From Calibration to Characterization: Enhancing Cone Penetration Test Interpretation via a New Calibration Chamber/Miniature Cone System and Spatial Variability Modeling
Advisor: Professor Scott M. Olson
Abstract
Among field investigation methods, the cone penetration test (CPT) is widely used to obtain in-situ soil properties because of its efficiency and reliability. Furthermore, CPT data typically are interpreted via empirical correlations to estimate soil properties for design. However, design reliability can be compromised by two major uncertainty sources: (1) empirical correlations used in point-specific parameter estimation; and (2) spatial variability across a site. This dissertation addresses both sources of uncertainty through controlled laboratory testing and advanced spatial modeling. A miniature calibration chamber (CC) system was developed to refine empirical correlations under laboratory-controlled conditions. The chamber utilizes a 150-mm diameter specimen and an 8-mm diameter miniature cone penetrometer, enabling efficient measurement of tip resistance and sleeve friction while substantially reducing time and material requirements compared to full-scale CC systems. Validation tests on clean Ottawa sand showed strong agreement with published full-scale CC data, confirming the reliability of the system and test procedures. Additional calibration chamber tests on silty sand were conducted to examine drainage effects on intermediate soils. The study demonstrates that normalized tip resistance decreases with increasing compressibility for similar state parameters. Correlation assessments illustrate the need to incorporate compressibility in empirical models for clean sand, and most existing correlations overestimate tip resistance in silty sands, particularly at higher effective stress levels.
To extend these insights to field-scale characterization, CPT data from 77 soundings at the Mallard Lake, Illinois site were analyzed. Horizontal spatial variability was modeled on field cone tip resistance measurements using semivariogram analysis and Integrated Nested Laplace Approximation (INLA). Conditional simulations based on spatial structure from semivariograms and Bayesian inverse analysis then were applied to estimate the undrained shear strength, integrating both spatial variability and model uncertainty. This framework can be transferred to other empirical correlations beyond the relationship examined in this study.
Overall, this research establishes a comprehensive CPT-based framework for geotechnical site characterization that combines laboratory calibration and spatial variability modeling. The integrated approach improves confidence in soil property estimation and geotechnical design by linking controlled testing with field-scale probabilistic analysis.