Can “too close” be ill-advised in fracture spacing?

Emblematic of the high-voltage development of unconventional resource plays is the growing attraction of beefed-up completions with high-proppant loadings and shorter distances between induced fractures. Exactly what constitutes optimum fracture spacing, however, depends heavily on economic considerations and developing a thorough understanding of the reservoir characteristics—insight that all-too-often falls to the wayside.

Fig. 1. A generalized plot of oil production from two fracture spacings, using similar/identical frac treatments.

Making an assiduous effort to get a handle on the permeability distribution of a targeted pay zone pays dividends with appreciable corresponding increases in reservoir drainage, Fig. 1. Frac Diagnostics' experience in modeling Delaware basin shale wells in the Permian, built around net pressure matching (NPM) of frac data, has resulted in fracture spacing progressively shrinking from 120 ft to 40 ft, with the prospect of 25-ft spacing in the Wolfcamp shales entirely plausible.

Building and refining a play-centric database and incorporating the frac data in an adaptive model, to analyze specific reservoir characteristics, avoids settling for the “whatever-comes-your-way” rationale that has typified the unconventional community's approach to completions.

THE SPACING CONUNDRUM

Generally accepted reasoning holds that narrowing the spacing between initiated fractures helps compensate for lower-than-expected source rock permeability, where the decreased flow of reservoir fluids through the tighter rock fabric generates rapid declines in production and profitability. Accordingly, contemporary unconventional well completions are typified with progressively shorter fracture spacing, in tandem with higher proppant loadings per lateral foot, to increase the contact frequency of the source rock.

Definitive answers to what constitutes optimum spacing, and if it can be “too close,” depends on myriad economic and reservoir considerations. As Morrill, et al, and others ^1,2,3 have pointed out, fracture spacing cannot be determined by looking merely at one or two properties. Rather, it requires an examination of all the important variables and identifying those that are most variable for the targeted reservoir. Yet, carefully analyzing the specific characteristics that optimize spacing in a given area, and designing the completion accordingly, invariably falls victim to scattershot planning, influenced largely by wide commodity price swings and the mantra to put wells on line quickly and at the lowest possible cost.

In this age of full-blown exploitation of unconventional reservoirs, the fundamental reality is that the pay zones the industry is completing today lack the permeability of reservoir rocks of one or two decades ago. Ultra-low reservoir permeability acknowledges that the subsequent migration of hydrocarbons from deeper in the reservoir now takes much longer to flow toward the hydraulic fracture, thus driving the strategy to place fractures closer together, to decrease the time required to produce hydrocarbons into the fracture and wellbore.

Moreover, a scan of investor material from many operators in the Permian basin and the Eagle Ford shale leaves the impression that the predominant frac metric touted is the pound of proppant used per lateral foot. Little information is revealed, however, on how exactly, and how often, fractures are initiated. While literature is replete with data on well spacing, the actual methodologies employed for placing 2,500 lb to 3,000 lb of proppant/lateral ft is rarely discussed.

Reservoir engineering principles dictate that the even distribution of propped fractures in a formation is the optimum method. This assumes, however, that the formation has a fairly narrow range of permeability, which is not necessarily the case. This uncertainty accentuates the need to develop a deeper understanding of shale reservoir permeability distribution, as well as its vital role in optimizing fracture spacing. It also is vital to planning the completion strategy to access the best pay with conductivity and perhaps rely on contact to drain the second-tier pay. While relatively straightforward on the face of it, collecting the necessary data poses a few challenges.

ASSESSING PERMIABILITY DISTRIBUTION

Permeability values are defined by Darcy's law for flow in porous media. Over the years, a number of associated methodologies and technologies have been employed to assess the capacity of fluids to flow through porous reservoir media.

Fig. 2. This view of passive seismic mapping illustrates that the fractures activated during pumping follow the natural fractures mapped pre-treatment. Thus, this process is a valuable tool in the development of an unconventional field.

Recent techniques developed by Frac Diagnostics incorporate gas fractions from drilling data with gas chromatograph or mass spectrometry, following the principals of the Pixler method, which describes the effect of gas ratios in determining permeability.⁴ By using a gas chromatograph and plotting the C4 and C5 values (heavy gases), as well as their ratios to all gas detected by drilling through reservoir rock, Pixler’s method allows for the determination of a qualitative basis of permeability detection. From a molecular aspect, this theory is supported by the fact that the molecular size of butane (C4H10) is 28 times larger than methane (CH4). Thus, butane requires a pore throat 28 times larger than methane to travel through the rock. Of course, using this method is limited, if the reservoir has been exposed to heat sufficient to eliminate any “heavies” in the formation, thereby making total gas peaks a preferred metric.

The detection of natural low-frequency earth movements, through the use of passive seismic, advanced steadily between 2008 and 2015, with intriguing implications for the unconventional completion sector. Passive seismic detects the natural “creaks and other noises” associated with formation movement during sun and moon tides. Theoretically, the more intense these noises, the higher the probability of natural fractures and other reservoir components that represent permeability. Consequently, with this technology, a reservoir or completions engineer can easily determine the frequency and best placement of fracture clusters and wellbores, Fig. 2.

On the downside, the use of passive seismic comes at a hefty cost, although the subsequent completion of multiple successful wells could outweigh the up-front economic restrictions easily. Many operators today employ the so-called “geometric” completion design, which involves setting a fracture initiation point on a set interval, regardless of what the drilling data indicate. Geometric design advocates argue that the drilling data used to determine wellbore reservoir quality are incapable of determining reservoir quality farther afield, and who is to say a productive reservoir lens is lying-in-wait 5 ft away from the wellbore?

Fig. 3. Much like snowflakes, no two wells have identical rock qualities. A straightforward geometric completion design does not adjust for changes in reservoir quality, while an engineered completion strategy applies appropriate conductivity to match the specific rock qualities.

The geometric completion protocol is biased toward the concept that the permeability distribution in a reservoir can be so vast that selecting the sweetest spots can actually eliminate the potential of stimulating second-tier, but quality, reservoir rock. In this design, the operator simply pre-sets a stage length and sticks with it, only to eventually shrink the spacing. While a safe method, the biggest attraction of this procedure is that it requires no science or analysis to use.

However, very few take the time to incorporate post-frac data to determine if a fracture did, in fact, enter a higher permeability zone. If the mostly overlooked art of net pressure-matching the frac data was employed, the geometric design practitioner could either justify or change the completion motif. Specifically, analysis could show that either reservoir permeability, completion contact percentage, frac fluid volume or proppant conductivity dictate project economics and warrant consideration in the completion design, Fig. 3.

NPM-INSPIRED MODELING

Extensive experience in net pressure matching of fracture stimulations, and the requisite building of a mechanical earth model that represents the reservoir, has allowed Frac Diagnostics to generate rather finely detailed reservoir characterizations. Elements of the completed model include stress, permeability, reservoir pressure, reservoir fluid compressibility of the reservoir fluid, water saturation, porosity and a few lesser relevant elements.

The next step in the NPM process is taking the stimulation data (minimum: rate, pressure and proppant concentration) on a 1-sec basis and running it through the mechanical earth model, to determine the effect of the frac fluid as it propagates a fracture. The predictive model is then calibrated to reflect the actual results, thus yielding the finished fracture.

During the process, and with some experience and understanding of the physics involved, one can determine the reservoir permeability.⁵ This is extremely important for a number of reasons, not the least of which is to understand if there is an actual change (usually an increase) in far-field permeability and the nature or source of said change.

Field data show that near-wellbore data more accurately reflect the stage-centric data than merely assuming the existence of a far-field productive zone. For instance, it has been observed that less than 10% of the NPMs of the Wolfcamp, Third and Second Bone Springs, as well as the shallower Permian-aged reservoirs in the Delaware basin, reflect any far-field permeability improvement. The drilling data correlation process, however, requires some non-linear analysis using broader databases to approach the elusive 99% accuracy rate in predicting reservoir permeability, to enable right-sizing frac treatments on a stage-by-stage basis. This approach requires several wells in the same reservoir, with extensive mapping, to determine if the wellbore wandered above or below the target, thus framing the comprehensive database to yield the best data.

CONSTRUCTING THE DATABASE

Building a relationship (correlation) database of near-wellbore and stage-centric data is an extremely useful tool. As the mining opportunities are driven to some extent by the amount of data put into a database, one should not be shy about incorporating even ostensibly less relevant data. At the onset, Totco MD and/or Pason ERD drilling data can generate a mechanical specific energy (MSE) value, gas chromatograph/mass spec data for gas portions, permeability and net pressure change from the NPMs and other perforation efficiency data. Consequently, the permeability readings can be matched easily with the drilling data to determine the variation in measurements and errors bars, after which the non-linear analysis can be used to understand potential production regions within the wellbore, thus denoting sweet spot indicators.

For the multi-well, non-linear analysis, wellbore mapping is the next level of maturity for this type of database, with broader depth of the data ultimately providing useful insights.

Finally, the analyst can build and shape the predictive model to meet the specific needs and objectives of the operator. The importance of this process notwithstanding, the ultimate goal is to understand well placement and to solve the question, “how many wells in this asset will it take to effectively drain the most reserves in a cost-effective manner?” The answer to that question requires the development of an adaptive reservoir model.

BUILDING, CALIBRATING THE RESERVOIR MODEL

Once the database has been assembled and the predictive model refined, the next step is transferring the data into a reservoir model. This model must be capable of degrading proppant permeability with time and possessing dual-porosity and dual-permeability capacity that can assist in understanding proper fracture spacing.

One appealing attribute of unconventional reservoirs, particularly the most prolific ones, is the presence of natural fractures. Obviously, flow through a crack is much more efficient than flow through porous media. Therefore, the ability to analyze and afterwards model the effect of natural fractures to well performance must be available. Another attribute of these pay zones is the ability to degrade conductivity in the hydraulic fracture with time. This property eliminates several very popular models, the reason being that the most popular models are built for conventional reservoirs, which typically do not require hydraulic fracing.

The main issue here is that the true fracture spacing will depend on several factors, some of which, unfortunately, may not be known. Some of these uncharted factors may actually involve economic elements, such as when to utilize a premium, high-conductivity proppant, as opposed to a lower-quality proppant that scours and potentially bridges fractures.

Here, three or four key factors could be found to drive project economics. For example, consider that reservoir permeability, completion contact percentage, frac fluid volume and proppant conductivity are found to directly influence reservoir performance. If analysis indicates that reservoir permeability on a new well is actually 50% of the original value, the model would subsequently indicate that contact must be increased 70%, with a corresponding reduction in proppant conductivity, or similar scenario.

The benefits in understanding proper well spacing are linked to fracture spacing in terms of effectively draining the reservoir around the wellbore and fracture. While the aforementioned unidentified economic factors—such as whether contact or conductivity most benefit production—need to be established, a viable rule of thumb is that the lower the permeability, the greater the need for contact via a hydraulic fracture.

POST-FRAC OBSERVATIONS

The analytical approach has helped remove many of the ambiguities prevalent in the typical completion designs of unconventional resource plays. Combining the engineered completion methodology with modeling of net pressure matching of frac data has enabled Frac Diagnostics to progressively shorten fracture spacing for its shale clients, with marked improvement in both initial production (IP) rates and estimated ultimate recoveries (EUR). To point, on average, the approach has delivered around a 24% increase in IP rates, with an estimated 8% improvement in EUR, as well as 90-, 120- and 180-day net gains in barrels of oil equivalent (boe) recovered.

The transition from beginning to now has been striking, considering the original spacing of 160 linear ft between clusters, with three clusters providing a single stage. That design has sequentially shrunk to range between 35 to 40 linear ft between clusters, with four clusters comprising a stage. It is evident that the old spacing between clusters is now the total stage length of the new design. The justification behind this transition to shorter spacing is the data-driven realization that the permeability of the formations now being completed is extremely lower than originally anticipated.