## Modeling simulfracs and subsurface implications

Simulfrac’s are growing in popularity (see 2021 JPT article for when the trend was just gaining momentum). The idea is that one pumping crew can treat two wells simultaneously versus one well at a time. As such, a frac crew may zipper four wells at a time versus two (see schematic below).

At ResFrac we are seeing an increase in simulfrac interest across our consulting and license customers. Simulfrac’ing wells within the ResFrac software is simple to set up without any complicated modifications – so this makes ResFrac an ideal platform to investigate the effects of simulfracs.

In this post, I investigate a couple questions that come to mind when I think of simulfracs:

1. Due to horsepower limitations, the treating rate into each individual well during a simulfrac has to be lower than if injecting into one well at a time.
• What is the impact on productivity if no modifications are made to stage design (i.e. if the same stage length, cluster count, hole count, etc are used in the simulfrac wells as would be used in zipper-frac’ed wells)?
2. If #1 limits productivity, can we modify stage design to compensate?
• We could shorten stage length in simulfrac’ed wells to maintain a higher injection rate per cluster.
• We could modify the limited entry perforating design to maintain the same or greater degree of perforation friction.

## Base case

To investigate, I begin with two wells, 600 ft apart, in a generic shale-oil type reservoir.

The minimum principal stress and permeability of the model are shown below. There are stress barriers above and below the pay zone. Permeability is 400 nd in the pay zone, and 100 nd above and below.

I start with a base case “zipper frac” scenario of injecting at 90 bpm and a total of 19 bbl/ft of slickwater and 1600 lb/ft of 100 mesh proppant into a 300 ft stage with 10 clusters per stage.

I simulate two stages in each of the wells, and use external fractures on the toe-ward side of the model to account for the unmodeled stages. After a ~10 day shut-in, the wells are then brought online at the same time using BHP control. Running the simulation out to five years, the well pair produces 40,732 bbl from 1200 ft of lateral (two 300 ft stages in each well), for an average five-year production of 33.9 bbl/ft.

That production is coming from 7.9 million sq ft of propped fracture area (orange line in image above). The maximum productive lengths are about 720ft (see below).

## Simulfrac

Using the same model, I modify it to represent a simulfrac scenario:

• Reduce pump rate to 60 bpm (and extend the injection duration to keep total fluid and proppant volumes the same).
• Shift Well Two injection controls to occur simultaneously to Well One

The resulting injection control sequence is shown below (because Well One and Well Two controls are identical, they are indistinguishable in the image).

The shut-in and production controls are left the same. Running the simulation out to five years, the well pair produces 35,310 bbl from 1200 ft (two 300 ft stages in each of the wells), for an average five-year production of 29.4 bbl/ft, 13% less than the equivalent zipper-frac scenario.

The total propped fracture area created from the simulfrac case is 16% less at 6.7 million sq ft. You will also note the different distribution of fractures, with the simulfrac exhibiting lower cluster efficiency and slightly longer maximum drained lengths (out to about 800 ft).

Comparing the five-year pressure depletion, we observe the higher uniformity in the zipper frac as well as more of the region between the wells drained.

The depletion gap between the wells in the simulfrac scenario is a notable consequence of the induced stresses of simulfrac’ing the wells. In the two videos below, you can observe the stress shadow differences between the simul- and zipper frac cases. The simultaneously propagating fractures in the simulfrac case push the fractures away from each other. While this is also true in the zipper frac case, the fractures from the adjacent well leak off quickly and stress reduces quickly ahead of the second well being fractured.

Zipper frac:

Simulfrac:

## Modified-stage-length simulfrac

There are two primary effects impacting the productivity of the simulfrac system. First, the reduction in pump rate with the same stage design results in less perforation friction and cluster efficiency decreases as a result. Second, the stress conditions push the fractures apart, toward the outside of the well pair, more so in the simulfrac case than in the zipper frac case.

I create a third case to address the first impact. In the case below, I shorten the stage length by 20% to 240 ft. I maintained the same cluster spacing, so the number of clusters was decreased from 10 to 8. I also reduce the duration of the injection periods such that the volume of fluid and proppant on a per-lateral-foot basis remain equal (19 bbl/ft, 1600 lb/ft, respectively) while maintaining the 60 bpm maximum injection rate.

In the pressure depletion image below, we see that the reduction in stage length improves the cluster efficiency of both wells. The well pair produces 29,900 bbls of oil in five years, or 31.2 bbl/ft, from 5.9 million sq ft of propped fracture area.

Shortening the stage length while maintaining the same proppant and fluid volumes successfully counteracts some of the negative productivity impact of the simulfrac, but not 100%.

In this scenario, would the simulfrac save money? I’m not an operations expert, so I don’t know. But some simple math indicates that it is plausible. In the 240 ft simulfrac scenario, total pumping time is 116 minutes. Assuming that downtime between stages is the same as pumping (you could be simulfrac’ing wells on the other side of the pad), that would be that 232 minutes per 480 ft of lateral (two 240 ft stages), or 29 seconds per lateral foot.

In the case of the zipper frac, pumping time is 85 minutes. Again assuming that pump and downtime are equal, that would correspond to 170 minutes per 300 ft stage, or 34 seconds per lateral foot, about 15% slower.

## Limited-entry-modified simulfrac

Due to the wellbore friction and less available horsepower, the injection rate into each individual well being simulfrac’ed is lower. Below is the equation for perforation friction (amount of pressure drop as fluid exits the casing – see Murphee et al., Huckabe et al., Barhaug et al., Ribiero et al., and others for great discussion on nuances of perforation friction):

Ppf = Perforation friction (psi)
ΔPp = Pressure drop across perforations (psi)
Q = Flow rate (bbl/min)
ρ = Fluid density (lb/gal)
Np = Number of perforations
Dp = Perforation diameter (in.)
Cd = Discharge Coefficient

A higher magnitude of perforation friction (higher “limited entry”) results in a greater pressure drop for fluid exiting the casing, and all else equal, will result in greater uniformity (even distribution of fluid exiting all clusters).

In the perforation friction equation, we see that injection rate is in the numerator and is squared. When testing zipper versus simulfrac, and  keeping our stage design the same, injection rate drops from 90 bpm to 60 bpm, resulting in a 55% decrease in perforation friction (1 – 60^2/90^2). In order to maintain perforation friction we could drop clusters or number of shots per cluster (decrease Np) or decrease the diameter of those shots (decrease Dp). Note that in the case where I shortened the stage length, I implicitly decreased Np (by cutting the number of clusters from 10 to 8).

Another way in which I could maintain higher perforation friction would be to decrease the number of holes per cluster or the diameter of those holes. My original design had 10 clusters, each with three 0.36” holes (for those doing the math, that’s ~1500 psi of perforation friction). By reducing the diameter of perforations from 0.36” to 0.3”, I increase the perforation friction of the 10-cluster, 3 shot-per-cluster design to nearly the same level as the zipper (about 50 psi lower, see Table 1 in summary).

The image below shows the created fracture geometries and depletion after five years of production.

We observe that cluster efficiency has improved versus the original simulfrac base case, and similar to the shorter-stage simulfrac, the fracture area biases outward, away from the center of the well pair.

However, on a per-foot basis, the limited-entry simulfrac generates more propped area (7.04 million square feet) than the initial simulfrac case, but still performs worse than the zipper frac, producing 30.1 bbl/ft of oil after five years (versus 33.9 bbl/ft for the zipper frac, and 31.2 bbl/ft for the shorter-stage simulfrac).

The performance difference between the limited-entry simulfrac and shorter-stage simulfrac can be explained by the increased injection rate on a cluster basis. The higher injection rate per cluster in the shorter stage (7.5 bpm/cluster) maintains a higher pressure inside the fracture, resulting in a more uniform distribution of fractures in the far-field than with the LE simulfrac design (6 bpm/cluster). We can quantify this by dividing the total created propped area by the lateral length to arrive at a “propped area per lateral foot” metric:

## Bounded depletion

In the three cases above, the well pad was free to drain from an unbounded reservoir region on either side of the well pair. This situation is becoming increasingly rare. Most often, this well pair would be part of a large well pad or placed between existing wells. So how do the results change if the well pair is bounded on either side?

To quickly investigate, I imposed no-flow boundaries 300 ft on either side of the well pair after the conclusion of fracturing to approximate that the situation where this well pair were in the middle of additional wells, all spaced at 600 ft (so 300 ft would be halfway to the next hypothetical well).

In the cases where no-flow boundaries have been imposed, productivity is systematically lower than in unbounded cases (as expected).

Examining the results, we see that not only does productivity categorically decrease in the bounded case, the relative impact for the base simulfrac case is larger than the base zipper case (25% reduction versus 23%). Due to the low cluster efficiency and outward-bias of fractures, relatively more fracture area is “cut-off” with the no-flow boundaries.

The impact of bounding on the modified simulfrac cases (shorter stages and increased LE) is a mixed bag. Fractures bias outward due to the simultaneous propagation, but the higher cluster efficiency results in lower injection rates into each frac, and consequently shorter total propped lengths (resulting in less frac area susceptible to being “cut-off”).

## Discussion

A table summary of the four cases is provided below.

We see that the success of simulfrac in this setting (geology, well spacing, fluid type, etc.) will ultimately be a function of a) what is the anticipated cost savings of simulfrac operations compared to zipper frac, b) are the subject wells bounded, c) do you anticipate infill drilling adjacent to the simulfrac wells at a later date.

B and C are important considerations. Consider the case where you have depletion on one or both sides of your subject wells. There will be a tendency for fractures to grow preferentially toward the depleted wells. Simulfrac’ing would exacerbate that tendency due to the interwell stress shadowing during fracturing operations. Or alternatively, in case where you are developing these wells in virgin reservoir, simulfrac’ing is likely to extend the depletion halo around the wells, increasing interference from eventual infill wells.

Other considerations:

• In the example geologic setting, the fractures are moderately-well contained within the pay zone (limited propdation above and below) and the wells are developed in the same bench. Generally, the more height constrained the fractures are, the larger the relative stress shadow between the wells. The results would absolutely be expected to change if fractures were more/less confined.
• This was a single-bench development example that demonstrated the stress interaction between the wells. Could this stress interaction be employed in multibench developments to “engineer” the propagation of fractures (i.e. incentivize more upward or downward propagation)? It’s an interesting idea that could be investigated.
• This example used only two wells. If drilling a pad of something like 10-16 wells, many more degrees of freedom are introduced. For instance, you could simulfrac well-pairs several well-spacings apart.
• One of the primary impacts on the simulfrac results is the simultaneous creation and interaction of stress shadows from adjacent wells. The results would change if we manipulated simulfrac operations. For instance, if you had a four-well pad (two wells pointed south, two wells pointed north), you could simulfrac one north-well and one south-well together, and then second north-well and second south-well together as depicted in the image below (where numbers next to fractures indicate the fracture order). From the perspective of the north or south wells, this would be the same as zippering the well pair (but at a reduced injection rate).

For example, in the image below I compare the short-stage simulfrac to a case where I use the same 60 bpm design and zipper the wells (as if you were simulfrac’ing opposite sides of a pad simultaneously). The result is production increases from 31.2 to 32.1 bbl/ft.

• I did not adjust or sensitize on well spacing. If optimizing the entire problem, I would expect that the optimal well spacing when simulfrac’ing would be different than if zipper frac’ing, as well as stage length and perforation design being different.

My goal in this post was to give a flavor of some of the considerations when designing for simulfracs versus zipper fracs. We saw that there were two primary mechanisms differentiating simulfracs: a) lower effective injection rate per cluster (all else equal) and b) stress interaction occurring simultaneously between adjacent wells, and that each effect can be addressed with thoughtful engineering.

As always, if you are a ResFrac user and interested in these simulation files, email me and I’m happy to provide a link to download.