Relating Cone Penetrometer Test Information to Geophysical Data: A Case Study

Anthony L. Endres1,3 and William P. Clement2,3

1Centre for Groundwater Studies and Department of Geology and Geophysics, University of Western Australia, Nedlands, WA, Australia

2Center for Geophysical Investigation of the Shallow Subsurface (CGISS), Boise State University, Boise, ID 83725

3Formerly Institute for Scientific Research, Boston College, Chestnut Hill, MA

Summary

Analysis of data from a recent experiment at Dover Air Force Base (AFB) has revealed a relationship between soil types determined from the mechanical properties measured by cone penetrometer tests (CPT) and the electrical properties that influence geophysical methods. This correlation connects two distinctly different types of physical properties and provides a petrophysical basis for combining information obtained from CPT and geophysical techniques governed by electrical properties. We observed this relationship through the use of semi-logarithmic crossplots of dielectric permittivity versus electrical resistivity to find that CPT soil types cluster in a systematic manner to form a linear trend from clay-prone to sand-prone lithologies. We obtained improved segregation of soil types when other factors, such as location relative to the water table and stratigraphy, were used to refine the analyses of these data. In addition, results indicate that the ratio of the permittivity to the logarithm of resistivity is a good geophysical discriminator of the engineering soil classification.

Introduction

The interpretation of geophysical data acquired to investigate environmental and geotechnical problems is greatly improved if subsurface information is available. One source of subsurface information in unconsolidated materials is the cone penetrometer test (CPT). This device measures mechanical properties (e.g. sleeve friction, penetration stress and pore fluid pressure) that are used to infer soil types. Closely spaced sampling is done as the device is hydraulically driven into the ground, resulting in a well log-type profile of the subsurface.

The CPT profiles provide information about subsurface composition and interfaces that should be are useful when combined with near-surface geophysics in site characterization. This point has already been empirically demonstrated by Wyatt et al. (1996) and Clement et al. (1997a,b). Conversely, CPT profiles, like conventional well-logs, provide detailed information that is localized to the immediate vicinity of the borehole. Further, CPT is an invasive technique; it may be necessary to limit the number of CPT profiles at sites where disturbance of the subsurface must be minimized. Near-surface geophysics is a non-invasion technique that can potentially provide the spatial distribution of the mechanical properties measured by the CPT probe.

However, the relationship between the CPT-determined soil types and the physical properties controlling the response of geophysical methods is not well established. The interdependence between these quantities is not apparent because mechanical properties sampled by CPT are fundamentally different from many geophysically important properties, such as electrical properties. To further our understanding of a possible connection, we have examined a data set from a recent experiment where detailed, in situ measurements of dielectric permittivity and electrical resistivity were done coincident with CPT profiling. While our work is ongoing, the analysis of this data set has revealed some interesting connections between CPT soil type and electrical properties. These relationships provide a petrophysical basis for the combined use of CPT-derived soil type profiles and near-surface geophysical surveys.

Experiment and Site Description

The data set used in this analysis was acquired as part of the site characterization of the Groundwater Remediation Field Laboratory (GRFL) at Dover AFB in Delaware. In addition to the CPT data, a variety of surface geophysical surveys (i.e., reflection and refraction seismology, ground penetrating radar profiling, electrical resistivity sounding and terrain conductivity) were performed. A discussion of the surface geophysical data is given by Clement et al. (1997a,b).

The site is located on a level, grass field. The underlying sediments are part of a Pleistocene fluvial system with lithologies ranging from clayey silt to coarse sand units (Clement, 1997a). The surficial aquifer system is underlain by an aquitard located at about 12 meters (39 feet) depth. From nearby observation wells, the water table was approximately 7.3 meters (24 feet) below the ground surface. The water table was not readily identifiable in the data from the CPT profiles or surface geophysical surveys, the except for the seismic refraction data.

Applied Research Associates (ARA), Inc. acquired the CPT data in 1995 and 1996. The CPT measurements were performed at fifty eight locations distributed across the site. A variety of instrument packages were attached to the CPT string to obtain dielectric permittivity, electrical resistivity, oxidation-reduction potential and seismic wave velocity measurements in conjunction with the mechanical property data used to infer soil type. The electrical resistivity measurements were obtained from a small scale, Wenner-type ring array. A resonant frequency system was used to measure the dielectric permittivity; the measurement frequency of this device is 100 MHz. The data analyzed in this study are the eight locations where both permittivity and resistivity profiles were obtained. Complete profiles through the surficial aquifer system into the aquitard were obtained at four of these sites. At three other sites, profiles were taken of the upper portion of the aquifer; the lower part of the aquifer was sampled at the remaining site.

Data Analysis

A variety of soil classification schemes are used in the interpretion fo CPT measured mechanical properties (see Zhang and Tumay (1996) for a discussion). We use the CPT soil classification determined by ARA using a scheme based on Robertson (1990). Two quantities, the normalized cone resistance and the normalized friction ratio, are used to group soils into nine classes that differ in terms of grain size and consolidation; these soil types are described in Table 1. A primary feature of this scheme is the progression from clay to gravelly sand as the soil classification increases from two to seven. Soil classes eight and nine are overconsolidated materials that lie outside of this progression. Both of these soil types are uncommon at this site and occur only within three feet of the surface; they will not be considered in our analysis. As well, soil class one, representing underconsolidated materials, is off the primary two to seven sequence; this class was not found at the site.

Our analysis of the data is based on crossplotting dielectric permittivity versus electrical resistivity in a semi-logarithmic fashion. Figure 1 shows the results of this procedure when the data from the eight CPT pushes are considered. Data points corresponding to specific CPT soil types cluster in a systematic manner, forming a linear trend on the seimi-logarithmic crossplot from clays with higher permittivity and lower resistivity to sands having smaller permittivity and larger resistivity. Petrophysically, this is a very interesting result because it implies an interdependence between two fundamentally different physical properties (i.e., mechanical and electrical).

We attribute this trend to the differences in the electrical properties of these two lithology types. Electrical resistivity in sands is principally governed by current flow through the conductive pore water. In clay-rich lithologies, surface conduction is a major mechnaism affection electrical currents (Guéguen and Palciauska, 1994). The effects of surface conduction are substantially larger than conduction through the pore water when water salinity is low, such as in our freshwater aquifer. The higher dielectric permittivity of the clays is due to their higher water content relative to the sands. The permittivity of water (approximately 80) is substantially higher than other components of soils; hence, the greater water content of the clays leads to their larger permittivity.

Unfortunately, substantial overlap of the points corresponding to the CPT soil classifications occurs, blurring the discrimination of similar soil types using only these two parameters. To obtain improved segregation of CPT soil classifications, we have used other parameters in our analysis. First, we considered the sample location relative to the water table; hence, the effects of water saturation could be determined. Figures 2a and 2b show the permittivity-resistivity crossplots for the sections above and below the water table depth at 24 feet depth, respectively. Comparison of these figures reveals the existence of two distinct subsets in the data due to the effects of water saturation. The subset containing the saturated section below the water table is shifted to lower resistivity relative to the equivalent CPT lithologies in the unsaturated section above. This effect is the result of cnductive water replacing the non-conductiong air as the water saturation increases.

In addition, the dielectric permittivity of the sand-prone soil types in the saturated section is significantly greater than the permittivity of the corresponding lithologies in the unsatruated zone. Little or no shift in the dielectric permmittivity due to changes in water saturation is eveident for the clay-prone soil types. The presence of high permittivity of water versus the permittivity of the sand and air (approximately 5 and 1, respectiveily). Due to the large amounts of hygroscopic water present in clays, this mechanism will cause much less change in the dielectreic permittivity of clay-prone lithologies.

While both data subsets retain linear trends on the crossplots, the scatter for the section above the water table (Figure 2a) is greater than that for the water saturated section below the water table (Figure 2b). This is probably due to the relative amount of lateral stratigraphic variation in these two sections. The comparison of the individual CPT profiles reveals that the lower section has substantially less lateral stratigraphic variation than the upper section. In particular, the upper section contains a thick clay unit in the northern part of the study area; the equivalent interval in the southern portion consists of interbedded clays and sands.

To investigate the effects of lateral lithologic variations, we have separately crossplotted the sections above the water table at 24 feet for the CPT profiles in the northern and southern parts of the field site; these are shown in Figure 3a and 3b, respectively. Comparison of these two figures shows that each contains a distinct population. While sand-prone soil types are comparable, the two populations diverge for the clay-prone lithologies. This difference suggests that these upper clayey units are disimilar in the northern and southern parts of this area. This distinction is further illustrated by crossplotting only the clay-prone units, as is done in Figure 4 where the lower aquitard clay is included for comparison. Each of these three clay-prone units is distinct in terms of their electrical properties.

Different clay types, as suggested by the electrical properties, could have significant consequences when correlating the upper clay-prone units in the study area. It could indicate that the southern clay units are stratigraphically different from the northern clays. This possiblity would have implication for the continuity of the clay units and, in turn, for the hydrogeology of the upper section. Since the CPT soil classifications make no such distinction between the clays, a very different interpretation of the subsurface geology could result based only on the CPT soil classifications.

The linear trend observed in the crossplots suggests that the ratio of dielectric permittivity to the logarithm of electrical resistivity is a good electrical property discriminator of the CPT soil type. To test this hypothesis, we have generated depth profiles of this quantity for the eight CPT locations; two example profiles are given in Figure 5. It can be seen that the clay-prone soil types have larger values of this ratio relative to the neighboring sand-prone units, although this comparison changes above and below the water table. In addition, a progressive increase in the ratio is observed between the depths of 20 and 30 feet where the CPT lithology remains sand-prone throughout. This ramping feature suggests a thick transition zone between the residual and fully saturated sections. This is consistent with the geophysical imaging results of Clement et al. (1997a). They attributed diffuse impedance contrasts due to a thick transition zone as the cause for the lack of water table reflections in the seismic reflection and ground penetrating radar profiles.

Conclusions

Our analysis has found that a relationship exists between CPT soil type based on mechanical response and geophysically important electrical properties, two distinctly different types of petrophysical properties. The relationship is best seen by the semi-logarthmic crossplotting dielectric permittivity versus electrical resistivity. Data points corresponding to specific CPT soil types cluster in a systematic manner, forming a linear trend on the crossplot from clay-prone to sand-prone CPT lithologies. However, substantial overlap of these clusters occurred, blurring the discrimination of similar soil types using only these two parameters. We obtained improved segregation of clusters when other factors, such as location relative to the water table and stratigraphy, were incorporated into the data analysis. In addition, we found that the ratio of permittivity to the logarithm of resistivity is a good geophysical discriminator of the engineering soil classification.

The observed interrelationship between CPT soil type and geophysical response has several important implications. First, it provides a petrophysical basis for the use of CPT-derived soil type profiles in the interpretation of near-surface geophysical surveys. Second, it implies that geophysical methods are a non-invasive method to spatially map CPT soil type, an important feature for some sensitive sites.

Acknowledgments

The authors would like to thank Chris Bianchi and Jim Shinn of Applied Research Associates, Inc. for very useful discussions about the cone penetrometer test methodology. This work was partially supported by the USAF Air Force Office of Scientific Research.

References

Clement, W. P., Cardimona, S., and Kadinsky-Cade, K., 1997a, "Geophysical and geotechnical site characterization data at the Groundwater Remediation Field Laboratory, Dover Air Force Base, Dover, Delaware," Proc. SAGEEP, pp. 665-673.

Clement, W. P., Cardimona, S., Endres, A. L., and Kadinsky-Cade, K., 1997b, "Site characterization data at the Groundwater Remediation Field Laboratory," The Leading Edge, v. 16 (11) , pp. 1617-1621.

Guéguen,Y., and Palciauska, V., 1994, Introduction to the Physics fo Rock, Princeton University Press, Princeton, NJ.

Robertson, P. K., 1990, "Soil classification using the cone penetration test," Canadian Geotechnical Journal, v. 27, pp. 151-158.

Wyatt, D. E., Waddell, M. G., and Sexton, G. B., 1996, "Geophysics and shallow faults in unconsolidated sediments," Ground Water, v. 34, pp. 326-334.

Zhang, Z., and Tumay, M. T., 1996, "Simplification of soil classification charts derived from the cone penetration test," Geotechnical Testing Journal, v. 19, pp. 203-216.