In the last several years a variety of new tools for interpreting interwell tracer tests have been developed. The new methods are based on residence time distributions of the tracer, where much of the previous work used only the mean residence time. Using the distribution of residence times extends the power of moment analysis by allowing for the determination of reservoir properties and flood performance as a function of time. Flow geometry and construction of flow capacity - storage capacity diagrams also follows directly from the analysis. Swept volume vs. time, and sweep efficiency are also determined from the residence time distribution, as is remaining oil saturation. One important key to these new methods is our use of the integrated tracer recovery histories. Estimating residual oil saturation is greatly simplified by our mathematical treatment of slug tracer injection. Examples are presented that show improved saturation estimates even at early times in a tracer test.
This paper describes the new analysis methods developed recently and shows by comparisons with analytical and experimental data that the methods are accurate and robust. The method is simple and can be done with a spreadsheet using only produced tracer concentration data; it does not require a reservoir model or numerical simulation. The equations are derived from first principles for a very general case that includes both conservative and partitioning tracers produced from any heterogeneous reservoir.
Introduction and Motivation
Successful fluid injection programs for improved oil recovery require detailed information on reservoir heterogeneity and connectivity, remaining oil saturation and its distribution, and estimates of reservoir volume contacted as a function of volume injected (i.e., sweep efficiency). Methods for assessing these reservoir and operational properties include repeat seismic interpretation, detailed reservoir simulation, production log analysis, and tracer interpretation methods. While all of these methods offer insight into secondary or tertiary recovery mechanics, only tracer methods provide information at the appropriate interwell scale, with less spatial averaging than typical in numerical methods.
Oilfield tracer testing is a mature technology, with more than 50 years of application (Du and Guan, 2005). An extensive literature review by these authors shows 43 tracer tests reported in the literature, 60% of which were aqueous phase tracers, and the balance gas phase tracers. They further report that nearly 70% of the test results were interpreted qualitatively, 25% used analytical interpretation, and 12% were interpreted numerically (in some cases more than one method was used, so these do not sum to 100%). Quantitative tracer analysis was first presented to the oil industry by Brigham and Smith (1965) for homogeneous repeat 5-spots, and Abbaszadeh-Dehghani and Brigham in 1984 for layered media. For unit mobility ratio, and neglecting gravity and capillary forces, they show a general solution to tracer transport for arbitrary well configuration, from which swept pore volume can be determined. They also show an iterative method to estimate individual layers' flow capacity, kh, and storage, ?h, by deconvolving the combined tracer signal into individual layer responses. The fundamental restrictions in their development were the assumption of intra-layer homogeneity and unit mobility ratio.
Tang and coworkers (Tang and Harker, 1991a, 1991b; Tang, 1995, 2003) developed a method to estimate residual oil saturation that was based on chromatographic theory. In their original work, they show a tracer breakthrough curve of conservative and non-conservative (partitioning) tracers can be collapsed to a single curve by "correcting" the partitioning tracer's residence time by its partition coefficient and the residual oil saturation. This correction is applied at various points in a tracer test (e.g., tracer breakthrough, 10% recovery, etc.). Tang (2003) also extended the Brigham and Smith (1965) model to estimating oil saturation in individual layers from tracer response. Again, these methods assume constant intra-layer properties.