Interpretation of DMT Results in Coarse-Grained Soils

The DMT is well suited for use in granular soils such as sands because of its robust construction. While there is some uncertainty as to the appropriate interpretation, there are currently a number of available correlations for use in granular soils. Table 6.11 presents a summary of some of the reported correlations suggested between various soil properties and DMT results for granular soils.

Table 6.11 Reported correlations between soil properties and DMT in coarse-grained soils

Relative Density (Dr)

Engineers may wish to use the results of the DMT to estimate the in-place relative density of sands, for example, to make an initial assessment of liquefaction potential or to evaluate effects of ground improvement. Even though the interpretation of DMT results in sand is still very complex, some suggestions have been made to estimate D_r from the DMT lift-off pressure via K_D.

Robertson & Campanella (1986) suggested a simple correlation between K_D and D,. in normally consolidated, uncemented sand as shown in Figure 6.34. Unfortunately, it appears that the penetration mechanics and value of P₀ in sands are highly complex and, in addition to D_r, depend on a number of other variables, including stress history, initial in situ stresses, compressibility, aging, and cementation. Konrad (1991) presented data from a dune sand that illustrated this complexity and the importance of stress history on the correlation between K_D and D_r. The test results indicate that in comparison to normally consolidated sands, overconsolidated sands show considerable scatter, making such a correlation difficult and perhaps unreliable.

A correlation for estimating D_R of sands using K_D and stress history and normalized DMT modulus (E_D/o'_vo) was presented by Lee et al. (2011) as shown in Figure 6.35.

Figure 6.34 Correlation between K_D and D_R in NC uncemented sands. (After Robertson & Campanella 1986.)

Figure 6.35 Correlation between K_D and D_R in NC and OC sands. (From Lee et al. 2011.)

State Parameter

The use of the state parameter to describe the behavior of sands was discussed in Chapter 4 as a rational approach combining the effects of void ratio and stress level (Been & Jefferies 1985). Konrad (1988; 1989) has suggested that the state parameter should also be considered to evaluate the results of DMTs in sand.

Results presented by Konrad (1989) demonstrated that for dry Ottawa sand the normalized state parameter (i.e., the state parameter divided by the value e_max-e_min) is related to the pressure difference Pi-P₀ normalized with respect to the mean or octahedral effective stress. Konrad (1989) suggested that the value of K_D obtained from the DMT should be related to both the mean effective stress and the normalized state parameter. Engineering properties of the undisturbed sand may then be obtained from existing correlations related to the state parameter.

Drained Friction Angle

It has been suggested that the penetration of the DMT blade in sands and other freely draining soils represents a drained bearing capacity failure approximating a plane-strain condition. The projected end of the DMT blade represents a long, narrow punch. Since the failure condition is controlled in part by the in situ state of stress in granular materials, it is reasonable to expect that the results from the DMT may be used to determine cp'.

Schmertmann Method

A procedure for obtaining the effective axisymmetric friction angle in sands was presented by Schmertmann (1982) using the wedge penetration theory and is attractive since it incorporates the horizontal effective stress that may be estimated from the DMT results. The method is iterative and requires measurement of the pushing thrust to advance the DMT, such that an estimate of the blade tip resistance may be obtained. This measurement may be made using a load cell located immediately behind the blade as previously described.

Campanella and Robertson Method

Campanella and Robertson (1991) observed that the DMT parameter K_D was directly related to both the normalized blade end bearing (q_d/o'._a), obtained from the measurement of thrust, which in turn could be related to parallel CPT-normalized tip resistance, i.e., q_c/o'_vo. Based on the fact that reasonably good estimates of cp' could be obtained in sands from q_c/a'_vo, they reasoned that it should be expected that a direct correlation should also exist between cp' and K_D. A chart for predicting peak friction angle was proposed by Campanella & Robertson

(1991) but requires an estimate of K„ from DMT results.

Marchetti Method

Marchetti (1997) suggested that the method given by Campanella & Robertson (1991) could give cp' directly in terms of K_D after first assuming some values of K₀ for different conditions: K_ONC = 1 - sin cp'; K„ = 1; and K₍₎ = K_P⁰-⁵. The resulting correlations between cp' and K_D are shown in Figure 6.36. The three curves shown in Figure 6.36 represent three different estimates of K_c, and it can be seen that for practical purposes, the correlation is only mildly sensitive to the value of K₀ for any given value of K_D (typically! 1.5° from the K„ = 1 curve).

To provide some conservatism, Marchetti (1997) suggested that an initial estimate of cp’, which is somewhat lower than all three curves of Figure 6.36, could be obtained from

This is shown by the lower solid line in Figure 6.36. This approach has been shown to be very reasonable based on triaxial compression tests on undisturbed sands (Mayne 2015).