The DFIT Industry Study kicked off in January 2018 with six major operators and one service company participating. We have been holding meetings every three months, and the study will conclude at the end of 2018. The detailed results won’t be shared until next year. However, I can now provide a general update on the findings. I am really looking forward to sharing the full results!
The motivation of the DFIT study is that conventional DFIT interpretations are based on ‘preclosure’ and ‘postclosure’ calculations. These calculations skip over the effect the of closure itself. Prior work published in SPE Journal showed that incorporating residual fracture aperture allows a simulator to describe a full DFIT: before, during, and after closure. This work led to a deeper understanding of how closure impacts the pressure transient. The work suggested that the conventional DFIT closure pick can sometimes be significantly inaccurate.
ResFrac provides a unique opportunity to understand DFIT because it is a fully combined hydraulic fracture, wellbore, and reservoir simulator. This allows us to simulate new processes, such as how multiphase effects affect DFIT interpretation. Integrating all processes into a single calculation allows us to see how they interact. Interactions and transitions in-between simplified ‘end-member’ states turn out to be very important. In many DFITs, most of the transient is occurring as a transition between end-member states. This is why conventional work, which has tried to isolate pieces of the problem like ‘preclosure’ and ‘postclosure,’ can sometimes be problematic.
The goal of the DFIT study is to design a new procedure for interpreting DFITs. How should we calculate stress, permeability, and formation fluid pressure? What qualitative interpretations are possible? What plots should we use?
The study integrates insights from ResFrac modeling with the experience and expertise of the industry participants. Participating companies have contributed a collection of DFITs from shale basins around the US. This allows us to apply proposed DFIT interpretation procedures to a diverse set of data. This allows us to put things in context – we can see when results are ‘typical’ and what is unusual.
In Q1 of this year, I ran about 70 ResFrac simulations, covering a wide range of conditions and scenarios. I wrote a python script that automatically reads the simulation results and applies a variety of different interpretation techniques. The interpretations can be compared to the known simulation inputs in order to evaluate accuracy. There were many cases where conventional methods of estimating permeability yielded significant inaccuracy. The simulations allow us to see when methods are inaccurate and why.
On the basis of the simulation results, we put together a step-by-step interpretation procedure. Choice of plots is partly a matter of personal preference. I prefer to use plots of dP/dG versus G, a log-log plot with derivative taken with respect to shut-in time, and an estimated stiffness versus pressure plot (similar to the plot shown by Wang and Sharma, 2018). The dP/dG plot can be used to identify closure and quickly assess the impact of near-wellbore tortuosity. I prefer the derivative plot to a plot of G*dP/dG because I find it easier to see when the derivative is increasing due to closure.
Pressure drop causes deviation from Carter leakoff, which invalidates the assumptions underlying the G-function. Therefore, a stiffness versus pressure plot can be used as a backup way to pick closure, because it is designed to deconvolve the effect of pressure change in the fracture over time. This approach is related to the method of Mayerhofer et al. (1995), which also uses a convolution integral to account for changing pressure over time.
The results show that conventional methods of estimating permeability have relative pros and cons and can be frequently inaccurate. Based on the results, I derived a new method for estimating permeability, designed to avoid potential pitfalls that were identified from the modeling. ResFrac was used to help guide the development of these procedures, but it is not required to do the interpretations themselves. The interpretations are based on analytical derivations, with assumptions designed to be consistent with insights from the more general numerical calculations. They are then confirmed by application to the simulations and field data.
In Q2 of this year, the methods were applied to a diverse set of about 20 DFITs provided by operators. The interpretations showed significant inaccuracy from conventional methods in most cases. In one example, an operator had found that the service company permeability estimate was 50-100 times higher than what the operator had inferred from rate transient analysis and other diagnostics. Our results predicted this inaccuracy, based on the conditions of the test, and explained why it occurred. We applied our newly developed method and arrived at a permeability consistent with the value that had previously been inferred by the operator.
We have also discussed how to assess confidence in the interpretation. If the transient deviates from expected behavior, this lessens confidence. It is better to say “I’m not sure” than to be confidently wrong! We have also investigated how the prior history of the well before the DFIT can sometimes be important to take into account.
In the second half of the year, the study will be looking at pressure estimation and tracing the implications of DFIT interpretations for fracture and well design. The implications can have a strong impact on economic performance.