Permeability and Saturation Distributions from Tracer Data

by
Tom Harneshaug, MSE

University of Texas at Austin, 1997
Supervisor: Dr. Gary A. Pope and Dr. Larry W. Lake

Interwell tracer tests commonly used in the oil industry to measure residual oil saturation in petroleum reservoirs have now been adopted by the ground water community as a means of estimating the saturation of contaminants in aquifers. So far, the use has been limited to measurements of average residual saturation, but many aquifer remediation techniques require knowledge about the spatial distribution of contaminants to be efficient.

In response to this need, this thesis describes how the spatial distribution of contaminants can be found from tracer data. The field tracer data were collected from a partitioning interwell tracer test at a contaminated site at Hill Air Force Base, Utah. Two methods were used to infer the saturation distribution. First, the method of moments was used to obtain an estimate of the saturation distribution. The second method was history matching with a computer simulator. History matching involves gradually adjusting one or more matching parameters as a means of matching observed data. The simulator used in this research, however, regresses on only one set of parameters at a time, so for the solution to be accurate, knowledge about the other parameters is necessary.

In our research, the spatial distributions of both permeability and saturations are unknowns, although good estimates of the average values are available. Therefore, a two-step approach was adopted. First, by using the non partitioning tracer which is independent of saturation, the permeability field was estimated. Several good matches of the non partitioning tracer data were obtained. Second, the partitioning tracer data were used to obtain the saturation distribution, now with the permeability from step one treated as known. Several estimates of the NAPL distribution were obtained, but the NAPL distributions were different for different permeability fields. To overcome the sensitivity to permeability, the simulator was rewritten to match the retardation factor rather than the individual tracer response data. The results from matching the retardation factors agree with the results from the first moments. Also, there is agreement between the estimated NAPL distributions for two different permeability fields. This agreement indicates that the NAPL saturation distribution can be estimated without much knowledge about the permeability.

Back to theses index