Analysis of a gas-phase partitioning tracer test conducted in fractured media

This work successfully applied the gas-phase partitioning tracer method to determine the NAPL, water and air saturations in the vadose zone at the field scale. This project was one of the first, and still one of the few, field-scale gas-phase partitioning tracer tests. This work differs from other work in that it was conducted in a high water content, fractured clay. There were three primary components of this work. First, gas-phase tracers were identified and their NAPL-air and Henry's Law constants measured. There were four types of tracers used in this study: noble gases or nonpartitioning tracers; alkanes, which were expected to be nonpartitioning tracers; perfluorides, NAPL partitioning tracers; and halons, NAPL and water partitioning tracers. A laboratory method for measuring NAPL-air partition coefficients was developed and TCE-air partition coefficients were measured for the perfluoride and halon tracers. The second component of this study involved conducting a field-scale gas-phase partitioning tracer test, the results which were used to estimate NAPL, water and air saturations. The NAPL saturation, calculated to be an extremely low value, resulted in an estimate of NAPL mass present that is similar to the amount that has subsequently been extracted from the test site via SVE remediation. The alkane tracers, which had been used previously in laboratory column studies as nonpartitioning tracers, were more retarded than the perfluoride tracers at this site. It was the alkane tracers, and not the halon tracers, that were used to determine the water content. The water content was estimated to be approximately 90%, which is unexpectedly high for a vadose zone. Additionally, the tracer response time, vacuum data, and other geological data indicated that the tracer test was performed in fractured clay. The third component of this work comprised an analysis of the tracer test data to determine transport parameters. The analysis employed matching eight simple mathematical models to the experimental data. All of the models tested: two porous, three double-porosity, and three fracture-based (single fracture, multifracture, fracture-matrix) models could reasonably match the experimental data and no one model resulted in consistently superior predictions than the others