A LABORATORY AND NUMERICAL ASSESSMENT OF THE GUELPH PERMEAMETER METHOD

Abstract

The Guelph permeameter method was assessed using laboratory and numerical procedures. Sand tank experiments using air-dry Caledon sand confirmed the theoretical basis of the method, that flow out of a small, uncased well into unsaturated soil rapidly decreases to a steady rate (Qs) and involves a finite wetting region. Saturated flow and field-saturated flow experiments found that the truly saturated hydraulic conductivity Ks, (no entrapped air present) for repacked Caledon sand was approximately a factor of 3.5 greater than the corresponding field-saturated hydraulic conductivity, Kfs (entrapped air present). Horizontal infiltration experiments, which were conducted primarily to measure the metric flux potential (φm) of air-dry Caledon sand, suggested that the discrepancy between Ks and Kfs was the result of an approximate 17% undersaturation due to the entrapped air. Application of the Laplace (K1fs) and Gardner (φ1m) analyses to the sand tank results yields approximately a factor of 12 overestimate of Kfs by K1fs, but only about a 10% overestimate of φm by φlm. This indicates that capillarity (parameterized by φm) dominated field-saturated flow (parameterized by Kfs,) in the sand tank experiments. A numerical model based on Richards' equation was able to accurately simulate the sand tank results using Kfs from the field-saturated flow experiments and φm from the horizontal infiltration experiments. Qs and C values were predicted within ≤2% of the experimental values. The numerical results suggest that Kfs, (rather than Ks) and φm are appropriate parameters for describing flow out of a well into unsaturated, structureless soil; and that the integrally correct representation of the K(ψ) relationship provided by the exponential form of Gardner (1958) and by the method of calculating α, can be adequate for describing both transient and steady flow. A three-member family of numerically derived, unsaturated flow C value curves representing “sand,” “loam and structured clay,” and “unstructured clay” appears to compensate adequately for the neglect of the gravity-capillarity interaction in the Guelph permeameter equations. Sensitivity analyses indicate that for the Richards analysis the correct choice of C value curve from the three-member family of curves is least critical for the hydraulic parameter of primary importance in describing the flow. When field-saturated flow is the dominant component of flow out of the well, the error induced in Kfs, by incorrect choice of C is on the order of ≤30% for the flow conditions analyzed. When capillarity is the dominant component of flow out of the well, the error induced in φm by incorrect choice of C is on the order of ≤10% for the flow conditions analyzed. The sensitivity analyses also confirm that the Laplace analysis is reasonably accurate when the field-saturated component of flow dominates, and that the Gardner analysis is reasonably accurate when the capillarity component of flow dominates.