Problems and possibilities with combined analyses of PLT and pressure-transient data from layered reservoirs are discussed, with standard single-layer methods used to estimate parameters, and model verification based on matches of available PLT and pressure-transient data used to complete analyses.
For commingled reservoirs it is shown that if the total flow capacity can be obtained from a direct analysis, then the flow capacity and skin of each layer from stable and short-time PLT profiles can also be obtained by simple two-rate methods. Assuming repeatability of the data, the PLT profile after shut-in can be determined from another test. Also, the rate of crossflow after shut-in relative to the production is shown not to be relevant for the success of this approach.
The main problem with layered-reservoir data is contrasts in skin values, both for commingled and crossflow reservoirs. With severe contrasts in skin values, analyses can become highly uncertain regardless of method used. This is shown by closed-form approximations of infinite-acting solutions for commingled reservoirs and illustrated with synthetic data. A field case from the Norwegian sector of the North Sea has also been included to highlight the problems.
Pressure-transient data from layered reservoirs used alone in analyses are highly non-unique if information on individual layers are sought. Results can also be misleading when such data are used to determine total flow capacity and average permeability of the system and to estimate a total system skin value. The point is that information on crossflow parameters and differences in skin values really are needed to judge the quality of the data, and hence of analyses.
Even if the pressure-transient response can be matched with synthetic data from a sufficiently accurate layered model, one cannot be sure of the analysis. What one really needs is some supporting information. Knowing the permeability distribution in a relative sense, e.g., from logs, certainly helps, but uncertainties in crossflow parameters and layer boundaries can still make analyses highly uncertain. On the other hand, if combined with stable and short-time PLT profiles, then the non-uniqueness can be reduced significantly in many cases.
Using as much information as possible is obviously a requirement to obtain a best possible interpretation of pressure-transient data from layered reservoirs. The problem being that various types of data can be matched with completely different models. A case in point is that any PLT profile can be matched with any not too extreme permeability profile by adjustment of skin values alone. With both stable and short-time PLT profiles known the situation can change considerably, at least with some restriction on crossflow. With easy crossflow, layering in a relative sense must be obtained from cores or log data.