Seismic Cone Penetration Testing

There is considerable interest in methods of geotechnical in-situ engineering which enable shear (S)and compression (P) wave velocities in the ground to be accurately estimated. These measurements provide insight into the response of soil to imposed loads such as buildings, heavy equipment, earthquakes, and explosions. The S-wave and P-wave velocities are desired because they form the core of mathematical theorems which describe the elasticity/plasticity of soils and are used to predict settlement, liquefaction, and failure. As such, accuracy in the estimation of shear and compression waves velocities is of paramount importance because these values are squared during the calculation of geotechnical parameters such as the Shear Modulus, Poissons Ratio, and Youngs Modulus (Ohya, 1982), among others. For example, from elasticity theory we know that the formula for the maximum Shear Modulus is Gi = VS2, where is the soil density and VS is the shear wave velocity. Figure 1 illustrates how Gi is used to predict Shear Modulus values at different strains during dynamic soil analysis (ie G = Gi [ 1 -r /rmax] .

Two methods have developed over the years and are widely used in geotechnical engineering to indirectly estimate the Shear Modulus: the Standard Penetration Test (SPT circa 1927) and the Cone Penetration Test (CPT). These tests use empirical correlations, developed with large amounts of data, to relate certain measurements taken during the course of these tests to values for the geotechnical parameters mentioned above. However, these empirical correlation are subject to large variances, and therefore require large factors of safety to be employed when using them in construction design. As an example, the SPT N-value/ Shear Modulus relationship is described as

Gi = 125 N0.611 kg/cm2 ± 500 % (Imai & Tonouchi, 1982) (SPT)

Also, CPT qC (bearing pressure)/Shear Modulus relationships are sometimes used in foundation design, albeit less frequently, where Gi = 0.8 qC. No real error estimate has been made on this empirical correlation and this parameter is more often used for soil type identification (Figure 2) or as support for SPT N-values. However, in both CPT and SPT measurements considerable error is involved, and industry therefore desires a better predictor of elastic constants.

For that purpose the Seismic Cone Penetration Test (SCPT) (an extension of the Cone Penetration Test (CPT)) was devised to measure seismic velocities directly through data obtained by an installed seismic sensor in the cone penetrometer, in addition to the standard bearing pressure, sleeve friction,

and pore pressure sensors (Campanella et al, 1986). As the cone penetrometer is advanced through the ground, using a pushing force, the advance is halted at one meter (or other such increment) intervals. When the cone is at rest, a seismic event is caused at the surface using a hammer

blow or explosive charge, causing seismic waves to propagate from the surface through the soil to be detected by seismic sensors installed in the cone penetrometer. This event is recorded and the penetrometer is advanced another increment and the process is repeated. By determining the seismic arrival times average velocities are calculated over the depth increment under study.

A further in-situ measurement is also desired by industry (especially in Japan), the factor Q. This parameter is required in order to fully model, and therefore predict, a soil areas response to dynamic loads caused by ground shaking, heavy equipment vibration, storm wave motion, or ice loading. Many software programs for dynamic modeling do exist, but all require accurate inputs, especially for attenuation values. In this respect, Dr.Finn (1984) states that dynamic analysis in geotechnical practice is intimately related to the capability of measuring the necessary soil properties.

Present techniques being used to derive P- and S-wave attenuation rely on power spectrum and spectral ratio techniques (Bruce B. Redpath and Richard C. Lee, 1986). These techniques have several inherent disadvantages associated with them. Primarily, due to the nature of obtaining power spectrums and spectral slopes from short duration seismic signals, these methods result in certain frequency spectrum effects such as "leakage" which are dramatically magnified when calculating the power spectrum, creating uncertainty. In addition, spectral ratios can be statistically biased when noise or other seismic body waves (than those of interest) are present in the trace (Pisarenko, 1970) and demand highly sensitive and calibrated seismic receivers. As a result of uncertainty, power spectrum and spectral ratio techniques can only be practically applied over large soil profile intervals (e.g., 16-20 meter intervals) where changes in attenuation would be dramatic and quantitative, albeit subject to error. To obtain higher resolution soil attenuation (e.g., 1 metre intervals) it is necessary to design a digital filter which does not work in the frequency domain, but in the time domain. Such a filter is a Kalman Filter. BCE implements more precise real-time time-domain Kalman Filtering techniques to derive attenuation estimations from SCPT data.

In the dynamic response analysis of soils in engineering practice the different states of shearing deformations are the elastic, elasto-plastic, and failure, where the corresponding soil models are the linear elastic, visco-elastic and load history tracing, respectively. The associated methods of response analysis are the linear method (linear elastic model), equivalent linear method (visco-elastic model) and step-by-step integration method (load history tracing type model).

In all of the above methods of analysis the fundamental input geotechnical parameters are the maximum shear modulus and attenuation characteristic of the soil profile under study. Some of the dynamic analysis computer programs developed include SHAKE (Seed, 1972), QUAD-4 (Seed, 1973), LUSH (Seed, 1974), FLUSH (Seed, 1975), MASH (Seed, 1979), CHARSOIL (Streeter, 1973), DESRA-1 (Finn, 1975), DESRA-2 (Finn, 1978), TARA (Finn, 1981) and TARA-2 (Finn, 1981).



Glossary of terms

attenuation: rate at which the soil decreases the amplitude of the waveform under study.

in-situ : Latin term meaning on-site or "in the situation".

SCPT: Seismic Cone Penetration Test: non-destructive method of testing soils on site for the evaluation of settlement, liquefaction, and failure potential. Uses electronic probe, equipped with seismic sensors, to measure pore pressure, sleeve friction, and seismic signals.

References

Baziw, E.J. (1986), "Concepts of Kalman Filter and the Application to Seismic Time Series", BASc Thesis, Dept. of Geophysical Engng., Univ. of British Columbia.

Baziw, E.J. (1988), "Application of Digital Filtering Techniques for Reducing and Analyzing In-Situ Seismic Time Series", MASc Thesis, Dept. of Civil Engng., Univ. of British Columbia.

Baziw, E.J., Campanella, R.G., and Sully, J.P. (1989) "Interpretation of Seismic Cone Data Using Digital Filtering Techniques", presented at the 11th International Conference on SoilMechanics and Foundation Engineering, Rio de Janeiro, August.

Campanella, R.G., Robertson, FTC and Gillespie, D. (1986). Seismic Cone Penetration Test. Proc. INSITU86. ASCE, Geot. Spec. Publ. No. 6, June, 116-130.

Finn, W.D. Liam (1984). Dynamic Response Analysis of Soils in Engineering Practice, Mechanics of EnGineering Materials. John Wiley & Sons Ltd. Chapter 13.

Ohya, 5. (1985). In Situ P and S Wave Velocity Measurement,unknown publisher, pp.1218-1235.

Baziw, E.J., and Weir-Jones, Iain, "Application of Kalman Filtering Techniques for Microseismic Event Detection", submitted to Journal of Pure and Applied Geophyics, May, 2000.

Baziw, E.J., "Digital Filtering Techniques for Interpreting Seismic Cone Data", Journal of Geotechnical Engineering, Vol. 119 No. 6 ASCE, June 1993.



Related References

Close & Fredrick, Modeling and Analysis of Dynamic Systems, Houghton Miflin Co., 1978.

GeIb, A., Applied Optimal Estimation, The M.I.T Press, 1978.

Ogata, J., Discrete Process Control, Cambridge University Press, 1987.