 3: designation CPT3 This is the most accurate class (Table 4.2) with very strict requirements ‘¢ Specially calibrated five-tonne piezocones or piezocones, with smaller measuring ranges. Narrower tolerances for the cone geometry than IRTP (ag Class 2). ‘© Application: all types of soil. In sands and other stiff soils the penetration is restricted. In cases where the load restrictions and accuracies cannot be fulfilled, the test class has to be lowered. ‘« Possible interpretation: stratification and properties in all types of soils. However, because of load restrictions, itis mainly limited to soft and medium stiff fine-grained soils « Limitation: the penetration is restricted in coarse and stiff soils and at stiff layers in fine-grained soils. It is the opinion of the authors that most commercially available cone penetrometers will not presently be able to satisfy CPT class 3. However, the most accurate devices may well be able to satisfy CPT class 2 ‘As summarized in Table 4.1, the SGF standard differ. entiates between the three different CPT classes as regards requirements in dimensions and tolerances for manufactur ing and wear. ‘An additional feature of the SGP standard is that it has introduced a special requirement for temperature stability, as follows: “All measuring elements and all other electronic devices shall be stable for temperature changes. When handling the equipment, care shall be taken to minimize tem- perature changes in the cone. ‘The required stability expressed in maximum zero shift is 2.0 kPa/°C for cone resistance 0.1 kPa/*C for sleeve friction 0.05-0.1 kPa/°C for pore pressure (transducers with ‘measuring ranges 1-2 MPa) This demand on stability, which applies to five-tonne capacity cones, shall be verified. For cones with higher ‘measuring ranges a proportional instability is accepted.” ‘The SGF standard requires that q- and f, should be corrected for pore pressure effects and that the following parameters should be presented after atest: fi My ReC@s filly in 9%) and DPPR = Au/g, or alternatively By. In the authors" opinion it is only possible to follow the requirement to correct the sleeve friction for pore pressure effects when both u and us are measured. If the correction cannot be done properly, f; should be used, As mentioned catlier, it is preferable that cone penetrometers with equal fend area friction sleeves be used in order to reduce the cffects of pore pressure on f. ‘As an example of the latest type of specification doct- ments, the SGF standard is included in Appendix B.DUTCH STANDARD: DETERMINATION OF THE CONE RESISTANCE AND SLEEVE FRICTION OF SOIL 43 4.3. NORWEGIAN GEOTECHNICAL SOCIETY (NGF): GUIDELINES FOR CONE PENETRATION TESTS (1994) ‘The Norwegian Geotechnical Society have produced their own document “Guidelines for Cone Penetration Tests”, ‘which is 2 relatively comprehensive document based on the IRTP and several of the ideas contained in the SGP docu- ment, Itis again applicable to both CPT and CPTU. The NGF Guidelines also differentiate between three levels of required accuracy: high, medium and low accord- ing to the type of soil: “With high accuracy the results can be used for precise evaluations of stratification, soil type and soil properties. With low accuracy the results ean only be _used for simple stratification and preliminary evaluations of soil type and soil properties However, the NGF Guidelines are not as strict asthe SGF document with regard to required accuracy, ‘Tho NGF Guidelines give examples of CPTU profiles for range of Norwegian soil types. 4.4 ASTM: STANDARD TEST METHOD FOR PERFORMING ELECTRONIC FRICTION CONE AND PIEZOCONE PENETRATION TESTING OF SOILS (1995) ‘This document gives detailed descriptions/instructions on a number of matters including: different combinations of load cells that may be used, filler types and fluids, fiction reducers, calibration of new and repaired cone penetrom- efers and calculation of parameters to be presented. The calibration of new and repaired cone penetrometers is detailed in an annex to the standard The ASTM standard also allows for 15 em? cone pen- ctrometers and includes manufacturing and operating dimension tolerances. Guidance is also given on how close a CPT/CPTU sounding should be to soil borings: “...should not be Performed any closer than 25 borehole diameters from any backiilled or uncased borchole”, Regarding tests in pre- bored holes, the ASTM standard states: “an estimate of the depth below the pre-bored depth which is disturbed by Grilling, should be made and penetration resistance data obtained in this zone should be noted. Usually this depth of disturbance is assumed to be equal to at least three borehole diameters.” ‘The ASTM standard does not give any requirements to the total accuracy of the measurements. However, for newly ‘manufactured or repeited cone penetrometer, the following, requirements are given: Linearity beter than 1% of Full Scale Output (FSO) zero load error better than * 14 kPa (1.0 Ib/in’) ‘The standard also reflects on the precision that, based on past. experience, may be achieved: ge, Standard deviation about 2% of FSO ‘fz, for subtraction type load cells, stdev. 15% of FSO. for independent type load cells, stdev. 5% of FSO 1, st dev. about 2% of FSO ‘The document relies on past experience in many cases and ‘may not fully reflect what ean now be achieved. 4.5 DUTCH STANDARD: DETERMINATION OF ‘THE CONE RESISTANCE AND SLEEVE FRICTION OF SOIL. NEN5140 (1996) This is the latest specification to be published and comes fom the home of the CPT. Itis restricted to testing for cone resistance only or for cone resistance and sleeve fiction. Like the Swedish and Norwegian documents it introduces classes of test based on required accuracy of the results, which are project dependent. There are four classes as outlined in Table 4.3. In general these are much coarser requirements than for the other two specifications, but it is supposedly based on practical considerations encompassing all potential inaccuracies in the testing which “allows a more practical and commercial selection of a CPT class on the basis of project requirements”. The resolutions ofthe meas- uring systems are to be beter than one third ofthe measure- ment uncertainty of the required class It willbe soen that the inaccuracies in inclination and penetration depth are also part ofthe classification. Classes 3 and 4 are seen primarily to establish stratig- raphy, Class 2 is said to permit the derivation of geotech- nical parameters in stiff clays and sands but requies rigorous procedures. Class I is for soft ground engineering including derivation of geotechnical parameters and the use of very well-calibrated equipment. This standard also requires the use of actual areas for the cone and frietion sleeve in calculating parameters for classes 1 and 2. I also requires chocks on depth measurements by recording periodically throughout atest the depth penetrated by reference to the Length of rods pushed, Table 43 Dutch NENS140 (1996) Classification system: allowable uncertainty (the larger ofthe values) Penetration Cone Sleeve Class resistance fiction —Ilination depth 1 SOKPaor3% 10kPaor10% 2° 0.2 mor 1% 2 280KPaorS% S0KPaor15% — 2° 02 mard% 3 S00KPa or 39% S0KPaor20% «5° 02 mar2% 4 ‘500 kPa or 39% 50 kPa or 20% not required 0.1 mor 1%44 ‘STANDARDS AND SPECIFICATIONS ‘When reporting results for classes 1 and 2, tabulated data, as well as plotted, is to be supplied for all parameters to @ time base and specified resolution. Cone penetrometers of other sizes are allowed for special projects and provided they are in the range 500mm? to 2000 mm®, no correction factors are to be applied for size effets, 4.6 RECOMMENDATIONS It is not the purpose of this book to set a standard for specification document. By its nature, any base specification document should not be too prescriptive in its content, but should contain essential requirements for equipment, calibrations and testing; it is then up to engineers specifying the test tg tailor the document to their own needs, The advent of test classes and accuracy levels that are project and test requirement specific is to be encouraged. Having decided on the precision desirable in the CPT results to allow satisfactory accuracy in the determination of geotechnical information or parameters, and having under- stood the requirements for good testing practice (discussed in Chapters 2 and 3), the engineer should be able to set a specification for testing that will give the required con- fidence level in the CPT resultsINTERPRETATION OF CPT/ PIEZOCONE DATA [As stated in Chapter 1, the CPT and CPTU have three main applications: 1, to determine sub-surface stratigraphy and identify mate- rials present, 2, to estimate geotechnical parameters, and, 3. to provide results for direct geotechnical design, This chapter covers the interpretation for items 1 and 2 above, while item 3 is covered in Chapter 6, The inter- pretation of geotechnical parameters is divided into three categories: |. fine-grained soil ~ penetration undrained (section 5.4) 2. coarse grained soil — penetration drained (section 5.5) 3. other or intermediate material ~ penetration process complex (section 5.6), This chapter also contains discussion on non-standard test procedures, unexpected test results, statistical treatment of data and software applications. ‘As an initial guide, Table 5.1 shows a summary of the applicability of CPT/CPTU for deriving soil parameters in sands and clays, The ratings shown in the table have been assigned in a very general way based on current experience and represent a qualitative evaluation of the confidence level, for the derived parameters. The ratings may well vary depending on the amount of experience available in specific, deposits. 5.1 GENERAL FACTORS AFFECTING INTERPRETATION Before analysing any electric CPT data, it is important to realize and account for the potential errors that each element of data may contain, Significant aspects that pertain to cone ‘Table S.1 Perceived applicability of CPT for deriving soil parameters Strength Deformation Flow Initial state parameter parameters characteristics charact yD, yk OR rotated k a 34 estate teats te 23 2 4548 2 24 24 23 Applioblty rating 1 High relay; 2 High to moderate reliability; 3 Moderate ibility, 4 Moderate to ow reliably: 5 Low reliability.46 INTERPRETATION OF CPT/PIEZOCONE DATA penetrometer designs are discussed, ‘The reader is encour- aged to investigate the details of the particular cone design before performing detailed interpretation of the data. A checklist for this purpose is included in Chapter 3 as Table 32. One should also be aware of the significant factors regarding soil conditions and how these influence the meas- ured cone data and thus the interpretation. This is discussed in the following sections, 5.1.1. Equipment design Section 2.1 outlined the significant factors regarding cone design that influence the measured parameters and therefore ‘the subsequent interpretation, As discussed in Chapters 2 and 3 the three major areas of cone design that influence interpretation are 1. Unequal area effects. 2. Piezometer location, size and saturation, 3. Accuracy of measurements. It is strongly recommended that cone penetrometers. be calibrated and the results corrected (as described in Chapter 3). The ertors associated with equipment design are usually most significant for penetration in soft, normally consoti- dated, fine-grained soils. Test results in sand are little influenced by the above factors, except possibly variations in friction sleeve stress, f, 6.1.2 Insitu stresses Theoretical models and chamber test studies have shown that the in situ horizontal effective stress, jy, has a dominant effect on the cone resistance, and the friction sleeve stress. Therefore, the stress (geologic) history of the deposit is of great importance in CPT interpretation, Unfor- tunately, there is often only qualitative data conceming ‘geological history and the techniques for measuring in situ horizontal stresses are still not very reliable, especially for sands (some potential additional techniques are discussed in Chapter 7), ‘An excavation will reduce the horizontal stress in adjacent soils. Even an open borehole, if closer than 10-20 hole diameters may reduce the horizontal stress, depending on the soil conditions. Both static and vibratory compaction or the installation of piles can change the horizontal stress. Applied surface loads (such as fiom surface fill o CPT equipment) can also inerease the effective stress, The inter pretation of CPT data should, at least qualitatively, account for such effects that may influence the horizontal stress. Subsequent sections will show that the relative density fand shear strength correlations for sand are significantly influenced by changes in horizontal stresses. 5.1.3 Compressibility, cementation and particle size The compressibility of soils can significantly influence 9, and f. Highly compressible sands tend to have low cone resistance and in some cases, high friction ratio values, Some carbonate sands have fiction ratios as high as 3%, ‘whereas typical incompressible quartz sands have friction ratios of about 0.5%. The compressibility of sand during cone penetration is also influenced by grain crushing. Subsequent sections will show that variations of sand compressibility have a significant influence on correlations with relative density but a somewhat smaller influence on correlations with friction angle. Cementation between particles reduces compressibility and thereby increases the cone resistance, Cementation is always a possibility in situ and is more likely in older soil deposits (Sehmertmann, 1991). When the particle size of a soil penetrated becomes « significant fraction of the cone diameter, then the cone resistance ean increase abruptly because of the decreased compressibility of the soil due to the need for the cone to displace these particles as rigid units, This effect tends to produce sharp peaks in the cone resistance profile when encountering gravel-sized particles Interseeting very large particles can abruptly stop penetra- tion of eause a sudden deflection, Penetration through grav- clly soils often produces a distinct sound up the cone rods. Cone penetration is often not possible through dense coarse gravel deposits without damaging the load cells or other elements of the cone penetrometer, 5.1.4 Stratigraphy Even ifthe CPT measures the correct mechanical character istics in uniformly soft or strong materials, the transition from one layer to another will not necessarily be registered as a sharp change. Cavity expansion and strain path theories as well as laboratory studies (Schmertmann, 1978; Tread- well, 1976) show that the cone resistance is influenced by the material ahead and behind the penetrating cone. Hence the cone will stat to sense a change in material type before it reaches the new material and will continue to sense @ ‘material even when it has entered a new material. Therefore, the CPT will not always measure the correct mechanical properties in thinly interbedded materials, ‘The distance over which the cone senses an interface inereases with material stiffness. In soft materials the diam- eter ofthe sphere of influence can be as small as two or three cone diameters, whereas in stiff materials the sphere of influence can be up to 10 or 20 cone diameters. Hence, the cone resistance can respond fully (that is, reach full value ‘within the layer) in thin soft layers better than in thin stiff layers. Soft layers thinner than 100 mm can be fally detected by the cone resistance, whereas stiff layers may need to be &s thick as 750:mm or more forthe cone resistance to reach its full value. The CPT will detect thin stiff layers but theGENERAL FACTORS AFFECTING INTERPRETATION a7 strength of stiff layers could be underestimated ifthe layer is jess than about 750 mm, It is possible to detect the presence of soft layers as thin as 75 to 100mm using the cone resistance. Therefore, care should be taken when inter- protng cone resistance in a thin sand layer located within a soft clay deposit. Based on a simplified elastic solution, Vreugdenhil et al. (1994) provided some insight as to how to correct cone data for thin layers. They showed thatthe error inthe measured cone resistance within thin tiff layers is a function of the thickness of the layer as well as the stiffness of the layer relative to that ofthe surrounding soil. The relative stiffness of the layers is reflected by the change in cone resistance fiom the soft surrounding soil (g.,) to that in the stiff soil layer (q.2). Based on this work, a corrected cone resistance (ge) is derived as der = Ke" Gea G1 where: K,=a correction factor for the cone resistance as function of the layer thickness as shown in Figure 5.1 ‘The corrections of Figure 5.1 appear to have a reasonable trend, but are rather large. Therefore, a more conservative correction is recommended (corresponding t0 q2/qei = 2) ay ? episiotor 3 10.0n' cone 8.5 & 8 s ‘Rlecommensde¢ ae § _ froin ° 7000 2000 "3000 Layer thickness H (mm) b) igure S.1 Suggested correction to CPT cone resistance in thin sand layers (based on Vreugdenhil eta, 1994). and is shown in Figure 5.1 and given by the following expression: a 2 1.45) +10 (5-14) 62) where: galga = 2 H = layer thickness, in mm ea = come resistance in the layer ei = cone resistance in the soil surrounding the layer: Thin sand layers embedded in soft clay deposits ae often incorrectly classified as silty sands based on CPT sil behaviour type charts. Hence, slightly improved classifica- tion can be achieved ifthe cone resistance is first corrected for layer thickness before applying the classification chats ‘There is further discussion of layering effects in section 522. Since the sleeve frietion measures an average value over a distance of 13 om (For 10 om? cone) it will tend to smooth out any effects of thin layers. The effect of layering on the recorded pore pressure will depend on the filter location. Filters on the cone will tend to reflect the changes from one layer to another more sharply than filter positions behind the ‘This effect of layering can also cause scale effects when using cones of a larger diameter (that is, a 15.em® cone penetrometer) The natural variability of many sand deposits produces qe profiles with many sharp peaks and troughs, A comparison of CPT data in sands from 10 ean* and 15 em? cone penetrometers shows that the 15cm? data will not reproduce the stiff peaks but will reproduce the softer troughs, Since the relative layor thickness for full response of cone resistance is smaller for softer layers, the average cone resistance profile tends to be slightly lower for the 15 em? cone in sands, The 10 em” cone may detet thinner layers than 15 em? cones. Generally speaking, however, in moderately uniform soil the results of a 15 em? cone are essentially the same as those forthe standard 10 em’ cone. Tn some cases it may be advantageous to use cone penetfometers smaller than the standard 10 m. Cones of 5 om” and even smaller have been used in special projects and research studies (see section 5.8 for futher details). 5.1.5 Rate of penetration Rate effects can be caused to some extent by cteep and pparticle rushing, In general, however, the pore pressure effects predominate and are of most interest, especially, ‘when using the piezocone in fine-grained soil. Normally a tenfold increase in rate causes 10-20% increase in cone resistance in stiff clays and 510% in soft clays (see, for example, Powell and Quarterman, 1988). ‘The recommended constant rate of penetration for an electric CPT sounding is 20 mm/s. The ISSMFE.48 INTERPRETATION OF CPT/PIEZOCONE DATA. Reference Test Procedure allows a penetration rate of 20 mm/s + $ mm/s. Traditionally cone penetration in sands has been considered to be drained and penetration in clays ‘undrained. However, for intermediate sols such as silts and clayey silts, the drainage condition during penetration is not well defined. The drainage condition can be approximated from the soil classification charts or by measuring the dissipation of excess pore pressure in a CPTU test. The effect of rate of penetration is discussed further in section 583. 5.1.6 Pore pressure element location ‘Most ofthe empirical or theoretical corelations discussed in subsequent sections are based on pore pressures measured immediately behind the cone and in front of the friction sleeve (v3). Much less has been published conceming other locations for recording cone penetration pore pressures (Roy et al, 1982b; Smits, 1982; Campanella et al., 1982; Batta- ‘lio etal, 1986). Recommendations concerning the location of the piezometer element have generally been based on considerations of either equipment and procedures of inter- pretation methods. On the basis of a reviow of cxisting experience, the following comments can be made about pore pressure measurements during cone penetration, In terms of equipment design and test procedures there has been a trend towards placing the pore pressure element ‘bchind the cone, normally in front of the friction sleeve (ua). This location has the advantage of good protection from damage due to abrasion and smearing and generally easier saturation procedures, The location behind the cone is also the required location to correct the measured cone resistance (q.) to total resistance (q) to account for the unequal area effects (section 3.1.1). In terms of intexpretation itis generally agreed that pore pressures measured on the face of the cone produce the ‘maximum values and provide excellent stratigraphic detail, provided problems with filter element compression, load transfer, abrasion and smearing can be removed, Interpretation of cone penetration pore pressures is espe cially beneficial in fine-grained soils in which penetration is essentially undrained and is directed towards the evaluation of undrained shear strength (s,), and stress history. To identify the preferred measurement parameter (q- 0 u) to be used for interpretation, it is necessary to distinguish between, soft, uncemented fine-grained soils and stiff, fine-grained soils with high OCR. In most cases, however, a combination of qe (org) and u will be best. Figure 5.2 presents a summary of the main differences in measurement parameters between these two extreme ranges of fine-grained soils. An example is, shown in Figure 5.3. The profile at Bothkennar, which is a soft clay site, illustrates thatthe pore pressure (u1) is close to the measured q. for the cone used (see section 3). The profile at Brent Cross, whichis a very stffelay site, illustrates that ue cean be negative and qe lage. Preferred measurements for correlations using CPTU Siifhigh GOA] «— Solltype —+ [Softiow OOR Lalge Cone resistance Stra (0 Large —— Face pore pressure — Large (iter compression?) (u,) Small ___ Shaft pore pressure — Largo (negative?) (w) Small __ Correction a, Large (negiigibley 4 Coreclations 4 Ge J+— (6, OCR), ce Figure 5.2 Profercod moasuroments for correlations using CPTU, (Robertson 1990), For cone penetration in soft, uncemented fine-grained soils the measured qe is generally small, whereas, the pore prossures on the face (u;) and behind the cone (u2) arv both relatively large compared with the qe. Generally, for cone penetration in so, low OCR, soils, the pore pressure 1 is typically about 80% of the pore pressure a. Both locations (u and 1) provide large pore pressures with good strati- ‘graphic detail. The pore pressures further up the shaft behind the friction sleeve (u:) tend to be somewhat smaller and provide less stratigraphic detail. Because q. is gencrally small in soft, low OCR, fine-grained soils and the pore pressures are of similar magnitude, the correction to qe is ‘generally significant (for instance, Figures 3.32 and 6). Honce, itis generally important to record the pore pressure just behind the cone (v2) so thatthe correct pore pressure can bbe applied to obtain q,. Because of a generally decreased accuracy in recording the small q. values and the need to make significant corrections to get q, the preferred measure~ ment for use in interpretation in soft sols is the penetration pore pressure (ether uot wn). Because of equipment and procedural consideration (saturation), the preferred location for the pore pressure measurement is just behind the cone (that is, 12) For cone penetration into stiff, high OCR, fine-grained soils, the measured q. is generally large. The pore pressure measured on the cone (1) is generally large, but problems with filter compression can be encountered and pore pres- sures may be unreliable (Battaglio et al. 1986). However, the pore pressures measured behind the eone (4) are often small and can sometimes be less than the equilibrium pore pres- sure (Figure 5.3) and in some cases cavitation can occur (see section 5.7.1), This can sometimes result in loss of saturation in stiff clays or dense silty sands. An exception to this can ‘occur in cemented and (or) sensitive stiff clays where large eeGENERAL FACTORS AFFECTING INTERPRETATION 49 tn pore pressures can be recorded due to the collapse of the soil structure during cone penetration. Because the q values are generally large and the uz pore pressures are generally small, the correction for q, is often small and negligible. Hence, the preferred measurement for use in interpretation in stiff, high OCR, fine-grained soils is the measured cone resistance, ge. During a stop in the penetration, any excess pore pres- sures start to dissipate and the rate of dissipation can be interpreted to evaluate consolidation characteristics of the surrounding soil. In soft, low OCR, soils the pore pressure dissipation data are generally good for pore pressure element Focations both on the cone (u) and behind the cone (1). However, in stiff, high OCR soils the dissipation behind the cone ean suffer from local equalization with the higher pore pressures on the face of the cone and interpretation of dissipation data can be diffcut. From the above observations it is clear that there is no single location for pore pressure measurements that meets all requirements forall sil types, Hence, the preference is to record pore pressures at two or more locations simultane- Friction Sleeve, (MPa) o__02 04 06" 08 10 Pressure, u(MPa) ‘Pore Pressure, u(iPa) Cone Resistance, (MPa) 7 2 3 S 4 J “4 pt afin Lo 16 18 ously (to give uw, Hay and so on), Cone penetrometers presently exist that can record pore pressures at two or more locations but saturation procedures may be complex. To avoid increased complexities with equipment and saturation procedures it is recommended to have flexibility in cone design so that pore pressures can be measured either on the cone of just behind the cone. Many cone designs already ‘exist that enable the filter location to be easily changed in the field For general piezocone testing itis recommended to meas- ure the pore pressure just behind the cone (u:) for the following reasons: ¢g00d protection from damage easy saturation generally good stratigraphic detail generally good dissipation data «# correct location to determine gy ‘The IRTP recommended by ISSMFE states that unis the preferred location. However, if a stiff, high OCR, fino- trained soil is encountered and the measured pore pressures Friction Sleeve, fs(MPa) 0002" “004 0,08" 0.08 _ 0.1 ore Pressure, uiMPa) oda 0S 08 10 Gone Resistance, a,(MPa) OZ" 04.06 08. 10 2 “4 AD} f rh { T} 1} { = ole al? alt § 3 3 16 a RK 20 Figure 53 Measured CPTU parameters for atypical: (a) Stiff 0.C. clay ftom Brent Cross, London; (b) Soft N.C. clay fiom Bothkennar, ‘UK (atter Powell eta, 1988),50 INTERPRETATION OF CPT/PIEZOCONE DATA behind the cone (us) become very small, it may then be beneficial to change the location in the field and record pore pressures on the cone (w:). For quantitative interpretation of the pore pressures measured on the cone during penetration in stiff soils itis important to avoid, minimize or be aware of, potential errors due to filter element compression. If, based on previous experience, sti, fine-grained soils are expected, it may be advisable to measure pore pressures oon the cone (1). 5.2 SOIL STRATIGRAPHY Figure 54 illustrates the excellent profiling capability of the piezocone. ‘The continuous monitoring of pore pressure during cone penetration can improve the identification of soil stratig- raphy. The pore pressure develops in response to the soil type being penetrated in the immediate area of the pore pressure sensing element, For a pore pressure sensing ole- ‘ment behind the cone (us): ‘¢ soft to medium stiff clays can give very high pore pressures @ very stiff overconsolidated clays can give very low or negative pore pressures fe very dense fine or silty sands can give very low or negative pore pressures « contractive silts can give high positive pore pressures ¢ dilative silts can also give low or negative pore pressures ‘The drainage characteristics of thin layers also become very important for identification of thin layers of sand, silt or clay ina thicker body of clay or sand (section 5.1.4). ‘The response time for a fully saturated piezocone is usually sufficiently fast observe pore pressure changes for very thin layers (<5 mm). Whether or not such thin layers are observed in practice depends on the response of the soil to the advancing cone, the depth interval of duta recording and the location of the filter element. “The ability of the cone resistance and pote pressure to respond to changes in the material type is not restricted to Sleeve Pore water pressure _—friction Cone resistance Friction Soil (MPa) (MPa) (MPa) ratio profile « . — a) Ag iesal,stesed oetuast 0.200) (ee 0PM awn gente 1ono0ueraigle 4120: 10 18 Depth (m) 20 30 Prevdiilled “(sand Figure 5:4 Example CPTU results showing excellent profiling capability (after Zuidberg etal 1982).SOIL CLASSIFICATION st the coarse or obvious stratigeaphy changes of say soft and stiff layers; itcan also be used to detect and map more subtle changes within a deposit. Powell and Quarterman (1995) show how, with a litle care in calibration, facies variations down a soft clay profile can be detected with confidence by the piezocone (Figure 5.5), and each facies can be assigned 4 piezocone signature which is then used to map the varia- tions across the site ‘Ata slightly larger scale Powell and Quaterman (1995) showed that the CPTU gave potentially far better mapping of the lithological units in a deposit than could be obtained by visual descriptions of sample profiles (see Figure 5.5b). ‘Variations in CPTU signatures have also been used to identify thin failure surfaces within a potentially unstable clay slope. 5.3. SOIL CLASSIFICATION Some of the most comprehensive work on soil classification using electric CPT data was presented by Douglas and Olsen (1981) (Figure 5.6). The correlation was based on extensive data collected from areas in westem USA. The chart con- firms early observations from Holland that sandy soils tend to produce high cone resistance and low fiction ratio, ‘whereas soft clay soils tend to produce low cone resistance and high friction ratio (Figure 1.6). Organic soils such as peat tend to have very low cone resistance and very high friction ratio. Sensitive soils, on the other hand, tend to have low cone resistance and low friction ratio. Soils with high horizontal stresses (that is, high OCR) tend to have higher cone resistance and ftition ratio. One important distinction made by Douglas and Olsen (1981) was that CPT classification charts cannot be expected to provide accurate predictions of soil type based on grain size distribution but provide a guide to soll behaviour type The CPT data provide a repeatable index of the aggregate behaviour of the in situ soil in the immediate area of the probe. In recent years, soil classification charts have been adap- ted and improved based on an expanded database (Robert- son ef al., 1986; Olsen and Farr, 1986). Also, research has illustrated the importance of cone design and the effect that ‘water pressures have on the measured cone resistance and sleeve friction due to unequal end areas (Campanella et a, 1982; Baligh et a., 1981) (Section 3.1.1). Thus, cone pen- ctrometers of different designs, but conforming to the IRTP (ISSMPE, 1989) (Appendix A) can give different values of Ge and f,, especially in soft clays and silts. Correction for pore pressure effects on cone resistance normally eliminates significant differences, when expressed in terms of q,, from ‘one cone design to another. Recent studies have shown that even with careful proce- dures and corrections for pore pressute effects, the measure- ment of sleeve friction (f) is often less accurate and less teliable than the cone resistance (Lunne ef al., 1986; Gille- spie, 1990). Cones of different designs will often produce different friction sleeve measurements, This can be caused by small variations in mechanical and electrical dé features, as well as small variations in tolerances (for ‘example, an oversize cone), To overcome problems associated with sleeve friction ‘measurements, several classification charts have been pro- posed based on g, and pore pressures (Jones and Rust, 1982; Baligh ef al, 1980; Senneset and Janbu, 1985). The chart by Senneset and Janbu (1985) uses the pore pressure parameter ratio, By, defined as: 53) where: tus = pore pressure measured between the cone and the friction sleeve up = equilibrium pore pressure im = total overburden stress cone resistance corrected for unequal end area effect ‘The original chart by Senneset and Janbu (1985) used qe However, it is generally agreed thatthe chart and By should use the corrected total cone resistance, 4 Experience has shown that, although the sleeve friction measurements are not as accurate as q;, and w, generally more reliable soil classification can be made using all three pieces of data (that is, ge, 1). A first attempt at defining a system that uses all three pieces of data was proposed by Robertson etal. (1986) and used q,, By, and Ry Figure 5.7) ‘A problem that has been recognized for some time with soil classification charts that use q, and Ris that soils ean change in their apparent classification as cone penetration resistance increases with increasing depth. This is due to the fact that q,, f, and u all tend to increase with increasing ‘overburden stress. For example, in a thick deposit of nor ‘mally consolidated clay the cone resistance (g.) will increase linearly with depth resulting in an apparent change in CPT classification for large changes in depth. Early classification charts were based predominantly on data obtained from CPT profiles extending to a depth of less than 30 m. Therefore, for CPT data obtained at significantly greater depths some crror can be expected using early CPT classification charts that are based on g,, (or qe) and Ry ‘Attempts have been made to account for the influence of| ‘overburden stress by normalizing the cone data (Olsen, 1984, 1995; Douglas ef al. 1985; Olsen and Farr, 1986; Olsen and Mitchell, 1995). These approaches require differ- ent normalization methods for different soil types, which produce a somewhat complex iterative interpretation proce- dure that requires a computer program, Conceptually, any normalization to account for increasing stress should also account for changes in horizontal stresses,82 INTERPRETATION OF CPT/PIEZOCONE DATA Pore pressure U, (MPa) Cone resistance q, (MPa) 0 02 04 08 08 1 d yf at 514} | Rectamed taal tat 4 i ial Beds | (Upper division) | 4 ae ) 4g ce4 | i 4 fees] Claret beds } oe va (middie division) a Feces E a] Weathered € 10 [gf mots | s ina F=|camincted | 9 42 Bedded | A \ ' Claret beds + — (Lower division) 16 | | 19 | ] Basal claret beds: Bothkennar gravel Reclaimed tidal flat Claret Beds (Upper Division) CSCSF1 Piezocones HwWye D4 Boreholes 2-2 0D —- Ism tom | “a | bol i i Xxx x Claret Beds (Lr Division fen a / Basal Clarét Beds Top of Bothkennar gravel Piezocone proposal -—-- on o 0m Paul ot al (1992) -x—x-&-2--9 404 Figure $5 Example of CPTU in a soft clay, (a) showing ability to detect facies changes (6) showing better definition of lithological units (after Powell and Quatermann,1995).SOIL CLASSIFICATION 109) 60 40 20} Cone resistance (MPa) o 1. 2 3 4 5 6 Friction ratio (%) Figure 5.6 CPT soil behaviour type classification chart by Douglas and Olsen (1981). 53 since penetration resistance is influenced by the horizontal effective stresses Gamiolkowski and Robertson, 1988). However, at present, this has little practical benefit without a prior detailed knowledge of the in situ horizontal stresses Even normalization using only vertical effective stress requires some input of soil unit weights and ground water conditions. Wroth (1984,1988) suggested that CPT data should be normalized using the following parameters: ‘Normalized cone resistance, f 64) where of, = effective vertical stress, dio — Mo Normalized friction ratio, oe (55) Fm (if possible f; should be used) Pore pressure ratio, au By 66) ate where Au = 12 — Up. Based on these normalized parameters and using the extensive CPTU data base now available in published 100 100 & & . = = 10 = 10 8 8 5 A 3 3 gy gy 2 2 8 8 on o 02 0 02 04 06 08 10 12 14 0 1 2 3 4 5 6 7 8 Pore pressure parameter B, Friction ratio (%) Zone: Soll Behaviour Type: 1. Senetivefine grained =. Clayoy sit toslly clay 8. Sand 2 Organ material 8 Sandy alto clayey et 10. Gravely sand o sand a Cay 7. Sitysandto sandy sit 41. Veryst fine gained 4 Sit olayto cay 8 Sandto sity sand 42, Sando cayey sanct * Overconsolidated or cemented. Figure 5.7 Proposed soil behaviour typo classification system from CPTU date (after Robertson et a, 1986).54 INTERPRETATION OF CPT/PIEZOCONE DATA, and unpublished sources, modified soil behaviour type Classification charts have been proposed by Robertson (1990) and are shown in Figure 5.8. The linear normal- ization of cone resistance (2. is best suited to clay soil, but is tess suited to sand The two charts shown in Figure 5.8 representa throe- dimensional classification system tat incorporates al three pieces of CPTU data. For basic CPT data where only qe and {face available the let hand char (Figure 5.8) can be used “The error in using uncorrected ¢, dala will generally only influence the daia in the lower part of the chart where normalized one resistance is les than about 10 This part of the chats for sof, fine-grained soils where q ean be sma and penetration pore pressure () can be ltge Included on the notmalized soil behaviour type classifi tion chart is zone that represents approximately normally consolidated soil behaviout. A guide is also provided to indicate the variation of norinalized CPT and CPTU data for changes in: overconsolidaton rato (OCR), age and sensitiv- ity (5) for fine-grained oils, where cone penetration is generally undrined, and OCR, age, cementation and friction angle ($’) for cohesionless soils, where cone Penetration is generally drained. Generally, soils that fall in zones 6 and 7 represent approximately drained penetration, whereas, soils in zones, 1, 2,3 and 4 represent approximately undrained penetration, Soils in zones 5, 8 and 9 may tepresent partially drained penetration. An advantage of pore pressure measurements, uring cone penetration is the ability to evaluate drainage conditions more directly Robertson (1990) suggested that the charts in Figure 5.8 are still global in nature and should be used as a guide to define soil behaviour type based on CPT and CPTU data Factors such as changes in stress history, in situ stresses, sensitivity, stiffness, macrofabric, mineralogy and void ratio, will also influence the classification. Occasionally, soils will fall within different zones on each chart; in these cases judgement is required to correctly classify the soil behaviour type. Often the rate and manner in which the excess pore pressures dissipate during a pause in tho cone penetration will significantly aid in the classifica- tion. For example, a soil may have the following CPTU 1000, T TT TTT 48 [5] ot Senshi? 040 04081 I teu, ate Zone Soil behaviour type ‘Soil behaviour type Zone Soil behaviour type 1. Sensitive, fine grained; 2, Organic 1 ‘Mayne (1991) compared predicted with measured profiles of OCR for many different sites with many different soils with where generally good results Figure 5.17 presents data from Mayne (1993) ilhustrating M = slope ofthe critical state line the comparison between measured and estimated OCR using ; the above correlation. Different values ofthe effective stress S sin (512), slow aa 4 hve Ben une to provide the bet 34 16 30 fal Z 26 8 12| © 2 Mo f 218 ge 2 S é a Te Ze a 6 4 a 5 1020 7y 5 10 20 | 14 16; | | 14 o — oa 1 5 1 20 “fF 5 10 20 | Overconsolidation ratio, OCR | Legend: 4 Troll © Haga X Cowden | © Brage ® Flo Y Brent Cross Haltenbankon © Vancouver = Onsay 5 Emmerstad © Drammen lean clay ‘© Drammen plastic clay Figure 8.14 Empirical corclations of OCR vs normalized CPTU parameters (after Lunne eta, 1989) bal SaeRnn tnt eRe 111 ec OUNMMneE 111 CC UUMEUNOE ttt .Eee GUeUien ttt .cee GUNGeee ents cee cece60 INTERPRETATION OF CPT/PIEZOCONE DATA. a) a: , L A & 8 : Wf Me 3s Us 5 uy 8 3 2 OCR 0.88 + 1.43 (PPD) B é © Robertson et al. (1986) © Loven gh 80 “ryote 8 Assully 1986) a Pore pressure difference PPD= (u,-u,)/u, b) sof London clay “ 7 Brom cromy yO oS Natiation due to “krone woh / § g B sop § 5 Taranto ctey 2 2 fram [impo Vey 3 wb Estimated OCR : & é ok ’ oRe0 68 +148 (PPD) 5 02 4 6 8 10 12 14 16 18 20 Pore pressure difference PPD= (U,-U,)/U, Figure 5.18 Effect of averconsolidation ratio on normalized pore pressure difference (after Sully etal, 1988). In soft, lightly overconsolidated clays the g, values are ‘often small whereas, the uz values can be quite large. Hence, the difference between cone resistance and pore pressure (@.~ m) is small and may not be reliable. Based on the above methods and comments, itis recom- ‘mended to estimate OCR in cohesive soils from CPT/CPTU, data as follows: 1. For deposits where litle experience is available, estimate OCR based on the shape of the normalized cone resist- ance profile (Q)) using (5.8) and shown in Figure 5.13, This relationship can be expressed by the following simple formula 6.13) 12 ocr OCR For6 Sig Su 1: Closely spaced discontinuities < cone Clay “diameter fairly random orientations Z| RBGO A: No macro escontinuties @PN iB. intermediately spaced aiscontinuties Ik: Discontinuities spaced > cone diameter b) B 39} a ¢ 20] ole ac x yu Cow z 2kfer' BRE SEH Vara rent Cross °9 10203040800 Plasticity index I, Figure $.24 Cone factors for fissured clay (after Marsland and Powell, 1988 and Powell and Quarterman, 1988). 3. s, estimation using excess pore pressure Using theoretical and semi-theoretical approaches based on cavity expansion theory, a number of relationships have been proposed between excess pore pressure Au and 5, (Vesic 1972; Battaglio et al,, 1981; Randolph and Wroth, 1979; Massarch and Broms, 1981; Campanella et al., 1985), ‘The relationships have the form of Aw seq (lua uu) (5.19) Based on cavity expansion, Nay is theoretically shown to vary between 2 and 20 Lunne ef al. (1985) found Nay to correlate well with By and to vary between 4 and 10, for North Sea clays taking triaxial compression (CAUC) strength as the reference strength. La Rochelle et a. (1988), using uncorrected eld vane strength as the reference strength, found that for three Canadian clays Ny, varied between 7 and 9, even if OCR ranged between 1.2 and 50. Using s, values from CAUC tests on block samples, Karlsrad e¢ al, (1996) obtained Na, values varying between 6 and 8 with no clear dependency on B, (Figure. 5.25). The above correlations for Nay were derived for normally to lightly overconsolidated clays and should not be extrapo- lated to heavily overconsolidated deposits where B, is small or even negative (Powell etal, 1988). The correlations referred to above relate to the excess pore pressure measured immediately behind the cone (1). ‘The use of the piezocone to correct the cone resistance for pore pressure effects, use of standard cones and reference test procodures and a more critical use of reference values of undrained shear strengths have helped in improving the “ . R 2 Boonnar 2 Enmerieg + Orsay Cone factor Ny, 80 05 08 09 12 Pore pressure ratio B, Figure 5.25 Cone factor Nay vs By (after Karlsrud ef a, 1996),68 INTERPRETATION OF CPT/PIEZOCONE DATA quality of correlation ‘wherever possible Use of site-specific empirical correlations still seems to be ‘the best procedure for interpretation of sy from CPT/CPTU. ‘general, Piezocones should be used Based on the above discussion it is recommended to evaluate s, in cohesive fine-grained soils from CPT/CPTU, data as follows: 1. For doposits where little experience is available, estimate ‘s, using the total cone resistance (q,) and preliminary ccone factor values (Nx) from 15-20. For a more con- servative estimate, select a value close to the upper limit. For normally and lightly overconsolidated clays Ny, can ‘be as low as 10 and in stiff fissured clay can be as high as 30. In very soft clays where there may be some uncer- tainty with the accuracy in q,, estimate s, from the excess pore pressure (Au;) measured behind the cone using Nj, from 7 to 10. For a more conservative estimate, select a value close to the upper value. 2. If previous experience is available in the same deposit, the values suggested above should be adjusted to reflect this experience. 3. For larger projects, where high-quality field and labo- ratory data may be available, site-specific correlations should be developed based on appropriate and reliable values of sy. 5.4.2.2 Sensitivity ‘Tho sensitivity (S,) of clay is defined as the ratio of undis- turbed undrained shear strength to totally remoulded undrained shear strength. The CPT sleeve friction is a function of the remoulded shear strength and Schmertmann, (1978b) proposed that sensitivity can be estimated from fiction ratio (Rj) in percent as: (6.20) where Ny is @ constant. ‘Schmertmann suggested a value of N, = 15 from mechan- ical CPT data, Robertson and Campanella (1988) based on. ‘comparison with sensitivity from piezocone tests and S, from vane tests suggested an average value of N, = 6, Rad and Lunne (1986) found N, to vary between 5 and 10 with 7.5 as the average value (Figure 5.26) It has been recognized for many years that the sleeve friction stress, f,, from an electric cone is approximately equal to the remoulded undrained shear strength. An ‘example is shown in Figure 5.27. Therefore, the sensitivity of a clay can be estimated by calculating the peak s, from either site-specific or general correlations with . or u and the remoulded s, from f, ‘Although the field vane isa better test to measure sensitiv= ity in situ, ff no vane measurements are available, 5; can be roughly estimated using CPT data. An average value of 15 * Onsay site| Haga site fAverage N=7.5 7 1 2 2 Friction ratio Ry= fYq,(%) Figure 5.26 Friction ratio, Ry vs sensitivity, §, (after Rad and ‘Lanne, 1986). N,= 7.5 seems appropriate to use, while N,=6 to 9 will give a likely range, As with other empirical correlations it cannot be expected to find a unique value of N, valid for all clays since it will depend on mineralogy, OCR and other Stress (kPa) o 50 100150200250 I I 1 Remoulded UU triaxial test (6,) IE. CPT sleeve friction (f,) Depth below seafloor (m) 160l Figure 5.27 Comparison of remoulded UU triaxial tests and CPT sleeve frietion (aller Quiros and Young, 1988).INTERPRETATION IN FINE-GRAINED SOILS 69 factors. The N, values recommended above should therefore preferably be calibrated with local experience. Tt has been observed that relatively sonsitive clays have very low sleeve friction values. The inherent difficulty in seasuring low sleeve friction accurately can influence the estimate of sensitivity using CPT data. 5.4.2.8 Effective stress strength parameters ‘An effective stress interpretation method has been devel- ‘oped by Senneset ef al. (1982, 1988), Senneset and Janbu (1985), and Sandven et al. (1988). A bearing capacity formula in terms of effective overburden stress, aj, can be expressed as: 4 ~ Gg = Na(io + 4) (5.21) where 6.22) B= angle of plastification as shown in Figure 5.28, a= attraction bearing capacity factor = tan St phe me ‘Au = excess pore pressure (measured immediately behind the cone) Ny = beating capacity factor 6 tan $' (1 ~tan g') Using this method requires an estimate of the soil attrac- tion, a, which can be estimated from the shape of the q, vs io diagram, from triaxial tests or from general experience, Table 5.6 taken from Senneset et al. (1989) summarizes typical values of soil attraction, a, and tan ¢" for various soil types. ‘The angle of plastification, f, also needs to be assessed. expresses an idealized geometry of the generated failure zones around the advancing cone, As stated by Senneset et al. (1990) the value of f is difficult to assess, both experi- mentally and theoretically. However, Senmeset ef al, argue that f° depends on soil properties like compressibility and stress history, plasticity and sensitivity. Sandven ef al. (1988) presented values of f found from experimental correlations between laboratory determined tan ¢” and interpreted values from CPTU as shown in Table 51. Information on soil type and soil consistency can be obtained from interpretation of CPTU results or from results of laboratory tests, if available. Figure 5.28 includes charts that can be used to estimate ‘Table 5.6 Typical values of soll attraction (a) and fiction (an $") (after Senneset eal, 1989) ‘Shear strength parameters a “ (kPa) tang’ degrees, Ny By Clay, sof 510 035-045 19-24 1-3 08-10 Clay, medium 10-20 040-055 1929 35 06-08 Clay, tit 20-50 0:50-060 27-31 3-8 03-06 Silt, sot 0s 050-060 27-31 Silt medium S15 055-065 2933 5.30 0-04 Silt, sift 15-30 0.60-0.70 31-35 Sand, loose 0 055-065 29.33 Sand, medium 10-20 0.60-0.75 31-37 30-100 <0 Sand, dense 20-50 0.70-0.90 35-42 Hard, stifsoil, >50 0810 3845 100 <0 emented friction angle (tan 4") from the pore pressure parameter By, and the cone resistance number, Ny (eg. equation 5.32). Chars for f from + 15° to ~30° are included. ‘The f-values and bearing capacity factors referred to above are however associated with large degrees of uncer- ‘ainties, Lunne et al, (1985) compared triaxial (CAUC) and CPTU results from North Sea soil investigations and found significant difference in laboratory values of effective stress strength parameters and those determined from CPTU data using Senneset ef al. charts AAs shown in Figure 5.29, Powell etal. (1988) found that for overconsolidated London (Brent) clay it was necessary tose f = ~30° to obtain a good correlation between tan ¢ determined from piezocone tests and tan ¢" determined from triaxial tests. Similarly in soft Bothkennar (Grangemouth) clay = 0 gave the bos fi. Keaveny and Mitchell (1986) reported good results for silts and overconsolidated clay, but rather poor results for normally consolidated clays. Sandven (1990) found that both friction and attraction were overestimated using the Keaveny and Mitchell approach. Conceptually itis difficult to estimate the effective stress (Grained) strength parameter (c',¢) from a complex ‘Table 5.7 Tentative values ofthe angle of plastifcation f in various soil types (afler Sundven et a, 1988) Soil type ‘Tentative f values Dense sands, overconsolidated silts, high- 20° to —10°| plasticity clays, low-compressible overconsolidaied clays Mealium sands and silts, sensitive clays, high- —S® to -+5° compressible clays Looe silts, clayey silts +10°to +20"70 INTERPRETATION OF CPT/PIEZOCONE DATA 1000 1000 Medium soll Medium sits Lightly OC sits poo a 4 Loose soil io Loose, fine silts Normally consolidated Bovis? 400] Pore pressure ratio B, 10 10| Cone resistance number N, Pore pressure ratio B, Cone resistance number N 0 02 04 06 08 10 02 04 06 08 10 ' Friction tan ° Friction tan ¢ 1000 +1006 Denso soil Dense, fine sand Sitly sand Bae 5° | Hard, sti soit Very dense sand | cemented, ‘overconsolidated B= 30° 4 Cone resistance number N, Pore pressure ratio B, Pore pressure ratio B. Cone resistance number N 02 04 06 08 10 (eo aFy We 0 Friction tan 4° Friction tan 9° Figure 5.28 Effective stress bearing capacity factors (after Senneset eta, 1988) 7 eoINTERPRETATION IN FINE-GRAINED SOILS m1 Tan ¢" Tan ¢ 02 04 06 08 10 40 02 04 06 08 10 a8 Range of E laboratory aa values eT | ope \ 30 kPa 12 | | 4 | 8 16 8, 18 (a) Brent Range of laboratory 10) 12 14] 16| 18 (b) Bothkennar Figure 5.29 Predicted and measured effective sres-strength parameters from Brent Cross and Bothkeanar clays (from Powell etal, 1988). uundrained penetration test like the CPT. Although theoret- ieal solutions have been proposed in an attempt to correlate CPT parameters with c’ and ¢, these methods are subject to serious problems, Any solution must make major assump- tions as to the distribution of total stresses and pore pressures around an advancing cone. Unfortunately, the distribution of stress and pore pressures around a cone is extremely com- plex and has not been adequately modelled or measured, except perhaps for a sof, normally consolidated clay. "The methods should therefore be viewed as highly empiri- cal and approximate. Wherever possible, laboratory or local experience should be used to update the approaches. For major projects where estimates of effective stress part- meters are important, itis advisable to obtain high-quality undisturbed samples to measure c’ and ¢ and to develop site- specific CPT correlations, 5.4.3 Deformation characteristics Deformation characteristics are generally inthe form of soil ‘moduli, Soil moduli are generally a function of the stress history, stress and strain level, drainage condition and the stress path direction, Any estimate of soil moduli is compli- cated by these factors. In practice the more commonly used tmodiuli for fine-grained soils are the one dimensional con- strained modulus, M, undrained Young's modulus, in com- pression loading, and small-strain shear modulus, G, ‘There are two approaches for estimating moduli from CPTICPTU data: 1. Indirect methods that require an estimate of another parameter such as undrained shear strength s,. 2. Direct methods that relate cone resistance directly to modulus, The following sections describe existing correlations to estimate soil moduli from CPT/CPTU data, 5.4.3.4 ‘The 1-D constrained modulus, M, as measured in an oed- ‘ometer test, has been expressed in terms ofa coefficient, &, and cone resistance: Constrained Modulus (6.23) where: be Ce my 6% Compression Index Mitchell and Gardner (1975) made a comprehensive review of the numerous correlations between cone resistance and constrained modulus. Sanglerat (1972) developed a comprehensive array of dy values for different soil types ‘with different cone resistance values. Table 5.8 shows a2 INTERPRETATION OF CPT/PIEZOCONE DATA. Table 5.8 Estimation of constrained modulus, M, for clays (Adapted from Sanglerat, 1972) (after Mitchell and Garciner, 1975) Clay of low plasticity (CL) Ge<0.7MPA 3<0q<8 07 4e52.0MPa 20 <5 G2 20MPE— 1<0y<25 q.22MPA3<0q<6 _ Sills of low plasticity (ML) Ge<2MPa 1 Yorktown Haga “dMadingley * (sensitive) 400 |.© Brent Cross Y 200 Constrained modulus ( M/p,) 10 20 40 , Gyo Net cone resistance Figure 534 General relationship between constrained modulus and net cone resistance (from Kulhawy and Mayne, 1990) INTERPRETATION OF CPT/PIEZOCONE DATA. 6.4.3.3 Small strain shear modulus The shear modulus is largest at very low strains and decrea- ses with increasing shear strain, It has generally been found thatthe initial maximum shear modulus is constant for shear strains less than about 10°26, ‘This initial, small strain modulus is often denoted G,, Mayne and Rix (1993) showed that the small strain shear modulus varied with void ratio (e) and cone penetration resistance (q,) for a wide range of clays and can be expressed an = 99.5 0) er (6.30) where: a= atmospheric reference stress in the same units as G, and 4, The strong dependence of G, upon void ratio (e) requires that CPT g. is only successful as a profiler of G, if comparison profiles of ¢, are known. This is not usually the case. However, elastic theory relates the maximum shear modulus, G, , 0 soil density, p, and shear wave velocity, F,, as follows: Gap Ve 631) where: p= mass density of the soil = y/g ‘and this supports the recommendation of making direct measurements of in situ shear wave velocity using the seismic CPT (see section 7.4). Based on these observations, Robertson ef al. (1995) suggested a chart to identify soil type using seismic CPT data, as shown in Figure 5.10. This chart can also be used to estimate G, based on an estimate of soil type from the basic, CPT soil classification charts. However, care must always be taken when using any of these charts or cortelations as it should be remembered that, G is not independent of the direction of shear (Powell and Butcher, 1991), Butcher and Powell (1995a) showed that the shear wave velocity in clays, and therefore the G, value deduced, was dependent on the stresses in the directions of propagation and polarization of the shear waves and can vary by up to 300% in heavily overconsolidated clays. 5.4.4. Flow and cons lation characteristics Flow and consolidation characteristies of soil are normally expressed in terms of the coefficient of consolidation, ¢, and hhydraulie conductivity or permeability, k ‘They are inter- linked through the formula 632)- INTERPRETATION IN FINE-GRAINED SOILS 75 where Mis the constrained modulus relevant to the problem modelled (that is, unloading, reloading, virgin loading). "The parameters c and k vary over many orders of magni- tude and are some of the most difficult parameters to measure in geotechnical engineering. It is often considered that an accuracy within one order of magnitude is accept- able. Nevertheless, ¢ and & are parameters that are often essential input in some geotechnical calculations, Due to soil anisotropy both c and k have different values in the horizontal (cy, ky) and vertical (¢,, k) direction. The relevant design values depend on drainage and loading direction, 5.4.44 Rate of consolidation parameters may be assessed from the piezocone test by measuring the dissipation or decay of pore pressure with time after a stop in penetration, Coefficient of consolidation 2000 No. Deseription 841 O49 1, Portsmouth 1000 CL Clay |, = 15 020 Bole UL =35% 600 2. Boston CL clay F LL=41%1,=22% 0.20 400 8. Bangkok CH clay F LE 65% |,=41% 0.27 ~ 200) 4. Maine CH OH clay ui LE= 65% |,=98% 0.29 5. AGSCH clay ua LL=71%1,=40% 0.26 20] eo] 6. Atchafalaya CH clay LL=95%1,=75% 0.24 _ 7. Taylor River L Peal w = 500% 20 (CK.U Simple shear tests 02 0408 08 a) APPlied shear stress ratio ,/s, Al os normaly consoidatod 1000 1000, 800 800 @ 600 7 600 uP 400 400 | 200 200-2 7 2 4 6810 co 2 4 6810 OCR= 0/09 OCR= op /O've ») Figure 5.35 stiffness ratio, a function of (adapted from Ladd etal, 1977)76 INTERPRETATION OF CPT/PIEZOCONE DATA 1600 533) 1400] where 11, = the pore pressure at time ¢ 1200) 11; = initial pore prossure at t= 0 uu. = in situ pore pressure before penetration +1000] . ‘The results of Figure 5.37a are replotted in normlized & form in Figure 5.37b. B00) Over the last 10 to 15 years, theoretical and semi- ur empirical solutions have been developed for deriving the 600 oon ofcomoldaton fom pore peste dissipation Table 5.9 presents an overview of the main solutions 400 TOO} available to calculate the coefficient of consolidation from ppiezocone dissipation data, eH Ot 50 Torstensson (1975, 1977) developed an interpretation es ee model based on cavity expansion theories. Initial pore i pressures were computed assuming an elasto-plastic soil t ‘model and spherical or eylindrical eavity expansion theory, I 370 ocr Figure 5.36 Stiffness ratio, His,, a8 fumetion of OCR (after Duncan and Buchignani, 1976). Figure 5.37a shows typical dissipation curves for a soft clay (Bothennar clay) plotted on a logarithmic time scale. ‘The results vary with the filter position. For interpretation it is best to normalize the pore pressure relative to the initial, pore pressure at the beginning of dissipation, uj, and the ‘equilibrium in situ pore pressure 1,. The normalized excess pore pressure, U, at time fis thus expressed as 0.80 as shown in Table 5.9, Torstensson then used linear uncou- pled one-dimensional consolidation to compute the dissipa- tion of pore pressures, ‘Torstensson suggested that the coefficient of consolida- tion should be interpreted at 50% dissipation from the following formu a 634) where the time factor Tsp is found from the theoretical solutions, fs is the measured time for 50% dissipation and, enetrometer radius (cylindrical model) or equivalent 8 Pore pressure (MPa) 0.00 aL 1 1 1 oor 01 4 10 100, Log time (min) 1000 Figure 5.378 Dissipation test results from Bothkennar 0.80 0.60 0.20 100 Log time min, Figure $37 Normalized dissipation test dta from Bothkenn.—_—_ INTERPRETATION IN FINE-GRAINED SOILS 7 ‘Fable 5.9 Available solutions to caloulate the coefficient of consolidation from piezocone dissipation tests Initial pore ‘Author Cavity ype Soil mode! pressure, Au, Consolidation Remarks Soverberg (1962) Cylindrical, radius R_Elastoplastic R LD Not in use orstensson (1975, Cylindrical/spherical Elastoplastic 2s, Ine) LD 1977) Au, =4s, InP) Randolph & Wroth Cylindrical lastoplastic As Torstensson’'s 1D iquiv. to (1979) Torstension Battagli eta. Cylindticalispherical Hlastoplastic Same as Torstensson 1D Fits entre curve (1981) Senneset etal Cylindrical lastoplastic Same as Torstensson LD = Equiv. to (1982) ‘Torstensson Lovadoux & Baligh Piezocone model Non-linear ‘From strain path method 2-D Most comprehensive (1985) 1,= 500 work Gupta & Davidson Piezocone model _—_-Elastoplastic Mosifiod cavity Linear (1986) expansion, some axisymmetic dissipation Soarés etal. (1987) Piezoeone model Non-linear Corrected by visual 2D examination and regression analysis Houlsby and Teh Piezocone model Non-linear Predicted by large strain 2-D Extension of Levadoux (1988) T.varies finite element analysis ‘and Baligh (1985) and strain path method 1s pore presnure; 9, —undrained shear strenth; r= adil distances,» radius of plastic zone: J, Gi ™ iit index ‘The selection of the appropriate model depends on the location of the porous element. The spherical solution may ‘be most applicable if the filter element is located somewhere on the conical part, If the filter is located somewhere on the cylindrical shaft some distance away from the cone the cylindrical solution will be the most applicable. ‘A comprehensive study on pore pressure dissipation has been performed by Levadoux and Baligh (1980, 1986) and Baligh and Levadoux (1980, 1986) who proposed an inter~ pretation method after evaluating predictions of dissipation tests in Boston Blue Clay (BBC), Levadoux and Baligh used the strain path method (Baligh, 1985) to predict the initial pore pressure distribution for normally consolidated Boston Blue Clay with rigidity index, 1, = 500. A finite element method was used for the sub- sequent coupled and uncoupled linear isotropic consolida- tion analysis. Some important conclusions from the study of Baligh and Levadoux (1986) were 1, The simple uncoupled solutions provide reasonably accurate predictions ofthe dissipation process. 2. Consolidation is taking place predominantly in the recompression mode for dissipation less than 50%. 3, Initial distribution of excess pore pressures around the probe has a significant infiuence on the dissipation process 4, Dissipation is predominantly in the horizontal direction Houlsby and Teh (1988) proposed an interpretation based on the results of large strain finite element analyses of the penetration pore pressures, and a finite difference analysis of the dissipation pore pressure. They used an approach similar to the Levadoux-Baligh theory but included the effect of varying the rigidity index, f, (=G/s,). Houlsby and Teh suggested using a modified dimensionless time factor, 7* given in Table 5.10, defined as follows: owt rea PvI (5.35) ‘Table 5.10 Modified time factors 7* from consolidation analysis (from Houlsby and Teh, 1988) Location Cylindrical extension FiveradiiTenradit Degree of Cone above cone above -— above consolidation (3) base (us) cone base cone base 20% © 0.014038 0294 ©0378 30% = 0.032 0.078 0503, 0.662 40% 0.063 0.142 0736” 0.995 50% ois 0245 Lit 1458 60% = 0226 0.439 165 2.139 0% = 0463 0.804 243, 3238 80% Lot 60 4.10 52478 INTERPRETATION OF CPT/PIEZOCONE DATA where: ck = coefficient of consolidation in direction perpendicular to cone axis, typically horizontal radius of cone, typically 35.7 mm igidity index, ‘These solutions are based on elastic plastic soil models and hence the rigidity index represents an equivalent clastic, normalized stiffness. Figure 5.38 shows the values of 7 derived by Houlsby and Teh (1988) and compares them with values by Tor- stensson (1977) for element locations immediately behind the cone (u:) and on the face of the cone (1). Its interesting to note that the simplified solutions by Torstensson (1977) provide essentially the same values as the more recent and comprehensive solutions by Houlsby and Teh (1988). Figure 5.39 shows a simplified diagram that can be used to estimate ¢, using the Houlsby and Teh (1988) solution. es Zz = os 3 10 S-¥~Torstensson (197) % Cylindrical 7 . Spherical 1, values: 25 to 600 Au, ge Torstensson (197 0.001 001 Ot 1 10 100 Time factor T* Figure 5.38 Theoretical solution of normalized pore pressure dissipation vs 7 (after Teh and Houlsby, 1991), Robertson ef al. (1992b) reviewed dissipation data from piezocone tests i predict coefficient of consolidation using Houlsby and Teh’s (1988) solutions with reference values from laboratory tests and field observations. The review showed thatthe Teh and Houlsby solution provided reason able estimates of cy. Results were evaluated for pore pres- sure data from different filter locations and the leat seater ‘was obtained with the pore pressure element location imme- diately above the cone (u:). Figure 5.40 shows some of the results presented by Robertson etal. (1992b). Powell and Qutermann (1997) showed that in «soft clay the normalized dissipation curves from different filter posi- tions were very close in shape to those of Teh and Houlsby but displaced relative to each other. Values of o, deduced using Teh and Houlsby were very similar for filter positions tu, and ts but somewhat lower for ty ‘Teh (1987) also proposed the interpretation of the con- solidation data on a root time scale, asthe initial seetion of the plot approximates closely toa straight line the pore pressure dissipation i plotted on a square-root time scale, the gradient of this linear section ism, as shown, cc 1.5 for a 18 cm? cone) 10 10° 0° | ™ L 400 1000 10.000 10% tyo(min) uo funny fee fuse fas 10" 107 108 104 rm (min“*) Figure 5.39 Chart for finding cy ftom ts (after Robertson et al 19928),y INTERPRETATION IN FINE-GRAINED SOILS 79 on Figure 5,39, Then cy, can be evaluated from the equation: c= (lM) Vie“? 536) sadient corresponding to the theoretical curve for a given probe geometry and porous element location m= measured gradient of the initial linear dissipation ( (time) units). Values for M are given in Table 5.11 ‘The square-root time method is useful for short dissipa- Fo Stra tn one) 10° = cy (om’/min) 10% Che 10 100 1000 10000 eo (min) Figure $40 Average laboratory cy values and CPTU results (ater Robertson etal, 1992). Table 5.11. Gradient of dissipation curve (M), root time plot (from ‘Teh, 1987), Cylindtical extension Cone above cone base Five ni Filter position (a) ) above cone base Gradient of 1.63 Lis 0.62, dissipation save (§M) tion tests and/or where initial excess pore pressure (wu) is uncertain, ‘As summarized by several authors (for example, Soares ef al,, 1987; Sandven, 1990; Robertson et al., 1992) the ‘meaning and use of theoretical solutions are complicated by several phenomena including: 1, Estimation of initial pore pressure distribution Experience has shown that the initial pore pressures around ‘penetrating cone vary from the cone and along the shaft as illustrated in Figure 5.41. In stiff, heavily overconsolidated clays there is a very large gradient in pore pressure going from the cone to the shaft — where negative pore pressures, may be observed. Before pore pressure dissipation starts a Jocal redistribution occurs which may result in an initial inerease in pore pressure behind the cone before radial dissipation staris (see Figure 9.26) Comparison of theorctical analysis with laboratory and field test results shows that to improve the reliability of the prediction of coefficient of consolidation from piezocone tests it is necessary to define the initial pore water pressure attime ‘To better define the correct initial pore pressure value, itis essential to record pore pressures at frequent time intervals, ‘which will be a function of soil type (section 2.3.8), at the beginning of the dissipation test. A linear projection of the square-root time plot can provide a reasonable estimate of Powell and Quatermann (1997) showed that in soft clays the use of u, derived from a square-root time plot sig~ nificantly improved the repeatability of the normalized dis- sipation curves. 2. Effects of soil disturbance due to penetration. ‘This effect can create a zone of disturbed soil around the piezocone during penetration which may have lower perme- ability than the undisturbed soil 3, Importance of vertical as well as horizontal dissipation. ‘Although it is believed that dissipation is mainly governed by the radial coefficient of consolidation, (e.g. Levadoux and Baligh, 1986) there will be some uncertainties related to the relative importance of c, and cy. The importance of this, ‘will also depend on soil anisotropy, as discussed below. 4, Soil anisotropy. For most soils the permeability and coefficient of consolida- tion is higher in the horizontal direction, but this may vary from one soil to another (see below). Based on the above discussion, the recommended procedure to estimate the coefficient of consolidation is fo use dissipa- tion data from the filter location behind the cone (us); however, other filter locations may be used even though the data may be somewhat less consistent80 INTERPRETATION OF CPT/PIEZOCONE DATA Figure 5.41 Pore pressure distribution in saturated clays (after Sully etal 1988). ‘The recommended procedure is as follows: (@) Plot the early part of the dissipation (less than 10% dissipation) at an enlarged scale, either log or square root time, and evaluate the initial pore pressure, (b) Define u, from available data on ground water level, piezometrie readings or data from piezocone tests in adjacent sand layers. (©) Plot normalized excess pore pressure = Wo us Wo u 637) versus time (¢) on log and/or Jt scale. (@) Define time for 50% dissipation (ts), (©) Use fay and curves in Figure 5.39 to predict cy. If no other data are available use an average J, between the range in Figure 5.39, (© If dissipation has not proceeded sufficiently long to define fo, then the slope of the straight line from the first part of u vs yé plot (mi) may be used in Figure 5.39 to predict cy Based on available experience, this recommended procedure should provide estimates of c, to within = half an order of ‘magnitude, However, the technique is repeatable and pro- vvides an accurate measure of the changes in the consolida- tion characteristics within a given soil profile. ‘Arough estimate of the coefficient of consolidation in the vertical direction can be obtained using the ratios of permea- bility in horizontal and vertical direction given in Table 5.12 o 6.38) Pw anh 6.39) If the soil compressibility is assumed isotropic (that is, m= me (6.40) Due to the uncertainties associated with interpretation of coefficient of consolidation from dissipation test data, the predicted value of c, can currently only be considered to be ropresentative within one order of magnitude 5.4.4.2 Coefficient of permeability (hydraulic conductivity) Baligh and Levadoux (1980) recommended that the hor- {zontal coefficient of permeability can be estimated from the expression: B= RR cy (41) ‘where RR is the compression ratio in the overconsolidated range. It represents the strain per log cycle of effective stress, during recompression and can be determined from labo; ratory consolidation tests (0.5 X10 7 10 Gravelly sand to sand 1x10°%t01 11 *Very stiff fine-getined soll 1X 10-9to 1 x 1077 LX 10"* to 1x 107% 12. *Very stiff sand to clayey sand ‘overconsolidated or cemented ‘Table 5.13(b) Estimation of soil permeability Q) stom CPT soil ‘behaviour charts. Based on normalized CPT chart by Robertson, 1990 (Figure 5.8) Range of soil permeability Zone. Soil behaviour type (SBT) (m/s) Sensitive fie grained 3x 10°03 x 107 Organic soils 1x 10™8 1x 10-F y 1x10" to 1x 10"? Silt mixtures 3x 107% t0 1x 10° ‘Sand mixtures 1% 1077 t0 1x 107% Sends 1x 107$t01x 10"? Gravelly sand to sand 1x 107 to 1 Very stil sand to clayey sand 1X 10"*to1 x 10° 1X 10-%t0 1 10-7 * Very aiff fine-arsined soil *Overconsolidated or cemented82 INTERPRETATION OF CPT/PIEZOCONE DATA Pore Sloove cone Felton prosaire fon resttanee rato” Diforntal PPS cara) tea) ite) ya) ate algy pete oder) Sr Oh! Sept 29 oN 2 go, bo Sonaay eam \ coaro sand E anes 310 0 ito sand 3 z —_ z Fairs 3 V /some silt" . > =60% i st comaty 3 (Baldi et al, 1982) \ elayey sit Bao zo} | & | 7 | a 1 | 80 soll Figure $43. CPTU profile from MeDonald’s Farm, Vancouver (from Campanella and Robertson, 1981), Sleeve Pore water friction “pressure f, (MPa) (MPa) resistance qo(MPa) 10] on 20 Depth (m) 25/5 oll Figure S44. CPTU data in dense Po river sand (Stor Bataglio and Bruzzi, 1987). (0002 1080604020 30 20 10 0 Relative density, D,, or density index, Ip, is defined as Ip=D,= (6.42) where: mx and ein are the maximum and minimum void ratios that can be determined in the laboratory using appropriate standards; e is the in situ void ratio, The problems associated with en and enn are well known and the results from one laboratory may be quite different from those from another laboratory. Also, recent research has repeatedly shown that the stress strain and strength behaviour of cohesionless soils is too complicated to be represented only by the relative density of the soil However, for many years, relative density has been used by engineers as a parameter to describe sand deposits and correlations have been developed relating D, to the angle of intemal friction, liquefaction potential, and so on. Thus despite recent trends to try to bypass the intermediate determination of D, (such as state parameter, section 55.1.2), relative density is still commonly used by many practising engineers,y INTERPRETATION IN COARSE-GRAINED SOILS Recent work in lange calibration chambers (Appendix C) fas provided numerous correlations between cone penetration resistance and relative density for clean, predominantly silica sands. Table C.1 in Appendix C pro- vides a summary of the major calibration chamber studies and shows the basic classification index properties of the different sands used in the calibration chamber studies, The calibration chamber testing has shown that the cone resistance is controlled by sand density, in situ vertical and horizontal effective stress and sand compressibility. Robert- son and Campanella (1983b) carried out a review of the available calibration chamber test results and showed that the correlations between cone resistance and effective stress ata given relative density were similar in shape but were strongly influenced by sand compressibility, as illustrated in Figure 5.45. Sands with a high compressibility (that is, a sand with a latge feldspar and/or mica content and/or angular grains) would have a lower cone resistance than @ sand at the same relative density with a low compressibility (that is, a sand with very litte feldspar and mica and rounded grains). Figure 5.46 illustrates the variation in cone resist- ance for a range of relative densities for different sands. The Ticino sand used by Baldi eral, (1986) was a clean, uniform siliea sand with subangular grains and appears 10 have a moderate compressibility Figure 5.46 shows the range for ive predominantly silica sands used under controlled laboratory conditions; field cases are likely to exhibit more variability Cone resistance q, (MPa) 36 © 10 20 4050 00) o4 | 03] | Vertical effective stress o',, (MPa) os _Schmartmarn (1970) ion Mines sand high compressblty @ Bale stal, (1082) Tao sand: moderate comprossbily © Vio & tot (108%) Montorey sarc ow campresebiy Figure S45 Effect of sand compressibility on gz, ao, D-relation- ‘hip in normalized form (based on Roberson and Campanella, 19838), 83 96+ 65 09,7 & Relative density D, (%) n ton no san sii 4 1000 Figure 5.46 Influence of compressibility on NC, uncemented, ‘unaged, predominantly quartz sands (after Jamialkowski er al, 1985), Based on extensive calibration testing on Ticino sand, Baldi ef al. (1986) recommended the following formula to estimate relative density: ain ogmlair) oH where: ‘or Cy and C; are soil constants effective stress in kPa, cither mean stress Ginna, ‘or vertical stress 0! {ge= cone penetration resistance in kPa, ‘The correlations based on the above formula and soil constants are shown in Figure 5.47. To estimate the relative density ofa given sand from the CPT requires an estimate of horizontal effective stress and the compressibility of the sand relative to those used in the calibration studies. For moderately compressible, normally consolidated unaged and uncemented silica sands, Figure 5.47a (using g,) can be ‘used to estimate D,.. For overconsolidated sands use Figure 5.47} and the horizontal offective stress oj... The compressi- bility of a given sand can be estimated based on the grain characteristics and index properties compared to those given in Table C3 in Appendix C. Figure 5.46 can be used to adjust the correlations for variations in compressibility. Previous experience has shown that an increase in grain size can increase cone resistance and an increase in coefficient of ‘uniformity may decrease cone resistance at a given relative density. A visual clasification of grain characteristics would improve the selection of the appropriate D, correlation. The compressibility of sands tends to increase with increasing84 INTERPRETATION OF CPT/PIEZOCONE DATA ‘uniformity in grading, angularity of grains, mica or feldspar content, carbonate content and fines content. Care should be exercised in interbedded deposits where the cone resistance may not have reached the full value within a thin layer Kulhawy and Mayne (1990) suggestod a simpler formula for estimating relative density: 4s Oa 44) where: ge: = Dimensionless normalized cone resistance (aspe) ipa Pe = Atmospheric pressure in same units as gq. (6.45) Q.= Compressibility factor 091 <9.<1.09 ocx = Overconsoidation factor ocr"* Q4 = Ageing factor. Kulhawy and Mayne (1990) suggested compressibility factors (Q.) a8 follows: 0.91 Low compressibility predominantly quartz sands, rounded grains with Tittle or no fines 1.00 — Medium compressibility ‘quartz sands with some feldspar and/or several per cont fines 1.09 High compressibility high fines content, mica or other compressible minerals. Based on experience with the SPT, Kulhawy and Mayne (1990) also suggested an ageing factor Q, as follows: : y= 12+ 005 tog 0 sioe() a6 Hence: 1 (oe , Posepaamg eG) 6 Cone resistance q, (MPa) 40 50 60 Vertical effective stress o',, (kPa) 500 2030 40 50 60 L 1 100 Valid for normally os consolidated sand, K,=0.45, 200+ 4 300 [- 4 400 |- +| 80 80 4 L L D=100% “| @ c. R=0.96 87 50.88 C,=2.41 Figure S47 qo, ojo, D, relationship for Ticino sand in terms of (a) normally and (6) overconsolidated behaviour (ater Baldi eal, 1986),‘The above correlations are basod predominantly on tests on unaged, clean fine to medium, uniform predominantly sila sands, However, most natural sands are to some degree aged. Schmertmann (1991) has demonstrated the importance of ageing on penetration test results; in most cases ageing tends to increase cone resistance. Currently there is no interpretation method which takes the effect of ageing into account in a rational way. Schmertmann (1991) recommends that when using charts like those shown in Figures 5.46 and 5.47, one should refer to the obtained D, value as “nominal” (the term “equivalent” is tecommended by Robertson and Campanella, 1988). ‘Based on the above discussion its recommended that, for sands that are reasonably similar to the sands used in the calibration chamber studies (Table C:3) and that can be assumed to be normally consolidated, use the normally consolidated part of the chart in Figure 5.47 to obtain a “yominal” value for relative density. If the sand is assumed not to be normally consolidated, estimate K, and use the overconsolidated scales inthe chart given in Figure 5.47. Correlations for relative density ate approximate and are sensitive to variations in soil eompres- sibility and horizontal stress INTERPRETATION IN COARSE-GRAINED SOILS 85 5.5.1.2 State parameter A recent trend is to interpret cone penetration test results in terms ofa state parameter, y, instead of relative density. The state parameter combines the influence of void ratio and stress level with reference to an ultimate (steady or critical) state, and can be used! to describe sand behaviour. ‘The state parameter y represents the difference in the void ratio of a soil at a given mean effective stress, O'oean = (Blo + 2043, and the void ratio on the steady or critieal state line, at the same mean effective stress (Figure 5.48), The state parameter is used to describe ‘whether the soil is contractive (that is, the state parameter is positive) or dilative (the state parameter is negative) at large strains Been and Jefferies (1985) suggested that the state parame- ter correlates well with large strain engineering parameters such as peak friction angle determined in a triaxial test (ection 5.5.2), The steady state line (SSL) and the slope of the steady state line in e-log p’ space (24) can be determined from a series of stress or strain controlled triaxial tests ‘Been ef al, (1986, 1987b), based on a study of calibration chamber tests, developed a procedure for estimating the Cone resistance q, (MPa) oo ww aoe 2 Noa (ager) 100 Vald for normally and z overconeoldale é sence | = 00] é ® 300] = 5 2 5 | < 400 | 8 | = | Lo 203040 50 60 70 80 90 D= 100% 00 3) 20.55 C=2.61 0.95 Figure 5.47 Continued,State parameter for A INTERPRETATION OF CPT/PIEZOCONE DATA 3. Estimate the n stu horizontal tess, oy and obtain oa and oa 44. Compute the sate parameter y from the equation $= Ge come Tana ome (6.48) ‘ a4, ‘The two requirements for this method are to take samples if 2 and to run triaxial tests to determine the SSL and to assess \ & 2}, For the latter parameter, sce section 5.6.1.3, Figure 5.49 3 illustrates the expected range in y- values for different 3 sands, ‘ ‘The state parameter approach is still relatively new and ! ‘may need further validation. This is highlighted by Sladen i 1989) who questions the use of the linear normalization a ; procedure as presented by Been e al. (1986). At this stage it appears that the state parameter may be a Mean normal stress o,” (og scale) very useful approach; however, che parameters should still m ‘be checked against the other better established methods Figure S48 Definition of state parameter (after Been and included in this chapter. Jefferies, 1985) ee state parameter in sand from cone penetration tests. The o sii procedure can be summarized as follows: ed 1, Carry out a seties of triaxial tests to define the steady ue Ottawa state line (SSL). Its slope Ax., varies from sand to sand. OF BSksuna , For preliminary interpretation it may be possible to use En ij results from tests on a previously tested sand if it can be assumed to behave similarly tothe one being investigated (Figure 5.49). ud 2, From Figuro 5.50, determine the normalized state para- raters m and i for the observed slope of the steady state 8 i fine. 0.01 4 10 ; oo, ga (0g scale) 1 ‘500! 50; a [a2 = com. cmp] 3005 sol POttwa = Le ' eal Monterey 1 ae 30] tH call Hokicsuna | 100 “ © crksak a “| ‘O Mi ff “| 10 aol ol 03-03 0 ° © aa ao State parameter yw X Figure 5.49 Steady state lines for a number of sands (from Been etal, 1986) Figure 5.50 State parameters K and m (after Been etal, 1986).INTERPRETATION IN COARSE-GRAINED SOILS 87 Both the relative density and state parameter correlations are influenced by soil compressibility. The state parameter approach has attempted to incorporate this by using the slope of the steady state line (2a) a8 an indirect measure of compressibility. Laboratory studies have shown that sands with a bigh compressibility have values of 2a higher than less compressible soils Plewes ef al. (1992) have incor- porated the variation of Zag directly on to a soil behaviour chat to avoid the need to Know Ay Robertson ¢f al. (1995) have suggested the application of insitu shear wave velocity to evaluate the sate parameter of ‘uncemented and una sands, This approach has the advan- tage that shear wave velocity is independent of soil comptes- sibifity. The work of Robertson et al, (1995) confirms the suggestion by Sladen (1989) that the linear normalization used by Been etal, (1986) is incorrect, Further details on the ‘measurement of shear wave velocity using the seismic CPT are given in Chapter 7 (section 7.3) Sand state from shear wave velocity Robertson and Fear (1995) suggested that the void ratio (e) of e sandy soil can be estimated from the normalized shear wave velocity (V,) over a given range of vod ratio by’ Vy =A = Bove 6.49) vere: 4 and Bare constants for a given sand, both in mis K,'s the coeficient of earth pressure at rest eis void ratio Vis in mis Ya=%e @) (650) V, is the measured in situ shear wave velocity ‘Peis the reference stress isthe vertical effective stress. ‘The in situ ¥, is controlled by the effective stresses in both the direction of wave propagation and the direction of particle motion. For the seismic CPT this means that ¥, is controlled by the vertical and horizontal effective stresses Hence, equation 5.50 should include the horizontal stress as well as the vertical stress. However, in practice, the hor- izontal stress is generally unknown; therefore, the rocom- mended normalization (equation 5.50) includes only the vertical effective stress, similar to that for the cone resist- ance. The error in excluding the horizontal effective stress is onerally less than 10%. However, when using in situ shear ‘wave velocity measurements to estimate in situ void ratio, it is advisable to account for K, conditions, as shown in equation 5.49. Based on the definition of state parameter, Robertson et al, (1995) developed the following expression to estimate sand state from shear wave velocity: = Larter afta 43)) es wa BK” ain} )) « ) where: intercept of USSL at p’ = 1 kPa Jy = slope of USSL in e ~ Inp’ space A and B = constants for a given sand, both in mvs typically A = 316, B ~ 166 ina = stress exponent; typically, na = 0.125 ¥, i in m/s and of is in KPA USSL = the ultimate steady (critical) state line. ‘The above approach based on shear wave velocity is new and is applicable only to unaged and uncemented sand, The ‘method should be used with caution Lf shear wave velocity is measured using the seismic CPT, itis possible to develop a site-specific conrelation between Y and cone resistance (q,) and hence develop a site-specific correlation between q- and in situ sand state. Since cone resistance and shear wave velocity vary with overburden effective stress indifferent ways, itis important to compare normalized values. ‘The relationship between normalized shear wave velocity and normalized cone resistance ean be given by the following: we (fo ; ‘The parameter ¥ appears to be controlled by grain charac teristies, sand compressibility, age and degree of cementa~ tion. For more compressible sands, the parameter Y will increase since the cone penetration resistance, ge will ‘decrease, while the shear wave velocity (V1) remains essen- tially constant. For aged or cemented sands, Y will also inerease since shear wave velocity will increase faster than, penetration resistance, Sand compressibility has a strong influence on cone penetration resistance, but little effect on shear wave veloc~ ity, so it would appear that sand compressibility could be identified by comparing the former with the latter in a sand of the same depth, ‘This can be accomplished using the seismic CPT in which both cone resistance and shear wave velocity are measured {000 yrs Kee oop Kee) OCR 659 100 syttrude Koa) ino’ oN) where: a tye Keto sin 4" (aky, 1944) Incompressiblo Lunne und Christophersen (1983) recommended 80 m= 0.45. However, Mayne and Kulhawy (1982) recom- mended m=0,65, Laboratory tests on @ subengular to yo subrounded clean sand (Al-Hussaini and Townsend, 1975) 010 20 30 40 60 60 _showed that mvaties with OCR and is smaller for loose sand he, Figure 5.51. Proposed correlation for estimating sand compress bility (Robertson and Fear, 1995). between 2 and the parameter ¥ has been developed and is, shown in Figure 5.51. Included in Figure $.51 are estimated contours of age. Figure 5.51 applies only to sands that are uncemented. 5.5.1.8 Overconsolidation ratio At present there are no reliable methods to estimate stress, history from CPT/CPTU data in sands. Geological evidence should be used if at all possible; for example, the thickness of the ice sheet during previous glaciation, Another possibility is to evaluate the over-con- solidation ratio (OCR) by laboratory or im situ tests in neighbouring clay layers (For piezocone see section 5.4.1). IF in situ K, can be assessed, itis possible to estimate OCR using some of the empirical correlations mentioned in section 5.6.1.4 below. An indirect estimate of OCR can be obtained using shear wave velocity results from the seismic, CPT and the chart shown in Figure 5.10. 5.5.1.4 Insitu horizontal stress In situ horizontal stress, ofo, oF the coefficient of lateral stress, K, are important parameters, both for use in design (uch as in assessing the bearing capacity of piles) and as an intermediate parameter in interpretation of CPT results (for, example, for the interpretation of relative density and the state parameter). Measurements of the in sit horizontal stress are very difficult to make with any device, and are usually unreliable. than for dense sand. X, is substantially lower duting recom- pression than during unloading, 2. Based on relative density estimation. Calibration chamber tests have demonstrated that cone resistance in sand is controlled by density and horizontal stress (e.g. Holden, 1971; Baldi ef al., 1986; Houlsby and Hitchman, 1988), The tests by Houlsby and Hitchmann, using Leighton Buzzard sand showed that: 4%) 654) nN wre p,~reference stress of 100 KPa and sa constant Sepending on sand density. Thus ithe sand density ea be evened te above conelaton ashe potent testa Kulhawy and Mayne (1990) reviewed additional calibra- fon chamber esas to produce te following more genta formule, Tho (Gelpe)' Pa 35 exp(D,/20) However, more validation is required before this approach may be used in practice. Based on calibration chamber data, Mayne (1991) pro- posed a modified form of the above by combining the calibration chamber test results to produce the following: (6.55) the K,=@ (edie) (aelpe)'* -——Tadpoloispors) © ma (Cae ) 122 (0cRy™INTERPRETATION IN COARSE-GRAINED SOILS 89 ‘The above equation requires iteration until a matching value of K, is obtained. For sands which have experienced 2 simple loed-unload history by mechanical overconsolidation, a compatible rela- tionship should exist between K, and OCR. As shown by Mayne and Kulhawy (1982), the adopted form is taken as Ky = Kotuy OCR"? in which Kegs) corresponds to the NC horizontal stress coefficient ofthe sand, Tentatively, Mayne (1992) suggested the following form for practical use: K,=0.35 OCR unless better information is available for the particular sand under investigation, The above equation can be rearranged in the form: OCR = 5.04 Ky ‘The use of the above equations requires iteration until matching values of K, are obtained. One side effect of this, procedure is that an estimate of in situ OCR is also made. Mayne (1992) illustrated the use of this approach on a sand site in Stockholm. Dahlberg (1974) reported that this, site originally consisted of 24 m of dry natural sand over- lying bedrock, presumed to be normally consolidated. Man- ‘made excavation removed 16 m of the sand resulting in 8 m of overconsolidated sand, Jn situ testing included CPTs and, pressuremeter tess, Figure 5.52 shows the results of the iteration procedure proposed by Mayne. The resulting K, profile (Figure 5.52a) compares well with values interpreted from pressuremeter tests, The resulting OCR profile (Figure 5.526) compares well with that calculated based on excavation of 16m of soil, 3. Based on direct lateral stress measurements Several attempts have been made to measure the lateral stress on the friction sleeve, and then to empirically correlate this measurement to the in sifu horizontal stress, che (c.g, ‘Huntsman et al., 1986; Jefferies et al., 1987; Campanella et 4l., 1990). So far there have been problems with the refiabil- ity of the measurements and also to establish reliable corre- lations (¢.g. Masood et al., 1990). More details are given in Chapter 7, ‘Sometimes the presence of high horizontal stresses can produce high friction sleeve values, f,. However, to quantify the stress level itis necessary to have some knowledge of the fiction sleeve value of the same sand under normally consolidated conditions. Thus, it is very difficult, if not impossible, to distinguish between a dense, normally con- solidated sand and a less dense, overconsolidated sand 5.5.2. Strength characteristics 5.5.2.1 Effective stress strength parameters The shear strength of uncemented cohesionless soil is usually expressed in terms of the peak secant friction angle, #". The concept used by Janbu and Senneset (1974) also includes attraction, a, which is an attempt to account for non-linearity of the failure envelope, Numerous methods for assessing ¢" from cone resistance have been published. Basically the methods fall into one of the following three categories ‘© cmpirical or semi-empirical correlations (mostly based on calibration chamber tests ~ see Appendix C) «¢ bearing capacity theory ‘ cavity expansion theory. Table 5.14 summarizes the most referred to methods within the three categories. Horizontal stress coefficient, K, ppeeae iene mee ouemnateer Profle based on 2 * known stress history = (Oahiberg, 1974) & a4 a a NP MTP, = oy + fe HE, = 1 (4 0'y OR) elt Overconsolidation ratio, OCR: 0 ss ; + q ea ‘Caleulated profile assuming = ‘sand is NC and 16 mis a4 /3 Tomoved from orginal é, overburden t [#ocn-1 (Ge oo» Ko) 7 fé att Figure 5.52 Results of iterative procedure for X, andl OCR in sand (eller Mayne 1991)| | 90 INTERPRETATION OF CPT/PIEZOCONE DATA Empirical De-approach Itis possible to estimate the peak fiction angle (f") based on 4 knowledge of the relative density (D,), soil gradation charaeterstes andthe in situ stress level, Figure 5.53 shows a relationship proposed by Schmertmann (1978) between the peak secant ftiction angle and relative density for different grain-size characteristics. It appears that the (#') shown in Figure 5.53 applies to an effective overburden stress of about 150 kPa. Ifthe overburden stress is higher than 150 kPa the (¢") values may need to be reduced, The typical variation of (§") with overburden stress for different D, values is shown in Figure 5.54 “The weakness with this approach is the approximate nature to which D, can be estimated from CPT data 3 2 S 3 $ 8 8 friction angle ¢' (degrees) 0 10 20 80 40 50 60 70 80 90 100 Relative density D, (%) wre 5.83 Relationship between 4’, and D, suggested by ‘Schmertmann (1978), ‘Table 5.14 Summary of methods to find drained shear strength Experience has shown that the D,~qe correlations are sensitive to soil compressibility and in situ horizontal stres- ses. Figure 5.46 illustrated the influence of compressibility and showed that D, could have an error of + 12%. This ‘would translate to an error in g” of 1,5° fora uniform fine sand, Empirical ~ calibration chamber data A review of calibration chamber test results was made independently by Lunne and Christophersen (1983) and, Robertson and Campanella (19836) to compare measured ‘cone resistance (q,) to measured peak secant friction angle (#’). The peak friction angle values were obtained from drained twiaxial compression tests performed at confining, stresses approximately equal to the horizontal stresses in the calibration chamber before cone penetration (that is, in situ je). ‘The resulting comparison is shown in Figure 5.55. Details of the sands used in the studies are given in Table C1 (Appendix C). The scatter in the results illustrate the influ- cence of soil compressibility. The correlation for uncemented ‘moderately incompressible, predominately silica sands pro- posed by Robertson and Campanella (1983b) is also shown, in Figure $.55. This correlation results in the rather simple set of relationships shown in Figure 5.56, where qe increases, linearly with ¢%, for constant ¢’. For highly compressible sands, the chart shown in Figure 5.56 would tend to predict low friction angles. Empirical state parameter approach Been et al. (1986) suggested that the CPT data could be used, to evaluate the state parameter y for sands (section 5.5.1.2), With an estimate of y, the ¢' can be evaluated from Figure 537. ‘This method incorporates the determination of the steady state line on reconstituted samples to evaluate the slope of Curvilinear Relevant Other strength —Compres- stress required ‘Type of method Reference Stressstrain envelope sibility tensor data Empirical Schmertmann (1978) [No assumption No No None D,,GSD Deapproach Baldi etal. (1986) No assumption No No a, Da GSD, Kleven etal. (1989) K Empirical Lmne & Christophersen (1983) No assumption No No a Robertson Se Campanella (19836) Empirical sate paramoter Been etal (1986) No assumption No No oy das Bearing capacity ‘anu and Senneset (1974) Rigid-plastic No Noe eo ea Lunne & Christophersen (1983) Durgunoghs and Mitchell (1975) Rigid plastic No No Gy Ky Cavity expansion Vesic (1972) Elastic perfelly plastic No Vea eee Get Baligh (1975) Elastic perfectly plastic Yes Yes he Koy Gye (GSD = grain size distribution; @~ sheet modulus; f= angle of plastfeation, = traction,INTERPRETATION IN COARSE-GRAINED SOILS 1 + T uy North Sea site ion angle 4, (degrees) 35 30F 25 Each symbol represent one T 7 T oe Peak drained fri 30 405060. 7 80 9 100 110 Relative density D, (%) Figure 5.64 Correlations between gD, and cn for lne to medium, uniform silica sands (ater Kleven e al, 1986). Bearing capacity number N,, 1009 800} LEGEND: 7 600} . ‘repay & Dona (1981) ‘7 lest 400} sn (1978), 7 © Panera (1900) 20) ‘4 Villet & Mitchell (1981), cr) ----f 4 80F durgunogu & Mitchell (1975) 47 q rey | ‘ Pi d es Of? ‘by Flobertson and 40} (1983), g 8 ; ne | Wi oS | : ee 00 02 04 08 o8 10 12 Tangent ¢' Figure 5.85 Relationship between bering capacity number and fiction angle from large calibration chamber tests (xfer Robertson ‘and Campanella, 1983b). Cone resistance q, (MPa) 5 a zim g & ; © 160 @ 200; g < 8 oxo | 2 g Bao - sz 350 4 Figure 8.56 07, qc, $" reationships (afer Robertson and Campa relia, 19836),92 INTERPRETATION OF CPT/PIEZOCONE DATA, Y Valginds sand (@jrnan eta, 1981) CD Hoktsuna sand (NCD 1 Moezeray no. 0 ard (Lad, 1872) 8 (© kogyok sand (0-10% es) 2 Beaulr sand A 210% ras) Oban sand B (5% nes) 12 Banang sand, Havchpato sand (Caso, 1868) Drained angle of shearing resistance ¢' (deg) “02 “Ot 0 oA State parameter y Figure 5.57 Correlation between state parameter, y, and peak fiction angle of sand, ¢ (after Been and Jefferies, 1985) the steady state line, The slope of the steady state line is, essentially a measure of the compressibility of the sand. Bearing eapacity methods Bearing capacity solutions are generally based on assumed failure mechanisms, incompressible material, linear strength envelopes and plane strain conditions. Model studies (Vesic, 1963) have shown how the material compressibility can significantly influence the shape of the failure mechanism, ‘The material compressibility is influenced by density, stress, level and grain characteristics. The two main bearing capacity solutions available were developed by Janbu and Senneset (1974) and Durgunoghu and Mitchell (1975) and are shown in Figure 5.55. The solution by Janbu and Senneset (1974) depends on the shape of the failure zone, described by the angle of plastcification, {B. The Durgunoglu and Mitchell (1975) method accounts for the effet of in situ horizontal stress and cone roughness. Figure 5.55 shows the Durgunoglu and Mitchell (1975) solution for Ky=1—sing’ and K,= 1.0. Figure 5.58 shows a useful graphical representation of the Durgunogh and Mitchell (1975) solution presented by Marchetti (1988). Research by Mitehell and Keaveny (1986) showed that the bearing capacity method provided reasonable predictions of " for most sands but under predicted for highly compres sible sands, Cavity expansion methods ‘The cavity expansion method developed by Vesic (1972) accounts for soil compressibility and volume change charac- teristics. Baligh (1976) developed this method further to incorporate the curvature of the strength envelope. Research, by Mitchell and Keaveny (1986) bas shown that the cavity, ‘expansion method appeared to model the measured response extremely well and could predict the response in highly compressible sands. Unfortunately, the cavity expansion approach requires a knowledge of soil stiffness and volu- metric strains in the plastic region, both of which are rr disicut parameters to estimate or derive. Hence, the cavity expansion methods, although good, have not been used a great deal, ‘Based on the above discussion, its recommended that for relatively uniform, uncemented, clean, predominantly quartz, unaged sands should be estimated empirically using Figures 5.54 andlor 5.56 and 5.58. Figure 5.55 can be used to adjust the estimated 4 to account for variations in compressibility and Ky. Sands with inoteased compressibility will fend to have slightly higher fiction angles fora given penetration resistance and sands ‘vith high K, will tend to havea slightly lower fretion angle fora given penetration resistance. ‘A reasonably conservative value of " should be selected based on the range obtained and the design problem in which # is required. For uncemented unaged sands the reliability of @° can be improved by running triaxial tests on samples. The samples should be reconstituted to the best estimate of D, (section 5.5.1) and consolidated to the best estimate of in situ stresses, Normally the triaxial tests should be sheared drained (that is, CAD of CID), However, in sotne cases with rapid loading it will be appropriate to run undrained tests (CAU or CW). 4000 TTT 2000, = tect ovrurden aass F kc asi sss cottons | |. p= Pras ses oatent cone roughness, §/¢'= 05 1000 eat Normalized cone penetration resistance ,/ Oy, %0 =z 1 = Ky Lae i 30 4 TT nel tea h[pbbeae r 8 we | fis oooas cet 2 8458 8 Coefficient of earth pressure at rest K, Figure 5.58 Chart for predicting peak fretion angle from CPT in tunaged, uncemented, NC, predominantly quartz sand, with Dorgu- oglu and Mitchell (1975) approach (after Marchetti, 1988)INTERPRETATION IN COARSE-GRAINED SOILS 5.5.3 Deformation characteristics ‘A reliable determination of sand stiffness in situ is of great practical interest in view of the difficulties in obtaining deformation modulus from tests on laboratory specimens. Undisturbed samples are extremely difficult to obtain and often more or less impossible using conventional techniques, in cohesionless soils Obtaining reliable estimates of soil stiffness from recon- stituted samples is considered far less likely than obtaining reliable large strain strength parameters Itis difficult to obtain reliable estimates of sand stiffness from in situ penetration tests because (1) modulus depends. on effective stress and on stress history, (2) the in situ test conditions, stress level, drainage and direction of loading, cannot be controlled and (3) reference modulus values are rarely or seldom documented. Any modulus value should refer to the drainage conditions, stress level, and strain level, over which the modulus is applicable. 5.5.3.1. Young's modulus Research using calibration chamber data has shown that the drained Young's modulus in sand mainly depends on rela tive density, overconsolidation ratio and current mean stress level Figure 5.59 presents @ chart to estimate the secant ‘Young’s modulus (2) for an average axial strain of 0.1% for, ° 200 ‘500 1000 72000 Normalized cone resistance q,, Figure §.59 Evaluation of drained Young's modulus ftom CPT {or silica sands (fom Bellotti etal, 1989), 93 a range of stress histories and ageing. This lovel of strain is reasonably representative for many well-designed founda- tions. The stiffness of normally consolidated aged sands (> 1000 years) appears to fall between that of very recent normally consolidated sands and overconsolidated sands. Robertson (1989) suggested a modified correlation for shallow foundations basod on the degree of loading (section 63.2) 5.5.3.2 Constrained modulus Most correlations between CPT results and the drained constrained modulus, M, refer to the tangent modulus, as found from oedometer tests. The reference value of M is normally based on the effective vertical stress, af, before the start of the in situ test; this value is denoted Ma, Based on a review of available calibration chamber tests, Lunne and Christophersen (1983) recommended that an estimate of M, for NC unaged and uncemented predom- inantly silica sands can be obtained from: M,=4q0 for ge <10 MPa Mg = 2g. 20(MPa) for 10 MPa 50 MPa Lunne and Christophersen also included OC sands in their study and recommended as a rough guideline to use: My = 54 for q.<50 MPa 150 MPa for g.> 50 MPa For an additional stress Ag), Lunne and Christophersen recommended Janbu’s (1963) formulation to compute M for the stress range 0}, t0 0%, + Ac: M=M, 657) Vo Gio Recently Bslaamizaad and Robertson (1996) suggested an alternative method to estimate M, from CPT results, based ‘on assessment of extensive calibration tests on quartz sands (Baldi ef al., 1986 and Fioravante er al, 1991). Curve fitting, techniques were used to develop correlations for sands with different stress history. ‘The method presents a correlation incorporating normmal- ized cone resistance and normalized vertical effective stress, inthe orm of w.=hun() Pe (6.58)94 where ‘n= stress exponent equal to 0.200 for normally consolidated sands, and 0.128 for overconsolidated sands ‘Pe = atmospheric pressure, in same units a5 0M, and ge ‘ky = dimensionless modulus number which ean be determined using Fig. 5.60, based on normalized cone penetration resistance (q./p,) and estimated ‘overconsolidation ratio (OCR), This method has the advantage that a prior knowledge of relative density is not required. On average, the predicted Me is betweon 75 to 125 per cent of the corresponding value ‘measured in the calibration chamber tes. 5.5.3.8 Small strain shear modulus The shear modulus is largest at vory low strains and deerea- ses with increasing shear strain, It has generally boon found that the maximum shear modulus is constant for shear strains less than 10~°%, This inital, small strain modulus is often denoted G,. Elastic theory relates the small strain shear modulus, G,, to soil density, p, and shear wave velocity, Vas follows: G.= pe where p = mass density of the soil = yig. ‘Therefore, G, can be found by measuring the shear wave (5.59) 2000] 1500 1000} 500] 200 400 Fe /Pa Figure 5.60 Constrained modulus number of sand as function of cone resistance and OCR (after Eslaamizaad and Robertson, 1996), 600 INTERPRETATION OF CPT/PIEZOCONE DATA velocity using the seismic CPT (Robertson ef al., 1986). Alternatively, G, can be estimated using empirical correlations. Jamiolkowski ef al, (1988) showed that the same vari- ables of soil density and in situ effective stresses controls both q- and G,. Hence, a correlation between qe and Gi ean be found for uncemented and unaged cohesionless soils. Cementation and ageing have different influences over q. and G, (Figure 5.10). Compressbility ean have a significant influence on the correlation between G. and q.. Based on calibration chamber results as well as field measurements, Rix and Stokoe (1992) suggest the correlation shown in Figure 561. The wide range of G,/q. at low values of normalized cone resistance is most likely duc to variations in soil compressibility. More compressible sands appear to produce lower values of normalized cone resistance and hence higher values of G, qu ‘A similar relationship was shown in Figure 5.10 5.6 AVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL Most references and interpretation methods are related 0 either sand (fully drained conditions) or clay (fully undrained conditions). However, very frequently other soil, types (for example, those with partial drainage during pene- tration) are encountered and CPT are also used in materials such 2s, silt, peat, mine tailings, permafrost and so on, 50 Unoemented quartz sands 6, &_ yor (ayy = 1634 (er) 40, 7 F v0. 4 Average + Mer Range Average + Average sof 4 s e zo}. a vob 4 ° eee eee 1 199200800500 1000 2000 3000 aoe Vee Figure $.61 Gaus/qe (after Rix and Stokoe, 1992),‘This section is therefore devoted to summarizing avail- able experience and, where possible, providing recommen- dations on interpretation in materials other than clean sand | or clay. Since the amount of experience varies considerably with material type, the coverage is more detailed for some ‘materials than others 5.6.1 | 5.64.4 Interpretation methods valid for sand or clay may not be applicable for silts since penetration in this type of material cean be partially drained, Bugno and MeNeilan (1984) sug- {gost that undrained response will occur for the standard ponetration rate (20 mm/s) if the permeability ofthe soil is less than 10°” to 10° cm/s. Soils having permeabilities between 10° and 10°* emis, which would include most silts, aro believed to behave in a partially drained manner, Bugno and McNeilan compared the results of piezocone tests to soil classification data shear strength parameters and deformation characteristics obtained from laboratory tests ‘on “undisturbed” piston samples Based on a comprehensive study of the soil investigation data from the North Sea site, Gullfaks C, Hight etal. (1994) confirmed that the clay content is very important for the penetration behaviour in clayey silts and clayey sands. Figure 5.62 shows the results ofa typical CPTU at the site, with g. (uncorrected) and pore water pressure measured on the cone (us) vs depth, Penetration through the clayey sands Intermediate soils (clayey sands to silts) Penetration behaviour Cone resistance q, (MPa) u (MPa) 20 Depth below seabed (m) 40 AVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL, Pore pressure 95 is marked by large variations in both q- and u which Hight et al, argued was due to variations in clay content which are also shown in Figure 5.62, ight etal, found that the relation in Figure 5.63 could be used to assess the in situ variation in clay content and the drainage conditions during penetration, For the interpretation of CPT data in silty soils it is {important to identify the drainage conditions expected in the design problem and during the cone penetration. If the design problem will involve the undrained shear strength and cone penetration is also undrained, the CPT data can be intorpreted in a manner similar to clay (Section 5.4). On the other hand, if the design problem will involve drained loading and the cone penetration is also drained, the CPT data can be interpreted in a manner similar to a sand (section 5.5). Interpretation is more difficult ifthe design problem is expected to involve drained loading but the cone penetration process is undrained or partially drained. In this case effee- tive stress strength parameters are needed for design. 5.6.1.2. Typical results and classification Figure 5.64 shows an example from the well-investigated Stjordal silt (Senneset ct al, 1988). Two types of piezocones, ‘wore used, one with the filter on the cone face (u) and one with the filter behind the cone (u2). The cone resistance _measured by the two piezocones were very similar, while the uz profile is consistently lower than the w profile. On Clay content (%) 06 10 15 20 25 30 35 Figure 5.62 Typical CPTU profile from Gullfaks and its correlation wit clay content (Hight etal, 1994),average tia ~ 0.6 1. Unfortunately sleeve friction measure- ‘ments were not obtained. Figure 5.64 also includes the soil, profile as given by the description of the soil samples obtained with the 54mm NGI piston sampler. Strength ‘parameters are determined from conventional triaxial (CTU) and oedo-triaxial (CK,U) test, and the effective stress para- meters, attraction (a) and friction (tan ¢") are given in the figure. Modulus number, m, and grain size distribution are also included. Figure 5.65 includes an example from Keilisnes, Ieeland where grain size data are also given. Pore pressure was ‘measured behind the cone, Below a 2 m thick layer of clean, sand, a silt (55-60%) is encountered. A 70 Low <= Clay content — High 60 Drained “Penetration. — Undrained 50 40 30 20 10 0 01 02 08 04 05 06 (Us = Ua) / (de = ve) INTERPRETATION OF CPT/PIEZOCONE DATA 5.6.1.3 Undrained shear strength The cone factor, Ny, (section 5.5.2.1) was found by Senne set eral (1988) to range from 15-30 when CAUC tests were used for the reference 5, However, the relevance of using undrained shear strength in silts isa function of the design problem. Senneset et al. (1982) reported that ,-values obtained by such correlations are particularly questionable for soils generating small pore pressures, especially corresponding to 3B, < 04. Such conditions may be representative of materials coarser than clayey silts 5.6.1.4 Effective stress strength parameters Senneset etal, (1988) suggested that effective stress strength parameters can be estimated using the same approach as for clay (section 5.4.6), Table 5.6 recommends values of the plastification angle f for calculations of the cone resistance number, Nj, and the pore pressure ratio, By. The attraction a must be estimated fist using Table 5.15. ‘The bearing capacity charts in Figure 5.28 can then be used to compute friction tan g’. Figure 5.66 shows a typical interpretation of the piezocone profile at Stjoerdal shown in Figure 5.64. From Figure 5.66, it can be observed that there is good agreement between the piezocone determined and labo- ratory values of tan ¢" Senneset ef al. (1988) presented in Table 5.15 typical values of attraction (a) and friction (tan g’) for silts that may be used for guidance. ‘The expected ranges of q,, in Table 5.15 are taken from Norwegian Geotechnical Society (1994). 5.6.1.5 Constrained modulus Figure 5.63. Assessment of in situ variation of clay content (Hight Under conditions where one dimensional loading applics ats 1938), (sich sina thin ayer undemeath large loaded rea on fale | [uae] | 2 wane] )egram || Omerggawe || roopossue fe] co |v | ada | ame [fixer] 9 | exes |] Seuss se (s| % ™ sano { leo ines [pam cn 4 | fe sly bd” la cs 3 ss Be aface || sfwl) | Py a 2 | 4 4 by EEL | ae eur 7 to 0 mcd Lane | om [a3 7. Figure 5.64 Soi profile and piezacone results ftom Stjoerdal sit ( fer Senneset etal, 1988)AVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL, 97 ae TEI FSI ET raw TE 21! corse ST canna 8 Is Be . 3 E é E 3 Sd 0 25 60 75 100 125 160 17.5 20 Constrained deformation modulus at 6’ ,, (MPa) from oedometer tests Figure 5.67 Constrained modulus (M,) vs cone resistance, q, for silly sols (after Senneset eral, 1988)98 INTERPRETATION OF CPT/PIEZOCONE DATA between sands and clays with typical Gi/g, values of between 20 and 40, This is still only a guide and it is recommended that Gy be established directly by seismic Friction f, Cone Resistance q, ‘cone testing (section 7.3) while noting the comments made (MPa) (MPa) in sections 5.4.3.3 and 5.5.3.3. 030201 0 4 8 12 16 20 24 5.6.1.7 Coefficient of consolidation To get a fist estimate of coefficient of consolidation the methods described for clay (section 5.4.4.1) may be used. However, it should be born in mind that the theoretical initial pore pressure distribution used in the models is, developed on the assumption of ideal undrained response daring penetration. This assumption may be questionable in, silty soils with B, < 0.4 and partial drainage considerations, fare of great importance in understanding and interpreting 4p dissipation data in silty soils, If the cone penetration is considered to be partially 14 rained it may be possible to change the rate of penetration to ensure either a drained or undrained penetration. This, procedure may be feasible in silt deposits where a reason- 18 ble change in rate of penetration may produce the required change in drainage. However, since permeability varies by ners of magnitude it may require changes inthe rate oF = pp penetration by orders of magnitude. Such large variations in & the ate of penetration ae often dificult achieve Ba 5.6.1.8 General experience ge i Discussions on use and interpretation ofpiezaconetestingin 28 fits have also been given by Villet and Darragh (1985), 2 95 oy | eo 2 eee Bo sy 5.6.2. Peatlorganic soil Su € + cry [As discussed by Landva (1986) peatland soils may be & i 8s subdivided into different groups: peats, peaty organic soils, organic soils and soils with organic content. Peats are derived from plants and can th ‘materials, As the degree of humification increases, the peat ‘becomes Jess and less fibrous until itis transformed into an amorphous mass without any discernible structure. Any testing in peat should be related to a classification as for 447 instance given by Landva et al. (1983). Peat | fore be very fibrous chy Sandine ey é 42 ‘Landva used @ 300 mm diameter bulbous cone and repor- 46 ‘ted negative excess pore water pressures on the face of the gg i cone. Landva explained the negative pore pressures by the fibrous nature of the pent tested: “negative excess pore 50 pressures would be gencrated at the location around the filter since the layers that are compressed by the cone point will tend to rebound as they are punctured and thus unload near 54 52 the point”, ‘Landva (1986) concluded that due to the fibrous nature of 66 peat, and the frequent presence of obstructions like stumps 68 : tnd roots, small-scale in situ tess like the cone penetrometer rthe vane tests are of litle engineering use forthe design of Figure 868 Example of CPT profile from Holland with pest road embankments He recommended using carefully selec- lay (air Vos, 1982) odAVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL 99 ted laboratory tests or large test fills. However, in organic soil of non-fibrous nature, CPT and vane tests can be very useful tools Vos (1982) showed examples of CPT profiles from Hol- land which included peat layers. One example is included in Figore $.68. Vos reported that in Holland with the eleetrical cone a friction ratio larger than five normally identifies peat. Figure 5.69 shows examples of a standard piezocone test a) Cone resistance q(MPa) O49 12 16 29 m4 28 oP -5 0 5 10 0 4 8 from the coast of Germany which identified a peat layer that vas later confirmed by sampling in a nearby borehole. As can be seen from Figure 5.69 the peat layer can be identified by a very high friction ratio combined with negative pore pressure in one case (Figure 5.69) and positive pore pres- sures in the other case (5.69b). In the profile with negative pore pressures some sand was mixed in the peat. Experience in the Vancouver area of British Columbia has shown that fibrous peat has highsftiction ratio and low- Pore Friction pressure, ratio, u, (MPa) Ri, (%) 2 4 = E 415 fle B10 8 12! Peat layer b) Cone resistance q, (MPa) 904 8 12 16 20.24 28 9 -5 0 5 10 0 4 0 01 0205 0408 06 07 08 Sleeve friction f, (MPa) Pore Friction pressure ratio U, (MPa) Ry (%) 0 at 4 = 6 Bs 10. ek. a4 a6 18 ah 1 00.1 02 08 0405 06 0.7 08 Sleeve friction f, (MPa) Figure 5.69 Example of CPTU profiles from Germany with peat layers.100 INTERPRETATION OF CPT/PIEZOCONE DATA penetration pote pressures. However, amorphous peat has hhighefriction ratio and high pore pressures. Also, the rate of pore pressure dissipation is slower in the amorphous peat. 5.6.3 Underconsolidated clay Underconsolidated soils are soils which are not fully con- solidated under their present overburden stress. This generally ‘means that the statie pore water pressure in the soil is higher than the corresponding hycirostatie pressure (that is, depth below water table x average unit weight of pore water) ‘The reason for a soi to be in this state can be that excess pore pressures have not had sufficient time to dissipate since its deposition or there could be other aspects related to its goological history. “Tanaka and Sakagami (1989) reported results of piezo ‘cone tests in undereonsolidated soft marine clay in Osaka Bay, Japan. The clay was underconsolidated because of placement ofa fill, nd the excess pore pressures caused by this fill had not had time to dissipate. Tests were carried out jn both normally and underconsolidated clay. Pore pressures were measured behind the cone (U2). “Tanaka and Sakagami showed that by plotting the results in terms of Au, q;- Gu» and s, (measured in UU triaxial tests) the underconsolidated clay could clearly be identified, as shown in Figure 5.70. By using reference values of the in situ excess pore pressure (measured in adjacent piezometers and by dissipa- tion tests) they developed a procedure for assessing this rising the measued penetration pore pressure. However, its necessary to have the corresponding dy and Aw values in NC clay with the same classification characteristics (0 use this procedure. Figure 5.70 also shows thatthe Ni factor to derives, from. ccone resistance is the same for both normally and under~ consolidated clay. The fact that some of the 5, values from the underconsolidated clay are higher than for the normally consolidated clay must be due to the fact that the latter are not available atthe same depth intervals, 5.6.4 Chalk Power (1982) presented a comprehensive study on the use of CPP in chalk, Based on testing in various types of chalk the classification of chalk grades according to q. and f, values aaiven in Table 5.16 were arrived at ‘An example from Power of a CPT profile in chalk at Mundford and the Chalk Grade description is given in Figure 5.71. The sharp peaks in the profile were thought to be the result of flints (which Frequently occur in chal), as well as the manner in which penetration resistance builds up, ‘thon is followed by grain crushing and/or closure of fissures, together with the possible effects of variability in density, degree of cementation and jointing and fissuring. ‘More recently Powell and Quarterman (1994) examined the use of the CPT on a number of chalk sites, including ge ‘Normal console clay ‘eUnderoonslted cay 8 AU= (Up - U,) (KPa) g 8 200 00 600 600 1000 1200 - Gyo (KPa) ‘9 Normally constdate clay (Under consolidated ly ee 2%. 7 Ok 0 20 40 60 80 100 120 140 8, (kPa) 140 oncmaty entity Under crete 420) me 100) ° z al z =, 60 e Up aol af 20) . 20 #0000600 10001200 G- Ge (KPa) Figure 5.70 Possible identification of undereonsolidated clay layers (efter Tanaka and Sakagami, 1989),AVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL, Mundford, where detailed logging of the chalk grade had been undertaken. They found that the trend in Power's work of increasing values of both g. and f, with improving chalk ‘grade (that is, VII) was valid, but that the actual ranges for any given grade varied from site to site. Figure 5.72 summarizes their findings and also shows Power's correla- tions of Table 5.16. It is noticeable that even for the ‘Mundford site much of Powell and Quarterman’s data would fall outside Power classifications. Powell and Quarterman also considered the potential influence of the chalk density on the CPT results and while site A in Figure 5,72 had a lower density than the others and ‘Table 5.16 Chalk grades rlated to CPT values (after Powe, 1982) Grade Brief description ae(MNin®) RA) 101 formed the lower bound of the daa, no significant difference ‘was evident amongst the other data Based on tests in a rango of chalk types, Power (1982) also presented a corelation between q- and SPT N-value as shown in Figure 5.73. Forone site (Mundford) Power also developed a tentative correlation between Young’s modulus and qe as shown in Figure 5.74 Te should be remembered that chalk is @ very variable material where the state of fracturing and fissuring and the nature of infill material is very important, Thus eare should be exercised in extrapolating correlations from one area and chalk type to another. Ideally, specific correlations need to bbe developed for each area. Also to facilitate comparisons botween sites, it is essential that CPT results should be accompanied by earefil desriptions of the chalk. VI Extremely softstrutuelesschulk =» <5 ~~ 6.5 Caleareous solls coniining small lumps of ntact chalk The word “calcareous” is being used asthe generic descrip- V-Stryturless remoulded chalk 5.15 075-10 tion of all materials containing calcium carbonate though ‘containing lumps of intact chalk stricter nomenclature is used for detailed classification. IV _ Rubbly partly weathered chalk 5-15 1.01.25 Table 5.17 shows a classification system proposed by King wil Bedding and jointing etal. (1980), Caleareous soils, which area feature of warm Joa 060 sam spar open. wor gen tite 30 degrees noth 1o 30 degrees south, te 201m, and fen ined tre formed largely from biological processes that involve aoas sedimentation of skeleton remains of marine organisms It Rabtisteblocky unveathered «15-20 4.25+1.89 om the upper waters of the ocean After deposition various halk, Joints 6D to 200 mm processes take place, the most important being cementation par, open to3 me, and due to precipitation of bonds. The main geotechnical fea- pa, op . sometimes infled with tures of calcareous sediments are (Semple, 1988) angularty fragments and structural weakness of their particles, variable cementa- Il Blocky medium-berd chalk. Joints > 20 1.5-20.0 tion and the occurrence of calcareous sediments at higher mat tan 20% pe od voids ratio than silicate sediments. In general, the engineer- close ing properties of ealcareous soils are difficult to assess due to ee eee 6 ee 4 ie a= |} | alee 4 ot - = i> —te eal te |= s Dy 1c 10} t= — Bro * i ge) a 0 12} = \ 12, ~ BE 14 $= ud 16} - “RRC SP 18 im 20! 010 0 0 01 020 05 os 0 8 0 OS Figure $89 Typical mine tailings CPT (Poland).AVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL 15 Cone Sleeve Friction Pore resistance, friction, ratio, pressure, Resistivity q, (MPa) f(MPa) —R, (%)--U(mwater) (ohm-m) 3 1 ot 9192939 5 2 dF z| E 4 qk q E oF sf J q Ew i ‘| £ zl ] a” q q 6" j fl : 16 4 18 20 qf F | 22! a) Cc Figure 5.90 CPT profile in the pond area for an oil sand tilings facility, layers. Locally the material is called Hogfuel Gillespie, _ CPTU results from a 10 m thick sawdust/wood chopping. 1994) {ill in Halden, Norway, gave very similar values, Some characteristic features in the Hogfuel layer are: high To the authors" knowledge, there is no information avail- friction ratio of about 4%6, penetration poro pressure very able for interpreting the results in terms of settlement close to hydrostatic. parameters, Sleeve Friction Cone Pore friction ratio resistance pressure Interproted f, (KPa) Ry (%) q, (MPa) u, (MPa) profile oP 520 2 2S 1018, 20.25 90,10 g Sand fil 26 25 25 25 28 E g Hogtuel 2 ci i 5 5 3 5 5 Sand 75 78 75 75 7sL_ ure S.91 CPTU profile from Westbridge, Vancouver, with Hogfue ill (rom Gillespie, 1994).116 INTERPRETATION OF CPT/PIEZOCONE DATA. 5.6.13 Dutch cheese During the losing dinner of the Second European Sympo- sium of Penetration ‘Testing (ESOPT II) in Amsterdam in 1982, a competition was held. The aim was to predict the CPT cone resistance of a large Dutch Gouda cheese meas- tured with a miniature cone (base area = 0.75 em*), see LGM Mededelingen (1982) Figute 5.92 shows the result of this test. Below the erust, the cone resistance was steady at about 192 kPa, ‘To the authors’ knowledge, information on CPTs ‘cheese types are not available. Strength of the crust ua) 3s ¢ Penetration resistance “i020 0A Depth below crust (mm) 5.6.14 Slurry walls Slurry cutoff walls are frequently used as containment systems for contaminated soils, The backfilling material for the cut-off wall is normally a cement bentonite (CB) self hardening slurry with a composition of 75-77% water, 19-21% blast furnace cement, and 4% sodium bentonite. For the cut-off wall to function properly it is important that it has a low hydraulic conductivity or permeability. Manassero (1994) has suggested that the CPTU is a very cfficient method for obtaining a continuous assessment of the hydraulic conductivity down a vertical profile of a slurry ‘wall, Figure 5,93 shows the results from one particular CPTU profile and also the range from eight other profiles. It should be noted that the pore pressure measurement sensor bad an upper limit of 2 MPa, therefore higher pore pressure values could not be measured. Manassero defined a parameter % “has (5.68) ‘which he argued should be a function of hydraulic con- ductivity, k He used empirical correlations between the soil, classification chart according to the definitions shown in section 5.4 and k as shown in Figure 5.94, Based on these empirical correlations, Manassero proposed that on average: Figure $92 Result of CPT in Dutch cheese Jog k= 2.61 VB, — 10.93 (5.69) @) (b) (ce) Corrected cone Excess Sleeve resistance _pore pressure friction q, (MPa) ‘Au, (MPa) 4, (MPa) o24 ° 6 8100 1 2 3 400102090405 ceryce| Depth (m) ‘opus Tey Bepaten | from 8 CPTUS Figure 5.93 Typical results of CPTUs in cut-off slury wall (after Manassero (1994).AVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL, 17 gL sree Pg Mgdium sande ‘Sandy ¢ ge cogel we tt 2 ot Fy sgn = ey tno : ee cay, giteaey, 8m, savsand, Sana Pt 9? tot) : Hydraulic conductivity k (rvs) Figure 5.94 Tentative hydraulic conductivity as Function of soil classification (after Manassero, 1994) Hydraulic conductivity (m/s) Outer 0 seed Ose meet 0 1 fom, sei cars 2 ‘Average valves} ‘onlog scale E ro By a g3 Jnasnedich chars Tage from : dissipation 0 k range from Itaion os 6 Figure 5.95 Hydraulic conductivity vs depth based on By parame- ter (after Manassero, 1994), Figure 5.95 shows hydraulic conductivity computed from this formula compared with other estimates off Using this method of interpretation, Manassero concluded that CPTUs are able to identify zones where hydraulic conductivity defects exist However, Manassero stressed that it is important to calibrate the proposed comrelation between By and & for any particular wall. While showing promise, this method needs further validation; much of Manassero’s data fell outside the range of his pore pressure transducer, although it was still used. Tedd et al. (1996) report little’ success in obtaining meaningful results of hydraulic conductivity from piezo- cones tests in slurry walls, 5.6.15 Volcanic soils Soils derived from volcanic activity exist in some parts of the world. These soils are often characterized by low specific gravity and crushable, angular grains. Takesue eral (1995) presented some examples of CPTU ina voleanie soil, known localy as “Shirasu”, which is widely distributed in southern Kyushu, Japan. This soil was composed of vesicu- late and easily crushable voleanic glass grains of angular shape and low specific gravity. A typical profile of CPTU data is shown in Figure 5.96. The cone resistance in the Shirasu layer is slightly lower than in the adjacent quartz sands indicating a slight increase in soil compressibility for the Shirasu soi. The friction ratio in the voleanic Shiresu soil is also slightly higher than the quartz sand. However, Takesue et al. (1995) showed that the soil classification charts by Robertson (1990) provided a reasonable elassifica- tion for these sols, as shown in Figure 5.97. Takesue etal also showed that the CPT-SPT correlations and pile design values for the voleanic soily were consistent with quartz sand values. Hence, it appears that this particular voleanie soil is similar to predominately quartz sols. There have been published cases where CPT data has been influenced by the high compressibility of some vol- canic sols. This will depend on the type and natare of the volcanic soil, Some coarse-grained pumice deposits can have high compressibility 6.6.16 Fuelash Coal-fred electrical generating power plants bum coal to produce electricity. A byproduct ofthis process is fuel ash Which forms due to the non-combustible mineral consttu- ents in coul. Bottom ash isthe portion of burned coal that falls to the floor of the power units; it has traditionally lite value end is often disposed of in ash lagoons by hydraulic deposition, Fly ash is collected by filters and cleaners inthe chimneys of the power units and is often sold as conerete additives or mixed with the bottom ash for disposal. CPTs have been carried out in fuel ash disposal areas to evaluate stability and remediation. The composition ofthe fuel ash is a function ofthe composition ofthe original coal from which it was formed. The non-combustible minerals from the coal18 INTERPRETATION OF CPT/PIEZOCONE DATA Particle Cone Sleeve Friction Pore SPT — density resistance friction ratio pressure 5 q Nevalue p, (g/om') q(MPa) (kPa) FR, (%) Up (MPa) ' o_20 4023 25 270 ‘10 20.300 °200 4000 112.9 400051015 ; fs ° F —(Shirasu Fi) Re 10 E al earl i [Clean sana 20 \(Quartz Sand)" 7] E i) = 2__ t €. |2 | = 301) sity sana 4 | I S = i a | | © 404 | sit 4 | | (hirasu) i sol] el §| Pumice 3|& 4 i | san | i | (Shirasu) | ] Figure 5.96 Typical profile of CPTU data in voleane sol (Takesue et al, 1995). Grain sizo Fines Mean Soil disirbution content grain size Solltype Soil typo layers (6) FE(%) Dg im) (oy F,-Q, chart) (PY BQ chart) | wo Peg ge se 7 f 08a oft tig tice te (Shirasu Fil) fe a ° e.88, et i Clean sand oe 20° (Quartz ‘| sang) i Ex] sity jd =) sand t § | stiney 0 i silt (Sbirasu) so : 4 Pumice el i i & 5 F sand ial (snirasu ray ' See tig. 55 Figure 5.97 Compatison of laboratory test results with CPTU classification in voleane soil according to chats in Fig. 5.8. (Takesue et al, 1995), | |AVAILABLE EXPERIENCE AND INTERPRETATION IN OTHER MATERIAL, 19 melt in the famace and as a result undergo physical and chemical changes. Clay minerals can form glass spheres of complex silicates as well as some quartz, calcite forms calcium oxide, calcium hydroxide, calcium silicate and calcium sulphate and pyrite forms iron oxides. Gases can be trapped in particles, giving rise to cenospheres. The exact nature of this process depends on the type and composition of the source coal. Bottom ash can produce sintering of particles and fly ash can be further altered by chemicals in either the collection or stabilization process, ‘The porous, angular and friable nature of the ash often produces a very compressible granular material. This, com- bined with the pozzolanic nature of the material, can pro- dace a compressible slightly cemented material similar to some carbonate soils, Hence, very low cone resistance and low friction values have been observed in some fuel ash deposits. An example is shown in Figure 5.98, The cone penetrometer used for the profile shown in Figure 5.98 was a ow capacity 25 KN cone. It is interesting to observe the very small sleeve friction values of about | kPa, which is close to the limit of accuracy for the cone penetrometer. The ealcu- lated friction ratio values are close to 0.1% and penetration pore pressures are close to hydrostatic indicating a drained penetration, Dissipation tests during pauses in the cone penetration showed very rapid drainage to approximately hydrostatic conditions. The ash at this site is predominately @ aravelly sand with a specific gravity of around 2.25 and a ‘maximum void ratio of around 2.5. The top 1m of the profile is a cemented crust, It is also interesting to note that the ash is classified as a “sensitive fine-grained soil” on the rnon-normalized CPT charts shown in Figure 5.7, whereas it is more closely classified as a “sand mixture ~ silty sand to sandy silt” in the normalized chart shown in Figure 5.8, The ‘normalisation required careful consideration of the low soil unit weight, The very low friction ratio values are an indication of a potentially metastable (sensitive) soil. The ash profile shown was carried out at an ash lagoon where a sudden flow slide ad occurred due to the filling of a now containment eell, Hence, the CPT appears to have identified fan important soil behaviour. 5.6.17 Loess soil Loess deposits cover large areas of Eastern Europe, Asia and ccontral parts of Northem America, It is generally accepted that the origin of these deposits are post glacial, Strong ‘winds blew the particles off'the areas that were covered with slaciers; deposition occurred when the winds got weaker. ‘The grain size distributions of loess deposits vary from location to location but often fall in the group of low plasticity clays (CL), with clay particles not exceeding 15%, ‘A special feature frequently encountered for loess soils is large pore volumes (Which may be up to a few centimetres in diameter), mainly in the vertical direction. This often results Cone resistance, Friction resistance, _ Pore Friction tio, i (kPa) 4, (kPa) pressure, _ratio, " (KPa) Rety/cy (26) o0___$00_1000_1500 2000,0, 2 4 68 1050 20 60 100901 2.94 1 2 = gq & 3 4 5 4 al uu Figure 5.98 Example of CPT profile in fuel ash,120 INTERPRETATION OF CPT/PIEZOCONE DATA in low unit weight, 11-12 kN/m*, Without change in water content this structure is stable. However, when the water ccontent ofthe loess soil with low dry weight is increased, the soil structure can collapse, resulting in large volume chan- ‘ges. Very large settlements ean take place if water becomes available in collapsible loess (Susic and Spasojevic, 1995). ‘The usual procedure for evaluating the collapsibility of loess soil is to run two parallel oedometer tests on undis- turbed samples: one on a specimen with natural water content and one on a saturated sample. Susie and Spasojevic (1995) proposed a methodology for estimating settlements due to wetting of collapsible loess soils based on CPT results from two locations in Beograd. ‘They assumed that the normalized cone resistance decreased linearly with the degree of saturation S,: + b(1—S,) 6.70) Ps where a and b are constants whose ratio has been found t0 be close to 1(p, = atmospheric pressure ~ 100 kPa). For the loess soils studied, Susic and Spasojevie found 'S,=0.5 at the natural water content. Testing soils that were 100% saturated, they found that on average, full saturation of loess sol had oceurred where the cone resistance before ‘wetting was more than 1.5 larger than qe after wetting, From a general knowledge of the water content of the soil in the area the authors deduced the dry unit weight of the soil at saturation point yu... Knowing, or estimating the dry ‘unit weight of the soil at natural water content, jun, setle- ment due to wetting only could be determined (assuming one dimensional conditions): a = Yo0 7 Loess soil, as other soils, will vary from one geographical area to another. The correlation referred to above cannot be directly used for all loess soils. The authors therefore rightly say that one more variable has to be introduced linking the CPT resulis to the dry unit weight at natural water contents, (3.71) 5.6.18 Lunar soil “Mitchell and Houston (1974) reported results of cone pene- tration tests on the moon during the Apollo lunar exploration, programme, Tests were done by a handheld simple cone penetration deviee to a maximum penetration depth of 0.76m below ground level. Cores obtained at the same locations as the CPTs showed that the lunar soil consisted of well-araded silty sands to sandy silts. Grain shapes ranged from perfectly spherical to extremely irregular, The specific, gravity varied from 2.90 to 3.24. It is important when ‘analysing CPT results on the moon to remember that there is, no atmosphere and thatthe gravity field on the moon is only ‘one sixth that of the earth. Figure 5.99 shows results from four of the tests during the Sect Geen 11 cee Wadena tt pee ieaeeeeeeeeeeaeee™ On Cone resistance (kPa) 9 Op 590, 1000 1500 E Maximum Sol seadable at 5 20 £ —= 540 é 60 Test 2 Cone resistance (kPa) 0500 1000 _ 1500 ) recordable | Penetration (cm) Tost 4 Figure 5.99 CPT profiles in lunar soil (Mitchell and Houston, 1974), Apollo 16 mission, Mitchell and Houston (1974) interpreted the results and concluded the relative density of the top 0,60 m varied between 65 and 95%, Several studies have been made simulating lunar soils by using ground basalt (for example, Goodings and Lin, 1995). 6.7 EXAMPLES OF UNUSUAL BEHAVIOUR Occasionally the results from a CPT/CPTU deviate from ‘what is expected. The reason for the unusual behaviour may ‘be duo to problems with either the equipment ot procedures, ‘or both, Checks on the quality of the data should be made as, outlined in Chapter 3, Alternatively, the reason for the ‘unexpected results may be due to some special aspects of the material being penetrated (as discussed in section 5,7) and related to the physical material behaviour, If unexpected results do occur, it is important to try to understand the reasons for them and the possible implica- tions for the interpretation of the results, "The purpose of this section is to give some examples of ‘unusual CPT/CPTU results, 5.7.1 Limiting negative pore pressures due to cavitation It is well known that if the total pore water pressure in a Soil ‘becomes equal to minus 1 atmosphere (100 kPa), cavitation occurs and it eannot become lower. However, much largerEXAMPLES OF UNUSUAL BEHAVIOUR negative values of Au can be obtained depending on the initial pore water pressure in the soil. For offshore soils, at a given depth below seabed, the ‘maximum negative excess pore water pressure change (Au) that ean occur depends on the water depth. Figure 5.100 shows an example of this phenomenon at the Sleipner field in the North Sea, With the filter position behind the cone (u2) large negative pore pressures relative to hydrostatic can, be observed, A constant value of minus 1,150 MPa= 1150 kPa relative to in situ hydrostatic pore water pressure is ‘measured. The water depth atthe site is 105 m, correspond ing to a backpressure of 1061 kPa; that is, the absolute nogative pore pressure is ~ 100 kPa~ I Atm. By evaluat ing all the three parameters in Figure 5.100 (qe, f,, #) the behaviour can be described as follows ‘© First, a sandy layer of about 0.2 m with zero pore pressure ‘generation, Then a hard sandy/clayey soil to about 3.6m with high cone resistance, Due to dilatance, high negative pore pressures develop until cavitation occurs. ‘The cavitation causes desaturation of the pore water pressure measurement system and when a 2.4m thick clay layer is encountered the pore pressure response is sluggish for about I m before it starts to react properly again. The sluggish response from about 3.6 m to 4.3 mis therefore not representative of in situ conditions but rather ‘caused by the cavitation above. Below 4.3m the pore water pressure response again Cone resistance q, (MPa) 2 aveee Geatle Se tA10 Depth below seabed (mm) 05 Sleeve friction f, (MPa) 5 eae 2 26 Figure 5.100 Results from CPTU in Sleipner cay. 121 appears reasonable until again cavitation occurs at about 78m. ‘This is again followed by sluggish response for a couple of metres before reasonable pore water pressure response is again achieved. ‘The problems with sluggish response after cavitation may possibly be alleviated ifa liquid other than water were to be used to saturate the filter stones. 5.7.2 Negative pore pressure measurement with filter on the cone In heavily overconsolidated clay, pore pressures measured behind the cone (vz) may be negative due to dlatance caused ‘by the change in shear stresses. On the other hand pore water pressures measured on the cone (1) are normally positive due to the predominant effect of changes in total stresses. One example was given in Figure 5.2 (Brent Cross), show ing a typical example from a CPTU in overconsolidated London clay Figure 5.101 shows an example of @ CPTU with fter clement on the cone (ui) where the pore water pressure change is deviating from normal behaviour by showing negative values. Tho reasons for the negative pore pressures with the u; positions are not filly understood, but one plausible explanation might be the frequent occurrence of| stones. When the piezocone hits a stone, the stone is pushed aside and a cavity is formed which creates low pressure ot Excess pore pressure Au,(MPa) 2 2 4 0 1 2 UW122 INTERPRETATION OF CPT/PIEZOCONE DATA. Pore pressure and Gone resistance (MPa) i 5 5 Cone resistance E = 10} Poe. % _[ prossur’ a 15 20 Figure 5.101 Results ftom a CPTU in Cowden clay (from Lunne et a.,1986t), suction recorded as negative on the pore pressure transducer, ‘This site is close to the shoreline where the sea is steadily ‘eroding and exposing the clay till as a cliff. It is therefore possible to get an idea of the distribution of stones in this, deposit as shown on the picture in Figure 5.102. Performing tests at a much slower rate than the normal 20 mm/s might reduce this suction effect by pushing the stones aside more slowly, allowing the clay material to move around the stone and go not creating the void. Figure 5.103 shows that one test performed at a rate of 2 mm/s did give more or less continuous positive pote pressure response throughout the profil 5.7.3 Effect of the weight of rig on shallow test results For many offshore applications the CPT is used extensively to derive soil design parameters in the upper 0.1 to 1.0 m « pipelines to be placed on the sea bottom «pipelines or cables that need to be placed in shallow trenches ‘¢ small structures that are placed on the seabed, ‘There are several problems related to interpretation of shallow CPT/CPTU results, For example the charts devel- oped for relative density interpretation (section 5.5.1.1) are based on calibration chamber tests corresponding to depths larger than 5 m below ground level. Only one important effect will be dealt with here and that js the effect of the weight of the CPT rig on the stresses, below it and hence on the measured cone resistance. The effect is illustrated in Figure 5.104. ‘The additional vertical stress, Ag, caused by one example CPT rig foot- print is plotted versus depth, For a typical case with a normally consolidated sand; Table 5.21 illustrates the importance of making this correction on the interpreted relative density, D,, and fric~ tion angle, 4". The correlations in Figure 5.47b and Figure 5,53 have been used for calculating the values in Table 5.21. "The overprediction caused by the weight of the rig is mainly applicable for cohesionless soils and will vary with, the geometry of the footprint. ‘The larger the opening in the centre and the smaller the ‘contact stress, the less important will this effect be. Further, the overprediction will be more important the looser the sand is. Below 2-3 m, the effect is not significant. For almost all onshore applications, the CPT pushing equipment has a large space between the CPT location and, the weight bearing contact points of the rig. Hence, this, problem is generally limited to offshore applications. Figure 5.102. Exposed clay ill at Cowden,‘THE USE OF NON-STANDARD EQUIPMENT OR PROCEDURES 123 Pore pressure and Sleeve friction Cone resistance (MPa) (MPa) o@_1_2 3 4 5 90 01 02 08 deows rota (Powell ef al, 1995). One of the main reasons for this is ‘that when combined with 10 em* push rods increased penetration can be achieved ~ see also section 2.2.3. ‘¢ In model tests in the laboratory cone penetrometers of | 5 em’ and smaller are used in order to increase the sample to cone diameter ratio and to reduce sample boundary effets (eg. Parkin and Lune, 1982; Baldi et al, 1985). ‘¢ Smaller diameter cone penetrometers have been used to increase sensitivity to detect thin layers (Torstensson, 1975), Figure $.103 Family plots of qe. f; and up fom six CPTUs et Cowden (from Lunne er al, 1986). 5.8 THE USE OF NON-STANDARD EQUIPMENT OR PROCEDURES In general, use of non-standard equipment or procedures should not be encouraged. However, in some cases impor tant additional information may be obtained by deviating from standard procedures. ‘Also it may happen that results from tests using non- standard equipment or procedures become available in a project, In such a case itis important to understand what the deviations from the standard may mean in terms of the test results ‘The purpose of this section is therefore to review the literature and the authors’ project files regarding non-stan- dardized cone penetrometer equipment or testing procedures. ‘The effect of different locations of the filter element for ‘the piezocone is not included in this section. This aspect is covered in section 5.1.6, 5.8.1 Cone size and scale effects Cone penetrometers with diameters differing from the stand- ard 35.7 mm are quite frequently used, Well-known exam- piles are: «# For offshore seabed tests 15 em? cones (dia, = 43.7 mm) dominate in practice (for example, Lunne and Powell, 1992). Also onshore in some countries such as Great Britain, 15 em? cone penetrometers are used extensively is | 3 Weight = 150 KN | Contact ang, =201Pa | Foot print of seabed rig. Vertical stress (kPa) ° 10 20 30 40 Vertical stress below centre point Depth below sea bottom (m) 30 Figure 5.104 Effect of CPT rig loading on CPT results124 ‘Table :21. Example of effect of rig on interpreted ¢" and Dy INTERPRETATION OF CPT/PIEZOCONE DATA. Depth below Measured Corrected seabed 4 oly Aa 4. m WPa kPa kPa kPa os 1322 5 62 883 Lo v2 oD 1249 20 2030 20 64 1766 40 2602 4 34 2098 Uncorrected Cortected %Over values: predietion i doge, De tag 45 34s 30 32 so 10 2 34 30 32 40 8 36 36 30 32 20 4 2 32 30 2 6 0 "gnorng eft of rig loading ‘e Large diameter cone penetrometers have been used to jncrease sensitivity for oblaining accurate measurements in very soft soils (e.g. Muromachi, 1981 ~ Figure 5.105). fe To perform CPTs during centrifuge flights very small diameter penetrometers are required. In the literature penetrometer for centrifuge use have been reported with trose-sectional areas of about 1 om™ (eg, Almeida and Pay, 1985; Tani and Craig, 1995) « For some offshore applications considerable cost savings can be achieved if light equipment can be used (that i, by reducing the siz ofthe vessel and hence the cost), Power find Geise (1995) reported the, development of & light “'Seascout” rig that uses a 1 em* minicone, which ean be used to 5 m penetration and hopefully to 10:m in favour able soils ‘To allow the engineer to judge if the test results from a penetrometer with & non-standard diameter can be used (vithout corrections, a summary of available comparative studies is given below. "A study reported by Lunne (1976) in stiff overeonsoli- dated clays concluded that no significant difference in cone a) ») ° g E 3 84d ~coneroxer 3 Nut gs Batteaing = toed et £ 20 € $ a ad 71m resistance was found between 5. om? and 10 cm* cone pen- ttrometers. Lunne et al, (1986) found that in soft clay and foose to medium dense sand a 1Sem* and 10cm" pene ‘ctrometer for all ptactical purposes gave the same cone esistance provided pore pressure corrections were carried ‘uit as described in section 3.1.1. ‘Powell and Quarterman (1988) showed that when com- paring relatively large penetrometers (up t0 150mm diam ter) to the standard 10 cm? cone, both significant rate and scale effects were evident in stiff heavily overconsolidated clays. Comparison between results of S and 10cm" penet- rometers have also been done in calibration chambers (Appendix C). However, from these types of tests it is difficult to separate any scale effects from chamber bound ary effects (see, for instance, Parkin and Lunne, 1982). ‘De Lima and Tumay (1991) carried out a systematic programme with cone penetrometers with cross-sectional Breas of 127em*, 10em? and 15 em’ in five sites in Louisiana, USA. The soil conditions included sands, sits and clays, Based on a detailed statistical analysis de Lima fand Tumay (1991) found that the cone resistance from the a (KPa) 4020 30 40 60 60 70 80 90 Hydrostatic line Estimated top surface of floating mud Figure 5.105 Example ofarge ameter cone to inerease sensitivity (Muromachi, 198).—_ ‘THE USE OF NON-STANDARD EQUIPMENT OR PROCEDURES 128 1.27 em? cone penetrometer was consistently larger than the reference 10 cm? penetrometer. For practical purposes they concluded that a multiplieation factor of 0.85 could be used to correct the 1.27em* cone resistance to obtain the reference 10 cm? cone resistance, They also concluded that, no significant correction was nevessary for eross-correlating ‘cone resistance and friction ratio of the reference 10 cm? and the 15 om’ penetrometers. However, there are three factors that will influence the conclusion of de Lima and Tumay, related in particular to the difference between the results ofthe small 1.27 em? cone and the 10 om’ standard cone: 1, The fact that g, was used and not g,. Since the area ratio of the small cone was 0.75 and for the standard cone 0.45 this will introduce a difference in the clayey soil with high generated pore pressure, 2. Several of the soils tested were quite layered. The larger come penetrometer needs a thicker layer to reach a steady cone fesistance. Therefore in several layers the small cone may reach a “plateau” of cone resistance while this may not be so with the larger diameter cone penetrometer. 3, The difference in cone diameter will introduce a differ- ence in relative rates of penetration of the two cones. The factors (1) and (2) are likely to be the most important and both will tend to give higher cone resistance for the smaller cone penetrometer. In conclusion it is suggested that in practice cone pen cetrometers ranging in cross section from Sem? to 15 em? vill give very similar corrected cone resistance values in ‘most material. For dimensions outside this range it is recommended that correction should be considered, prefere- bly based on site-specific corelations. For a theoretical elaboration of possible scale effects in sand refer to De Beer (1963). Scale effects can also be important in highly interbedded soils section 5.1.4); 5.8.2 Cone penetrometer geometry ‘The variables that will be covered below are related to those parts of the geometry of the cone penetrometer that influ- ences the measured cone resistance and sleeve friction. Different pore pressure measurement locations are not, included in this section, 5.8.2.1 Length of the cylindrical portion behind the cone included in ge Tests in overconsolidated stiff to very stiff clays in the UK showed that this effect can be significant (Lunne et al, 1986). ‘According to the IRTP (Appendix A) the Tength of the cylindrical pat behind the cone included in the measured should be between 7 and 10mm. In the test programme described by Lunne ef al. (1986) tests were done with cone penetrometers from four different organizations. One of these (DSML) had a cylindrical length of 22 mm while the others all had cylindrical lengths within the “standard” of 7 to 10 mu, Figure 5.106 shows statistically treated data from the Brent Cross and Madingley sites. At Madingley the “non-standard” (DSML) cone resistance is consistently about 10% higher than the “standard” cone resistance, At Brent Cross this difference is about 8-9% or 200 KPa from 5 to 15 m depth. The higher qe measured by the “non-stand- ard” cone penetrometer is believed to be caused by friction acting on the longer cylindrical section, Assuming the fiction of the cylindrical part of the cone to be equal 10 the measured sleove friction, and the fiction ratio to be 5%, a difference in cone resistance of about 10% is to be expected 5.8.2.2 Reduced area behind the cone ‘When the electric cone penetrometer started to be used in the late 1960s (Heijnen, 1973; de Ruiter, 1971), the mechanical cone hacl already been used for several decades (section 1.3). A large amount of empirical data had been collected with the ‘mechanical cone. Classification and design charts that were valid for the mechanical cone could not be directly used for the electric cone due to the potential differences in geometry and also in the measutement system. In particular, Delft Geotechnics in Holland were very concerned about this and carried out a large research programme using electric cone penetrometers with different geometries to find one that gave the same results as the mechanical cone in a range of soil types. From this study Hefjnen (1973) concluded that a 60°, 10 em? cone with a narrowed part of 28 mm diameter and 200 mm length placed directly behind the cone would give results that were fully compatible with the mechanical cone penetrometer. For more details on this comprehensive research reference is made to Heijnen (1973) and Viasblom (1985), Due to the difference in geometry, mechanical cone penetrometer give different results from electric cones. The effect of friction on the inner rods is also important. The mechanical cone penetrometers shown in Figures 1.4 and 1.5 have a reduced diameter behind the cone. In clays the cone resistance from mechanical cones is generally slightly larger than for electric cones, while the reverse is true in sands, Rol (1982) presented the results of a comprehensive study on sand sites in Holland, He found that on average the ratio of cone resistances from mechanical cones (Gen) and electric (g..) is given by the following formula: 085+ 0,00035(D+2) gee (5.72)126 INTERPRETATION OF CPT/PIEZOCONE DATA where D = depth of penetration in metres and qe is in MPa. "The (D+ 2)dce partis introduced to take into account the effect of inner rod friction in the mechanical cone, 1000, 500F (a/)9= 047 (AdP.dn" nop. 775 x05, $0.-007 Z 100 8 50 g 2 gw 2 ead e 8 Botemamn 8 2 Ga” oe 1 Kuthawy and Mayne (1990) reviewed data from 14 sand Cone resistance (MPa) 4 | oem E== Fuge ===" Uc fotand toe Fees eal 5 E a 10 | i #ist, de fe 15 = Brent Coss, London ley | 20 Cone resistance (MPa) 9 H 2 3 4 of ; 7} — == DSH] [= Fug si = 10} S a 15 —t Madingley Gault lay oo Sauteny| J Figure $.106 Comparison of mean and (.C. UK clays (from Lunne eta, 1986). 1 st dev. ing, in two 128 10 20 80 100200 5001000 {gg (Mechanical cone) /P, Figure 5.107 Relationship between q- derived from electrical nd mechanical cones (after Kulhawy and Mayne, 1990). ‘and 10 clay and silt sites (Figure 5.107) and suggested the following relationship: (:)-en(e)" 6.73) 5.8.2.3 Non-standard position and area of friction sleeve ‘According to the International Reference Test Procedure the friction sleeve should be placed immediately behind the cylindrical part of the cone. "The maximum distance due to slots and dirt seals should be 5 mm, Further the sleeve area should be 150 em* and a length of 13.37 em for a cross-sectional area of 10 cm. ‘Studies by Campanella and Robertson (1981) showed that in sand the sleeve friction and hence Ry varies significantly with the distance of the friction sleeve above the cone (Figure 5.108). From the tests done by Campanella and Robertson the sleeve friction was found to vary significantly to about 35cm or 10 cone diameters behind the cone. This is expected to be valid for other soil types but to varying degrees. Hence there would be some advantages in having the friction sleeve 10 cone diameters ot more behind the cone where mote stable conditions are reached. ‘Virely et al, (1995) reported the design of the Laval piezocone where the fiction sleeve starts about 5 cone diameters behind the cone, The prediction of pile skin friction may represent one situation where it may be advan tagoous to have the sleeve some distance behind the cone. Due to the large vatiation in f, just behind the cone, even small deviations from the standard position and standard ‘area could result in significant differences in f,. However, as discussed in section 3.1.1, pore pressure effects on both ends. of the friction sleeve can make it difficult to single out the effect of sleeve area and position iff is not corrected.‘THE USE OF NON-STANDARD EQUIPMENT OR PROCEDURES 127 oo cy a ol a EU aol. 2 2 5 5 5 = 8 8 5 § || 29 $ 3 a a 4 oo q V aa Y \ 1° Sle’ve friGton f, {kPa) STFS riction ratio (%) Figure 5.108 Sleeve friction and friction ratio along shaft in sand at MeDonald’s Farm (from Campanella and Robertson, 1981). 5.8.2.4 Cone apex angle For research purposes it may be of interest to use cone penetrometers with different cone apex angles. For instance in clays, the work of Baligh and Levadoux (1986) indicated that whereas the soil distortions due to blunt 60° cones involved large deformations concentrated in the vicinity of the cone, the mechanism of penetration by a sharp 18° cone was quite different and consisted of cutting instead of pushing the soil apart. In order to shed more light on the deformation mecha- nisms involved in penetration in clay, Silvestri and Fahmy (1995) carried out laboratory tests with a range of cone penetrometers with apex angles from 7.5° to 180°. They found that for apex angles less than 35° the net cone resistance was almost constant, For higher apex angles the net cone resistance increased linearly as a function of the cotangent of half the apex angle. Studies on the effect of ‘cone apex angle in sand were carried out by Durgunoglu and Mitchell (1975). 5.8.3 Rate of penetration ‘Tho International Reference Test Procedure (IRTP) and ‘most national standards or guidelines (Chapter 4) require that the rate of penetration should be constant at 20 mm/s, ‘with quite a narrow tolerance (typically * 5 mm/s). For fully undrained and fully drained penetration rate effects will not bbe of concern when the standard penetration rate is, ‘maintained. However, much has been written on rate effects and in some cases additional understanding about soil behaviour can be obtained by deviating from the standard rate of penetration, Examples where rate effects may be of concern inchude (but are not limited to}: © Correlations between penetration tests and laboratory tests the time to failure is normally much shorter in the CPTICPTU than in, for instance, a triaxial test. Rate effects can therefore significantly influence parameters such as the Ny factors (section 5.5.2.1) ‘When using cone penettometers of non-standard diameter the mite should ideally be adjusted to account for differ ences in the time to failure (rate of strain). «By varying the rte of penetration it may be possible to obtain an insight into soil behaviour for undrained, partly drained and completely drained conditions for a particular soil. This may be important for intermediate soils, such as silts (ee section 5.6.1). ‘© The loading situation in a practical design ease may require knowledge of rate effect on undrained shear strength. For instance, for an offshore platform, earth- quake induced loads will be atleast 10 times faster than ‘wave induced loads. Cone penetration tests with different rates may assist in evaluating this rate effet, In the following a summary is provided on CPT/CPTU rate effects based mainly on a comprehensive literature survey bby Danziger and Lunne (1997), which is summarized in ‘Table 5.22, From the table it can be observed thet a range of soils has been tested with rates of penetration within the range 0059-3210 mm/s. However, most of the tests were performed wit rates slower than and up to the standard rate of 20 mm/s, Most of the tests were aimed at studying the8 oa of Ng a 02 Cone resistance (MPa) oa 1 10 Rate of penetration (mm/s) 100 200 Figure 5.109 Influence of rato of penetration on cone resistance (Bemben and Myees, 1974), effect of rate of penetration on qe and only in a few cases, hhave measurements of w and f, also been reported. For clays, most of the CPT rate effect studies have shown fan inerease in q. with an increasing rate of penetration. However, two studies deserve special attention, since they ‘cover many different rates over a wide variation. Bemben and Myers (1974) performed tests in a lightly overconsolidated varved clay using nine rates of penetration between 0.2 and 200 mm/s; Figure 5.109 shows, ‘4 summary of results from one depth. A minimum cone resistance was obtained at a rate of 2 mm/s, Bemben and Myers attributed the shape of the curve to a combination of viscosity and pore pressure drainage effects. They argued that up to about 0.5 mas drained conditions apply and that, above about $0 mm/s undrained conditions apply. Roy et al, (1982) performed tests in a sensitive, soft, slightly overconsolidated silty clay using seven penetration rates varying between 0.5 to 40 mm/sec. As can be seen from Figure 5,110, their tests showed a very similar trend to that found by Bemben and Myers (1974). The scatter in the test results are mostly due to the fact that cone penetrometers, with different tip shapes were used. A clear minimum can be observed in the range 1~2.5 mms, : ea tad ei) q; ble : 2g 4h 8 8 4 Rate of penetration (cm/min) Figure $110 Influence of rate of penetration on cone resistance in avarved clay (Roy et al, 1982a). INTERPRETATION OF CPT/PIEZOCONE DATA When inspecting results of other CPT rate effects studies more closely it is possible to obtain an indication of a ‘minimum cone resistance (Campanella ef al., 1982; Powell ‘and Quarterman, 1988) even if itis much less pronounced than that found by Bemben and Myers (1974) and Roy et al (1982a), The reason that other studies have not clearly, shown this effect may be due to the limited range of rates seleoted, However, it seems reasonable that the shape of the ‘curve found by Bemben and Myers and Roy et al. represents ‘a gonoral behaviour for clays. For very slow rates of penetration q. is predominantly of a drained nature. As the rate of penetration increases, qe decreases due to the docrease in effective stress and reduction in strength. As the penetration rate increases, the viscous forces offset the strength reduction and the curve will pass through mini- ‘mum, Then viscous forces will tend to dominate the process and q. will increase, Based on reported studies included in Table 5.22, Danziger and Lunne (1997) hypothesized that the minimum g. would occur at different rates for different types of clay depending on variations in OCR, Jp, clay content, water content and other material characteristics. ‘Most testing in sands indicates that for rates a litte slower than a penetration rate of 20 mm/s there is litle effect on the ‘cone resistance (Table 5.22). For faster rates of penetration fan increase in cone resistance may occur as a result of dilatancy and higher negative pore water pressures, 5.8.4 Setup tests In order to check how much pile- or skir/side friction will increase due to stop in penetration socalled set-up tests can te done. They ean be cartied out in conjunction with dissipation tests. Figure 5.111 shows the ratio ofthe sleeve friction ater a setup test to the sleeve ftition before the start of set up ina Bea. ="3.5} 30) S25 8 29 5 215 = 10} Z os 2 L @ 0.97005 2000 3000 4000 5000 6000 Delay (sec.) © Moasured ratio sloove friction alter and before dissipation * Ratio calculated on basis of measured sleeve friction after dissipation and average sleeve friction of seabed CPTUs. Figure 5.111 Set-up effect on friction sleeve as a funtion of time for a North Sea clay (froth Norwegian Geotechnical Insite, 1992), ee‘Table $.22 Summary of data related tothe influence of rate of penetration on cone behaviour ‘THE USE OF NON-STANDARD EQUIPMENT OR PROCEDURES 129 Cone Author penetrometer Jéxéquel (1969) mechanical and electical | Ladanyi and eetrical Eden (1969) Bemben and mechanical Myers (1974) Marsiend electrical agmy | Dayal and Allen electrical as) Vivatrat clectrical | 1978) Bruzzi and cletzical | (Cestar (1982) 1 Campanella _piezocone etal. (1982, 1983) ‘Measured perameter oa % ra Goh a % ohn Insitul laboratory insitu laboratory insite insite laboratory insite insite Type of so clay, silt, sand sensitive clay varved olay Assured clay pottery clay (clayey silt) and siliea-70 sand NC Boston Blue and EABPL clays NCsilty clay clayey silt Penetration rte nmis) 220 0.059-2.1 02-200 0-20 13-814 1-200 3-20 025-20 Main conclsions ala enn value) ee, cone 135-145 for soft clays 0.81 for sl clay, 0.8 fr sil, .08 for loose non-saturated sands, 0.71.10 for loose satura sends, 053 for cemented éalearous loam, Mee. cons 1.14-1.26 for sot clays, 1.01 for slay, 081 fori ‘A tend inezease in penetration rate resulted in an inerease of about 7.5% in Both drained and undrained conditions can be obtained with rates from 0.5 to $0 mm/s, The standard rate of 20 ms did not provide a fully undrained condition ‘ge inereases with rate of| penetration; q.!4e01 = 1.3 For the clay the inerease in penetration rte causes an Increase in g anf; g/ge.s and Saif ean be represented by og m/.3, the constant ofthe relations defined asthe soil viscosity coeticient, Ky. Kz is higher for the fiction sleeve than forthe cone. A sudden increase in sleeve fiction values was lbserved fr ates greater than 140 mmis. Ki increases as tho strength of the soil decreases, For the san ate effet ong, and ‘were insignificant ge inereases by 10% fora tenfold increase in rate of penetration Values of gas were higher than aaa for most ofthe profile Penetration is essentially ‘undrained dowa to a penetration rate of mv. As penetration rate decreases below this rate w decreases, while q, and f: ‘ncreases; the increase is greater inf. Rfective cone resistance is130 INTERPRETATION OF CPT/PIEZOCONE DATA Table $22 Continued Cone Measured Insitw Penetration rate Main ‘Author penetrometer parameter laboratory ‘Type ofsoil——_(ammis) conclusions Te Kamp (1982) electical desk laboratory’ dense fine sands 0,033-100 In situs there isa tendency for insine ‘decreasing penetration rae to be associated with decreasing qe and Sisdaaa™ 08-09) qe, possibly «due tothe effect of dilatancy. No ‘obvious differences between ‘onshore and offshore tests were found. Laboratory: possible effect of penetration rate was not proven Lacasse and —_—piezocone eo nin clays 2100 ga~(09-0.95) gen for both ‘Lune (1982) clays testeds qe ™ deo for the Onsoy clay, quo ™ (1-2-1.5) dann for the Drammen clay; 2 wo for both clays; ujo~*(1.02-1.1) tio for Drammen clay Rocha Filho piezocone laboratory fine silty sand 20, static test qaqnie/4e increases a8 d.20 (1982) increases, being ~ 1 for gao~ 5.5 MPa, Regarding rate effect, mobilization of viseous resistance predominates on Loose to medium leposits, while eduction of effective stresses is likely 10 ‘cur in medium to dense soils Roy eral piezocone des Insitu—sensitiveclay ——0.5-40 gap. is a minimum fora rate of (19828 & b) tlbout 2.5 mm/s and inereases when the rate increases up to 20 mmis ot deercases below 1,0 mms, Variations are not linear, Penetration rate has no inflaenee on the induced u ‘measured on the cone but a small cffeot may exist when wis ‘measured behind the cone and along the shaft ‘Almeida and electrical and = gesrstt ‘laboratory Kaolin,clay—0.2-20 For the kaolin: for OCR = 1, de Parry (1985) __piezocone centrifuge increases withthe rate within the range 6-20 mm/s, and itis nearly ‘constant within the range 1-6 mms, The penetration rate has vory Tite influence at OCR = 3 and 10. For the clay: for OCR = 1, 4. {aad q) and increase with the rate within the range tested, (0.2-4 mum's, but w seems to be mote affected than 4. (and 4) by the penetration rte. For OCR = 1.9 and 7, ge (and g,) and tare not much influenced by variations in penetration rate betwoen 0.1 and 20 mis‘THE USE OF NON-STANDARD EQUIPMENT OR PROCEDURES 131 Cone Measured Insitul Penetration rate Main Author penetrometer parameter laboratory Type ofsoil (mm/s) conclusions Roch Filho and piezocone a insitu clay 05-15 uw was not affected by penetration ‘Alencar (1985) rate in the range 5-15 mms, but presented a significant reduction forthe rate of 0. minis. (qs — uy) 4, increased 50-65% when the rate deereased from 15 t0 05 mms Lunne ea piezocone des fin insitw glacial ti 220 AtCowden, w measured on the 1985) (Cowden), ‘cone was negative along 10:m highly OC clay with th rate of 20 mvs, but (Brent Cross) positive withthe rate of 2 mm/s ‘measured just behind the cone was negative with both rates at ‘Cowden and at Brent Cross. At Cowden gez = (0.85-0,90) qex0 in the upper 10 m; i= (0.8-0.9) Foo for 2types of plezocones, but this was not clear for another pierocone Konrad (1987) piczocone dfn tt insitu clay 5-20 measured 4D above the cone base showed no significant change within the range of rates used May (1987) piezocone dou laboratory NC kaolin 25-2210 Approximately constant values of ‘Nii the range 25-238 mms, an increase of about 40% of NB for 321 mils. Approximately constant Nf. forthe whole range of ates Powell and clectrical Gord insitw glacial tl, OC 0167-20 Rate effects are evident, ‘Quarterman lays controlled by ground conditions, (1988) eg. permeability und Comparing results from CPT and ‘other in situ and lab, tests, it has ‘been found that rate effects alone ‘ean account for a significant proportion of the difference ‘between theoretical and deduced values of Np Juran and piezocone Ge insitu clay, sand 2-100 Penetration rate was found to ‘Tumay (1989) have no appreciable effect on ge, ‘but it did on , mainly in the ease ‘of tho sand. In this case approaches ua, whereas to9 reached 4p measured cone resistane; 4, = conested cone resistance taking into account pore pressure effect onthe cone grooves; f= measured sleve fiction; comosted sleeve fiction taking into account pore proseue effect onthe cone grooves = measured pore pressure; y= hydrostatic pore water ssi dan uf sa Coe resistance, eve riton, and pre pressure, respectively, measire athe rate of penetation equal nem; Ny~ (un cone fcr Nf — cane fate, coreting for radial sires inside the tanks Na = Pre pressure fctor, NB, = pore pressure fitor,comecting for adil stresses inside the tank.132 lightly overconsolidated North Sea clay (NGI, 1992). ‘Although there is quite a bit of scatter there isa clear trend of increasing sleeve friction ratio with time. Michi and Matsumoto (1995) measured sleeve fi after dissipation tests and compared this to measured friction, of test piles after set-up. Their results indicated that the difference in f, before and after dissipation tests could be used to evaluate pile set-up effects on side friction, 5.8.5 Applying water during penetration Bayne and Tjelta (1987) and Yagi et al. (1988) have reported the design of cone penetrometers where water could be injected into the soil through channels in the penetrometer tip. The cone design described by Bayne and ‘Tjelta was intended to be used for investigation on how much skirt penetration in a dense dilatant sand can be reduced by water injection. High negative pore pressures and increased effec tive strosses, resulting in high cone resistance, may be neutralized by adding water during penetration. This cone penetrometer was not used in practice. Yagi er al. (1988) used their cone penetrometer to carry out tests at various ‘effective stresses. They controlled this by adding water or air ‘through the tip during cone penetration The potential exists for the evaluation of the collapsibility of soils due to wetting using a cone penetrometer that can inject water into the soil during penetration. 5 6 Vibratory cone penetrometer ‘Tokimatsu (1988) described the work of Sasaki etal, (1985) related to development and use of a vibratory cone pen- trometer for liquefaction analysis. The penetrometer is, shown in Figure 5.112 and is similar to a static cone penetrometer. The diameter is 44 mm to allow for a built-in, vibrator, The centrifugal force and frequency of a typical vibrator are 80 kgf and 200 Hz. The cone resistances with and without vibration, gay and ge, ate measured during penetration, Ifthe soil is loose, the penetration resistance with vibration will be considerably lower than the static resistance due to the induced excess, pore water pressures, For dense soils in which no excess pore pressure is generated by vibration, both resistances will, be similar, For clayey soils that are not sensitive, measured resistances are also expected to be similar since drainage conditions will be the same for both tests. ‘Comparing resistances with and without vibration can ‘therefore be used as indicators of liquefaction susceptibility. Tokimatsu (1988) defined the dynamic-static ratio of cone penetration ratio as, et Ros= 6.74) Ge Figure 5.113 shows typical results of vibratory cone tests INTERPRETATION OF CPT/PIEZOCONE DATA Th al @ Porous metal 1. @ Pore pressure transducer } @ Load transducer for li cone rsitance S¥ Sep © © Take-out cabo for ransducor TAA Oversee ky @ Power source cabo for vibrator j @ Push rod 3| Be a0) Figure .112 Japanese vibratory cone penettometer (after Sasaki eral, 1988) at a site where extensive settlement occurred due to soil liquefaction during an earthquake, and for a site where no apparent settlement was observed. Despite similar character- istics of static penetration resistance for the two sites, the difference in dynamic resistances are significant. The ratio Ras is very much lower for the site that liquefied compared to the site that did not liquefy. Tokimatsu concluded that sands with Rps<0.2 have high liquefaction potential, which he also found to be consistent ‘with results from laboratory tests. ‘Work on vibratory cones for evaluation of liquefaction potential has also been done in Italy (Mitchell, 1988) and the USA (Mayne, 1996). 6,9 STATISTICAL TREATMENT OF DATA Reliability-based techniques were introduced in geotechnical, engineering with the objective of providing a complement to deterministic analyses, The approach aims at accounting for the effect of uncertainties in soil properties and calculation ‘models on the results of geotechnical analyses. ‘An important part ofa reliability analysis is the statistical treatment of the data used to evaluate soil parameters, The statistical estimates quantify a mean value and the uncer tainty in the data. The statistical estimates should be used ‘as a complement to engineering judgement to select the parameters for design. Statistical approaches are especially relevant for appli tions to in sifu test results that generate a lot of data in parallel soundings, such as the CPT and CPTU.STATISTICAL TREATMENT OF DATA 133 Statistical methods belong to either traditional statistics or geostatistical approaches. Different types of statistical Ae OF oy (MPa) ca !Ae approaches that can be applied to CPT/CPTU data are 0 5 40. 48 0 05 1 summarized in Table 5.23. The list is not an exhaustive list J] __ofexstng approaches, but e brief survey of techniques tht are known to have had successful geotechnical applications ‘The application of geostatistical approaches should only bbe carried out after a careful review of the data in terms of quality and geologic evidence. Quality refers to the identi- fication of possible measurement errors, such as calibration errors, zero load (baseline) shifts, saturation problems for fl CPTU, location errors and so on. The geologic evidence is | also important, in that the geostatistical approach is only a) Liquetied site applicable within “homogeneous” layers. It is important to | first identify the main soil or geologic layers before per- | 4, 0F dgy (MPa) acclMg forming geostatistics, However, geostatistics can be used to ° 5 40 ts 0 os 1 Meni layer boundaries. 0 This section looks into the application of statistical tech- niques to the results of cone penetration tests. Examples of| the use of statistical methods for site description, spatial averaging and site investigation strategy are given. Because some of the terminology may be new to the reader, a few definitions are also given, Depth (m) Depth (m) 5.9.1 Definitions él Autocorrelation Distance expressing the rate of decay of bb) Non-liquefied site distance the autocorrelation function, The auto- correlation distance depends on the auto- Figure 5.113 Results form vibratory cone tests (ater Sasaki etal, correlation function used (exponential, 1985), triangular, spherical and so on.) ‘Table 5.23 Statistical tools and their application to geotechnical analysis (after Lacasse and Lamballeri, 1995) density function (e.g. Ang and Tang, 1975) Distibution function obtained from plotting on probability paper or goodness- offi tests Cluster and grouping methods «© Study of local variations ftom distance Use stochastic interpolation (Mlynarck and Lunn, 1987) between data points Goostatitics with stochastic interpolation _@ To do statistical site description © Apply ifenough data are available (Matheron, 1963; Nadim, 1988) 1 Software adapted to geotechnical © Use for any soil property or parameter parameter exists (etso dept, ayer thickness) ¢ Tinport advantage in data presentation ee ie are Ore fe fee er ts tr cise tise tise te ee134 Autocorrelation function Coeflicient of variation Correlation Distribution funetion Histogram, Linear regression ‘Mean Median Monte-Carlo simulation Population Probability density function Probability distribution Random process Random, able Residuals Seale of uctuation Standard deviation INTERPRETATION OF CPT/PIEZOCONE DATA Funetion describing the correlation of the residuals about a trend. Ratio of standard deviation to mean. Mutual dependency between two or more variables, See probability distribution. Graphical presentation of a range of ‘measured or observed values and of how frequently these values occur (also called frequency diagram). Linear relation between two random vari- ables expressed in terms of mean and variance of one random variable as a function of the other variable. Measure of the most likely value of a random variable (also called average). Value of a variable at which values above and below it are equally probable. ‘Numerical model of an equation based on a set of assumptions and yielding a ‘measure of response. Monte-Carlo sim- ulation involves repetitive calculations using a different set of a values for each random variables contained in the equa- tion simulated. Set of data points considered, Probability distribution for a continuous random variable. Law for describing the probability asso- ciated with each of the values of a random variable. Process associated with the numerical out- come of a random vatiable(s). Variable which exhibits scatter or dispet- sion and which value cannot be predicted with certainty. Algebraic measure of distance between the value of a data point (pair in two dimensions) and the value of the trend at tho same location, Distance within which 2 soil property shows relatively strong correlation between two values of a random variable. ‘Measure of dispersion or variability of a random variable and of the closeness of the values; the standard deviation is the square root of the variance, Trend Direction or tendency of a pair of vati- ables (often the slope ofa function. Variance Measure of the dispersion ot variability of a random variable and the closeness of the values; dispersion is taken with respect to the mean value. 5.9.2 Sources of uncertainty and variability of soll properties Two types of uncertainties affect soil property within a geologic layer: 1. Aleatory (or inherent) uncertainty, which represents the natural randomness of a property. For example, the variation in the ocean wave height and the variation in a soil property in the horizontal direction are aleatory uncertainties, This type of uncertainty cannot be reduced. 2, Epistemic uncertainty, which represents the uncertainty due to lack of knowledge for a given property. Measure- ‘ment uncertainty and mode! uncertainty are epistemic "uncertainties. This type of uncertainty can be reduced, by for example increasing the number of tests, improving the measurement method or evaluating a calculation procedure with model tests. For any extensive volume of a natural soil layer, the charac- teristics Muctuate spatially. The soil properties are then ‘considered to be controlled by a random process, with a random pattem of variation and at times an overall trend, The variability may be small or large. There is @ greater tendency for the properties tobe similar in value at closely neighbouring points than at widely spaced points. Soil properties are expected to show dependence both laterally and with depth. However, in the vertical direction, soil density and strength for example depend on the effective overburden stress. The spatial variability of soil properties in situ within a given geologic layer is often modelled by trend surfaces and ‘Table 5.24 Short-cut estimates from range of values in popilation (Snedecor and Cochran, 1964) Number Weighting Number Weighting ‘of points factor of points factor 1 un 0315 A 0.886 2 0307 3 0.591 3 0300 4 0.486 4 0.294 5 0.430 18 0.288 6 0395 16 0.283 7 0370 0 0279 8 0351 18 0275 9 0337 19 0271 10 0325 20 0.268Y residual variations about the trend, With the advent of rapid ‘computers, these aspects can now be routinely accounted for. 5.9.3 Statistical treatment ‘Short-cut estimates ‘The mean is obtained by averaging all available data. For an apptoximate estimate of the variance with only a limited number of data, one can use the short-cut estimates sug- gested by Snedecor and Cochran (1964), The standard deviation is obtained by multiplying the range of the avail- able values by @ weighting factor as shown in Table 5.24, ‘The range is defined as the difference between the largest and smallest values in the data population. For example, for five data points, the standard deviation would be the range multiplied by 0.43. The approach is a good estimator for symmetric data populations. ‘Mean, variance, histogram and probability density function Conventional statistics include developing @ histogram and obtaining mean and coefficient of variation. An idealized, example for cone penetration resistance is shown in Figure 5.114. When enough data are available, the probabil- ity distribution function should be established. In geo- technies the normal and lognormal distributions are the most, commonly used, It is often the relation between two parameters which is of relevance fora calculation, for example correlation between, corrected cone penetration resistance and undrained shear strength from triaxial compression tests via a cone factor, or Variation of undruined shear strength with depth. For random variables, rogression analysis is used to establish the statistics ofthe relationship. Ang and Tang (1975) described, in detail the mechanics of linear regression analysis. Exam- ples are shown in Figure 5.115 for constant and non-constant, vvarianee with depth. The goodness of fit is given by how close the correlation coefficient (r) isto unity, Lognorma rem nan 60 MPa Si = Sadar elton Frequency of occurrence Cone penetration resistance (MPa) Figure 8.114 Histogram of cone penetration resistance. STATISTICAL TREATMENT OF DATA 135 Corrected cone resistance q, Undrained shear strength s, (triaxial compression) a) Linear regression analysis Undrained shear strength s, (kPa) (triaxial compression) 100 200 Ew N £ £ 0.192 B 20 ‘Standard deviation 3 & 30 b) Constant variance with depth Undrained shear strength s, (kPa) (triaxial conor 0 100 -0 0.045 + 0.65+2 Mean) id £02402 es ‘Standard deviation) & 20 a 20] ©) Non-constant variance with depth Figure 5.118 Linear regression analysis (afer Ang and Tang, 1975). a136 INTERPRETATION OF CPT/PIEZOCONE DATA Friction ratio Cumulative curve Friction ratio Ry (%) la @ rae R, (%) oo 2 4 layering 9 2 40 2 40 2 4 ° es = 7 3 Hy st SOS & . Z N z = & § 10) $ 10 s s 5 () a 18 15 « af i 2) Raw data b) Cumulative cure ction 6) Fltred data) Shortened data e) Smoothed data Figure 5.116 Interpretation of fietion ratio date by filtering and smoothing techniques (Harder and van Blob, 1988) Estimates of cone penetration parameters have also been obtained with filtering and smoothing techniques (Vivatrat, 1978; Harder and von Blob, 1988; Tschuschke ef al., 1993) to obtain an unbiased average representation. Figure 5.116 prosents an example of the approach, as applied to the raw Gata for the friction ratio from cone penetration tests. The example includes a cumulative friction ratio curve [Zg(@)] 10 determine soil layering, then filtering, shortening (where every x measurements are replaced by their mean), and finally smoothing of the data by a moving average. Mortensen ef al. (1991), in a careful study of the correla- tion of cone penetration and field vane test for clay tills, also used a smoothing of the in situ cone resistance curves and ‘obtained a frequency histogram for cone resistance and for the cone factor correlating cone resistance and vane shear strength, The results are shown in Figure 5.117 with mean ‘cone resistance of about 1.7 MPa and mean cone factor, Nis of about 8, A lognormal distribution within the clay layer from 11 m to 23 m appears to fit the two histograms very well. The quality of the correlation can and should be enhanced by setting “quality criteria” (Figure 5.118), for example by rejecting anomalous data points or data that do not fit in the specified criteria, For a cone penetration investigation in clay tills, Mortensen et al, (1991) set the criterion of friction ratio between 1,0 and 6.0% otherwise the data were dis- regarded. To enable a representative calculation of the standard deviation, the number of data points within a depth interval was set to a constant. Autocorrelation ‘The variation of a soil property in space is illustrated in Figure 5.119, as a function of a trend, T(x), and residuals, ‘94) (Vanmareke, 1977, 1983; de Groot and Baecher, 1993). The residuals over a large volume of soil are assumed to have zero mean, The trend function is obtained by regression, analysis; the residuals are correlated, unless the data are very widely spaced. ‘The spatial variation ofa soil property can be modelled as, the sum of a trend component and a residual term; where: Y(x) = Tex) + 6) ¥(x) = measurement at location, X (6.75) Tex) = trend component (x) = residual (deviation about the trend) ‘The deviations about the trend tend to exhibit spatial correlation, ‘The degree of spatial correlation can be expressed through an autocovariance function: Ce) = FEO ~ TO) HD) = TW) Flee] "where ris the vector of separation distance between points i andj, [is the expectation operator, Y() is the data taken at location i, and 7() is the value ofthe trend at location i “The normalized form of the autocovariance function is known as the auto-correlation function: ow) = oy ‘where C(0) is the autocovariance function at r= 0 6.16) The properties of an extensive volume of natural soil layer inevitably fluctuate. Soil properties tend to exhibit a strong spatial correlation structure, which appears as a waviness about the trend. The properties tend to be more similar in value at closely spaced points than at widely spaced point. ‘The larger the width, length or depth over which a parameterSTATISTICAL TREATMENT OF DATA 137 Cone resistance q, (MPa) (Creeomennaie ne 810 120 | 100 T s i | - g Lognormal distibuton| 60] ro | a FH) teorormat eistribtion & | 8 ao) \ 40|— & & of} | jt 20] ptf 7 ° 1 2 3 4 6 6 o 10 OO Cone resistance q, (MPa) Cone factor Ny =( de 6,0)! 8, (ey) Statistics of cone resistance and N,- values Figure 5.117 Statistics of cone resistance and M; value (Mortensen eta, 1991)138 Measured q.-curve ‘Smoothed calculated evalu (each 0.2 m) High peaks due to interbedded gravels, stones or harder strata i Ee Low peats due to > inleedted sot strata ‘ or local pore pressure build-up o2motd ‘Eset proba toa owe" |—__ > Increasing strength (MPa) ° Figure $118 Example of estimated reliability of smoothed 4: ‘Value in clay tll after spurious data is taken out (Mortensen and Sorensen, 1991), is averaged, the more the fluctuations tend to cancel each other in the process of “spatial averaging”. This correlation structure can be important both for improving estimates at ‘unsampled locations and for assessing the reliability of a soil parameter. ‘The degree of spatial correlation among residuals is expressed by an autocovariance function: C(r), where r is the veetor of separation distance between two points. The normalized form of the autocovariance function {C(r/C(O)} is known as the autocorrelation function, where C(O) is the autocovarianee function at a distance r of 0. "The three autocovariance functions best suited for soil properties are shown in Figure 5.120. In the exponential Fhodels, the distance at which the aufocovariance function [C(r)] decays to a value of 1/e (where e is the base of the natural logarithm) is called the autocovariance distance, ro ‘This characteristic length describes the extent of the spatial Ye) ° x Spatial variation of soll property Figure 5.119 Spatial variation of soil propery. INTERPRETATION OF CPT/PIEZOCONE DATA @) eh : r (b) () ; 3 7 eet] all 5 & oS ot Common autocovariance functions: (@) Exponential; (b) Squared Exponential; (c) Spherical Figure 5.120 Common autocovariance funetions: ) exponential (>) ‘squared exponential; (c) spherical correlation. Figure 5.121 illustrates how the autocovariance distance and the variance influence the fluctuation ofa soil property (de Groot and Baccher, 1993), “The uncertainties associated with soil characteristics are generally attributed to two primary sources: (1) inherent {natural soil) variability, and (2) sampling and testing erors (Gdentified as “noise”), The soil variability component and hoise are separated ‘with the use of an autocorrelation fanetion. Nadim (1988) ond Keaveny et a. (1989) derived autocorrelations functions in the vertical and horizontal directions for various sets of in situ and laboratory dats. a ) T &) nF High fo Variance © s = Lowry Figure 5.121 Influence of variance and autocovariance distance (2) 00 variation of sol property eee eeeonet eeeSTATISTICAL TREATMENT OF DATA 139 ‘The autocorrelation is the basic reason for a reduction in the seater in the data (standard deviation) as the averaging dimension decreases, Figure 5.122 gives an example of the autocorrelation structure of cone penetration data at a depth of 9 m in a dense sand. The scale of fluctuation, which is a function of the shape of the autocorrelation function, is directly related to the autocorrelation distance, 7,, and represents the distance within which the soil property shows strong correlation ‘in the example in Figure 5.122, the autocorrelation func- tion had the form: Chr) = 0.986 € 6.77) ‘whore r and 37,5 are distance in metres. The closeness of the factor 0,986 to unity indicates that there was little measure- rent noise in the recorded data, ont Geostatisties Properly accounting for the variability in soil properties when predicting geotechnical performance may reduce 550 to “Rare e 7 =0.986 6 § PT cate. | cor cata have 10 sat craton 505 a 5 le 2 cr ata Ss val £00 = 8 ls e g oe 0,5 Liston 0 50 100 150 200 250 Horizontal distance r (m) Figure 5.122. Autocorrelation structure of 17 cone resistance data points ata depth of 9m in dense sand (Keaveny eta, 1989), 2275 500) 480 400 }¢375 350 ~ 300 250! 017s 200 713, g § 8 es 3 Figure 5.123. Regional site investigation (only cone penetration tests are shown; coordinates are in metres)140 INTERPRETATION OF CPT/PIEZOCONE DATA. substantially the uncertainties in the soil parameters and therefore the uncertainties associated with a design. Unfor- tunately, one is never able to gather enough subsurface data to get an exact picture of the variation of a soil property for fan engineering structure, One must therefore interpolate the soil properties within a large volume, and sometimes one needs {0 extrapolate. Traditional methods of interpolation used in geotechnical ‘engineering give little regard to the uncertainties associated ‘with soil properties and the fact that soil properties tend to exhibit a spatial correlation structure, ‘A stochastic interpolation technique that is well suited for this is the kriging approach (Matheron, 1963). Kriging is @ stochastic interpolation method that can account for the ‘uncertainties associated with the soil properties and mini- mize the vatiance in the observed data. To do the kriging interpolation, one nocds to first identify the spatial structure of the soil characteristics or the autocorrelation function. ‘The analysis method is described by Nadim (1988). Geostatistical analysis is illustrated with an example, In the neighbourhood of a shallow foundation, several cone penetration tests were run at different times in the lifetime of the structure, Figure 5.123 illustrates the locations of all available soundings in the regional investigation, Figure 5.124 presents the points selected for a local analysis very close to the circular foundation, while Figure 5.125 gives ‘examples of some of the results from the cone penetration tests, showing q., versus depth. ‘The soil profile consists of a top sand (710 m) over a weaker, partly laminated elay. In this case, the location of the weaker layer, and the undrained shear strength for the ATT oad green ae NE 440 430 Figure 5.124 Local site investigation (only cone penetration tests are shown; coordinates are in metres). Cone resistance, q, (MPa) o 4 8 0 4 8 0.4 8 Depth (m) oc Figure 5.125 Example of cone penetration test results, weaker material were the determinant factor for the feasi- bility of the foundation, Using minimum values would result in large added cost. Figure 5.126 shows a comparison of the contours of geat a depth of 10m, for the regional and the local data point. ‘Although the scatter in the data is large, which is t0 be expected in a layer where both sand and clay can be intermixed, the kriging mean is significantly higher than what would have been assumed in design at that depth (about 1.5 to 2 MPa). By illustrating the spatial distribution in three-dimensions ofthe measured cone resistance, as done in Figure 5.127 and 5.128, one gets a much better insight into the possible variation in the cone resistance beneath the foundation and the most likely values, "The plots show that there does not seem to be a con tiouous layer with weaker shear strength directly beneath the structure, At this depth, the best fit autocorrelation function hhad the following exponential form: CH) = 10° (5.18) where r and 9.6 are distances in metres. The factor of unity indicates no noise in the measurements, However, the coef ficient of variation for the data was about 50%. Figure 5.129 illustrates the autocorrelation function, Figure 5.128 gives a graphical representation at each metre between 7 and 12 m, It is possible to make statistical analyses at as many depths as required for a problem. Caleulation and plotting are done rapidly on a PC with ‘computer programs such as developed at NGI by Nadim (1988). In this particular ease study, the geostatistical analysis cenabled the designers to adjust the assumed position of theSTATISTICAL TREATMENT OF DATA 141 400 £ 5 | of io Raman | co § 45 [mengcaeur | a 8 8 € @ 8 (a) Regional analysis 40 450 440 OS es 7 = 7 mre Sea Sea) ee i RR ER 8829338 8 (b) Local analysis Figure 5.126 Stochastic intrprotation at 10 m (coordinates in m).Cone resistance (MPa) igure 5.127 Cone penetration resistance (9m) clay layer below the depth that had originally been assumed and to use slightly higher cone penetration resistance values, and therefore higher shear strength in design. In another example, Nadim (1988) applied the kriging technique to obtain the variability of soil parameters and layer thickness from cone penetration tests. Figure 5.130a presents the results of two cone penetration tests at a location in the North Sea in a traditional manner. Figure 5.130b shows the variation of cone penetration resistance at depths, of 7 and 9 m at that same location on the basis ofthe 17 cone penetration tests available at the site pable 5.25. Autocorrelation distance for eone penetration resistance Aato- conrelation Soil Direction distance (m) Reference Offshore Horizontal © 30-—-“Hseg and Tang (1977), soils ‘Tang (1979) (Offshore Horizontal «14-38 Keaveny et al, (1989) sand Silty clay Horizontal S-12_Lacasse and Lamballerie (1995) Cleansand Vertical =—««3-—— Alonzo and Krizek (1975) Mexico clay Vertical «1 Alonzo and Krizek (1975) chy Vertical 1 Vanmareke (1977) Sensitive clay Vertical 2——Chiasson eal, (1995) Silty clay Vertical. = «1—_etcasse and ‘Lamballrie (1995) INTERPRETATION OF CPT/PIEZOCONE DATA Autocorrelation distances Several workers have studied the spatial autocorrelation, structure of CPT cone resistance. A brief review of the srature suggests the autocorrelation distances for cone resistance given in Table 5.25. Cone resistance (MPa) Figure 5.128 Variability of cone resistance between 7 and 12.STATISTICAL TREATMENT OF DATA 143, CPT data Se Autocorrelation function + Oo 3 6 9 12 18 18 21 24 27 90 Distance (m) Figure 5.129 lustation of the autocorrelation funetion at 8m depth Summary ‘The advantage of the kriging interpolation technique is that the geotechnical parameters forthe analysis of a structure are more clearly defined with respect to their mean and variance. On the other hand, to do such characterization a relatively large quantity of data is needed. FFigure $.131 illustrates how one obtains the reduction factor to account for spatial averaging, The reduction factor is obtained from the square root of the autocovariance function. Within the scale of fluctuation, the reduction in variance can vary by a factor as much as 04 to 08. ‘Techniques todo spatial averaging exist and were described by Vanmareke (1977, 1983). They should be applied, when a sufficient quantity of data is available, fo obtain representa tive standard deviations of soll variables. ‘Vanmarcke (1983) demonstrated that the autovariance for ‘most autocorrolations functions used in geotechnical engi- neering could be approximated by a unique curve, which results in a simple relation between reduction factor and distance over which the sol parameter is averaged. 5.9.4 Site Investigation strategy and Bayesian updating techniques Statsties can also be used to optimize soil investigation programs. The optimizing strategy presents decision paths for the sampling, in stu testing and laboratory testing of soil, Initially, the site investigation strategy was solved manually With flow charts. The newer versions of the system are solved on personal computers with a knowledge-based system to make decisions when required, providing recom mendations for optimum sampling, in sit testing, and laboratory testing programs (Haarvik and Lacasse, 1990) ‘The uncertainty in a geotechnical calculation is often related to the uncertain presence of an “anomaly”, for example boulders, soft clay pockets or even a drainage layer a) ‘Sleeve friction (MPa) 028050 07 | — Cone resistance (Mal 1p eet ‘Sleove fiction (MPa) 025020 _ 07% one resistance (MPa Cone resistance at 7m depth Cone resistance at 9m depth ») Figure 5.130 Kriging results for cone resistance (Nadim 1988).144 Statistical approaches can be used to establish the cost effectiveness of additional site investigation to detect such “anomalies”. Statistical Bayesian updating can be used to combine objective measurements to subjective estimates or informa tion from other sources. Such techniques can usefully be applied with cone penetration test results. Problems such as the presence and size of boulders, the presence of gas or weaker layer, the probability of not detecting a drainage ‘Squared exponential a a ingle exponential E 5 Vanmarcke's model g a 3 oa 2 [S = Scale of actuation B o2 L = Averaging dimension < |= Autocorrelation distance ° l o 1 2 8 4 &§ 6 7 Distance ratio 3/L Figure 5.131 Reduction of standard deviation by spatial averaging 3 3 2 & 1.0 T T T : 5 ool. | £ 8 $09; change of detection Bool ——Sergchee | : = 3 3 cor chohoa 6 ey o SS Number of CPTs Probabi INTERPRETATION OF CPT/PIEZOCONE DATA layer or an anomaly, the increased safety margin against prosence of anomaly for each additional cone penetration test, can also be solved via statistical approaches. Figure 5.132 presents an example where the procedure developed by Tang (1987) was used. In this example, having ro drainage layer present at a depth of 35 m affected the estimate of lifetime settlement. A settlement of less than 50 em would mean a reduction in costs by millions. Ifthe probability of no drainage layer at a depth of 35 m was less, than 2%, the settlement would not exceed 50m. With drainage layer detectability for each boring of 30% or 80% ‘and distribution of drainage layer extent as shown on Figure 5,132, one would need three to six additional cone penetra- tion tests to ensure the cost reduction indicated on the figure (Lacasse and Nadim, 1994), 5.9.5 Recommendation Basie statistics represent a useful way to establish mean and variance. If a soil parameter is obtained from a complex calculation with a number of random parameters, one may ‘want to include a Monte-Carlo simulation scheme. In geotechnical problems that involve large soil volumes, the “spatially averaged” properties govern foundation analysis, as local fluctuations average over a large soil volume, One should then use a site description strategy, including (1) identification of the correlation siructure and autocorrelation function of the soil property. with given geologic units, and (2) use of the stochastic, ‘p=Probabilty of no drainage layer 2 2 3 2 (t-p) tla) function f(a) Extent of drainage layer prior probability of | ho drainage layer at 55m Updated probability of 02 4 6 8 10 12 Number of CPTs not detecting drainage layer ure 5.132 Cost reduction with inereased number of boreholes to detect drainage layer (Lacasse and Nadim, 1994),‘SOFTWARE APPLICATION 145 Depth (m): 9.95 Duration: 1705 s Pore pressure dissipation record 500 400 = = soo 4 200l 2 g é 100 es cae rr 0 7000 7000 Time (sec) Figure 5.133 Computer-generate print of disipation test interpolation technique (kriging) to estimate the soil prop- erty atthe location of interest, The added knowledge on the uncertainties in the soil parameters and from the subsequent reliability analyses accounting for these uncertainties should lead to safer designs. Geostatisties provide useful complementary information to help guide on the interpretation of soil profiles, stratig- raphy and horizontal and vertical variability in the soil properties. ‘The analysis will also give an estimate of the noise in the measurements and help discem actual in situ trends from ‘anomalies, The approach is well suited for the interpretation of CPT data since a lot of data points are generated, and hopefully it will become more usual to run a large quantity of profilings ‘The application of geostatistics requires that a large quantity of data be available for generating the autocorrela- tion function and obtaining reliable mean and standard deviation or variance values. Statistical methods can also be used to plan site investiga- tion programmes and for optimizing the location and ‘number of additional tests 5.10 SOFTWARE APPLICATION With the rapid developments in fast microcomputers, many software packages have been developed to collect, process, interpret and present penetrometer CPT data. Most of these developments have been made by the speciality manu- facturers and contractors supplying the equipment andor serviee of eone penetration testing, The software has become a value added component for these companies. Software for data aequisition purposes has been discussed in Chapter 2, section 2.4. This section describes some of the developments to process, interpret and present CPT data Most data acquisition systems collect and process the CPT data so that values of cone penetration resistance (q.), slee friction (/,) and pore pressure (u) can be stored and viewed during the penetration process. The visual presentation of the data is achieved either graphically on computer screens cor on paper in strip charts or printers. Strip charts are used less today than computers and printers. As computers ‘become more reliable, rugged and less expensive this trend will continue. Some data acquisition systems show the complete CPTU profile on the computer monitor. Software is sufficiently sophisticated that the screen will update with automatic seale changes when dense layers are encountered and data moves off serven. During dissipation tests there is software that will show the pore pressure dissipation in real time on the computer monitor. This provides the operator swith better control on when to stop the dissipation tests. A screen print of a dissipation test is shown in Figure 5.13 ‘This software can also estimate the coefficient of consolida- tion based on the time for 50% dissipation and the corrcla- tions given in seetion 5.44.1 Figure 5.134 shows an example of a CPT plotting soft- ‘ware package. This profile is produced immediately after completion of the CPT sounding. The plot shows the meas- ‘ued parameters of cone resistance (q.), sleeve friction (f,) and penetration pore pressure (u) versus depth as well as the calculated values of friction ratio (R). The software also allows the corrected cone resistance (q) to be plotted, if requested (as shown in Figure 5.134). Included in the plot are profiles showing the interpreted soil behaviour type (SBT) based on the charts proposed by Robertson et al. (1986). The software provides a visual representation of the soil type using colour and length of bar. Colour monitors and printers ean be provided in the CPT vehicle to enhance the presentation. Alongside the coloured soil profile the soft- ‘ware produces an automatic description of the soil profile using standard sol type descriptions. The software automat- ically tracks the layers and provides break lines where appropriate. The level of detail can be adjusted based on a minimum layer thickness criterion, Figure 5.135 shows the same profile to include interpreted parameters, such as tundrained shear strength and equivalent SPT values. Other parameters can be estimated using the correlations described in earlier sections and presented in a form similar to that shown in Figures 5.134 and 5.135. Software packages that can produce profiles similar to those shown in Figures 5.134 and 5.135 and interpretation profiles are now becoming more common. Interpretation can also be provided in tabular form, as shown in Figure 5.136. I is now possible for engineers or field personnel to haveINTERPRETATION OF CPT/PIEZOCONE DATA 146 pes puespues pues fs pues AeA posses Aas puegpues As ody ANOINBYEG 11S. Ao / 5 i = , mooitiit o 0 (0°H Jo w) *n ainsseid e104 08 (Ge) cones uowouy ose (ea) § uonouy anaeig 0 oz : (ean) *b eourjsisel 2u0g os 00! os oo n (w) wideq Figure 5.134 Example of computer-generated output of a CPT test.‘SOFTWARE APPLICATION 147 ‘Soil behaviour type SBT = =f OP reer j E | le E | | Ube | 3 2 &c ug E Bee 5 c I g 2 of | FE LP en ob Af MN LL Btn | a ° 9 o eI g 8 (w) wdeq Figure 5.135 Example of computer-generated output of interpreted parameters from a CPT tes.148 complete profiles of CPT results including interpretation of soil behaviour type and estimated geotechnical parameters in the field immediately, after completion of a CPT sound- ing. Some CPT vehicles now come with high-quality printers, cellular phones and fax machines to allow rapid transfer of results to remote offices for evaluation and analyses, This has allowed greater flexibility in site inves~ tigation supervision and control. A senior project engineer can now review the results from a remote location and provide rapid advice regarding future sounding locations and additional testing. This has become very useful on ‘geoenvironmental projects where groundwater and/or soil sampling locations need to be adjusted based on the inter pretation of soil type and stratigraphy from adjacent CPT soundings, INTERPRETATION OF CPT/PIEZOCONE DATA Software developments have also been made in the area of geotechnical analyses using CPT data. Software packages are now available to estimate pile capacity and load seitle- ‘ment response, shallow foundation settlement and sol lique- faction analyses using CPT data, These programs make use of the near continuous CPT data to provide, for example, profiles of pile capacity with depth and liquefaction resist ance with depth, Data are taken directly from the CPT files and design profiles produced. This can significantly enhance the speed of design and allow the engineer greater time for engineering judgement and to carry out sensitivity analyses. Itis clear that developments in software will continue and that these developments will further enhance the flexibility of site investigations and the interpretation and application of in situ testing.DIRECT APPLICATION OF CPT/CPTU RESULTS Chapter 5 provided details of interpretation of CPT and CPTU results in terms of soil parameters which can then be used as input parameters in analysis, An altemnative approach is to apply the in situ test results directly to an ‘engineering problem. A typical example of this approach is the evaluation of pile capacity directly from CPT results ‘without the need for soil parameters. Asa guide, Table 6.1 shows a summary of the applicabil- ity of the CPT for direct design applications. The ratings shown in Table 6.1 have been assigned based on current experience and represent a qualitative evaluation of the confidence level assessed to each design problem and gen- eral soil type. Details of ground conditions and project requirements can influence these ratings. In the following sections a number of direct applications of CPTICPTU results are described. These sections are not Table 6.1 Perceived applicability of the CPT/CPTU for various, rect design problems Pile Bearing Settle- Compaction Lique- design capacity ment control_—_faction Sand 2 120 23 12 12 Clay eae tasty es Intermediate 1-2 23 34 © 23 soils Rolibiity rating: | High; 2 High to moderates 3 Moderate; 4 Moderate to Tow; Law. intended to provide full details of geotechnical design, since this is beyond tho scope of this book. However, they do provide some guidelines on how the CPT can be applied to ‘many engineering applications. A good reference for foun- dation design is the Canadian Foundation Engineering ‘Manual (CFEM, 1993). 6.1 CORRELATIONS WITH SPT The Standard Penetration Test (SPT) is the most commonly- used in situ test in many parts of the world, especially North ‘America. However, despite continued efforts to standardize the SPT procedure there are still problems associated with its repeatability and reliability. However, many geotechnical engineers have developed considerable experience with design methods based on local SPT correlations. When these engineers are first introduced to the CPT they initially prefer to see CPT results in the form of equivalent SPT N values. With this intial introduction of CPT data there is a need for reliable CPT/SPT correlations so that CPT data can be used in existing SPT-based design approaches. There are also situations where SPT data are available in a project and engineers may wish to convert this to equivalent qe values. ‘The factors affecting the SPT test have been reviewed (Skempton, 1986; Robertson, 1988; Decourt, 1989; Clayton, 1995) and correction factors to be applied to measured V values have been identified. The single most important ‘correction factor is the energy delivered to the SPT sampler. ‘This is normally expressed in terms of the rod energy ratio150 DIRECT APPLICATION OF CPT/CPTU RESULTS (ER,), An energy ratio of 60% has generally been accepted as the reference value which represents the approximate historical average SPT energy. The value of BR, (%) delivered by a particular SPT equipment depends primarily ‘on the type of hammer/anvil system and the method of hammer release. Values ofthe correction factor (ER,/60) can vary from 0.3 to 1.6 (that is, 20% 0 100% energy). ‘Additional correction factors are also required for a rod Tength less than 10m, a borehole diameter outside the recommended interval (65-115 tmm) and a sampler without ‘an intemal liner. Sines the SPT N value also varies with the cffective overburden stress level, an overburden stress cor- rection factor is often applied to provide a consistent refer- cence (that i, (Nec). ‘A number of studies have been presented over the years 10 relate the SPT N value to CPT cone resistance g.. Robertson ef al, (1983) reviewed these correlations and presented the relationship shown in Figure 6.1 relating (q./p,)N ratios ‘with the mean grain size, Dso (varying between 0.001 mm to 1 mm), Values of g. are made dimensionless when dividing by the atmospheric pressure p.. It is observed that (./p.)/N increases with increasing grain size, Values of N used by Robertson et al. correspond to an energy ratio of about 60% (that is, Na). The correlation shows an increasing seatter with increasing Dso (particularly for Dso greater than 0.1 mm). Kulhawy and Mayne (1990) provided additional data, as shown in Figure 6.2 and greater seater was noticed. However, this new data did not define the energy ratio and part of the increased scatter is due, in part, to variations in SPT energy. Much of the new data was from Greece (Zervogiannis and Kalteziotis, 1988) where the SPT equip- rent produces a low energy ratio and hence a low ratio of CPT to SPT. Hence, the correlation by Robertson ef al (1983) is recommended to estimate Ney. In other studios (qupa)IN has been correlated to fines, content (FC) as shown in Figure 6.3, There is also scatter in Clay Clayey sits Sandy sit Silty sand Sand 10 -Ssllyclay _ sit Data fron’ 18 sites ° 0.001 ot on 1 ‘Mean patticle size D,a(mm) Figure 6.1 CPT-SPT correlation with grain size (Robertson et a. 1983), ‘Qrevation ans Canon, 188 Senate kana oe BE Semetiowel ot 186 0.001 0.07 oa 1 10 Mean particle size Dzy (mm) ‘Figure 62 CPT-SPP correlation with grain size (Kulhawy and Mayne, 1990), this correlation, but a deerease of (q/p.YIN with increasing fines content is quite clear. “The above correlations eed the soil grain size curve in order to determine the mean grain size Dep (ot the fines content, FC). To convert CPT data to equivalent SPT value, Ds (or FC) can be estimated directly using the soil classification charts presented in Figures 5,7 and 5.8 if no train size curves on adjacent samples are available. The CPT classification charts show a clear trend of increasing fiction ratio with increasing fines content and! decreasing sain size. Robertson ef al. (1986) suggested (qP.YNeo ratios for each soil classification zone shown in Figure 5.7 and illastrated in Table 6.2. "These values provide a reasonable estimate of SPT Noo values from CPT data, For simplicity the above correlations are given in terms of qe. For fine-grained soils the correla- tions should be applied to 4. 10 ‘¢ Jamiolkowski ot 2 Kasim etal, 1886 © Muromachi, 1981 2Chin et al, 1988, Za é 34 Seca (Pa). 425-FE (pee, fonts 80-089) © vidnaas. 2 (nin Aaete 80-089 a a a) Fines content, percent passing No. 200 Sieve FC Figure 63 CPT-SPT correlation with peteent fines content (Kuhawy and Mayne, 1990),Table 6.2 Soil classifications and ratios DEEP FOUNDATIONS 151 ‘Table 6.3 Boundaries of sol behaviour type Zone Soil behaviour type (apaNeo Soil behaviour type Indes, L Zone Soil behaviour type (igure 5.7) igure 58) sensitive ine grained 2 heist 7 Gravely sand ongonie teil 1 131360 2 _Orgaiesils— peas sand 5 gravelly sand to sand 6 Se eee a results, Jefferies and Davies (1993) suggested tha the above The main disadvantage of the above approach is the discontinuous nature of the conversion. Often a soil will have CPT data that crosses different soil behaviour type zones and hence produces discontinuous small changes in predicted SPT Ney values. Jefferies and Davies (1993) combined their soil classification chart (Figure 5.9) and the CPT-SPT correlation shown in Figure 6.1 to provide a continuous variation of (q./p,)/Neo with soil type. This approach has been modified for use with the Robertson etal. (1990) normalized CPT soil classification chart shown in Figure 5,8, The boundaries between soil behaviour type ‘zones (210 7) ean be approximated as concentric circles, and ‘the radius of each cirele can be used as a soil behaviour type index, The soil behaviour type index, J, can then be defined as follows: 1.= (G47 ~ log Q) + (log F, + 1.227)" 6.1) where: Qris the normalized cone penetration resistance and is dimensionless Fis the normalized frietion ratio and is (For definitions, see Figure 5.8) percent. ‘The boundaries of soil behaviour type are then given in terms of the index, J, as shown in Table 6.3. TThe soil behaviour type index does not apply to zones 1, 8 and 9. ‘The soil behaviour type index can then be combined with the CPT-SPT ratios given in Figure 6.1 to give the following, relation: (aeipadiNeo = 8.5 (1 = 1/46) 2) Equations (6.1) and (6.2) are a modified version of those tiven by Jefferies and Davies (1993) since they apply to a different CPT classification chart. These equations form a simple method to estimate the SPT N value from CPT. ‘methods provide a better estimate of the SPT N values than the actual SPT test due to the poor repeatability of the SPT. 6.2 DEEP FOUNDATIONS ‘The determination of pile capacity is one of the carlest applications of CPT data (Chapter 1). The prediction of pite capacity and load setlement response is complicated by the large variety of ple types and installation procedures as well as soil type. The method of pile installation can have a significant influence on pile response. Driven large displace rent ples ean densify granular sol, whereas bored ples can foosen the same soil. Hence, the design of piles has often been empirical in an effort to include these many effects. 6.2.1 Axial capacity The total ultimate pile axial capacity, Qu, consists of two components: end bearing load (or point resistance), Q, and side fiction load (sometimes referred to as shaft or skin fiction), Q,, as follows: Qn = Q,+ Ov 63) In sands, the end bearing, Qs tends to predominate, ‘whereas in soft clays, the side fiction, Q, is usually domi- nant, The end bearing Q, is calculated as the product between the pile end area, y and the unit end bearing 4), ‘The friction load, Q, is the product between the outer pile shaft area, A, by the unit side friction f, Quit = fre + IpAo (6.4) Obviously different f, values are m cent parts of the pile, s0 that in practice the calculation is performed as a summation of small components. For open tended piles some consideration should be made regarding152 whether the pile is plugged or unplugged (de Ruiter and Betingen, 1979), but the procedure is essentially as outlined above. The allowable or design pile load, Quy will then be given by the total ultimate axial capacity divided by a factor ‘of safety. Sometimes separate factors of safety are applied to and O, Many methods have been proposed to predict the axial capacity of single piles. Some ofthese methods use the CPT, while others use parameters derived from other in situ tests or from laboratory tests. Some non-CPT methods, such as the effective stress ff approach (Burland, 1973), have been ‘ety useful in providing insight fora better understanding of pile performance. Randolph and Wroth (1982) related f to the overconsolidation ratio (OCR) and produced tentative design charts, Effective stress concepts may not radically change empirical-based design rules, but can increase con- fidence in these rules and should allow extrapolation to new situations. Robertson ef al. (1988a) used 13 methods to predict the axial capacity of eight full-scale pile load tests. The piles were steel pipe piles driven into deitaic sols. The measured axial capacity, determined by the method of Davisson (1973), ranged from 170 KN to 8000 KN. The result of this exercise showed that the procedures based solely on CPT results gave much better predictions of pile capacity than those that did not use CPT. The method that provided the best predictions was that of Bustamante and Gianeselli (1982), followed by those of de Ruiter and Beringen (1979) and Schmertmann (1978), Other CPT methods also assessod bby Robertson etal. included those of Zhou etal. (1982), Van Micerlo and Keppejan (1952) and Van Impe (1986). ‘Another interesting study of CPT methods was provided by Briaud (1988), who analysed 78 pile load tests in various soils using many techniques including six CPT methods. ‘The CPT data used by Briaud was obtained with a mechan- ical cone. The large majority of piles were steel and conerete driven pies, but also six bored piles were analysed. Ultimate loads varied from 307 to 3890 kN. When load settloment ‘curves did not present a clear ultimate load, the wltimate load ‘was defined as the load corresponding to a settlement of one tenth of the pile diameter plus the elastic compression of the pile, Overall the best method was that of Bustamante and Gianeselli, In common with Robertson et al, (1988), the ‘methods by de Ruiter and Beringen (1979) and by Schmert= ‘mann (1978), were also assessed by Briaud (1988) and performed quite well. Briaud (1988) also evaluated the direct application of the CPT sleeve friction, f, and cone resistance, qo to represent the unit side ftiction, f, and unit end beating, 4p, respectively. Pile capacities were greatly overpredicted by this direct method. ‘Tand and Funegard (1989) also applied CPT plus other methods to predict the axial capacity of 13 pile load tests. The pile load tests were performed on concrete and steel driven piles driven into stiff overconsolidated clays in the Gulf of Mexico. Ultimate loads varied from 670 to 4880 KN is DIRECT APPLICATION OF CPT/CPTU RESULTS and were defined as the maximum load at plunging faifure, according tothe authors’ interpretation. The procedure by de Ruiter and Beringen (1979) provided the best predictions of the eight methods analysed. Some conventional methods ‘were also assessed by Tand and Funegard (1989), but apart from the method of Vijayvergiya and Focht (1972), they ‘generally produced predictions less satisfactory than CPT methods. Sharp et al. (1988) evaluated two CPT methods and throe SPT methods for a database of 28 pile load tests, They concluded that all procedures overpredicted pile capacities (defined by Davisson’s method) but the two CPT methods performed better, the best one again being Bustamante and Gianeselli’s (1982) method, which overpredieted pile capacities by about 30%, a low overprediction compared with the SPT methods. ‘The brief review above shows that CPT methods gen- erally give superior predictions of axial pile capacity com- pared fo most conventional methods. The main reason for this is that the CPT provides a continuous profile of soil response, Almost all CPT methods use reduction factors to measured CPT values. The need for such reduction factors is due to a combination of the following influences: scale effect, rate of loading effects, difference of insertion tech- nique, position of the CPT friction sleeve and differences in horizontal soil displacements. The early work by DeBeer (1963) identified the importance of scale effects. Scale effects have also been discussed in section 5.8.1. Notwith- standing these differences, the CPT is still the test that gives the closest simulation to a pile. Superiority of CPT methods over non-CPT methods have been confirmed in other studios (for instance, O'Neill, 1986). ‘Table 6.4 Bearing eapacty factors, k, 1982) (Bustamante and Gianseli, Factors ky 4¢ Group Group Nature of soil (Pa) To Soft clay and mud <1 04 05 Moderately compact clay 1S 035 045 Silt and loose sand 5 04 05 Compact to stifFelay and compactsilt = >S 0450.55, Soft chal ‘Moderately compact sand and gravel ‘Weathered to fragmented chalk Compact to very compact sand and gravel =5 02 03 S12 04 05 SS 02 04 S12 03 04 {Group plain bored piles; mad bored piles mero piles (groued unde ow presse) eed bored piles; hollow auger bored piles piers bares ‘Group I: cat sarewed pls rion proeas piles; prestressed tabular ies, riven cast piles; jacked meal piles; micropils (small diameter piles, ‘routed under high pressure wih diameter < 250 mm); driven grouted ples {low pressure grouting); driven metal ples; driven rammed piles; jacket umole piles high presure prouted pile of large diameter,‘The two main CPT methods, Bustamante and Gianeselli (1982) and de Ruiter and Beringen (1979), are outlined below. The method by Schmertmann (1978) is similar to but, more complex than the method by de Ruiter and Beringen (1979) and is not described here. Bustamante and Giasenelli (1982) method ‘The method by Bustamante and Gianeselli was based on the analysis of 197 pile load (and extraction) tests with a wide range of foundation and soi types, which may partly explain the good results obtained. The method, also known as the LCPC method, is summarized in Tables 6.4 and 6.5. “The equivalent average cone resistance, der, atthe base of| the pile used to compute the pile unit end bearing, gp, isthe mean q. value measured along two fixed distances a (a= LSD, where D is the pile diameter) above (~a) and below (+a) the pile tip, The authors suggest to calculate dex in three steps, as shown in Figure 6.4. The first step is to calculate qa, the mean g_ between —aand ‘+a, The second step isto eliminate values higher than 1.3q/. along the length ato +a, and the values lower than 0.742, along the length =a, which generates the thick curve shown in Figure 6.4, ‘The third step isto calculate q.q, the mean value of the thick “The ple unit side friction, fs calculated from measured qe values divided by a friction coefficient acec. The pile Table 6.5 Friction coefficient, « (Bustamante and Gianeseli, 1982) Coetiicients, % Nature of sol ary AB Soft clay and mud <1 3090 ‘Moderately compact clay Ios 40 80 Silt and loose sand <5 60 190 Compact to stifelay and compactsilt = >S—— 60.120, Soft chalk 35 100 120 Moderately compact sandandgravel Sto 12100 200 Weathered to fragmented chalk >s BO Compact to very compact sand and gravel >12 150 300 DEEP FOUNDATIONS 153 e 0.7dea4@ <8 —t 13a Foleo > a.| Depth igure 6.4 Calculation of equivalent average cone resistance (Bustamante and Gianesell, 1982). Category ‘Maximum limit off, (MPa) u I 0 Mm A BA B A B AB 0015 0.015 01S O15 0.035 40 80 0085 00385 0.035 0035 0.08 = 0.12 (0.08) (0.08) (0.08) 60 120 0035 0.035 0.035 0.035 0.08 60 120 0035 0035 0.035 0035 008 = 0.20 (0.08) @.08) (0.08) 100 120 0.035 0.035 0.035 0.035 0.08 100 200 008 0035 008 008 0.12 = 0.20 (0.12) (0.08) (0.12) 60 80 012 008 0.12 0.12 O1S = 0.20 (0.15) 12) .15) 150 200 0.12, 008 012 0.12 OS = 0.20 15) @.12) @.15) (Category TA: plain bored ples, mu bord piles; hollow auger bored piles; micoples (growed unde low pressure) as screwed piles piers; bares 1B; eased bored ples; driven cst piles IA; driven precast piles prestressed tabula piles jacket concrete ples. 13: riven metal piles; jcked metal pls. TILA: driven grouted piles; cven rammed piles, 113: high pressure routed piles of large diameter > 250 mm; microples(grouted under high press). ‘Note Maxim lint unit kin etn, j,: racks value apply to cael execution and minimum disturbance of sol due to construction154 unit end bearing, g), is calculated from the calculated ‘equivalent average cone resistance, q.., multiplied by an end bearing coefficient, k. So= qdorcvc 65) p= keMee 66) Maximum f, values are also recommended based on pile and soil type. Only the measured CPT q, is used for the calculation of both side friction and pile end bearing resist- ance. This is considered an advantage by many due to the difficulties associated in interpreting f, in CPT data. ‘A somewhat simplified version of the friction coefficient of Bustamante and Gianeselli is given in the French code of practice (MELT, 1993); @ review is given by Frank and Magnan (1995). De Ruiter and Beringen (1979) method ‘The method devised by de Ruiter and Beringen (1979) is sutnmarized in Table 6.6, Different procedures are used for clays and sands. In elays, the first step is to compute the undrained shear sttength sy from cone resistance q.. Then pile side friction, ‘and end bearing are computed by applying suitable multi- plying factors tos, Regarding sands, de Ruiter and Beringen have found that the pile end bearing is governed by the cone resistance over, a zone of 0.7 to 4 pile diameters below the pile tip. The procedure to compute gp is depicted in Figure 6.5. De Ruiter and Beringen also observed that driven piles in ‘overvonsolidated sands might experience some reduction in capacity due to driving, The recommended corrections, ‘based on a number of pile load tests in overconsolidated sands (de Ruiter and Beringen, 1979) are shown in Figure 6.6, However, in many cases it is difficult to estimate the ‘overconsolidation ratio in sands. In sands with a high ge, some conservatism in this respect is desirable, particularly at, shallow depths where scale effects predominate. The infiv- cence of overconsotidation in sands has led de Ruiter and r Beringen, 1979) Je 6.6 Furopean CPT design method (after de Ruiter and Sand Chay: Unitskin Minimum of: friction, fi = 0.12 MPa fro, by CPT sleeve friction, “where: 17300 (compression) 1 for NC. clay 11400 (tension) 05 for 0.0. clay Unitend Minimum: bearing, gp from Figure 6.6 p= Ne'Se o where: No=9 sy = ge, N,= 151020 DIRECT APPLICATION OF CPT/CPTU RESULTS Cone resistance q, Te tpahabe) sume. ‘asin bot the donrard {eam 0) and upard {fam be) declons, Ue ‘aan logge 2 rath nia ah ae ‘Sengpamb= Compu, ‘eryren tion arent (ean. Depth Heico | ‘hemnniram pah ne ae ‘minim a ae feepamb-omthe inmomp in ly. Figure 6.5 Calculation of average cone resistance (de Ruiter and Beringen, 1979), Beringen to propose a limit value for pile end bearing, qp, of 15 MPa, irrespective of the cone resistance, q.. In normally consolidated sands, g- can be greater than 15 MPa, but dp ‘values greater than 15 MPa have not yet been confirmed by Toad tests (de Ruiter, 1982), Recently Aimeida et al, (1996) reviewed results from 43, pile load tests on driven and jacked steel piles at eight clay sites using CPT data, The sites varied from soft to hard clays and the piles from 0.102 m to 0.812 m in diameter. Most of the tests were carried out in tension to evaluate the side ‘unt value 15 Pa for al cohestnass sale Fre-coaise sang Chey ery gravely coarse ‘sand ona sand wih OCR invargo 2104 Fine gravel and sand ‘nh GOR In renga 8 to10 Limit value of ultimate end-bearing capacity (MPa) S10 15 =<20 28 Calculated value of unit end bearing (MPa) 3085 Figure 6.6 Limiting unit end bearing for overconsotidated sands, (de Ruiter and Beringen, 1979).friction component. The pile weight was ignored and was negligible compared tothe total load. Most of the piles had a length to diameter ratio (L/D) of from 27 to 66, Almeida er 1, (1996) suggested the following method to estimate unit pile resistance: f= (q~ CoV on p= Qi oul (68) ‘Where &; and kz are constants. The constant k, varies with (0, a8 shown on Figure 6.7. Almeida ea. (1996) recommend that the average relationship shown in Figure 67 should be wed to estimate the sdefrtion of driven or jacked steel pipe piles in clay, The average relationship ean be expressed as: hy 12+ 14.9 log((q,~ oon) 9) However, in the ease of low plasticity index (Ip < 20%) and/or low clay content soils it is recommended not to use the above expression but to apply the upper limit shown in Figure 6.7, Alternatively, a general value of f; around 40 for low (4, GNtee and around 45 for stiffer soils can be applied, The same relationship is recommended for coated steel piles. A reduction in k; must be made for L/D > 60 based on the recommendations of Randolph and Murphy (1985). The values suggested by Almeida et al. (1996) for the unit side friction ate similar to those that are obtained using the traditional total stress (a) approach for piles in clay, (Ran- doiph and Wroth, 1982). ‘The recommended value of is as follows: = Nd (6.10) a ae fe — of J: Carenal al L 1 2 4 7 40 “ 100 Average (“7S) =) Figure 6.7 Friction coeflicont, fy for dtiven or jacked stoel pipe piles in clay (Almeide eta, 1996). DEEP FOUNDATIONS 155 ‘where Nj, is the cone factor for undrained shear strength and varies from 10 to 30 (section 5.4.2). Few test results were available to fully evaluate this ky relationship, In general, the recommendations of Almeida ef al. (1996) are less conservative than those by Bustamante and Giane- selli (1982). Iis recommended to use the methods by Bustamante and Gianeselli (1982), de Ruiter and Beringen (1979) and (for clays only) Almeida er al, (1996) and to adopt the lowest value for the evaluation of axial pile capacity from the CPT, If local experience exists, it may be possible to apply only the method that has been shown to provide the best prediction, 6.2.2 Factor of safety In order to obtain the design load, factors of safety are applied to the ultimate load and a deterministic approach is ‘usually adopted to define these values. The selection of an appropriate factor of safety depends on many factors, such as: reliability and sufficiency of the site investigation data, ‘confidence in the method of calculation, previous experience ‘with similar piles in similar soils and’ whether or not pile {oad test results are available. Factors of safety are generally about 2.0, although real values are sometimes greater, as partial factors of safety are sometimes applied during calculations (particularly to soil strengths) before arriving to the ultimate pile capacity. Factors of safety suggested by the two main methods are shown in Table 6,7. Almeida ef al. (1996) did not recom- mend any specific factors of safety for their method, The design of high capacity large diameter bored piles in stiff clay or dense sand can be difficult because it is settlement criteria that usually control, rather than capacity Hence, very high factors of safety are often applied to limit, settlement. 6.2.3 Settlement Although the installation of piles changes the deformation and compressibility characteristics of the soil mass govern- ing the behaviour of single piles under load, this influence usually extends only a few pile diameters below the pile base, Meyerhof (1976) suggested that the total settlement of a group of piles at working load can generally be estimated ‘Table 6.7 Recommended fuctors of safety for axial capacity of piles from CPT. Method Factors of safoty ‘Bustamante and Gianeselli (1982) 20@,) 3.0(01) dde Ruiter and Beringea (1979) 2.0 (static loads) 1.5 (static + storm loads)156 DIRECT APPLICATION OF CPT/CPTU RESULTS assuming an equivalent foundation. Fora group of predom- nantly fiction piles (that is, Q,> Q2), the equivalent foun- dation is assumed to act on the soil at an effective depth of| two thirds of the pile embedment. For a group of piles of| predominately end bearing (that is, Q) > Q,), the equivalent foundation is taken at oF lose tothe base of the piles. The resulting settlement is calculated in a manner similar to that of shallow foundations. Sometimes large capacity piles are installed and used as single ples and the load settlement response of a single pile is required, This response is controlled by the combined behaviour of the side resistance (Q,) and base resistance (Q,). The side resistance is usually developed at a small settlement of about 0.5 per cent of the shaft diameter and generally between 5 to 10mm, In contrast to the side resistance, the base resistance requires much larger move- ‘ments to develop fully, usually bout 10 to 20 per eent ofthe base diameter. Hence, an estimate of the load settlement response of a single pile ean be made by combining the two components of resistance according tothe above guidelines. In this way, a friction pile (that is, Q,> Qs), will show a clear phinging failure ata small settlement of about 0.5% of the pile diameter. On the other hand, an end bearing pile (thats, Qs Q.), will not show a clear plunging faiture until very large settlements have taken place and usually sette- ‘ment criteria govern before failure occurs, In both cases the side ftiction is almost fully mobilized at working loads. Hence, it is often important to correctly define the propor- tions of resistance (Qy/0,). ‘Methods have been developed to estimate the load-trans- fer (2) curves fiom CPT data (Verbrugge, 1988). How- ever, these methods are approximate at best and are strongly influenced by pile installation and soil type. The rocom- mended method to estimate load settlement response for single piles isto follow the general guidelines above regard- ing the development ofeach component of resistance 6.2.4 Skirt penetration resistance Installation of foundations for fixed offshore structures or ‘anchors for floating structures usually entails the penetration of a skirt system below the seabed, Measurements from CPT. cean be used to estimate skirt penetration resistance. Large gravity structures (GS) have large stecl or conerete skirts, that must penetrate into seabed soils. The design of these skirts is often critical to the overall design of the platform. ‘The depth of skict penetration required will vary with the soil conditions and other factors, The main purpose of these skirts is to move the effective foundation level down to stronger soils. But frequently they are also used as closed compartments underneath the platform where grouting can be performed, In sands the skirts also can act as scour protection. ‘An important part of geotechnical analyses for GS instal- lation is to predict the force required to penetrate the skirts into the sea bed, The skirts should penetrate at least to the ddopth presumed in the stability calculations. For steel skirts with thickness of 20-30 mm, Lunne and St. John (1979) compared measured penetration resistance from the installation of seven large gravity platforms in the North Sea with cone resistance, q., measured during soil investigation at the same sites Lunne and St, John found it convenient to correlate both skirt tip resistance, g,, and side friction, q,, to the measured cone resistance, ga, by using two empirical coeflicients, K, and K; defined by the following equations: AD) = KX gD) .u) GAD) = Ky qaD) 6.12) where: AD) = unit skirt tip resistance at depth D gAD) = unitskin friction at depth D q{D) = cone penetration resistance at depth D LLunne and St. John describe in detail how the CPT results should be treated before calculating the skirt penetration resistance at sites with soil layers of varying thickness across the platform base, Using average cone resistance profiles, 30 > 15 6.10 Approximate range of settlement for footings on sand (Burland et a. 1977. ‘The total vertical settlement is given by: L satay (6.17) where: C1 = correction for depth of embedment correction for ereep correction for shape of footing net foundation pressure (same units as q.) yundation pressure (p) minus effective ‘overburden pressure atthe foundation level, ai (Figure 6.11) 1, = strain influence factor Az = thickness of sublayer Equivalent Young's modulus de In this method (Figure 6.11), the sand is divided into a number of layers, , of thickness, Az, down to.a depth below the base of the footings equal to 28 for a square footing and 4B for a strip footing (length of footing, Z.> 10B). A value of qc is assigned to each layer, as illustrated on Figure 6.12. ‘Strain within each layer is taken as /.Ap/E, where E = age. ‘Schmertmann (1978) suggested the following values for the shape factor correction, C3Fora square footing, Cs = 1.25 Fora strip footing (> 10B), C3 = 1.75 In the calculation, the strain distribution diagram is con- siructed such that the peak value of J is obtained from the following: Typ = 0.5 + O41 (Aplo')* (6.18) ‘Where 0! i the effective stress atthe depth ofthe peak value of f. The values of J/F are divided by Cs then summed, and multiplied by Ap, C; and C;, For 1 2B ! 38 rd ° 4 8 2 16 0 02 04 06 08 10 Cone resistance q, (MPa) Strain influence factor |, Figure 6.12. Application of Schertmann Method for settlement of footings on sand (Sehertmann, 1978). ‘Average shear 4 strain,y(%) 25. 20) FForaged NG gande reduce Ebylacer of For ocan NC sand and ls edu Ebva tito 200 (dense) 0. 1 Degree of loading (q/qy,) Figure 6.13 Estimation of equivalent Young's modulus for sand ‘based on degree of loading (Robertson, 1991). Most deep compaction techniques involve cyclie shear strosses in the form of vibration to induce an inerease in soil, density. Vibratory compaction is generally more effective in soil deposits with a friction ratio Jess than 1%, When the fiction ratio exceeds about 1.5% vibratory compaction is, usually not efficient (Massarsch, 1994), These recommends tions apply to average values in a soil deposit. Local seams or thin layers with higher friction ratio values are often of little practical importance for the overall performance of a project and their effect should be carefully evaluated when compaction specifications are prepared. Soils with an initial cone resistance below about 3 MPa can be highly compres- sible (for example, if they have a high shell content) ot contain organic matter, silt and clay. The zone of soil behaviour where vibratory compaction is most applicable is shown on the CPT soil behaviour charts in Figure 6.14. Soils with a high initial cone resistance will not show significant ‘compaction and generally do not need compaction, Its also important to establish the level and variation of the ground- ‘water table before compaction since some compaction meth- ‘ods ate less effective in dry or partially saturated soils, The CPTU provides the required information on groundwater conditions. Often the aim of deep compaction is for one or more of the following: I, Increase bearing capacity (that erease shear strength).GROUND IMPROVEMENT ~ QUALITY CONTROL 161 2, Reduce settlements (that is, increase deformation modulus), 3, Increase resistance fo liquefaction (that is, increase density) Cone resistance in non-cohesive soils is governed by a number of factors including soil density, in situ stresses, stress history and soil compressibility. Changes in shear strength, deformation modulus and density can be docu- smonted with changes in measured cone resistance. The need for deep compaction and the geotechnical conditions will be project specific and itis important that design specifications. fake nccount of these site specific requirements. The required effect of any deep compaction technique can be set directly in terms of measured cone resistance or in terms of “equivalent relative density” using one of the charts given in section 5.5.1. The real relative density may be difficult to assess accurately, but changes due to compaction can be ‘monitored using “equivalent relative density”. The relation- ships between cone resistance and relative density involve a normalization with effective overburden stress, since the ‘cone resistance will grow with increasing overburden stress for a sand of constant relative density ‘A common problem in many deep compaction projects is to specify a minimum value of g. for compaction over a large depth range. This results in a variation of relative density with depth, with the required degree of compaction near to the surface being much higher than at depth. For certain projects, a higher degree of compaction close to the ground surface may be justified. However, this can be very difficult to obtain with certain deep compaction techniques and this decision should be based on engineering judgement related to the project geotechnical requirements. It is gen- erally more logical to specify a minimum normalized value of gu (section 6.5.2). An important aspect of deep compaction, which is not yet fully understood, isthe increase in cone resistance with time after compaction. This time effect has been observed in different granular soils, No measurable change in pore water pressure has been observed and the increase takes place Cone resistance q, (MPa) 458 Friction ratio R, (%) Zone: _ Soil behaviour type or + Overconsolidated or cemented Zone: __ Soil behaviour action 1 Sensitive fine grained oon 1 Sensitive ine grained 2 Organic materal ie 2 Organk sols poate 3 Oly By Meroinaty 3 Clays - clay to silty clay 4 Si clay clay =] compactable 4 Sitmatures = clayey sit 5 Clayey ato ally lay to sity cay & Sandy sito clayey sit 5 Sand rixtures -sity sand 7 Sity eand'o sangy sit to sandy sit & Sandia sity sand 6 Sande~cloan sandto sity sand 9 Sand 7 Gravely sand io sand 19 Gravely sandito sand 8 Vary st sand clayey sand 11 Very stitto line grained” 9 Very sf fine grain 12 _ Sandto clayey sand” * — Ovarconsolidated or cemented igure 6.14 Guideline for soils suitable for vibrocompaction techniques.162 DIRECT APPLICATION OF CPT/CPTU RESULTS without visible ground settlements. Mitchell and Solymar (1984) showed that immediately after compaction a docrease in q. was observed, while with time a significant increase was measured (Figure 6.15). This time effect appears to be more pronounced after deep compaction by blasting and seems to be related to the change in structure and cementation of the sand. Figure 6.16 shows the variation with time of CPT profiles at asite in Vancouver, BC, Canada where dynamic compac- tion was carried out, The profile consists of approximately 6 of clayey silt and sandy silt underlaid by a medium sand. e upper Sm of soil was removed and replaced with dredged river sand prior to the start of dynamie compaction. PTs were carried out before, three days and 28 days after dynamic compaction. The effect of the dynamic compaction is clearly seen in the q. profiles, although there is little evidence of ageing in the native soil. A significant change in the pore water pressure response before and after compac- tion is also to be noted, Although this aspect of change of penetration resistance with time after deep compaction is now appreciated by many engineers (for example, Mesti ef af., 1990; Charlie et al., 1992), methods are not yet available to predict this time effect accurately. The current recommended procedure is to perform field trials at the start of the project by performing CPT at different time intervals after compaction to evaluate the significance of any time effect and to develop site specific corrections. Schmertmann ef al. (1986) suggested thatthe time correction for CPT cone resistance measured five days after dynamic compaction was as high as 1.35. Figure 6.17, from Schmertmann (1991), shows compar- ative cone resistance values after dynamic compaction of a 10 m thick silty sand layer in Florida, ‘Te ratio q/goo isthe ratio of q. at time ¢ 10 qo, measured immediately after compaction. The cone resistance increased significantly for up to about 60 days and also increased with the number. of successive drops per location used for the dynamie compac- tion, Schmertmann reported that as a practical matter the engineers involved used the q- values as an acceptance criterion for the results of the dynamic compaction. The curves in Figure 6.17 were used to correct the qe data obtained shortly after dynamic compaction to the values expected 60 days after compaction; this was then used for acceptance or rejection Chatlie etal. (1992) studied a number of cases where ge ‘was measured with time after compaction, A range of ‘compaction techniques was used in the data they studied. The results ofthe study are shown in Figure 6.18 and shows a lange range of rates of inerease of q. with time. The cases shown in Figure 6.18 were representative ofa wide range of climates and geologie conditions with average temperatures varying from — 10°C (Beaufort Sea) to +27°C (Nigeria) Cone resistance (MPa) 10,20 30 0 10.20 30,9 10.20 30 0.10.20 80,0 10.20 90 0 10.20 30 ne Aa rs) sa 84 1 10 10 10 4 0 20 20 20 2 20h 29 iS & oof A loo) 20 30 29 29 ) AC 40] |40} | 40 se | 4¢ 40} Le [sof [le so fsot = jsot sot || 2" ont {eo ” i 60! ‘60! ‘60! 6 ‘60! ‘60l Before — @days alter 96 days after Before 8 days after 25 days after, Test layout © LEGEND: (B holes - 3.3kg per hole) im im 8 ° A-2, BD December 29 -¥-water table 8 A A3,B-3 1980, March 25 At, Bt Blasting 1978, December 20. + Blast hole December 21 © CP test hole} Figure 6.15 Influence of time aftr basting on cone resistance (after Mitchell and Solymar, 1984),GROUND IMPROVEMENT ~ QUALITY CONTROL 163 Cone resistance (MPa) Pore pressure (m) Shear wave velocity (m/s) O58 1015 20 0-10 0 10 20 08190200300 sk 4 tor Batre toh to | 15 15 L L Figure 6.16 Infuonce of time on penetration resistance after dynamic compaction (after Woeller eta, 1995), (Charlie er al, found that the variation of qe measurements, with time could be fitted to the following empirical Mitchell, 1984) sin expression: (@.) N weeks/(q.) 1 week = 1+ K log N where: number of weeks an empirical constant. Figure 6.19 shows K plotted against temperature indicating that log K is a linear function of average air temperature (in °C). Accortting to Charlie eta this may support the idea of the involvement of chemical reactions as a major cause of| 25 19) cementation bonds between particles (as suggested by chemical reactions generally occur faster at higher temperatures. For projects where deep compaction is recommended to either increase resistance to liquefaction or decrease future settlements for shallow foundations, the seismic CPT can nal data, The shear wave velocity jonal measure of liquefaction resist- ance, especially in silty sands (section 6.5). The shear wave Mee an Sober (1064) Mehl ana secmpcten fsa ome 20 Cone resistance at N weeks : Foriham ot (1261) a. 7° 18 bactng Goo “4 drops| ‘Sonmertmann (1987) 18 5 yumi ovat : oe char wat *2arops| acta 10 10 au ° 2a 60 80 ° 10 100 Days after dynamic compaction Figure 6.17 Normalized increase in q. after dynamic compaction, (efter Schertmann, 1991). Time after disturbance (weeks) Figure 6.18 Normalized cone resistance in saturated sand versus time after disturbance (after Charlie etal, 1992),164 DIRECT APPLICATION OF CPT/CPTU RESULTS a oe 4 OU Mitchell and < 'Sotymar (1984): s a f 8 fs 3 el 3 and E char ta. (1982) oe “0 0 10 20 9 40 60 Temperature (°C) Figure 6.19 Rete of increase of normalized cone resistance versus temperature (efter Charlie ef a, 1992) velocity is also directly related to the small strain shear modulus (G,) and. is therefore a direct measure of the stiffness ofthe soil. Hence, a compaction erterion can also be given in terms of a minimum normalized shear wave velocity (section 6.5). Alternatively, a site specific correls- tion between cone resistance and shear wave velocity can be developed and the compaction requirement given in terms of normalized cone resistance. Ground improvement can also include many other tech- niques, sich as grouting, soil mixing and stone colurnns. The CPT can also be used to evaluate some of these techniques although this will depend on soil conditions and the selected ‘ground improvement method “The CPT has also fourd limited use in monitoring surface soil compaction. Since surface compaction is often eatried out in thin Inyers with frequent quality contro, the CPT has not found extensive application in this area, 6.5 LIQUEFACTION Soil liquefaction is @ major concem for structures con structed with ot on sand and sandy silt, The major earth- quakes of Niigata in 1964 and Kobe in 1995 have illustrated the significance and extent of damage caused by soil lique- faction, Soil liquefaction is also a major design problem for Tange sand structures such as mine tailings impoundments and earth dams, To evaluate the potential for soil liquefaction it is impor- tant to determine the soil stratigraphy and in situ soil state. ‘The CPT is an ideal in situ test to evaluate the potential for soil liquefaction because of its repeatability, reliability, continuous data and cost effectiveness. This section presents fa summary of the application of the CPT to evaluate liquefaction potential ‘Several phenomena are described as soil liquefaction. Hence, a set of definitions for soil liquefaction are given here, 6.5.1 Liquefaction definitions Figure 6.20 shows a summary of the behaviour ofa granular soil loaded in undrained triaxial compression, A soil with an initial state higher than the steady/critical (ultimate) state ‘will strain soften (SS) at large strains, whereas a soil with an initial state lower than the steady/critical (ultimate) state will, strain harden (SH) at large strains. Itis possible to have a soil ‘with an intial state higher than the steady/critical (ultimate) state but close to steady/eritical state. For this soil state the response can show limited strain softening (LSS), but even ‘ually at latge strains the response strain hardens (0 the ultimate state, For some sands very large strains are required SS: Strain softening response SH_ Strain hardening response imited strain softening response qe: Static gravitational shear stress : Ultimate undrained steady state shear strength SS/0S: Steady or Critical state Shear | sey Cotapse Shears Cotas ‘tan stain (58) sysheae sin Lies tala bohening (38) | Peat strat omens q Phase transformation Sean raster em Figure 6.20 Laboratory evidence of flow liquefaction.to reach the ultimate state, and in some cases conventional itiaxial equipment cannot reach these large strains (> 20%). fa soil slope or structure, such as an earth dam or tailings dam, is composed entirely of a strain softening soil and the in situ gravitational shear stresses are larger than the ultimate steady/critical state strength (that is, relatively steep slope consisting of very loose sand), a catastrophic collapse and flow slide can occur if the soil is triggered to sirain soften, Sasitharan ef al. (1993) have shown that the collapse surface is a stato boundary and controls the onset of soil structural collapse. The collapse can be triggered by either cyclic or monotonic undrained loading as well as, certain types of draincd monotonic loading (such as a slow rise in groundwater level). Ifa soil structure is composed entirely of strain hardening soil, undrained collapse cannot occur. Ifa soil structure is composed of strain softening (SS) and strain hardening (SH) soil and the SS soil is triggered to strain soften, a collapse and slide will occur only if, after stress redistribution due to softening of the SS soil, the SH soil cannot support the ‘gravitational shear stresses. The trigger mechanism can be cyelie, such as earthquake loading, or monotonic, such as a rise in groundwater level or a rapid undrained loading, ‘During cyclic undrained loading, almost all granular soils, develop positive pore pressures due to the contractant response of the soil at small strains. If thore is shear stress reversal, the effective stress state can progress to the point of essentially zero effective stress, as illustrated in Figure 6.21. For shear stress reversal to occur, ground conditions are ‘generally level or gently sloping. When a soil element reaches the condition of essentially zero effective stress, the soil has very little stiffness and large deformations can occur during cyclic loading, However, when cyclic loading stops, the deformations essentially stop, except those due to local pore water redistribution. There is evidence to show that pore water redistribution can be important when the sand Mean otectve sess,» ‘Shear strain ‘Shoar sess reversal —e Zoo etoctve sie ‘Shear sta yeas 7 Ketotaaion No.of cies, N Figure 6.21, Laboratory evidence of cyclic liqueftction, LIQUEFACTION 165 deposit is overlain by a layer exhibiting low permeability. If there is no shear stress reversal, the stress state may not reach zero effective stress and cyclic mobility with limited deformations will occur. Based on the above description of soil behaviour in undrained shear, the following definitions of liquefaction are suggested: Flow liquefaction This applies to strain softening soils only. ¢ It requires a strain softening response in undrained load- ing resulting in constant shear stress and effective stress (Figure 6.20), ‘# It requires in situ shear stresses to be greater than the ultimate or minimum undrained shear strength ‘¢ Flow liquefaction can be triggered by either monotor cyclic loading. « For failure of a soil structure to occur, such as a slope, a sufficient volume of material must strain soften, The resulting failure can be a slide or a low depending on the ‘material characteristics and slope geometry. The resulting movements are due to internal causes and can take place after the trigger mechanism occurs. ¢ Itcan occur in any metastable saturated soil, such as very loose granular deposits, very sensitive clays and loess, Gilt) deposits or Cyclic softening Cyclic softening applies to both strain softening and strain hardening soils, Two terms ean be used to define it: eycic liquefaction and eyelic mobility Gyelic liquefaction ¢ This requires undrained eyclic loading during which shear stress reversal occurs or zero shear stress can develop (that is, where in situ static shear stresses are low com- pared to cyclic shear stresses), as illustrated in Figure 621, It also requires sufficient undrained cyclic loading t0 allow effective stress to reach essentially zero. 1 At the point of zero effective stress, no shear stress exists. ‘When shear stress is applied, pore pressure drops but a very soft initial stress-strain response can develop result- ing in large deformations. ‘@ Deformations during cyclic loading can be large as the effective stress passes through the zero condition, but deformations generally stabilize when cyclic loading stops. The resulting movements are due to external causes and only occur during the eyclic loading.166 DIRECT APPLICATION OF CPT/CPTU RESULTS It can occur in almost all. sands provided the cyclic loading is sufficiently large in size and duration, ‘¢ Clays can experience cyclic liquefuction but deformations at zero effective stress are generally small due to the cohesive strength at zeto effective stress. Deformations are often controlled by rate effects (ereep). Cyelic mobility ‘© This requires undrained cyclic loading during which shear stresses are always greater than zero; that is, no shear stress reversal develops. ‘© Zero effective stress will not develop. # Deformation during cyclic loading will stabilize. The resulting movements are due to external causes and only ‘occur during the cyclic loading. It can occur in almost any sand provided the cyclic, loading is sufficiently large in size and duration and no shear stress reversal occurs ‘# Clays can experience cyclic mobility but deformations are usually controlled by rate effects (creep). ‘Note that strain softening soils can also experience cyclic softening (cyclic liquefaction or eyclic mobility) depending (on the ground geometry. Figure 6.22 presents a suggested flow chart forthe evalu- ation of liquefaction according to the above definitions, The fist step is to evaluate the material characteristics in terms of [Material characterisation] ‘Stain soni ‘Strain hardoning behaviour behaviour Flow liquetaction Monotonatyelo ‘Size aa aration ‘goer ‘ot eyele aang T ‘Gravlonalavossoa> | [Sheat areas] [No shone sooo] lundrained shear stengtn] "reversal | |" reversal Contained | [Unsontained] | Oyete Gyo [sstormaton] [detormaton| [iquetacton] | mobilty I T I Potertal er Large Smal progressive Faure | [deformations] | detormatons yet ¥ [Datermations can continue Deformations essontilly ‘afar the Wiggor ovent stop alter cya loading Figure 622 Plow chart for liquefaction (Robertson, 1994), strain softening or strain hardening response. If the soil i strain softening, flow liquefaction is possible ifthe soil can bbe triggered to collapse and if the gravitational shear stresses are larger than the ultimate steady state strength, Whether a slope or sol structure will fail and slide will depend on the amount of strain softening soil relative to strain hardening soil within the structure and on the britleness of the strain softening soil. At high effective stresses some strain soft. ening granular soils appear to become less brite with increasing confining stress. The resulting deformations of a soil structure with both strain softening and strain hardening soils will depend on many factors, such as distribution of soils, geometry of the structure, amount and type of trigger mechanism, brittleness of strain softening soil and drainage conditions. If the soil is strain hardening, flow liquefaction will not occur. However, cyclic softening can occur due to cyclic undrained loading. The amount and extent of deformations during eyclic loading will depend on the size and curation of the cyclic loading and whether shear stress reversal occurs If'shear stress reversal occurs, it is possible for the effective stresses to reach zero and, hence, cyclic liquefaction can take place. At the condition of essentially zero effective stress large deformations can oceur. If shear stress reversal does not take place, it may not be possible to reach the condition of zero effective stress and deformations will in ‘general, be smaller; hence, eyetic mobility will occur. If cyclic liquefaction occurs and drainage paths are restricted due to overlying less permeable layers, the sand near the surface can become looser due to pore water redistribution, resulting in possible subsequent flow liquefaction, 6.5.2 Application of CPT for liquefaction assessment 6.5.2.1 Cyclic softening Most of the existing work on soil liquefaction has been related to cyclic softening and has been primerly for earth- quake loading, The late Prof, H.B. Seed and his co-workers developed a comprehensive methodology to estimate the potential for cyclic softening due to earthquake loading. ‘The methodology requires an estimate ofthe eycic stress ratio (CSR) profile caused by a design earthquake, This is usually done based on a probability of occurrence fora given earthquake. A site specifi seismicity analysis can be earied out t0 determine the design CSR profile with depth. A simplified method to estimate CSR was also developed by Seed! and Idrss (1971) based on the maximum ground surface acceleration (dgg.) at the site, The simplified approach can be summarized as follows: “= 0,65(MWF) oe g crsaverage cyclic shear stress [MW = Magnitude Weighting Factor wuy**73. ‘M= earthquake magnitude, commonly M= 7.5 = maximum horizontal acceleration at ground surface _g= acceleration due to gravity = 9.81 m/s* Oy = total vertical overburden stress ‘ly = effective vertical overburden stress 2 = dopth in meters (forz > 25 m) ‘rg stress reduction factor, typically (10.015 2) Seed etal. (1985) also developed a method to estimate the cyclic resistance ratio (CRR) for a sand with level ground conditions (that is, eyelic softening) based on the Standard Ponetration Tost (SPT). This method was based on extensive field performance data where sites either had or had not experienced cyclic softening (liquefaction) due to earth- quake loading. The CSR from the earthquake can then be compared to the estimated CRR for the soil deposit. Ifthe CSRs greater than the CRR, cyclic softening (liquefaction) is possible. The Seed methodology is the most commonly- used technique in most parts of the world to evaluate soil liquefaction potential due to earthquake loading, ‘Some of the sites in the above (Seed et al., 1985) database almost certainly contain sites with very loose sands that could experience flow liquefaction. However, the database applies only to level ground conditions where flow liquefac- tion generally cannot occur due to the very small statie shear stresses. Robertson and Campanella (1985) proposed a modified chart to estimate the CRR for clean sands and silty sands based on the CPT using normalized penetration resistance. The cone penetration resistance can be normalized as follows: Got = QP MPa)? 6.21) where 4) = normalized cone penetration resistance (Gimensionless) ‘Pa = atmospheric pressure, usually = 100 kPa 04, = vertical effective stress. imilar CPT-based charts wero also developed by Seed and de Alba (1986) and Shibata and Teparaska (1988). A comparison between the different CPT charts is shown in Figure 623, In recent years, there has been an increase in available field performance data, especially for the CPT (Ishihara, LIQUEFACTION 167 1993; Kayen et al 1992; Stark and Olson, 1995; Suzuki et dl, 1995). The recent field performance data have shown thatthe existing CPT-based correlations to estimate CRR are sgzenerally good for elean sands. The recent field performance data show that the correlation between CRR and qa: by Robertson and Campanella (1985) for clean sands provides a 004 estimate of CR. Based on discussions atthe NCEER ‘Workshop (1996), the curve by Robertson and Campanella (1985) has been adjusted slightly to be more consistent with the SPT curve. The resulting recommended CPT correlation for clean sand is shown in Figure 6.24, This curve is considered to represent a probability of oveurrence of cyclic liquefaction of about 20 per cent or less. Occurrence of liquefaction is based on level ground observations of surface manifestations of cyclic liquefaction, For dense sand (qe: > 75) this could involve the development of large pore pressures, but the effective stress may not fully reduce to zor0 and deformations may not be as large as in loose sands Hence, the consequences of “liquefaction” will vary depend- ing on the soil density as well as the size and duration of loading. The field observation data used to compile the curve in Figure 6.24 are basod on the following: Cyclic stress ratio tyay/0y' ° 100 Normalized cone resistance q,; Figure 623 Cyclic liquefaction resistance from CPT. Bee,———— TTT, 168 DIRECT APPLICATION OF CPT/CPTU RESULTS 06 400 a7s foa8 < Dag =] Recommended robs Range based on OPT design. os 1 a0} field observations 2 {Stark'& Olson, 1995) of 8 7 goa ms 60 Ble 4 g s [No fqvotaction| 4 Boa Pesan 00 2 4 5 | ga 2 Soa Sae% 5m of ® 20 ‘Seed and de Alba é “a n (i926) oa a 2 Tig Weck 1 20 Teliperiormance _{LIq- i 10 30 40 50 Stan kOton (985) Fines content FC (%) 1 0-50 100 180 200 260 300 Normalized cone resistance q. igure 6.24 Recent field evidence of eyelic liquefaction with CPT resistance curve for clean sand (Robertson and Fear, 1995). ‘¢ Holocene age, clean sand deposits ‘# Level or gently sloping ground. fe Magnitude M = 7.5 earthquakes. fe Depth range from I to 15 m (84% is for depth < 10). 1s Representative average CPT ge values for the layer that was considered to have experienced cyclic liquefaction. Caution should be exercised when extrapolating the CPT correlation to conditions outside the above range. An impor~ tant feature to recognize is thatthe correlation appears to be based on average values for the liquefied layers. However, the correlation is often applied to all measured CPT values, which include low values below the average. Hence, the correlation can be conservative in variable deposits where a small part of the CPT data could indicate possible liquefaction. ‘Based on data from 180 sites, Stark and Olson (1995) also developed a set of correlations between CRR and ge: for ‘various sandy soils based on fines content, and mean grain Size. The combined database is now larger than the original SPT-based database proposed by Seed etal. (1985). One reason for the continued use of the SPT has been the need to obtain a soil sample to determine the fines content of the soil. However, this has been offset by the unreliability of SPT data. With the increasing interest in the CPT due to its ‘greater reliability, itis now possible to estimate fines content fand grain size from CPT data and incorporate this directly into the evaluation of liquefaction potential, The following js a summary of the approach suggested by Robertson and Feat (1995), based on the CPT. Based mainly on the comprehensive database presented bby Stark and Olson (1995), Figure 6.25 shows the suggested Figure 6.25 Correction for ines content. correction (Ada) to the measured CPT normalized penetra tion resistance (gc) required to obtain an equivalent clean sand normalized penetration resistance (qr based on the train characteristics of fines content (PC) and grain size. The Correction is not constant for all values of penetration resistance, but the suggested correlation can be defined by the following: Age: = 60 Ager =9 (622) Age, = 24FC-5) 5% 0) the soil could be strain softening in undrained shear and flow liquefaction could be possible. Ifthe saturated sand {s loose to medium dense, cyclic softening is possible and will depend on the size and duration of cyclic loading. The range of potential CPT penetration resistance profiles to evaluate these different phenomena are illustrated in Figure 6.30. The profile to estimate the potential for flow liquefac- tion is influenced by the location of the USSL, direction of loading and the sand characteristies such as compressibility ‘The profile to estimate the potential for cyclic softening (cyclic liquefaction) is controlled by the size and duration of the cyclic loading as well as the sand characteristics. The available field performance data for cyclic softening sug ‘gests that sand characteristics have only a minor influence ‘on the correlations, provided they are applied to unaged, ‘uncemented, predominately silica sands, ‘An inereased amount of field performance data is now available to add confidence to the empirical charts based on CPT penetration resistance to estimate the eyclic resistance ratio (CRR) of sandy soils for eyelic softening (Figure 6.24) Existing CPT soil behaviour charts can be used to estimate the fines content and grain size of sandy soils (Figute 6.27) Shear wave velocity from seismic CPT can also be used t0 provide an independent assessment of the potential for cyclic softening ‘The addition of shear wave velocity measurements to the CPT (see section 7.3) can also provide valuable additional independent data to estimate the in situ state of a sand (Chapter 5). ‘The CPT and SCPT provide valuable data for the evalu- ation of liquefaction potential. The CPT provides. the roquired level of detail to evaluate the soil profile and to {identify even relatively thin layers of sand that could liquefy. The pore pressure data provides the required information regarding groundwater conditions as well as additional information to estimate grain size and fines content in sandy deposits. The shear wave velocity and cone penetration resistance provide valuable information to estimate the in situ state of sandy soils, From these it is possible to provide continuous profiles of liquefaction potential in a repeatable and cost effective manner. A detailed description of the evaluation of liquefaction potential is given by Robertson and Fear (1995).ADDITIONAL SENSORS THAT CAN BE INCORPORATED For some projects, the information from a CPT or CPTU ‘must be supplemented by including additional sensors. Moder electronics, sensor technology and data acquisition systems have made this possible, This chapter reviews the developments that have taken place over the last few years. Geoenvironmental sensors are discussed in Chapter 8. 7.4 LATERAL STRESS MEASUREMENTS ‘Many of the theories that exist for evaluating soil strength requite an assessment of the in sift lateral stress, ajo, ot coefficient of earth pressure at rest: Fl cn) Ge Several attempts have been made to get an independent measurement during a penetration test, notably the lateral stress cone and the cone pressuremeter. TAA ‘Several researchers have instrumented the friction sleeve, or an element behind it, in order to measure the lateral stress during cone penetration, Some examples of different designs are shown in Figures 7.1a~T.1d. Figure 7.1a shows the first development at the University of California, Berkeley, UCB, (Huntsman et al., 1986) ‘where the lateral stress on the sleeve was measured via the Equipment hoop strain by strain gauges. To achieve sufficient sensitiv- ity the thickness of the sleeve was reduced at the location ‘where the strain gauges were mounted. This desiga was also used by Jefferies etal. (1987) for tests in the Beaufort Sea in the Arctic Ocean, However, they experienced frequent fail- ures of the friction sleeve at the reduced section. Further development work at UCB by Tseng (1989) resulted in the lateral stress sensing element shown in Figure 7.1b. This design measures the lateral stress in a different manner from that used by Huntsman er al. (1986). The Iateral stress section is 25mm long and consists of a two- ring arrangement: an outer active ring and an inner passive ring, both fabricated from stainless steel. Four identical curved steel pieces 1.3 mm thick are joined by polyure~ thane compound to form the outer flexible ring. The flexible ring is formed over the rigid innet passive ring which contains a strain gauged stainless steel diaphragm. The 63 mm diameter thin-walled diapbragm performs as a pres sure transducer. A sealed rubber membrane isolates the inner and outer rings. The cavity between the membrane and Figure 7.4a First design of UCB fateral stress_cone (from Huntsman etal, 1986)the inne rng is filled with de-aired water. The saturation of| the pressure cavity is vital tothe performance ofthe measur- ing system (Figure 7.15). Two lateral stress sensors were jncorporated into the UCB cone designed by Tseng and used by Masood (1990); one located 1 D and the other 7.5 D behind the cone, where D isthe cone diameter. “ampanella et al, (1990) described a lateral stress piezo- cone designed and built at University of British Columbia (UBC). This comprises two separate measurement systems, a piczocone unit followed by a lateral stress module 0.69 m (15.6 D) behind the cone shoulder (Figure 7.16). The lateral stress sleeve is 88 mm long and 44 mm in diameter (same as tone penetrometer) with @ wall thickness of 3mm. At the centre of the sleeve, a 20 mm section has reduced wall thickness of Imm, An arrangement of strain gauges is oriented at this locaton to measure the hoop stress induced by the lateral stress acting onthe sleeve. Sully (1991) recommended that the UBC lateral stress piezocone should be modified to achieve better sensitivity and data quality, This could be achieved by using a thinner instrumented section, though this will reduce the robust ness of the device. Bayne and Tjelta (1987) described « special cone pen- trometer where the total lateral stross could be measured using a load cell with two active faces set on opposite sides of the element, flush with its face. Each active face is circular with a projected area of 403 mm”, The load cell is a spool-shaped post, which is instrumented with foil gauges. miles Be—— 11" - 12 thread |— Delrin seal ring | Prossure cavity | _ Rubber ‘Active ‘membrane fing |_ Pressure diaphragm Oning Inner ing ( Masood, 1990). LATERAL STRESS MEASUREMENTS Figure 7.1 improved lateral stress cone design at UCB (from 173 Pore pressures are measured with three porous stone dises set at 120° spacing around the cylindrical shaft, Figure 7.1d shows the arrangement of the lateral stress cell, porous elements and also an additional friction sleeve on an element that can be placed at various distances behind the standard friction sleeve. 7.4.2 Typical results Figure 7.2 shows the results of a test in Drammen clay at Lierstranda in Norway, with the Tseng (1989) version ofthe UCB lateral stress cone with the two lateral stress measure- ments at | and 75d, Figure 7.3 shows a result from atest ata (Not to scale) Lys pore pressure |~ tansducer 1 Lateral stress fiction sleeve Lateral stress, Lateral stress, oe Temperature sensor Cone electronics Us pore pressure transducer Plezocone unit Friction gleeve (225 em’) Up pore pressure Uy, pore pressure Figure 7.1e UBC lateral stress cone (ftom Campanella et a, 1990),174 ADDITIONAL SENSORS THAT CAN BE INCORPORATED North Sea site with the lateral stress cone described by Bayne and Tjelta (1987). 7.1.3 Interpretation ‘The insertion of a cone penetrometer into a soil will cause & change in the lateral stress, The lateral stress measured, Ties }.— Sleove friction, fg No. Porous elements Pore pressure, Uy 2No. Stress cells ‘Total lateral streS8 Ope Various connections Lengths from 1 to3 m Figure 71d Lateral stress cono (Bayne and Tjelta, 1987). will be the result of the initial im situ lateral stress, oy, and ‘the change eaused by probe insertion, Aap: ho + Aon 2) For cohesionless soil, Huntsman (1985) showed that cavity expansion theory predicts an increase of lateral stress, in the range 22 to 58 times the pre-penetration lateral stress, Calibration chamber and field tests have been studied to tty to relate an amplification factor, c/o, 10 sand density, or state parameter. Jamiolkowski and Robertson (1988) used data from Jefferies et al, (1987) to establish the following empirical relationship: eg ooargo |= 2] oars ae | 3) Fhe This correlation is showa in Figure 74. It ean clearly be seen that the estimated value of Ke is extremely sensitive to the measurement accuracy of oe This may be the result of the extreme stress relief and disturbance that occurs as elements of sol pass the tp ofthe penetrometer (Hughes and Robertson, 1985). The large stress relief may imply thatthe horizontal siess measurement (04) is not an independent reasurement in elation te Measured lateral stress (kPa) 9-100_200200 400 500 600 _700__ 800 T T T 22 a ool. | Figure 7.2 UCB lateral stress cone profile at Lierstrand, Norway (fro Masood eral, 1990),CONE PRESSUREMETER 175 u, (MPa) uy (MPa) Gp. (MPa) i 208 0 1-2 1 —< 10 ' 20 \ + 130 ya y \ a + 40, Ot 02, 03 f, (MPa) ‘aio'n=200 390 ‘p0 1 2 8 4567810 20 Wigure 7.4 X, from lateral stress cone for Monterey a.c. sand (fiom Jamiotkowski and Robertson, 1988). Figure 7.3 Lateral stress cone profile from Gaullfaks “C” ste North Sea (from Skomedal and Bayne, 1988). Sully (1991) on the other hand found that for the Molipag sand structure in the Beaufort Sea, measured horizontal stresses with the lateral stress cone gave good correlation with self-boring pressuremeter data, Fora range of densities present in the island fil, the amplification factor was found to be unity In clay, Sully and Campanella (1990) assumed that J corresponded to the cylindrical cavity expansion limit pros- sure, p., and that a could be found from: ire [sd + In) a) where J, = soil rigidity index, Gls,. Taking s, ftom vane tests and the shear modulus from pressuremeter data, they computed oq and found good agreement with measure- ments from push-in-spade cells. However, J, values are not normally available, so this method is difficult to use in practice. In conclusion, the lateral stress cone is still nota tool that has found practical use. There remain problems with the instrumentation of this type of device in maintaining a robust cone while obtaining the required sensitivity of the readings. The interpretation also needs to be improved 7.2. CONE PRESSUREMETER 7.2.1 Equipment ‘The concept of mounting a pressuremeter module behind a cone penetrometer was first applied in the early 1980s, Jezequel et al. (1982) developed a pressiopenetrometer for shallow offshore surveys, as shown in Figure 7.5. The device had a diameter of 89 mm and was installed using a vibrating hammer device. Robertson ef al. (1984) placed a| 176 ADDITIONAL SENSORS THAT CAN BE INCORPORATED 60° solid cone onto the base of a75 mm diameter self-boring. pressuremeter. Because of the large diameter of these devi- ces, special equipment was required for their installation, ‘The pressuremeter data proved to be successful for the design of laterally loaded piles. ‘The early dovices described above have been superseded by devices of smaller diameter known as cone pressure meters, which consist of a pressuremeter module mounted behind a standard electrical cone penetrometer. These devi- ces can be installed by standard CPT jacking equipment (cone truck or seabed frame) and enable pressuremeter tests to be performed as part of the CPT operations. ‘The frst cone pressuremeter (CPM), and the one for which the most results have been published, was designed and built by Cambridge In Situ, originally to a specification of Fugro Netherlands. This device is described by Withers ef al. (1986) and is shown in Figure 7.6. The equipment has been slightly modified since to improve the operation. ‘The pressuremeter module is of 43.7 mm diameter with a length/diameter ratio of 10 and it is operated behind a standard. 15 em? CPT or CPTU. The pressuremeter cell comprises a cylindrical rubber membrane inflated by nitro- gen gas. Protection of this membrane during insertion is provided by an additional steel reinforced rubber membrane inthe form ofa “chinese lantern”. Measurements of inflation | Rod @ 114 mm |___ Pressuremeter colt 9.89 mm, lengthio = 4 + im Plezometer coll (porous window 125 mm*) [=] ~|— Friction sieeve (area 581.5 om) The LPC pressiopeneirometer (after Jezequel et a, pressure and cavity strain are recorded at mid-height of the module by instrumentation at three locations, 120° apart. ‘The maximum radial strain is 50%. Details of the pressure ‘meter module are shown in Figure 7.7. ‘The other cone pressuremeters which exist have been developed by the University of British Columbia, UBC, (Campanella and Robertson, 1986), ISMES (Ghionna etal, 1995) and Fugro Netherlands (Zuidberg and Post, 1995). The UBC seismic cone pressuremeter was built to com- ‘bine the features of full displacement pressuremeters with those of the seismic cone penetrometer. The pressuremoter portion is of 44 mm diameter and operates behind a standard 15 em? cone. Incorporation of accelerometers in the cone nd just below the pressuremeter unit allows measurement of shear wave velocity, F,, and hence ealeulation of Gy, the small strain shear modulus of the soil. The pressuremeter portion is expanded by oil and has a length to diameter ratio of 5. It is eapable of radial expansion of up to 27%, radial Push head_Electro/hydraulic hose ‘Cone rods Condueting hose ‘Standard cone rod Cone rod adaptor Amplifier housing eas Contraction ring Pressuremeter module 7 Contraction ring | __ connector 08 Piezocone Figure 7.6 The Fugro-MeClelland cone pressuremeter (after Withers eta, 1986).displacements being measured atthe centre ofthe pressue- ineter by three strain arms of 120°, The pressuremetet membrane is made of natural rubber and is protected by a fantem consisting of overlapping stainless steel strips. "The ISMES cone pressuremeter is a newly developed probe. It consists of a pressutemeter module of 35.7 mm fiameter and an approximate length of 800 mm mounted behind a standard 10 em® piezocone. It is inflated by nitto- gen gas and expansion of the membrane is measured at mid- height by two strain arms at diametrically opposite positions, The membrane is of rubber reinforced externally bya “chinese lantern” of vulcanized stainless steel strips. Fugto Netherlands have recently developed another cone Connection to Gontracton ‘amptier sub fing Chinese Fantern Membrane clamp ring Membrane ‘Atm cover sleeve Strain gauged spring 8 strain sonsing| farms at 120' Spacing Instrument. —| body Membrane —} Membrane lamp ing Chinese fantern Contraction ring 243.7 mm =| Connection tg one spacer and Figure 7.7 Components of the CPM presburemeter module (after Withers e a, 1986). CONE PRESSUREMETER 7 pressuremeter which consists of a pressuremeter module mounted behind a standard Fugro 15 om’ CPT. The pres- suremeter module has a diameter slightly less than the penetrometer diameter in order to reduce friction on the membrane during penetration and has a length-to-diameter ratio of 10, The pressuremeter is inflated with a fluid and the ‘membrane is internally reinforced. 7.2.2 Testing procedure Cone pressuremeter testing proceeds as follows: ‘The device is pushed into the ground at a rate of 2 emis At the required depth (which can be chosen from a real-time study ofthe piezocone data), cone penetration is stopped and 4 pressuremeter testi performed, The membrane is inflated until maximum strain is reached, At two or three intervals during the test, inflation is stopped, and a small unload- reload loop is performed in order to measure the elastic response of the soil. Deflation of the pressuremeter is then cartied out until the pressure variation with strain becomes small, at which point the membrane is vented. The device can then be pushed further, and the provess repeated ata new depth, It is important that data obtained in the field is corrected for the effects of the membrane stiffness and compliance, These can be determitied respectively by inflat- ing the pressuremeter in air, and in a very stiff steel casing prior to CPM insertion, Typical pressure expansion curves obtained inthis way in stiff clay are shown for the Madingley site in England in Figure 7.8. Such curves in clay soils ate usually charac- terized by a high lift-off pressure (the pressure associated with the first increase in strain) relative to the limit pressure 1900 +1000 Pressure (kPa) 8 Ce ee) Strain (%) 30 35 Figure 7.8 Pressure-strain curves from CPM tests carried out in ‘overconsolidated clay at Madingley (after Powell and Shields, 1995),i i: ii 178 ADDITIONAL SENSORS THAT CAN BE INCORPORATED 200 700 600 00 400 300 Pressure (kPa) 200 100 o 4 8 12 16 20 26 Strain (In P/ Ry) % 7.9 Pressure strain curve from CPM tests in Holmen sand (alter Powell and Shields, 1997), 28 82 (Constant pressure with inereasing strain), although in soft clays the test curves show a much mote gradual rise to the limit pressure after lift-off than in stiff clays (Powell and Shields, 1995). In contrast, sandy soils usually exhibit a low {ift-off pressure relative to the limit pressure; an example of which is shown in Figure 7.9 for Holmen sand. 7.2.3. Interpretation ‘The cone pressuremeter has the disadvantage thatthe pres- suremeter testis carried out in soil which has already been disturbed by penetration ofthe cone, Hence, any analysis for deriving soil parameters from the test must account for the instalation process. In clay the initial installation of the cone pressuremeter ccan be modelled as the expansion of a eylindrical cavity within the soil. The expansion phase of the pressuremeter test is then modelled as the continued expansion ofthe same cylindrical cavity, and the contraction phase as a cylindrical contraction. Assuring the clay to behave as an incompressi- ble linear-elastc, perfectly plastic Tresca material, Houlsby ‘and Withers (1988) developed a large strain solution for cavity expansion and contraction. The contraction solution is used to obtain soil parameters as the expansion solution predicts all expansion to take place atthe limit pressure. The Houlsby and Withers (1988) solution is presented in Figure 7.10 in the form of a geometric construction to determine undrained shear strength, s,, soil shear modulus, Gao, and in situ horizontal stress, O. ‘This analysis has been used to interpret the results from tests in both soft and stiff clays at sites in England (Houlsby and Withers, 1988; Powell, 1990; Houlsby and Nutt, 1992; Powell and Shields, 1995), Canada (Campanella et al., 1990) and Holland (Zuidberg and Post, 1995) Itappears that t+in) Figure 7.10. Geometric construction to determine 55, Gee and ay inlay (after Houlsby and Withers, 1988) generally, shear modulus calculated from cavity contraction (and from unload-reload loops) gives shear modulus values in reasonable agreement with results from other in situ and laboratory tests. Undrained shear strengths appear to be ‘generally of the same order as those from other test methods but their relationship with values from a specific test method seems to vary with soil tested. The Houlsby and Withers analysis has not yet been found to give reliable estimates of 1 (Powell and Shields, 1995). Itis pointed out that to date, too little published information exists to discuss the applica- bility of the Houlsby and Withers analysis in other than general terms. Other theoretical (Ferreira and Robertson, 1994) and empirical (Baguelin and Jezequel, 1983; Campanella etal, 1990; Lutenegger and Blanchard, 1990; Powell and Shields, 1995; Zuidberg and Post, 1995) methods have been exam- ined to determine soil parameters from cone pressuremeter results in clay. However, these methods have been applied in isolation to a limited database of results and no general conclusions have yet emerged on their applicability. Jn sands there are, to date, no analytical or numerical solutions which correctly model the expansion of the CPM because it has so far been impossible to correctly model the stress conditions around a CPM following installation Therefore, the interpretive methods for soil parameters range from methods based on theoretical considerations, through methods which combine theoretical considerations and measured observations, 1o purely empitical methods, ‘The methods for determining soil parameters in sand cur- renily arise from 3 sources: © the methods of Oxford University, which are mainly empirical and based upon a substantial programme of calibration chamber testing (e.g. Schnaid and Houlsby, 1992; Nutt, 1993), for D,, 4", and‘SEISMIC MEASUREMENTS 179 fe the methods of Yu et al. (1996) based on theoretical solution for limit pressure together with laboratory obser- vations and experimentally determined material proper- ties for the state parameter model, for y, and ¢”; fe the methods derived by Ghionna et al. (1995) based on a ‘combination of theoretical considerations with field and laboratory experimental work, for pi, Ky, Yo» andy" ‘The methods are reviewed in Powell and Shields (1997) for Holmen sand, ‘Whereas in clays the interpretation relies solely on the pressuremeter results of the CPM, all the proposed inter- pretation methods in sands, withthe exception of those for CPM limit pressure and unload-reload modiulus, involve the use of both pressuremeter and cone measurements. TWO ‘methods have been proposed for extrapolating the test results to the limit pressure (Ghionna er al, 1995), these gave similar results but with one consistently lower than the other. Examples of some of the correlations are given below: ‘The work of Oxford University has resulted in the correla- tion for relative density (Nutt, 1993) as: Ge~h = 339+ 104D, 3) Pu Fre Yu et al. (1996) derived @ correlation for initial state parameter as: Wo = 0.4575 — 0.2966 In = 06 Pi where pis the effective limit pressure pz ~ t Ghionna et al. (1995) investigated an expression for Ke Which links it top, and q. and obtained the expression: wg Owe [2st fla.—wyanl | 3 where a and b are constants. All three sources have proposed methods for determining the angle of friction (sce Powell and Shields, 1997). From a study of CPM testing and interpretation in sand, Powell and Shields (1997) concluded that the CPM showed significant potential to determine soil parameters, However, ‘much work is still needed to fully appraise the interpretation ‘methods and the CPM’s use in a wider range of sands. Fstimates of the soil shear modulus from the unloed- reload loops carried out during expansion, G,,, can be made in both clays and sands according to the following relationship: K, an Ap 2G are 78) ‘Where Ap is the change in cavity pressure over the loop, and Avis the corresponding cavity strain amplitude (logarithmic strain), Tt has been found generally, in both sands and clays, (except in the soft structured Bothkennar clay, Powell and S007 zy iene crn : Geil (res 8 vn) sea Sse K SHH Pov 660 tA a & 200 : é ated 100 oli Coot 01 Strain amplitude, Ae Figure 7.11 Comparison of shear moduli derived from different types of pressuremete tsts (Powell and Shiolds 1995) Shields, 1995) that Gy from cone pressuremeter tests agrees with G,, from self-boring pressuremeter test, i.e. Gy is not generally sensitive to the initial disturbance caused by insertion of the cone pressuremeter. However, it should be noted thet Gy. varies with both the strain amplitude and the stress level. Hence, in clay it is desirable to perform all unload-reload loops in an investigation at consistent strain levels. From the research in sands itis emerging that itis desirable to perform the loops over a stress range such that of = opatthe end of unloading. Also, Gy can be related to shear moduli from different tests. Research has been carried out in both clays (Campanella cetal., 1990; Nutt, 1993; Powell and Shields, 1995) and sands, (Houlsby and Schnaid, 1994; Howie, 1989; Nutt, 1993; Ghionna ef al., 1995; Powell and Shields, 1997) on the variation of G with stress and strain level from cone pres- suremeter G,, measurements, The research to date shows good agreement between cone pressuremeter and other pressuremeter results Figure 7.11). In conclusion, the cone pressuremeter is a device which has potential for the evaluation of undrained strength and shear modulus in clays, and relative density, horizontal stress inital state parameter, angle of friction and shear modulus in sands. Further testing is required at this stage to confirm or improve the current methods of interpretation especially for horizontal stress in clays. 7.3 SEISMIC MEASUREMENTS Geophones andor acvelerometers have been added to cone penetrometers to measure compression (P) and shear (S') waves. Elastic theory relates the small strain sheer modulus (Gq) end constrained modulus (Mo) usin: Go= pe (79) Mo= HV, (7.10) ‘where p is the soil mass density.! 180 ‘The small strain shear modulus is an essential input for prediction of ground-surface motions from earthquake exci- tation, for evaluation of foundations for vibrating equipment and offshore structures behaviour during wave loading and, ‘more recently, for deformations around excavations. Zn situ seismic measurements have traditionally been ‘made with the erosshole method which requires two or more boreholes, Recently the seismic cone has facilitated more ‘economical downhole measurements with the prospect also of use for more cost-effective crosshole measurements. 7.3.1 Equipment and procedures The modem version ofthe seismic cone was developed at the University of British Columbia (UBC) (Campanella et a,1986; Robertson et al, 1986), The seismic cone usually consists ofa piezocone unit with a receiver above it. One txample were the receive is a miniature velocity sein i ometris shown in Figure 7.12. Le ‘A schematic diagram with the layout of the standard technique using @ seismic cone is shown in Figure 7.13. The ‘Seago fiting to lock ‘A conductor cable Wire spliced to cable inside tube Seismometer ‘Slope sensor Quad ring Equal end erea friction leave (150 om’ area) i Strain gauges for i ‘tition load call, Temperature sensor: ‘Strain gauges for cone t Er Pressure transducer oad colt i O-rings Quad } Porous plastic * ‘Small cavact | iy Figure 7.12. The UBC seismic cone (Campanella et al, 1986). ADDITIONAL SENSORS THAT CAN BE INCORPORATED oxsiosepe ® roger Sate al anor br es Shear wave source (perpendicular to cone) J — Founvae Seismic ‘cone penetrometer Figure 7.13 Principles of the seismic cone survey technique (Campanella ta, 1986), cextra equipment needed, in addition to the built in seism- ‘ometer, is a memory oscilloscope and an impulse source ‘with a trigger for the oscilloscope. The source can consist of 4 stcel or aluminium beam for shear wave generation or a flat plate for compression wave generation. The shear wave source usually consists ofa beam pressed against the ground by the weight of the CPT vehicle and a hammer with an electronic trigger. The shear wave is generated by hitting the beam ends horizontally with the hammer in the ditection of the long axis, Commercial equipment exists where the memory oscillescope has been incorporated into the field ‘computer, which collects and processes all the data from the CPT or CPTU, For offshore testing Figure 7.14 shows the set up used by Fugro McClelland (de Lange et al., 1990). The seismic source is a hydraulic underwater shear wave box (HUSH box). Normally the seismic cone penetrometer is pushed into the ground and penetration is stopped at 1 m {ntervals, During the pause in penetration, a shear wave is, generated at the ground surface and the time required for the shear wave fo reach the seismometer in the cone pen- etrometer is recorded. Butcher and Powell (1995a) cartied out tests which showed that the use of two receivers either 0.5 or I m apart can greatly improve the quality of the data obtained by climinating problems related to triggering times. The data will then also be specific to that particular depth. Larsson and Mulabdie (19916) stress the importance of incorporating an inclinometer in the cone penetrometer 80 that the verticality of the sounding can be checked,Baldi et a, (1988) used two cono penetrometer trucks and two cone penetrometers (one with a source and one with a receiver) to carry out crosshole seismic piezocone penetra- tion tests as shown in Figure 7.15. They reported good agreement with traditional erosshole methods though cali- tation of tigger errors and the verticalty of both source and receiver soundings must be quantified to ensure good quality data, 7.3.2. Typical results and interpretation Several studies have been earried out which showed that in normally consolidated soils the shear wave velocities meas- ured with the seismic cone are very similar to those from traditional crosshole measurements (Campanella et al. 1986; Eidsmoen etal, 1985; Larsson and Mulabdie, 1991) Figure 7.16a shows an example of the good agreement of derived shear moduli obtained at the Holmen sand site in Drammen, Norway. However, Butcher and Powell (1995b) showed that in heavily overconsolidated clays the shear wave velocities measured using different techniques can give significantly different velocity profiles (Figure 7.160). ‘They showed that the differences were dependent upon the stress anisotropy of the soil and the relative importance of Umbilical nee ee ‘Shearwave ‘generator ‘Computer er, Comput [FF 9) Seiemic input unit ‘Wison Selsmic piezocone L[rapaf ror [mon Penetrometer Figure 7.14 Offshore seismic cone testing (from de Lange et al, 1980), ‘SEISMIC MEASUREMENTS 181 TO RECORDING AND PROCESSING ‘SYSTEM | |e 20tons 4 =!i_| Penetrometer trucks % d f, | u fg] ly hy (peer Receiver Source 40 om* cone 15 om* cone Figure 7.15 Crosshole seismic piezocone tests run by Ball etal (1988). the in stu stresses in the directions of shear wave propaga- tion and polarization. Since the source (bear) is normally placed some distance away from the rods the travelling paths of the shear waves are not try vertical and a correction is usually applied as indicated in Figure 7.178. However, even this correction can result in different derived data as shown by Butcher and Powell (1995a) in Figure 7.17. The correction is filing to take account of the fact that asthe distance from the seismic cone increases so docs the likelthood that the first shear wave arival will have been refracted from a straight travel path, To minimize non-linear ray paths ofthe seismic wave, the source should be located close tothe push rods (les than 1m). To minimize near field effects at shallow depth the source should generate high frequency seismic waves. The seismic cone can be a very repeatable test when all potential errors are taken account of and, when coupled with piczacone testing, becomes a very powerful test. However, as mentioned above, the results of any geophysical test will be a function of the particular test method and in situ stresses and should be interpreted with this in mind ~ Gy is not a "unique property of the ground. While doing the seismic cone testing at a particular depth, a dissipation test can be performed simultaneously Tn connection with dynamic analysis the damping param- ter D is frequently needed in addition to Gp. A limitation with seismic cone testing has frequently been that only Go has been obtained, while D has been determined from laboratory tests. Recent work at University of British182 ADDITIONAL SENSORS THAT CAN BE INCORPORATED Shear wave velocity (m/s) 50100 150200 _ 260 BL Unitorm & medium to Erol cease J BT sen a 2 4 16 18 20 + Rayleigh waves (BRE) 1B Downhole (seismic cone) Crosshole (vertically polarized) Figure 7.164 Shear wave velocities measured by crosshole and seismic cone techniques in Holmen sand, Columbia (UBC) has shown that damping parameters may also be obtained from seismic cone testing using a special analysis technique called the spectral ratio slope (SRS) ‘method (Stewart and Campanella, 1991, 1993). The material damping and also shear modulus parameters that can be obtained from seismic cone measurements are only for small strain amplitudes, Damping parameters and shear modulus for larger strain may be obtained from other in situ tests or from laboratory tests. 7.4 ELECTRICAL RESISTIVITY MEASUREMENTS Measurement of electrical conductivity (or resistivity = (conductivity) ') in soils has been used for many years for ‘the purpose of estimating in situ porosity or density. More recently electrical resistivity has also been used as an indicator of soil contamination as outlined in Chapter 8. Elcetrical resistivity is also an important input parameter for the evaluation of the corrosive potential of the soil (Bryhn, 1989). ‘This seetion describes principles, equipment and proce dures for cone penetrometers that include modules for ‘measurement of electrical resistivity. Further interpretation ‘in terms of porosity or soil density is discussed, and other possible geotechnical applications indicated. For environ- ‘mental applications, see Chapter 8, 7.4.1 Principles for measurement Electrical resistivity of soils is not measured directly, but is inferred from the measured voltage across an electrode pur ata constant supplied curent (I). According to Ohm’s law, soil resistance, R can then be computed as: v ret ny ‘This resistance is not a fundamental soil property, but depends on the current path Tength (Z) and the cross- sectional area (4) of the effective resistive unit. With the assumptions that the soil acts as a homogeneous isotropic ‘medium, thatthe electrodes function as perfect conductors and the resistivity module circuitry acts as a perfect current supply source, the ratio A/L is constant and equal to K. The electrical resistivity (p) is then a fundamental soil property which can be computed as ‘Shear wave velocity (m/s) 0 60 100150200250 300 ol ab ws sag 2 feathood sity shee ote a at 5 F sett are = obs E Spsivethy z = 7}—— s ° bess" oe ae sof tLe 1 ey at ate | fetes ety Lar 13 14 16 © Vyy (Rayleigh waves) V Via (downhole selsmic cone) 1 Vjy (orosshole - vertically polarized) © Vig (tosshole- horizontally polarized) Figure 7.160 Shear wave velocities measured by various geo- ‘physical techniques fu heavily verconsolidted clay at Madingley (Butcher and Powell, 1995.ELECTRICAL RESISTIVITY MEASUREMENTS, 183 1 constant temperature solution of known resistivity. This calibration is normally carried out in a calibration chamber . ‘Yeung and Akhtas (1995) have presented correction factors ‘Kis acalibration factor which is affected by the geometry of that can be used to take the size of the chamber into the dimensions of the electrodes. K can be found from direct account. newman” gay i calibration of the measurement module totally submerged in measurement of sol bulk resistivity (or conductivity) as outlined above is a function of both the resistivity ofthe pore 0.65m fluid and the soil particles and how they are arranged. The ‘dominant factor by far is electrical conduetivity through the pore water. As discussed later, probes also exist which ean. Ground surtace be used to measure electrical resistivity of the pore water only. Source beam 7.4.2. Equipment and procedures In Holland electrical resistivity measurement combined with ‘a cone penetrometer has been used since the mid 1970s (de Graaf and Zuidberg, 1985) (Figure 7.18). The probe consists € | Travel path assumed of normal 10 em? cone and fiction sleeve. Behind this, é i current and measurement electrodes are mounted isolated i ffom each other, normaly by plate. A voltage difference is applisd othe wo outer (curren) electrodes andthe esistv- ip ofthe sol is determined with the aid ofthe two inner electrodes. The probe shown in Figure 7.18 has two ses of rent and measurement elecrodes, The distance botween Conectd tne = th cent ceude fet A i Sa and of "3" z 100 mm, The electrically influenced zone of set B is much t Uopthy + (65/1 smaller than the coresponding zone of set A. The B.readings will to a large extent be influenced by densifica- Figure 7.17a Correction of measured travel time for non-vertical tion around the probe, while the A-readings will be more travel path (fiom Eidsmoen er al, 1988) inclined towards measuring resistivity in the intact soil, The Shear wave velocity (mm/s) Shear wave velocity (m/s) Shear wave velocity at 0.4m 050150 100200 250 900 08 1012418 o Tr or i 7 T af fab 4 ab 4ooap sf J J sk J Er / JE 4 + s gs 1 sf j S 8 8 ook 8 et 4 7 Source distance ee Source distance | 0.4m .\ 0.4m er 40m 8 } ==-- tom J e 1.9m -a-— 1.9m | 4.8m os i === 4am ] jCs tol L Figure 7.17b Effect of correcting travel path on derived data (Butcher and Powell, 1995a)184 ADDITIONAL SENSORS THAT CAN BE INCORPORATED advantage of close spacing is that changes over small depth intervals may be detected. As discussed below, measure- ments in the disturbed zone close to the rods and in the ‘undisturbed zone further away from the rods may be used to infer dilatancy of cohesionless soils (Campanella and Kokan, 1993). As outlined by de Graaf and Zuidberg (1985) tho penetration is halted every 0.2 m while resistivity meas- ‘urements are taken, In parallel, a water resistivity probe (Figure 7.18b) is penetrated, At every 0.2 m, during a stop in penetration, a small water sample is sucked into a small ‘chamber and the resistivity of the pore water is measured. Many organizations have developed cone penetrometers which includes a resistivity module. One example with single electrodes is given by Zuidberg et al., 1987 (Figure 7,19) while Campanella and Kokan (1993) use a piezocone ‘with resistivity module, (called RCPTU) and two sets of clectrodes (Figure 7.19b), Instead of carrying out fests with the “water probe" it is possible to recover samples of pore water and measure resistivity in the laboratory. Electrical cable Current electrodes A. Measuring Measuring electrodes 8 electrodes A (Current electrodes B trl i hose roe Eccl cbla ator eity Mansuing cl Fier Friction sleeve ‘Cone 106m" 10cm? Soil Water resistivity resistivity probe probe Figure 7.18 Delft Geotechnies soil and water resistivity probes (after de Graaf and Zuidberg, 1985). 7.4.3. Typical results and interpretation ‘The ratio between the bulk resistivity of the soil, pp, and the resistivity of the pore fluid, py, is defined as the formation factor, F: , 7.13) "The formation factor (F°) was linked to soil porosity, n, by Archie (1942): a 14) where A and m ate constants which can be found by laboratory calibration on obtained samples. According to Campanella and Kokan (1993) A = 1 for unconsolidated soils and m = 1.5 for sands. As the fines and clay content increase, m increases and can be as large as 1.8 to 3.0. Itis to ‘be expected that A and m also depend on sand structure. i One example of such a calibration curve for Drammen sand is shown in Figure 7.20, which indicates a linear relationship bebween » and 1/7. Insulation er B cone rods Electrodes void 5 A ig Plastic ‘ ; insulation \ 3 Curront f BA: a electrode ‘ % ada At ef a ig z aie Grounded electrode Ro ‘Standard piezocone-10cm’ ares Undisturbed | Friction sol sleeve Cone. a) b) Figure 7.19 Two examples of piczocones with resistivity module: (a) single elcctrodes (afer Zuidberg et al, 1987); (b) double clectrodes (after Campanella and Kokan, 1993)Figure 7.21 gives a typical profile of measured resistiv- ities in Drammen sand (at Holmen site) with the two probes shown in Figure 7.18, The Drammen sand is located in the ‘beginning of the Drammen Fjord where freshwater from the Drammen River is coming out; thus the salinity of the pore ‘water is zero of vety little in tho upper layers and increases 50 nm - Calibration x cure ze : 3 35 80 0.18 0.20 0.25 0.30 Resistvly pore water Py. Ap, Ratio: Sash pore water Pw: On atvy 80 py, Figure 7.20 Formation factor, versus porosity, n, for Drammen sand (Koning, 1982). ELECTRICAL RESISTIVITY MEASUREMENTS 185 to that of seawater with depth, This is very clearly reflected in the resistivity profiles in Figure 7.21. Using the calibration curve shown in Figure 7.20, porosity of Drammen sand has been computed for both the A and B readings. Since Drammen sand is loose to medium dense, ‘compaction occurs close to the penetrometer and rods, hence porosity ofthe intact material is less (A-readings) than the disturbed soil (B-readings). ‘A. discussion on how representative the porosities obtained from the electrical resistivity test (ERT) may be for in situ conditions is included in section 7.7 where compar- isons with radioisotope measurements are also made. Campanella and Kokan (1993) proposed that the differ- cence in resistivity measured by inner and outer electrodes (Figure 7.196) could be used to infer the dilatancy of sands. Their approach is that the outer electrodes measure intact undisturbed porosity of sand while the inner electrodes measure porosity of remoulded sand during penetration induced shearing. Campanella and Kokan proposed a dila- tion parameter, D, which is the ratio of bulk soil resistivity for closely spaced electrodes to widely spaced ones. As D increases and is larger than 1, the sand is becoming more contractive during shear and as D decreases, and is less than 1, the sand is becoming more dilative. Figure 7.22 shows the results of an RCPTU profile performed by Campanella and Kokan at Laing Bridge in ‘Vancouver, Canada; only resistivity measurements using the ‘outer electrodes are shown, Using correlations between relative density, D,, cone resistance and overburden stress (section 5.5.1) they com- puted relative density versus depth. Figure 7.23 shows the resulting plot of dilation parameter, D, versus relative density for the Laing Bridge site. The results suggest a reasonably good relationship between D, Cone Sleeve resistance — friction Resistivity (Q,,) Porosity n (%) (MPa), (kPa) cf24seiaz 0° 109.0 Pw 10 20 90__4o gs 50454 Ps 5b 160 bo" E 5 5 5 E19} Ho] 9} 19 = Pw 3 B15} 5| 15| 19 20] 20 po) 20} Figure 7.21 Results of clecticl resistivity tests in Drammen sand,186 ADDITIONAL SENSORS THAT CAN BE INCORPORATED and D, with a mean value of D,=45% at D=1 as the boundary between dilative and contractive behaviour. The value of D will of course be very dependent on the spacing of the inner and outer electrodes. Thus any correlation obtained between D and D, is valid for the specific equip- ment geometry used. 7.5 HEAT FLOW MEASUREMENTS Fugro McClelland developed a heat flow probe (Zuidberg et al,, 1986) which combines the accurate measurement of temperature with the measurement of thermal conductivity. {As yet the probe is not combined with a CPT ot piezocone ‘unit, but this could well be a future development. ‘The probe is shown in Figure 7.24 and has, according to Zuidberg et al, an accuracy of nominally 0.01°C and a resolution of 0.001°C with a fast thermal response time. ‘The work by Zielinski et al, (1986) showed that reliable ‘measurements required a probe that was as small as possible and this could be achieved by installing a thermistor in the point of a thin stainless steel tube (Figure 7.24). ‘The measurement of thermal conductivity is done by heating a wire in the steel tube. The temperature rise in the middle of the tube is monitored and plotted on a log time scale. The straight portion of this temperature rise curve is, used to calculate the thermal conductivity. A typical plot is, shown in Figure 7.25 where the computed thermal con. ductivity is also noted, ‘The probe has been used on the Norwegian continent] shelf as part ofa survey of hydrocarbon potential It ean also bbe used in pipeline investigations and in studies concerning the burial of heat generating materials, such as nuclear waste. 7.6 RADIOISOTOPE MEASUREMENTS For many engineering purposes it is essential to have a reliable determination of the in situ density or porosity of a soil. This can be difficult to obtain with standard CPT or CPTU measurements, This is especially so for soils which differ from clean, fine to medium quartz sands where the D,, Gis, qe correlations described in section 5.5.1 can apply. ‘This has been one of the main arguments for the develop- ‘ment of various probes or cone penetrometers with a radio active source and a detector. 7.6.1. Equipment, measurement principles and procedures Delft Geotechnics, Holland has developed a nuclear density probe which has been used for both onshote and offshore soil investigations (Nieuwenhuis and Smits, 1982; Tjeta et al, 1985), Pore Resistivity Cone resistance Friction ratio pressure 9.5mm Interpreted 4, (MPa) F, (%) us(m of water) (chm-m) Profile : 25 5 glo 0 0, sir f den 10 10 10 10] tof SAND E € 1 S p stratified é ; 20 20 ol! 20 20k coy, f Sir 30 30 80 n 30. 30. Figure 7.22 Result of RCPTU sounding at Laing Bridge site, Vancouver (after Campanella and Kokan, 1993).100 RADIOISOTOPE MEASUREMENTS 187 Relative density, D, (%) Sand 07 08 0. Kokan, 1993) cq oe Omens oes) Dilation parameter, D Figure 7.23 Interpreted value of relative density vs dilation patameter for sand at Laing Bridge site (after Campanella and, ‘ampliior housing jx —Plastic nose Figure 724 Heat fow prob Stainless steel tube ‘comprising heat source and thermistors (after Zuidberg etal, 1988). TEC) Tae (CD 087 |= ‘Apaled power: ‘harmal sonducity: ra earl, 100 1000 Time (sec) Figure 7.25 Result of a heat flow test (afer Zuidberg et a., 1988), Figure 7.26 shows the principle of the 44 mm diameter Delft nuclear density probe. During penetration the gamma photons are emitted from the radioactive source near the tip of the probe, The emitted gamma photons are colliding with atomic electrons of the material irradiated by the probe. Afer collision, two phenomena occur depending on the nergy level of the photons: 1, Electric absorption 2. Compton scatter Data transmission through cabie Detector Nay ort ahd phenmutpten Lead shild 300mm, 21 Basse soo, yi tops Gacsim) CA 7 : 44mm. 4am Figure 726 Delft Geotechnies nuclear in stu density probe (after ‘Tielta etal, 1985),188 ADDITIONAL SENSORS THAT CAN BE INCORPORATED ay He Hoe ‘The Delft Geotechnies nuclear back scatter density system I was designed on the basis of the latter. The Compton scatter E) phenomenon deseribesthe energy dissipation ofthe photons § a5 a function of the photons energy level and the atomic density ofthe material (Tjelta eta, 1985) j__—___ttnm With this probe (Figure 7.26) the bulk density of the soil 28 = 8 5 ft i ‘ aol ‘average Vy * ¢ Avorage og | 727 ACOUSTICNOISE ‘As a cone penetrometer is pushed into a soil, acoustic ‘emissions are generated by soil particles rolling and sliding over the penetrometer, by particle rearrangement, and by particle crushing, Tis phenomenon was used in the 1970s in Japan which could measure (B-readings) 30 1 Figure 732 Porosity ofloose to medium dense Holmen sand ftom laboratory, ERT and NDT measurements (from Lunne ef 1989), design cone penetrometers ithe intensity and pulse of frictional noise (Muromachi, 1981). Menge and van Impe (1995) experimentally demon- strated that for a steady penetration the acoustic noise is jnfluenced by a number of factors: the grain size ofthe soil, the porosity or density, mineralogy, sand structure, ageing and the in situ scesses ‘Tringale and Mitchell (1982) instrumented a cone pen- etrometer with a microphone in the tip and measured the acoustic response with depth simultaneously with cone resistance and sleeve ftiction data Figure 7.34 shows a typical result from a sand site in California where the amplitude profile from the microphone measurements corresponds remarkably well withthe change in material type. The amplitude of the acoustic signal increased as the average grain sizeof the material increased, Itshould be mentioned that Tringale and Mitchell worked in ‘the audible frequency range where the microphone also can register surrounding ground disturbanee. Nevertheless Trin- gale and Mitchell proposed that the data from the acoustic, cone penetrometer can give improved layering information and characterization of soil properties. This was confirmed by Massarsch (1986) who instrumented a penetrometer with a highly sensitive acoustic sensor working in a much higher frequency range than the penetrometer used by Tringale and Mitchell (1982), Using such high frequency avoids the problem of acoustic noise from the mechanical penetration equipment, Massarsch showed by field tests that even very thin sand and silt layers (less than 1 mm) can be detected by this ACOUSTIC NOISE 191 Cons Friction Amplitude at resistance ratlo. microphone, (MPa) R%)_—_(O-t KHzfitered) So Depth in meters 3 Figure 74 Acoustic cone penetrometers used by ‘Tringale and Mitchel! (1982). device. Menge and van Impe (1995) demonstrated in laboratory calibration tests that layering was more clearly detected by the acoustic response than qo. Ifthe identifica ‘on of sand structure and mineral type is important, acoustic, «mission measurements could prove to be very useful meas- ‘urements in addition to qo, w and f,GEO-ENVIRONMENTAL APPLICATIONS OF PENETRATION TESTING In recent years there has been a steady increase in geo~ environmental engineering projects where geotechnical engineering has been combined with environmental con- coms, Many of these projects involve some form of con- taminant in the ground; these can take the form of vapours, liquids and solids. Hence, in recent years there has been a change in site characterization techniques to accommodate these environmental issues related to contaminants. Drilling techniques have been modified to account for possible contaminated ground. However, drilling techniques ‘generally produce considerable disturbance to the materials surtounding the drill hole, which can have a significant effect on subsequent sample quality. With increasing appli- cation of data quality management, drilling and sampling techniques are becoming less acceptable, Also, drilling and sampling methods produce cuttings of the material removed from the drill hole, If these cuttings are contaminated they ‘may requite special handling and disposal methods. In many states of the USA there are regulations that require all drill, cuttings removed from geo-environmental site investiga tions to be disposed of or stored in an acceptable manner. ‘This can increase the cost of a day of drilling by as much as, US$1,000. Hence, there have been clear incentives to develop techniques that do not produce cuttings from the subsurface. The most rapidly developing site characterization techniques for geo-environmental purposes involve direct push technology, that is, penetration tests. The direct push devices gonerate essentially no cuttings, produce litle disturbance and reduce contact between field personnel and contaminants, since the penetrometer push rods can be decontaminated during retrieval 'A variety of penetrometer tests exist for both geotechnical and geo-environmental investigations. These tests can be divided into three main categories: logging, specific and combined. The most popular logging test for geotechnical investigations in soil is the Cone Penetration Test (CPT). ‘The CPT provides @ continuous profile of measurements, ‘and it is rapid, repeatable, reliable and cost effective. Specific tests include the field vane test and the pressure meter test since these measure specific soil parameters and are often carried out in locations identified by the logging test. Combined tests typically combine the features of log- ging and specific tests into one test, examples of which are the seismic CPT and the cone pressuremeter. 8.1 OBJECTIVES OF A GEO-ENVIRONMENTAL SITE INVESTIGATION ‘A geotechnical site investigation determines the following: 1. ‘The nature and sequence of the subsurface strata (geo- logie regime), 2. Groundwater conditions (hydrogeologic regime). 3. ‘The physical and mechanical properties of the subsurface strata, For geo-environmental site investigations where contami nants ate possible, the above objectives have the additional requirement to determine:4, The distribution, composition and concentration of the contaminants ‘The above investigation should be carried out in sufficient detzil as required by the project, For geotechnical projects this is usually a fumetion of the proposed structure and the associated risks, The geotechnical engineer is often in control of the risk process and hence, selection of the required site investigation detail. For geo-cnvironmental projects the extent of detail required for the determination of the distribution and composition ofthe contaminants may be controlled by various regulatory agencies, over which the engineer may have little control. With the rapid improve- ments in measurement technology, the in site detection limits required by some agencies for certain contaminants are decreasing at an alarming rate, For geotechnical investigations the information is often ‘obtained at one instance in time and projections are made to predict changes in ground conditions due to such factors as seasonal rainfall, For major projects where the observational ‘method may be applied, critical parameters such as deforma- tions can be monitored to evaluate the changing mechanical conditions. For geo-environmental projects where potential contaminants are identified, long-term monitoring and sam- pling may be required for design and either remediation or containment, Hence, the objectives for geo-environmental site characterization can be quite different from those for a more traditional geotechnical site characterization, 82 CPT TECHNOLOGY FOR SITE CHARACTERIZATION ‘The Cone Penetration Test (CPT) has become an ireportant ‘nsitu test for the characterization of soils where penetration is possible, Penetration can be difficult through cemented materials and materials with large particle sizes. The CPT provides excellent near-continuous profiles of soil type and detailed stratigraphy. If pore pressures are measured (CPTO), improved stratigraphic detail can be obtained as well as important additional information on equilibrium groundwater conditions, consolidation characteristics and hydraulic conduetivity. Empirical and semi-theoretical cor- relations are available to estimate a full range of mechanical properties, as described in earlier chapters. The CPTU ‘measures the mechanical response of the ground or material to the penetration process through cone penetration resist- ance (q.), sleeve friction (f) and the pore liquid pressure (1). Ifa solid contaminant has mechanical properties signifi- cantly different from those of the surrounding soil then the CPT can identify the presence of the material. However, the CPT cannot identify tho chemical composition of the con- taminants. Hence, sensors have been developed that can be audded to cone penettometers in. an attempt to identify certain contaminants, ‘The measurement of equilibrium pore pressures can be an important part ofan investigation to evaluate the direction of GEO-ENVIRONMENTAL PENETROMETER LOGGING DEVICES 193 groundwater flow and vertical pressure head distribution and hence the hydrogeologic regime. Most cone penetrometers, that measure pore pressures contain high capacity pressure transducers because penetration pore pressures can be very large in sof soils, To improve the measurement of the equilibrium pore pressure some cone penetrometers include 1 low pressure transducer connected to the outside of the probe via a control valve so that equilibrium pore pressures can be measured to a very high dogree of accuracy (for example, 30mm head of water). This can minimize the number of possible permanent monitoring wells to measure groundwater flow regimes. However, care is needed when penetrating soft fine-grained soils since it can take a con- siderable time to dissipate the high penetration pore pres- sures. The use of small diameter probes (such as 1 em”) can bbe advantageous for speeding up dissipation time in fine grained soil in cases where knowledge of in sit pore pressure is vita. 8.3 GEO-ENVIRONMENTAL PENETROMETER LOGGING DEVICES 8.3.1 Temperature ‘The CPT and CPTU are excellent logging devices that provide near continuous profiles of mechanical parameters. Sensors have been added to cone penetrometers to enhance their application for geo-environmental site characteriza- tion, One of the earliest sensors added to a cone pen- etrometer was a temperature sensor, Initially temperature sensors were used to aid in either calibration corrections or to locate zones of different ground temperature, such as frozen soil. Recently temperature sensors have been used to aid in the identification of contaminants that generate heat due to biological and/or chemical activity. 8.3.2 Electrical resistivity and conductivity ‘The next major sensor that has been added to the CPT is for the measurement of electrical resistivity or conductivity (see also Chapter 7). The conductivity is the inverse of resistiv- ity, with the following as a useful guide: ‘Conductivity (uS/om) = 10,000 + Resistivity (Q-m) ‘As mentioned in Chapter 7, the measurement of electrical properties was first developed to evaluate in situ density of sands (Kroezen, 1981) but more recently it has been used to evaluate contaminated soils (Horsnell, 1988; Campanella and Weemees, 1990; Woeller etal, 1991; Strutynsky et al, 1991). The rationale for making electrical measurements is that in many circumstances, the electrical properties of the soil will be changed by the presence of contaminants. Therefore, by measuring soil resistivity, the lateral and vortical extent of soil contamination can be evaluated. Unsaturated soils and saturated soils with many non-aque- ous-phase-liquid (NAPL) compounds exhibit very high ale eee eee eee194 GEO-ENVIRONMENTAL APPLICATIONS OF PENETRATION TESTING electrical resistivity (low conductivity). Dissolved inorganic compounds, such as those contained in brines and landfill, leachates, significantly decrease soil resistivity. Table 8.1 presents some typical values of resistivity and conductivity for some soils and contaminants ‘The resistivity CPT works on the principle that the measuted voltage drop across two electrodes in the soil, ata given excitation current, is proportional to the electrical resistivity of the soil (see also Chapter 7). The resistivity electrodes are typically steel rings from Smm to 15 mm ‘wide that are set apart by distances that vary from 10 mm to 150 mm. The larger the spacing the greater the depth of penetration for the electrical field into the surrounding soil. ‘Some probes have several electrode spacings so that lateral penetration varies. Some devices use small circular electro- ‘des mounted around the circumference of the probe, Figure 8.1 shows a typical resistivity cone penetrometer with two ring electrodes (Woeller ef al, 1991a). The electrodes are designed to be reasonably wear resistant and have a high electrical conductivity, A non-conducting plastic ot other material is used as the insulator separating the electrodes. ‘The resistivity measurements are typically made by apply ing a sinusoidal current across the electrodes and measuring, the resultant potential difference between the electrodes, ‘The current can be regulated by a downhole microprocessor that adjusts the current when the resistivity changes appre- ciably to ensure a linear response. This enables resistivity ‘measurements between 1 and 250 ohm-m to be made with ‘an accuracy of * 0.2 ohm-m. A 1000 Hz source is typically used to avoid polarization of the electrodes. Electrical resistance is not a material property but a function of the electrode spacing and size. To convert from resistance to resistivity, which is a material property, a laboratory calibra- tion of the probe geometry is necessaty. The resistivity of soil is for the most part influenced by the resistivity of the pore liquid, which in turn is a measure of the pore liquid chemical composition. Hydrocarbons are non-conductive and will therefore exhibit high resistivity. The resistivity CPT has been used suecessflly in acidic ground conditions, ‘The main disadvantage with cloctrcal measurements is that the bulk resistivity or conductivity is not diretly controlled by the chemical properties ofthe surrounding material. The measurements are strongly influenced by the background soil and pore liquid. Henee, it is important to obtain meas- urements of the background uncontaminated ground for ‘comparison. For relatively uniform soil conditions it is possible to develop local site-specific correlations between the bulk resistivity and selected contaminants (Campanella etal, 1994). The primary advantage ofthe resistivity CPT is that it provides continuous profiles of bulk resistivity along with the full CPT data ina rapid, cost-effective manner. The profiles of resistivity measurements can then be used to identify potentially critical zones where detailed sampling and/or monitoring can be carried out. An example of a resistivity CPTU profile is shown in Figure 8.2. This profile was obtained at a site where the main contaminant was creosote from a timber treatment plant (Campanella etal, 1994). ‘The measured bulk resistivity values are larger compared to the background values in zones with the contaminant, The fiee product was verified by monitoring ‘well sampling 8.3.3 Dielectric measurements ‘The electrical measurements discussed above relate to resis- tivity. However, electrical measurements can also be made to measure the dielectric constant of the material surround ing a penetrometer. The resistivity is somewhat insensitive to contaminant type, whereas the dieleetric constant can be very sensitive to contaminant type. The dielectric constant is frequency dependent and is therefore dispersive. However, above about 50 MHz the dielectric constant is essentially constant. Delft Geotechnics have developed a high-fre- ‘quency-impedance-measuring (HIM) probe to measure both ‘Table 8:1 ‘Typical resistivity and conductivity values of bulk soil mixtures and pore fuid (modified from Cempanetia eta, 1994) Bulk Fluid Bulk Fluid Material type resistivity resistivity conductivity conductivity (Om) (am) (Siem) (usiem) Seawater - 02 50,000 Drinking water >is = = 665 Mine tailings sand with acid drainage 140 221 10,000-250 5,000-370 ‘Mine tailings sand without acid drainage 70-100 15-50 145-100 665-200 APL contaminants in sand 0545 03-05 20,000-6,000 33,000-20,000 ‘APL contaminants in sand 125 48 80 210 100% ethylene dichloride (ED) 20,000 os 50% EDI50% water in sand. 700 : 4 17% ED/83% water in sand 25 _ 36 Clays 1-100 10,000-100 ‘Alluvium and sands (son-marine) 10-800 1000-12 Oilsands 4-800 2 2,500.12yy GEO-ENVIRONMENTAL PENETROMETER LOGGING DEVICES A Insulation NE Electrodes Seismometer —o ‘Temperature sensor inclinometer |«— Friction sleeve Pore pressure transducer J— Cone Figure 81 Example of two ring electrode resistivity cone penotrometer (after Woeller etal, 19910. Pore Buik pressure Cone _——Fctlon_resietvty th resistance ratio" ROYO (motwater) (MPa) R= (%) (ohm) 0% #09 25 025 020 E 8 = fv. g S a 6 a J | Figure 8.2 Example resistivity CPTU sounding at an organic, contamination by heavy oil (after Campanella etal, 1994), 195 dielectric constant and conductivity of soil samples and hence, detect the presence of contaminants. The probe is a cone penetrometer-based instrument with a retractable cone. "The cone is pushed into the ground in the closed position and atthe required depth the tip is retracted so that a soil sample fills the spaco. The sample chamber contains a central pin ‘which acts as an antenna and the rim of the sample chamber acts as a receiver, as shown on Figure 8:3. A high-frequency electromagnetic field pulse is generated at the ground sur- face and transmitted through a coaxial cable to the sample in the HIM probe. The resulting dielectric constant and con- ductivity are determined as @ function of frequency (10-500 Mz). After the measurement the sample is pushed cout of the inner eylinder, re-establishing the cone shape. At the next depth the procedure is repeated. Accuracy of the HIM probe is 5% for both the diclectric constant and conductivity, although the issue of eross-contamination may be a problem. Contaminants such as light and dense NAPL, zones, if in sufficient quantity, can be detected by the low values of dielectric constant and conductivity in relation to the background surrounding water-saturated soil, The loca- tion of the contaminant in the sample cup showed some effect on the measured values (Delft Geotechnics, 1994), 8.3.4 pH sensors The acidity of a material can be measured using pH sensors ‘mounted either inside a cone or on the surface of the probe. ‘The major disadvantage of sensors mounted inside the cone is that a sample of pore liquid must be drawn into the ‘measuring cell and then expelled. This process can be difficult in low permeability soils such as clay and the cell and sensor must be cleaned after each reading, More details ae 1, Coaxial cable 2. lsolator 3, Retractable rim receiver 4, Soil sample 5, Central antenna e—O @ © Figure 8.3 Cross Section of HIM-probe (after Delft Geotechnis,, 1994)196 about sampling techniques are given in a later section, Sensors mounted on the outside of the cone have the advantage of direet exposure to the surrounding material, although abrasion and damage to the sensor can be a major problem, Several CPT manufacturers and operators have placed a pH sensor in a small recess a short distance behind the friction sleeve, This recess is designed to produce a small vortex for pore liquid to enter during cone penetration. The sensor is then well protected from abrasion and damage from the surrounding solid material and can measure con- timuous variations of the pH of the pore liquid. The pH measurements are sensitive to temperature changes and generally a temperature sensor is mounted adjacent to the pH sensor to allow automatic correction for temperature effects. The continuous measurement of pH can be a usefil guide for detecting certain contaminants with significantly different pH values from that of the background soil. A piT sensor was successfully used to differentiate between acidic, tar material and waste drilling mud at an old disposal site in California. The temperature profile was also elevated over the acid tar depth interval and confirmed that chemical reactions were occurting in the acid materials. At this site the resistivity profiles could not delineate between the acid tar material and the drilling muds (Bratton et al., 1993). 8.3.5 Redox potential ‘The oxygen exchange capacity of a material can be meas- ‘ured with a sensor for redox potential. A CPT probe (Chem- probe, Figure 8.4) described by Olie er al. (1992) carries, out the measurements of redox potential, pH and con- ductivity. These three parameters are major variables of Conductivity cell pHlredox pressure Gauge Filter 44mm tre 84 Chemicone (after Delft Geotechnies, 1994) GEO-ENVIRONMENTAL APPLICATIONS OF PENETRATION TESTING chemical equilibrium for inorganic substances. ‘The sensors are mounted inside the cone and a sample is drawn into a ‘measuring cell located a short distance behind the cone tip. ‘The measurements are made under a nitrogen atmosphere to reduce the exchange of atmospheric oxygen with dissolved ‘gases from groundwater. A slight excess pressure of nitro- igen is used during penetration of the probe to stop the flow of liquid into the 15 ml measuring cell. The nitrogen is, supplied from a small nitrogen cylinder in the probe. The stainless steel porous filter is cleaned by the excess nitrogen, which in turn cleans the sensors. Demineralised water is also used to clean the sensors and check the calibration, A pressure sensor is also included to monitor the flow of liquid into the measuring cell and to estimate the hydraulic con- ductivity of the surrounding soil. Olie er al. (1992) showed that the measurement of redox potential along with pH and electrical conductivity enabled the detection of a floating layer of versatic acid beneath a storage tank at a petrochem- jeal plant in the Netherlands. A similar probe (chemicone) was described by Woeller ef al. (19912) and is shown in Figure 8.5. Further details of devices that sample the sur- rounding pore liquid or vapour are given in later sections. 8.3.6 Gamma and neutron sensors As mentioned in Chapter 7, gamma and neutron sources and sensors have been added to cone penetrometers in the past in ‘an effort to measure in situ density and moisture content (ee, for example, Mitchell, 1988; Sully et al., 1988). However, these devices have not become popular because they contained active radiation sources which ean present significant problems if the probes should become lost in the ‘ground and require expensive recovery. Recently there has been a trend toward adding passive gamma-ray sensors in an effort to detect radioactive contaminants (Marton ef al, 1988). A variety of different passive sensors are available, the selection of the appropriate sensor is based on efficiency, range of gamma-ray energies expected, temperature depend- ence of the sensor and sensor ruggedness. An example of a radiation detection cone is shown in Figure 8.6, The applica tion of gamma-ray sensors is clearly limited to environments ‘where specific gamma-ray emitting contaminants are possible. 8.3.7. Laser-induced fluorescence The most recent sensors to be added to the cone pen trometer for environmental applications are those that involve laser-induced fluorescence (LIF). Hydrocarbons are fone of the most common ground contaminants and most hydrocarbons, because of the polyaromatic constituents, ‘produce fluorescence when irradiated with various forms of light. LIF technology applied to the environmental field is, relatively new, with the first published work conducted by Hirshfield et al, (1984) and Chudyk eal. (1985). The initial work and much of the recent work has centred around thedevelopment of feld-portable LIF systems. The US Army Engineer Waterways Experiment Station (WES) under the sponsorship of the US Army Environmental Center (AEC) developed a site characterization and analysis penetrometer system (SCAPS) for investigating and screening sites for ground contamination using LIF, SCAPS incorporates exist- ing cone penetrometer technology with LIF. A 6.35 mm diameter sapphire window is mounted flush on the side of the cone approximately 60 em above the tip. This window provides a view por for te fibre opic-based LIF system. A pulsed nitrogen laser light (337:nm wavelongth) is sent down to the window over a 400 micron diameter 60 m long silica optic fibre. Fluorescence generated in the surrounding material is carried back to the surface by a second fibre ‘where it is dispersed using a spectrograph and the intensity quantified with a time-grated, one-dimensional photo diode auray, Readout of a fluorescence emission spectrum requires, approximately 16 milli-seconds. A microcomputer-based dala acquisition and processing system controls the ffuo- rometer system, acquires and stores sensor data once a o |_-Pue mde Samping and Temperature. —s teating chamber denser Condutivty. — Sat or pH cal Downhole siepe sensor ~ | | —setamomete Temperature gonsor Fiction sleeve Pore pressure transducer Cone Figure 88 Example of sampling cone penetrometer-chemicone (after Woeller eta, 19918) GEO-ENVIRONMENTAL PENETROMETER LOGGING DEVICES 197 second, and plots the data in rel time as vertical profiles on a CRT display. A schematic diagram of the SCAPS probe is, shown in Figure 8.7. Initial field trials at petroleum. Iubricant (POL) contaminated sites have been promising, although results have been qualitative, since calibration for specific contaminants is difficult, Research has shown that common fuel contaminants such as heating oil, jet fuels, gasoline and diesel fuel marine exhibit strong fluorescence signatures, with the degree of fluorescence depending on the excitation wavelength (Gil- lispie and St, Germain, 1993; Chudyk et al, 1985). How- ever, common chemical contaminants such as chlorinated hydrocarbons (for example, TCE and PCE) do not fluoresce and are not suitable for the fluorescence technique. The intensity of fluorescence is a function of excitation wave- length and recent efforts have been made to develop tune- able laser fluorimeters, that is, systems that can vary the ‘wavelength of the laser light source (Bratton ef al, 1993), and hence, detect a greater range of contaminants. More information about the contaminants can be obtained if the complete wavelength-time-intensity matrix is recorded, although this measurement takes a little longer to perform and a pause in the penetration is required. Fluorescence Integral signal cable Grout we ——£4 Sliding ting —+ Grout ports Moisture soal Scintiation gamma sensor Mud block Water seal Two-axls tit sensor Fiction sloove Pore pressure ‘gauge Fluid-tiled portal Water seal Saturated porous filter 60" cone Figure 8.6 Example of a radiation detection cone (after Applied Research Associates, Inc, 1994)198 research to date has concentrated on the aqueous phase and very little work has been carried out on evaluating the LIF characteristics of contaminated soils (Apitz ef al., 1992). ‘The intensity of the LIF signal is strongly dependent on soil type, with sands having a stronger signal for a given concentration than clayey soils, Bratton ef al. (1993) suggest that LIF research in soils is in the infaney stage and even, standard laboratory procedures for evaluating LIF response in soil materials have not yet been developed. The calibra tion and detection limits of LIF in soils are complicated by soil type, soil grain size effects, natural organic compounds, (humic acid) and the influence of time on the contaminant degradation, There are also problems of low signal to noise ratios for the LIF systems, Currently, field correlations of LIF intensity to contaminate concentration are preferred Probe window enlargement 400-micron Y~ fused siica ¢ fre opts waveguides e Spring loaded ; {~~ fibre optic Grout tube | ema Sapphire to0-mieron epteal window used silea waveguides @ Fibre ; algnment Load cell H ton wring i Grout injector Block 1 Grout ube |+— Friction sleeve load call it — Point load cell, }<— 60-deg cone point Ejectable grouting tip Figure 8:7 Schematic of SCAPS probe (efter Liebermann er a., 1991), GEO-ENVIRONMENTAL APPLICATIONS OF PENETRATION TESTING since laboratory calibrations are still uncertain (Bratton er al,, 1993). Concerns also exist over the long-term durability and maintenance ofthe fibre optic cable. lie etal (1994) have also developed a fluorescence CPT probe (hydrocarbon probe, Figute 8.8). The probe is 55 mm in diameter and contains a UV light source as well as the fluorescence detection system, No fibre optic cables are required since the complete sensing system is located in the probe. During penetration, measurements are made by illu ‘inating the material surrounding the probe with a small mercury larap to produee the UV light source placed behind 1 clear window. The fiuoreseence emitted by the hydro- carbons is detected in the probe by a small photomultiplica- tor tube. The signal is conducted through the electrical CPT cable to a data procossing system at the ground surface. A detection limit of 50 mg/kg dry weight for light NAPL is claimed. The intensity of the radiation emitted by the Detector Window Uv-sourcel Figure 8.8 Hydrocarbon probe (aler Delft Geotechnies, 1994)Yr GEO-ENVIRONMENTAL PENETROMETER SAMPLING DEVICES 199 contaminant is an indication of the concentration of the product in the soil, although specific calibration is required. ‘The detection system can also handle other wavelengths for {identification of other contaminants by using filters to con- trol the excitation wavelength, Preliminary results at a demonstration site where NAPL layers of domestic fuel oil ‘were present show excellent results (Olie etal, 1994). An example profile from the hydrocarbon probe is shown in Figure 8.9. These results show excellent baseline stability and clear sharp peaks in a series of alternating NAPL layers. Olie et al. (1994) also investigated the effects of smearing and the displacement of the measured contaminant due to the penetration process. The limited results indicate thatthe soil effectively cleans the window on the surface of the cone and that the displacement of the detected layer can be as much as 5 cm. 8.4 GEO-ENVIRONMENTAL PENETROMETER ‘SAMPLING DEVICES ‘The cone penetrometers deseribed in the above section are primarily sereening devices that log the ground profile for Intensity fluorescence signal (output in V) 3 2 4 @Lo ce naan Depth below ground loval (mn) Depth (m) NAPLInec ot MR Sorte tel Description of soil profile Soil profile from hydrocarbon probe Figure 89 Example profile from hydrocarbon probe (after lie er a, 1994), geotechnical and chemical measurements. Based on these measurements it is often possible to identify potentially critical zones or regions that may require more selective testing to measure ot monitor specific contaminants. Sam- pling and monitoring wells are usually installed for this purpose and penetrometer technology has been used 10 develop a complete range of short- and long-term sampling, probes. Sampling probes have been developed to sample either vapours, liquids or solids. The following section describes some of the main developments in tis area Most of the available vapour and liquid samplers have some common features. Almost all the samplers are pushed to the desired depth based on adjacent CPT profiles. Some- times the sampler is pushed down the same hole as the CPT, and since the sampler is of a larger diameter (typically ‘50 mm) than the CPT, contact can be maintained with the surrounding material. Generally, the push rods are pulled back to expose a filter to the surrounding material, which avoids contamination of the filter before reaching the required depth, Selection of the appropriate filter material is, based on the type of sample (gas or liquid) and the expected contaminant type. A sample is pulled into the sampler using, either a vacuum or the natural in situ fluid pressure, The sampler can then be either withdrawn to the surface for sample retrieval and sampler decontamination or the sample can be taken to the surface via a tube or by wireline and the process repeated at a greater depth, The following describes some typical samplers. 8.4.1 Liquid samplers ‘The most common discrete depth in situ water samplers are the Hydropunch and the BAT Enviroprobe. The Hydro- punch and its variations are based on a simple sample tool that is pushed to the desired depth and the push rods Withdrawn to expose the filter screen. The filter screens ean ‘be made from a range of materials although stainless steel is the most common. Sereened intervals ean vary in length from 100 mm to 1500 mim depending on ground conditions, required sample depth, contaminant type and hydraulic conductivity of surrounding material. A small diameter baler is lowered through the hollow push reds and body of the sampler to collect aliquid sample, Figure 8.10 illustrates & typical Hydropunch sampler. A peristaltic pump ean also bbe used to pump larger volumes of non-volatile liquids to the surface, This type of push-in liquid sampler is very common, simple to use and can produce large samples. However, there is ftle control over the sampling process and liquid samples are often turbid (that is, they contain suspended solids), especially when using a coarse screen, Modified versions of the Hydropunch concept have been developed to install long-term monitoring wells, an example of which is shown in Figure 8.11. These inelude innovative techniques to seal and grout the sampler into the ground after the sampler is pushed to the required depth.200 GEO-ENVIRONMENTAL APPLICATIONS OF PENETRATION TESTING ‘The BAT Enviroprobe, shown in Figure 8.12, was devel- \ oped by ‘Topstenson (1988) and consists of thee basic * | sam 00, 264mm 1D Rod components: ' 1. A sealed filter tip with erable sleeve atached tothe | LL pperous prs rds Upper O-ring 2. Evacuated and sterilized glass sample vials, enclosed in a ; housing and lowered fo te fer tp via wireline system | ‘A disposable, double-ended hypodermic needle which ‘makes a hydraulic connection with the pore fluid by puncturing the sel-sealing flexible septum in the filter ‘260mm polypropylene tip. porous iter (variable length, typically 180mm to 300mm) Brass connector barhead {iting on one end and threads on the other Lower O-ring acted ficial co sample | | Sacificial cone tip sample Figure 8.10 Typical Hydropunch ; Protective cap. Pushin i Protective casin casing Cement pad exible oo Wire. PVC pipe Iu Double-ended oe hypodermic 4 needle } ‘Cement! bentonite = samp " q amin grout probe with ’ pve Foxtbie vacuated washer ise container Bentonite Extension — rings cap pipe Doublo-ende: ubbor oderrie washer needle Pvc i soreen BAT filter tip, INSTALLATION WITHDRAWAL. 1 Figure 8.11 Fxample of a modified Hydropunch (after Fugro- i MeCelland, 1993) Figure 8:12 BAT sampler (after Torstennson, 1984)“The filling rate ofthe sample vials can be monitored using a poe pressure transducer atached fo the vial. The monitoring Bows when the fui infiltration is complete, assuring that presure inside the vial i equal tothe dn situ fluid pressure. ‘These measurements can be used to estimate the hydraulic conductivity of the surrounding material. Various modified ‘ersions of this concept are now in use for ground! fluid Sampling, The system is most applicable for retrieving fuid samplos where limited volumes are sufficient (< 150 mb) ‘The small filter size allows for disrete sampling intervals, fut can create Jonger sampling times in less permeable rnaterals, The protective sleeve and fine filter sereen pro- duces fuid samples with low turbidity. However, sometimes the needle ean become blocked with fine material passing through the filter, resulting in incomplete filling of the vial ‘Modified versions of the above samplers have been devel- oped to provide a mixture of the two techniques. Samplers exist that allow large volume (> 1200 ml) samples to be taken under a back pressure of argon or nitrogen gas to avoid volatilization, The measurement of pressure with time can be used to monitor the sampling process. O'Neill ef al, (1995) describe a cone penetrometer that can also take in a liquid sample and pump itt the surface for analyses 8.4.2 Vapour samplers ‘Vapour (gas) samples can be obtained in a manner similar to that described above for liquid samples. However, special cae is required to purge the sample tubes and store the samples, Typically, the sampler is pushed to the required depth, the filter element exposed and a vacuum applied to draw a vapour sample othe surface. The volume of gas can be monitored using special monitoring equipment ancl typ- ical sample containers are Tevlar bags, gas tight syringes and glass or steel sampling vessels. Special disposable plastic tubing is used to draw the sample to the surface. ‘Vapour sampling modules have also been added to cone ponetrometers to allow sampling during ¢ CPT. ‘The sam- pling module is typically located a short distance behind the cone, Samples can be taken during short pauses in the penetration, ‘To minimize eross-contamination for subse- uent samples within one vertical sounding requires ether a two-line sampling design for purging of the vapour collee- tion lines between samples or a single line system with nitrogen or argon gas purging. A small positive intemal gas pressure can stop the inflow of gases during the penetration Process 8.4.3 Solid samplers ‘A variety of push-in soil (solid) samplers are now available. ‘Most are based on designs similar to the Gouda or MOSTAP soil samplers. The samplers are pushed to the required depth in a closed position. The Gouda type samplers, shown in Figure 8.13, have an inner cone tip that is retracted to the locked position leaving a hollow sampler with 31 mm ‘SEALING AND DECONTAMINATION PROCEDURES 201 diameter stainless steel or briss sample tubes. The hollow sampler is pushed to collect a sample and the filled sampler and push rods are then retrieved to the surfice. The MOSTAP type samplers, shown in Figure 8.14, contain & wite to fix the position of the inner cone tip before pushing to obtain a sample, Modifications have also been made to include a wireline system so that solid samples can be retrieved at multiple depths rather than retrieving and re- deploying the sampler and rods at each interval. 8.5 SEALING AND DECONTAMINATION PROCEDURES ‘The hole produced by the penetrometer requires sealing, usually with a special grout, The grouting can be carried out cither after the push rods are removed or during the removal process, Typically the grouting is carried out after rod removal using special grout push rods with disposable tips ‘The hole is grouted from the bottom up as the grout rod is pulled from the ground, The grout rods generally follow the previous penetrometer hole since the rods follow the path of least resistance, Some operators use penetrometers that allow grouting during penetrometer retrieval, as illustrated in Figures 8.6 and 8,7. The grout is usually delivered to the Pushed with CPT push-rod connection Easy-to-use ball lock action - push to depth, retract 260mm fo open sampler, push 240mm to capture sample, 316 SS sample tubes Rotractable tip Figure 8.13 Typical Gouda type solid sampler.202 GEO-ENVIRONMENTAL APPLICATIONS OF PENETRATION TESTING penetrometer through a small diameter grout tube pre- threaded through the push rods. The grout can exit the penettometer through either a sacrificial cone tip or through ports above the cone located on the friction reducer. A special high torque pump is used to provide the grout. A detailed account of grouting of CPT holes is given by Lutenegger and De Groot (1995). ‘To clean the push rods and equipment after a sounding, all, downhole equipment can be pulled through a rod-washing, decontamination chamber. The chamber ean be mounted at the base of the hydraulic thrust cylinders. The push rods can be steam cleaned as they pass through the chamber, before handling by field personnel, Rubber wipers at the entry and cxit of the chamber control water leakage. ‘The waste water is collected for later disposal. Many of the special truck- ‘mounted penetrometer systems also include stainless steel sinks for decontamination of samplers and penetrometer. 8.6 FUTURE TRENDS ‘The potential cost savings of penetrometer-based geo- environmental site investigations has fostered considerable research expenditure into additional sensors. Considerable research is underway on improvements to the fluorescence “Tip-release cable (pute fom Surface to bogin sampler push) QIN a Tip released to Tie fixed remain in place ts as sampler is pushing aavanced 65mm 4m Cuting edge“ | 35mm Figure 8.14 ‘Typical MOSTAP type solid sampler. techniques. Research is also under way at various centres around the world to investigate new sensors that could be incorporated into cone penetrometers, such as Ramon spec- troscopy, fibre optic chemical sensors, laset-induced-break- down-spectroscopy (LIBS), _time-domain-reflectometry (EDR), ground penetration radar and integrated optoclec- tronic chemical sensors Ramon spectroscopy involves a powerful laset light ‘which is focused onto the reverse side of a glass of quartz slide that has been coated with a species-specific chemical, When a contaminant vapour is in contact with the coated surface of the sensor, the reflected light can be interpreted using spectral analyses of the shift in wavelength. The shit is a chatacteristie of the contaminant, although the shift is generally very small. Improved levels of detection can be ‘made by pressurizing the vapour and a large range of contaminants can be identified using different chemical coatings. Fibre optics can be used to carry the laser and reflected light to the processing data acquisition system at the ground surface. Laser-induced-breakdown-spectroscopy (LIBS) can be used to identify hazardous metal compounds, The laser light must be focused directly onto the material to form a short-lived plasma which emits light that can be collected using fibre optics and analysed. Time-domain- reflectometry (IDR) sensors are under development to ‘measure soil moisture content, which can be an important parameter in unsaturated soils. Ground penetrating radar (GPR) technology for surface measurements has developed considerably in recent years and systems are under development that will incorporate the technique into penetrometer probes. GPR responds to chan- ges in diclectric constant of the material which can be sensitive to contaminant type. However, the dielectric con- stant of soils varies over a wide range and is strongly influenced by the presence of water. The depth of penetra- tion is also controlled by soil type and excitation frequency. Penetration of the GPR can be very limited in saturated clayey soils ‘One ofthe most interesting areas of current research is on the development of integrated optoelectric chemical sensors. They contain a small diode laser which is focused on & chemically selective overlay. These sensors can be small and in modular form and can be inexpensive. ‘A major problem with the development and application of chemical sensors relates to the intoraction of the sensor with the contaminant. Most of the sensors require thatthe con- taminant (usually in vapour or liquid form) be pulled into the probe so that the chemicals can interact withthe sensor. This produces problems related to the cleaning of the sensor, ‘measuring cell and filter element to avoid eross-containa- tion, as well as the time required for this process. Little research has been carried out to evaluate these problems and the issues related to the interaction of the contaminant and soil, the interaction of the contaminant and measuring device, the contaminate state and contaminate mixtures.SUMMARY 203 8.7 SUMMARY ‘To better characterize potentially hazardous sites, improved investigation devices and methods are being developed ‘which use cone penetrometers. The CPT gathers high- quality, im situ geotechnical information in a rapid cost- offective manner, Electrical and chemical sensors have been adapted to cone penetrometer probes to enable mapping of subsurface contamination in sufficient detail to reduce the reed for more costly invasive subsurface sampling and ‘monitoring points. Traditional drilling and sampling meth- fods when compared to CPT methods have high waste ‘management costs from handling and disposal of contami- nated materials. Also, CPT methods minimi: field personnel to hazardous environments. Significant developments have been made in recent years to improve geo-environmental site characterization using penetrometer technology. Sensors have been added to the cone penetrometer to enhance the logging capabilities for both mechanical and chemical measurements, Bratton and Higgins (1992) describe a synergistic approach to 3-D site characterization by utilizing a combina- tion of surface geophysical technology and direct-push (that is, penetration test) technology. By taking advantage of the synergism between the two, a significant improvement can be achieved in the interpretation of the geophysical data exposure of