From Bureau of Economic Geology, The University of Texas at Austin (
For more information, please contact the author.

Bureau Seminar, March 10, 2006

The Use of Fracture Surveys and Hydraulic Tests for Improving Predictions of Flow and Transport in Fractured Chalk

Daniel Kurtzman


A field test site located in a fractured chalk aquitard in the northern Negev, Israel, was designed to investigate flow and transport phenomena at a scale of 10 to 100 m. Outcrop surveys of fractures, as well as single-borehole slug tests in packed-off intervals, were conducted to statistically characterize the geometric and hydraulic properties of the fracture population on site. Multi-borehole pumping tests and a forced-gradient (multi-borehole) tracer test were carried out to improve our understanding of flow and transport at the network scale.

A new methodology that helps researchers decide between different possible conceptual flow models was developed and applied. In this methodology a homogeneous expected dilution factor (HEDF), which is calculated from the pumping test results, is compared with the actual dilution factor (DF) inferred from the tracer test. Applying this methodology to the pumping and tracer tests performed in a pair of boreholes at the investigated site resulted in large differences between the HEDF and the actual tracer DF, reflecting the deviation of the actual flow from homogeneity. These results imply that flow is significantly channeled to narrow sections of the fractures' planes. Combining this conclusion with the results of the fracture surveys and the other tracers' breakthroughs suggests that the dominant features controlling flow at the site are horizontal networks of channels.

A statistical characterization of the geometric and hydraulic properties of the fracture population formed the basis of discrete fracture network (DFN) models designed to simulate the multi-borehole tests. Problems with calibration of stochastic DFN models to the multi-borehole tests' observations shifted the modeling effort toward equivalent deterministic DFN models. Two models were tested: the vertical fractures (VF) model, consisting of only vertical fractures, and the fractures' intersections (INT) model, consisting of vertical and horizontal fractures having enhanced transmissivity at their intersections. The average accuracy of transient drawdown predictions of both models was 66%. Breakthrough predictions of the INT model did a better job of capturing the fast first arrival and peak recovery time than the VF model. These results are in line with the previously conceptualized channel-network flow derived from the macroscopic analysis of these tests.