For operators who have established a core development area, drilled their wells out and have been experimenting with bigger frac jobs, what’s left?  While longer wells drilled in the best rock with the biggest fracs may have the highest rates, our industry will keep pushing the limits of completions.

An increasingly popular technique in US shale plays that has recently been adopted by some operators in the Duvernay and Montney is “extreme limited entry” (XLE) fracturing and the more targeted “aggressive limited entry” (ALE).  In these completions, the frac job is pumped through a very small number of perforations, resulting in a high pressure drop (ΔPpf) across each perforation during the frac which helps to ensure each perforation receives high volumes of fluid.  This assists with frac initiation, reduction of near wellbore tortuosity, better chance of proppant transport, and helps to establish a more uniform fracture growth along the wellbore.  In short, XLE and ALE offer the possibility of higher cluster efficiency during multi-cluster plug-perf completions.

Utilizing GLJ’s extensive completion database, we can utilize high measurements of ΔPpf as a proxy for application of XLE technology (ΔPpf ≈ 15-20 MPa). If we consider higher rates to be indicative of better well performance, logic dictates that if higher rates are observed with higher ΔPpf, XLE —  that’s a good thing. In the plots below, we can compare the results of increased ΔPpf on initial gas rates for gas wells (left) and oil rates (right).

There’s a stronger positive response to the use of XLE with oil rates compared to gas rates.

Of course, there are hundreds of factors that will influence rates — a simple cross plot glosses over this complexity.  This is a great opportunity to use ML to tease out the effects of individual factors.  Using a simple boosted tree model, unsurprisingly, we see that factors related to overall fracture size matter greatly.  This includes the fluid pumped per metre and proppant placed per metre.  We also see from this bulk analysis that the terms most associated with XLE are relatively unimportant for production.  For example, shots per meter (spm) and perforation pressure drop (ΔPpf) don’t even show up in the summary plots.

However, XLE techniques have mostly been applied to liquid rich areas.  Thus, our dataset is unbalanced and it’s still early days in terms of trying out XLE or ALE on the gas rich areas.  It would be more prudent to limit our analysis to those areas in which XLE has been applied and to pick areas of similar fluid and reservoir properties.  We also want to remove redundant or overlapping features.  For example, proppant slurry density will be highly correlated to the combination of proppant placed per metre and fluid placed per metre.  When we implement these best practices, there’s  almost an inversion of results.  Parameters most associated with XLE become far more important, such as shots per meter (spm) and perforation pressure drop (ΔPpf).

From this ML exercise, of which we’ve only shown a few steps, XLE techniques are associated with better production rates, especially in oil rich areas. Unfortunately, ML models can struggle when datasets are limited, making prediction quality moderate at best. But as more data comes in, the ML models rapidly improve. It just requires time and patience.

If you need immediate clarity, we can go back to what we know and use the rules of physics to help guide your answers. For example, GLJ has already presented work demonstrating the importance of landing intervals on how to best characterize fracability and optimize well placement.

With regards to this study, the models presented below provide a juxtaposition of ‘traditional’ completion techniques to XLE style completions utilizing GLJ’s proprietary STimsIM software for an existing Montney wellbore near Kaybob. The first design case represents a non-XLE approach in which fracture clusters are perforated using 20 spm.  Using field pumping rates and rock mechanics, the simulation shows that only about half of the clusters are expected to have any significant fracture growth.  And the fractures created appear to grow upward with limited half-length.  Not ideal!

However, if we rerun the simulation but drop the spm from 20 to 2, we estimate the pressure drop across the perfs during frac operations increases to 20 MPa.  This helps frac initiation in clusters #2, #5 and #7.  Better still, these fractures have much more balanced growth with better downward and outward propagation.  Overall, we would expect higher cluster efficiency and more uniform frac growth within the SRV, which should provide higher well productivity.

Combining our ML and simulation work, it appears that XLE and ALE completion techniques can offer meaningful performance improvements.  While bigger is often better, it’s not always best!

Published On: July 13, 2021Categories: Gas


  • Mike Morgan

    Mike Morgan holds the roll of Vice President - Analytics, who, since 2007, has focused on tight gas evaluations and resource studies in northeast British Columbia. Projects include multi-billion dollar, multi-Tcf, resource play acquisitions for major independent producers and NOCs. He has completed a PhD thesis titled, “Modeling Multi-Fractured, Multi-Well, Tight-Gas Reservoirs”. His current research work concerns applications of stochastic differential equations and Itô calculus to analytical well production models.