Analysis of Cross-hole Tests in Fractured Systems

The estimates of transmissivity (T) and storativity (S) obtained from analysis of observation-well responses to constant-rate pumping tests are affected by the correlation between the specified (measured) flow rate and the subsequent pressure response, with both the magnitude and rate of change of the pressure response being a function of the diffusivity (D) – the ratio of T to S. Traditional well-test analysis methods assume that the fluid withdrawal during a constant-rate test is evenly distributed within the domain, but in a heterogeneous/fractured system, this assumption will be violated to some extent. For example, drawdown in higher T zones within a heterogeneous system will be increased relative to drawdown in a homogeneous system with the same “high” T value, simply because the water is preferentially withdrawn from the higher T zones in the heterogeneous system. This relative increase in drawdown in a heterogeneous system, coupled with the implicit assumption of evenly distributed pumping, will result in underestimating the T value in the “high-T” areas of the heterogeneous system. It is because of the uncertainty in the effective pumping rate at any observation point within the tested domain that D, rather than T and S individually, is actually the parameter that can be determined. In addition, errors in assumed flow-path length between pumping and observation wells may render even the estimate of D meaningless. Analysis of synthetic data sets obtained from numerical simulations of flow in binary heterogeneous fields (analogous to fractured systems) shows that individually estimated T and S values can be in error by orders of magnitude. These systems, like fractured systems we have tested, often exhibit a subradial flow signature, and the correlation between the assumed system geometry and the estimates of T and S makes those values virtually meaningless. Traditional well-test analysis of complex flowsystem observation-well responses should be largely abandoned and the observed responses should instead be used to calibrate areal (2D or 3D) models. *Sandia is a multi-program laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy's National Nuclear Security Administration under contract DEAC04-94AL85000. This research is funded by WIPP programs administered by the Office of Environmental Management (EM) of the U.S. Department of Energy.