This work is motivated from issues that arise during the analysis of dynamic fracture injection tests (DFITs). It often occurs with DFITs that the rate history before shut in is somewhat uncertain, but the total injection volume is measured accurately. The analysis method (PTA) commonly used by the reservoir engineering community is always assumed to require an accurate and detailed rate history. In this paper two main questions are addressed:
What is the impact of both poor rate measurement and imprecise knowledge of the times rates change on the shape of the resulting log-log diagnostic plot and subsequent interpretation/analysis?
If the rate history is defined in any way that is consistent with total volume injected, is that sufficient to perform some useful analysis of the data?
In this study, we design a DFIT for a well and reservoir with known properties, then we use analytical and numerical simulation to produce pressure responses for the test (during injection and fall off). In the next steps, we analyze the pressure and derivative response under the conditions below:
Right rate history and total volume injected.
Wrong rate history, but right total volume injected.
Wrong rate history and total volume injected.
The method above was repeated for a vertical well with a hydraulic fracture and a vertical well with no fracture, in order to confirm similar observation for a more standard well test when compared to a DFIT.
The results presented here imply that if the imposed rate history is incorrect with respect to rate values or times when the rate changes occur then the early time response shape is distorted in comparison to the case where the rate history is defined correctly (e.g. in a DFIT wellbore storage and linear flow before closure do not reflect the correct derivative shapes). However, if the rate history is defined in any way that reproduces the correct total volume injected, then the late time radial flow period remains unaffected and, hence, the ability to estimate reservoir properties (e.g. transmissibility and reservoir pressure) from the test is preserved. Thus, it appears that the late time part of the response is controlled by total volume injected and that total volume injected is sufficient information for estimation of reservoir properties for radial flow. This agrees with/confirms the common analysis approach used by the completions community based on the instantaneous line source solution presented by Gu, et.al. (1993)