Modeling aquifer heterogeneity using cone penetration testing data and stochastic upscaling methods

Abstract

Cone penetration testing (CPT) has become an increasingly popular characterization method for subsurface investigations under 60 m (200 ft) depth in the Atlantic coastal plain of South Carolina. The shallow Tertiary sediments consist primarily of interbedded and interfingering fluvial, deltaic, and shallow-marine sediments. Cone penetration testing is relatively inexpensive and does not require disposal of drilling fluid or cuttings. At the Savannah River Site, CPT is typically used to obtain depth-discrete groundwater samples and small-diameter permeability samples, and to define hydrostratigraphic horizons. The focus of this study is an environmental waste site where CPT at 139 locations was used to define contaminant plumes and hydrostratigraphy over an 8-km2 (3-mi2) area, instead of conventional borehole techniques (e.g., monitoring wells, cores, electric well logs, slug, and pumping tests). This investigation used the CPT lithologic data to predict hydraulic conductivity variations in hydrostratigraphic zones of the uppermost aquifer unit. The method developed involves correlating tip resistance, sleeve resistance, and pore-pressure measurements to fines (mud, silt, and clay) content and hydraulic conductivity. Predicted fines content at the scale of the CPT measurements (0.03 m; 0.1 ft) are then categorized into high, medium and low conductivity and upscaled to the flow model resolution using a geostatistical approach. The resulting model conductivity field provides a realistic representation of aquifer heterogeneity in coastal-plain sediments and significantly improves groundwater modeling predictions. Subsequent contaminant transport simulations compare well with field data.