Input Sensitivity on DFIT Analysis
By Chad Cluver |
Wed, 10 Mar 2010
The results of any computer model are only as good as the inputs. This is true in any professional field, and true for any model. At Data Retrieval Corp. we have been using our proprietary model for WHP to BHP conversion to conduct traditional PTA work for the past 25 years, with great success. However, as new resources are being developed with much lower permeability, traditional PTA tests are becoming less desirable due to the extremely long time required before any useful information is obtained. This has led to an increased number of diagnostic fracture injection or DFIT tests (Minifrac, MFO, Datafrac, mini pump-in etc.) being performed, and here at DRC we have seen a marked increase over the past year in the number of these tests where our SPIDR gauge is being used. Now that we are offering analysis of this data to our customers, for use in comparison to their own or the pumping company’s analysis, we felt it important to illustrate that the results of these tests are sensitive to the inputs in much the same way as we see in traditional PTA testing. There are many inputs that go into a DFIT analysis, and we will focus on the ones utilized in Meyer & Associates MinFrac software. It is assumed the reader possesses basic knowledge on performing the analysis, and using the MinFrac or similar software.
The basic analysis of the data, using various plots, is only sensitive to a couple inputs, and most of these should be well known. To start, using the Horner plot, we determine P* and the pore pressure gradient. Obviously these are only dependent upon the depth (TVD) and the specific gravity of the fluid used. Clearly, only specific gravity would likely have much room for error, and it is important to know exactly what fluid was pumped in so that the proper specific gravity is used. It is also important to use common sense here, and refer to drilling reports and other sources of information before just accepting what the computer calculates as reservoir pressure. If the reservoir pressure calculated is higher than what would have been balanced by the mud weight used to drill the well, and no connection gas or kicks were observed, then the P* calculation should be further scrutinized.
Next, using regression analysis, we determine when closure happened, what ISIP at the surface is, what the surface closure pressure is, what bottom-hole ISIP and closure pressure is, the net pressure, stress gradient, and efficiency. Once again, really the only relevant inputs are the TVD depth and specific gravity of the fluid. The most important information obtained from this analysis is determining when closure happened, what ISIP is, if the well reached radial flow after closure, and what the most appropriate regression analysis is to use for the after closure analysis.
After closure analysis is where reservoir pressure can also be calculated, but more importantly it is where formation permeability can be determined, and it is determined from the pressure response of the well during the infinite-acting time period. The formation permeability is very dependent on the total leak-off height (h or net pay, the total height penetrated by the leak-off) and the reservoir fluid viscosity. Most often the total leak-off height will be some value smaller than the total hydrocarbon pay thickness used to model production, or it may be equal to that thickness. The injection rates and other information can be used to more accurately model this input, but it is important to understand there is an inverse relationship between permeability and the total leak-off height.
The software also provides information about the fracture using various computer models (simulations) such as the PKN, GDK, and Ellipsoidal models. Here, many more inputs (such as Young’s modulus, Poisson’s ratio, and fluid properties such as n’ and K’) are taken into consideration to determine things like fracture half-length, fracture width, fracture net pressure, efficiency, and formation permeability. Table 1 below shows how a 5% change in any of the input values affects the values calculated by the computer models. In the table, the notation “I” stands for an inverse relationship, and “D” stands for a direct relationship. The TVD depth and specific gravity were not considered, as those values should be well known.
Table 1) Sensitivity Analysis
It can be seen in the table that correctly determining the leak-off height, fluid properties such as n’, and the reservoir fluid pressure have the largest affect on the permeability of the formation as calculated by the computer models. The results of these DFIT tests (closure pressure, near wellbore effects, fracture dimensions, fluid efficiencies, and permeability) are used to determine if a larger fracture stimulation job should be done, and if so how to properly design it for maximum effectiveness. It is always important to be as thorough as possible when determining values to use as inputs for computer models in order to get the most reliable information out of the testing being performed. Data Retrieval Corp. has been in the well testing business for over 25 years, and we are happy to work with you on all your well testing needs, including DFIT testing. Our consultation in these matters is always free of charge.
