Numerical Evaluation of Flowmeter Test Interpretation Methodologies

Abstract

We evaluate systematic errors inherent in two flowmeter test interpretation methodologies. A physically consistent two‐dimensional groundwater flow model is used to numerically simulate the single flowmeter test and the double flowmeter test in 25 different two‐layer confined aquifers of a priori known horizontal hydraulic conductivities, Ki; specific storativities, Ssi, and hydraulic diffusivities, vi=Ki/ssi, in each layer (i=1, 2). Values selected for the layer hydraulic parameters span those encountered in natural sandy formations with the ratios of the corresponding parameters in the two layers falling in the following ranges: 1 ≤ K1/K2 ≤ 100, 0.0001 ≤ Ss1/Ss2 ≤ 100, and 1 ≤ V1/v2 ≤ 10,000. We find that the size of the hydraulic diffusivity contrast rather than the hydraulic conductivity ratio is the dominating factor for the parameter estimation accuracy in the two flowmeter methodologies. For the single flowmeter test methodology the ratio of the estimate over the corresponding true value falls within the range , whereas for the double flowmeter test it falls within . The double flowmeter test also provides estimates of layer specific storativity, Ssi. These are accurate only for the special case of equal layer hydraulic diffusivities. The test yields order‐of‐magnitude estimates of Ssi for layer hydraulic diffusivities differing from each other by up to an order of magnitude. For larger differences the estimation errors are much larger. Hydraulic parameters of a given layer are better estimated when the flowmeter is placed either exactly at the interlayer boundaries or, ideally, at two different points within each layer away from the boundaries. The results from simulated flowmeter tests in a five‐layer system are consistent with those for the two‐layer aquifers. This implies that the presented results most likely apply to flowmeter tests in arbitrary multilayer aquifers. Existence of significant errors in aquifer parameters estimated from synthetic flowmeter data demonstrate that the Theis [1935] model, which is assumed to be valid in each layer by the two considered interpretation methodologies, does not fully capture the flow dynamics in layered aquifers. This is illustrated by numerically calculated examples of simultaneously nonuniform and transient well face flux distributions in a layered aquifer. A model more sophisticated than that of Theis [1935] needs to be found.