Ellen Scott and Dr Kenneth D. Mahrer, SIGMA3, discuss three, side-by-side stimulations in the Cotton Valley which show notable production differences.
The oil and gas industry is constantly asking, “Do different stimulation techniques in unconventionals affect production?” And, if so, how do stimulation costs affect total costs and the decision of which technique to employ? To approach answers, Beusa Energy compared three stimulation techniques on four neighbouring wells. The wells are parallel, horizontal laterals in the Cotton Valley (Figure 1): the 1H-1 (#1), the 1H-2 (#2), the 1H-3 (#3), and the 1H-4 (#4). As Figure 1 shows, each well’s lateral was ~4000 ft long and ~600 ft from its neighbour(s). Wells #1, #3, and #4 were landed in the same sub-strata of the Cotton Valley. Well #2 was landed 100 ft deeper.
Figure 1. Map view of the four stimulated wells and the three microseismic observation wells. On the stimulated wells, coloured bands indicate the injection points. These colours repeat after 10 stages.
Figure 2 shows the cumulative production responses of the four wells. Note that after three months, two wells, the #1 and the #3, yielded ~20% more gas than the #2 and the #4. The #1 was stimulated using coiled tubing-actuated sleeves; #2 and #3 were stimulated using plug-and-perforate in ‘zipper fracked’ strategy; and, #4 was stimulated using conventional ball-actuated sliding sleeves. Of these three techniques, conventional ball-actuated sliding sleeve stimulations are typically the least expensive. But, of the two best producers, the coiled tubing-actuated sleeve stimulation total cost was the least. This stimulation used half of the total proppant of the similarly producing plug-and-perf stimulation.
Figure 2. Cumulative gas production showing only commercially produced gas. Graph does not include times of flowback or shut-in.
Well #1, the western-most well, was stimulated first (8 - 15 September 2014, including two days of downtime). This stimulation was pumped down the annulus between casing and coiled tubing. Pumping down the annulus increases the friction pressure of the flow, limiting the flow rate, compared to the other stimulation techniques. After all the stages were complete, post-stimulation well preparation activity was minimal (i.e. limited cost). Well #1’s stimulation included 43 stages (i.e. single entry points), spaced ~100 ft apart. Each stage used ~100 000 lbm of proppant pumped at ~24 bpm. After the stimulation, the well was immediately flowed back and began commercial production three days later.
Well #4, the eastern-most well, was stimulated next (16 - 20 September, including 12 hours of downtime) and used ball-actuated sliding sleeves (i.e. single entry port between swell packers). Well #4’s stimulation included 14, ~300 ft long stages. The first eight stages used ~400 000 lbm of proppant and the last six stages used ~600 000 lbm. The stages were pumped at ~65 bpm. Similar to Well #1, Well #4 was flowed back immediately after the stimulation and began commercial production five days later (i.e. minimal post stimulation preparation).
Wells #2 and #3, the two middle wells, were stimulated last (22 - 30 September, including two days of downtime) and used a plug-and-perf, ‘zipper frac’ stimulation. After the stimulation, bridge plugs separating stages were drilled out prior to production (i.e. additional post-stimulation well preparation). Both Well #2 and #3 had 22 stimulation stages, spaced ~200 ft apart. Each stage had between three to five perforation clusters. Stage sizes varied between 400 000 to 600 000 lbm of proppant and were pumped at ~65 bpm. Well #3 was flowed back on 4 October, four days after the stimulation, and began commercial production seven days later. Well #2 was flowed back on 13 October, 13 days after the stimulation, and began commercial production three days later. It needs to be noted that this stimulation occurred while Wells #1 and #4 were producing. On 25 September, Well #1 loaded up with sand; as will be shown in the microseismic section below, the sand intrusion was most likely caused by the #2-#3 stimulation.
Comparison of the stimulations
All stimulations used the same fluid, proppant type, and proppant size. The major variation between the three stimulations is the combination of the number of entry points per stage, the stimulation rates per stage, the total volumes per stage, and the total volume per stimulation.
The major difference between Well #1 and the other wells’ stimulations is the number of injection points. The coiled-tubing actuated sleeve technique uses a multitude of single-point injections. As a result, the number of stimulation stages is greater and the volumes of the stimulation stages are smaller than the other wells’. By stage, #1 used ~25% of the proppant compared to #4, #2 and #3. In total, #1 used ~50% of the proppant as Wells #2 and #3. Well #1 stage spacings were half those of Wells #2 and #3 and one third of those for Well #4. The treating rate was 25 bpm, ~38% of the other stimulations.
Another point of comparison is the instantaneous shut-in pressure (ISIP). ISIP is the pressure when pumping ceases. To compare stimulations at different depths, this pressure is given as a gradient (i.e. pressure divided by depth). During Well #1’s stimulation, each stages’ frac gradient was ~10% lower than those of Wells #2’s and #3’s. Assuming similar rock properties, the lower frac gradients may be affected by one of two causes or a combination of both. The first is the smaller fluid and proppant volumes per stage. The second is a less complex fracture geometry.
Similar to Well #1, Well #4’s stimulation used one sleeve per stage. However, conventional sliding sleeve stages are not single injection points. Conventional sliding sleeves are essentially open-hole completions between packers. For Well #4, the packers were spaced at 300 ft. Therefore, Well #4’s stimulation, with potentially multiple entry points into the formation, is more like the Well #2 and #3 stimulation than the Well #1 stimulation. Techniques of multiple entry points require higher treating rates and larger stage volumes. Higher treating rates and larger volumes give a greater potential to divert fluid and proppant across the treated interval.
Well #4’s stimulation had a similar amount of proppant per stage as Wells #2’s and #3’s stimulations, which was approximately four times that of Well #1. With fewer stages, Well #4 used ~70% of the proppant of Wells #2 and #3.
It was not possible to determine fracture gradients for Well #4 stimulation. Sliding sleeve operations are continuous and do not shut-in between each stage.
Wells #2 and #3
Similar to the Well #4 stimulation, Well #2’s and Well #3’s plug-and-perf stimulations are multi-entry point techniques. As previously mentioned, Well #2 and Well #3 had three to five perforation clusters per 200 ft stage. In an effort to divert fluid and proppant equally across the multiple perforation clusters, Wells #2 and #3 used the same treating rate as Well #4, approximately three times that of Well #1. Wells #2 and #3 also used similar proppant volume per stage as Well #4. This amount is approximately four times that of Well #1. In total, each zippered well used twice the proppant of Well #1.
As a stimulation response diagnostic tool, SIGMA3 recorded and processed the microseismicity induced by the three stimulations. Prior to the stimulations, all participants understood that the microseismicity induced by hydraulic fracturing is not at this time a proxy for the production.
As shown in Figure 1, the acquisition set up included one deep vertical observation well and two shallow observation wells. The shallow wells were approximately 1000 ft deep. In each observation well, SIGMA3 deployed a 3-component geophone array. The deep well had a 40-level array. In this array, the deepest geophone was ~1300 ft above the treatment interval. The shallow wells had 20-level arrays. The geophone spacing for all three arrays was ~50 ft. After deployment, it was found that the two shallow wells were noisy, precluding microseismic event detection for these data sets. However, as shown in Figure 3(A), using the deep array, the company recorded and located events near Well #1’s toe, ~5000 ft from the middle of the deep array.
As noted above, conventional thinking indicates a stimulation with a single entry point and a low treating rate is expected to create a smaller and less complex stimulated volume than a stimulation with multiple entry points, using a greater rate. As noted below, the microseismic responses from the three stimulations support this model.
Figure 3 shows the individual microseismic responses of the three different stimulations in order of stimulation from left to right. Figure 3(A) shows the microseismic response of Well #1; Figure 3(B) shows the microseismic response of Well #4; and Figure 3(C) shows the microseismic response of Wells #2 and #3.
Figure 3. Map views of each stimulation response: (A) shows the microseismic response of Well #1; (B) shows the microseismic response of Well #4; (C) shows the microseismic responses of Wells #2 and #3. The microseismic event symbols are coloured by stimulation stage. Event symbols are the same colour as their corresponding injection points. The colours repeat after 10 stages. The event symbols are scaled by the relative sizes of the microseismic events. The white length scale at the top of (A) is the same in each graph. The circle wellhead at the bottom right of each panel indicates the well with the deep array.
Well #1’s stimulation induced 45 microseismic events. Of its 43 stimulation stages, 26 induced events; the remainder induced no events. The lack of events for these stages is not a function of the microseismic recording geometry; rather it is most likely a function of the small stimulation sizes and low treating rate. In general, the microseismicity was consistently sparse along the entire lateral. Events occurred both on the east side and the west side of the lateral. The event distribution located laterally and vertically within 500 ft from the well.
Well #4’s stimulation induced 272 microseismic events. All 14 stages induced events. Toe-half stages produced a different microseismic response than the heel-half stages. The microseismic event distribution from the toe-half stages overlapped the other three wells. Some of these events occurred up to 1800 ft west of Well #4. Only a few events occurred east of the lateral. The event distribution from the heel-half stages was more symmetric and more constrained than the toe-half distribution.
The Wells #2 and #3 stimulation induced 336 microseismic events. The deep geophone array required maintenance at two points during the stimulation, hence 7 of the 44 stages were not microseismically monitored. Of the 34 monitored stages, 26 stages induced microseismicity. As with Well #4’s microseismic response, the #2-#3 stimulation showed a difference between toe and heel stages. Toe-half stages induced many events, some of which were large compared to the general population. Within the general event population, events located across Well #1 to the west and Well #4 to the east. Heel-half stages induced fewer events. These event distributions had less extent and less overlap of the neighbouring wells.
As mentioned above, Wells #1 and #4 were producing while Wells #2 and #3 were stimulated. Also, Well #1 loaded up with sand during Stage 9 of Well #3. The events induced by this stimulation are cyan in Figure 3(C). The figure shows several events around Well #1, most likely indicating a flow path or flow network between the stimulation and Well #1.
Well #1’s microseismic response is distinctly different from the responses of the other two stimulations. SIGMA3’s working hypothesis is that this difference is consistent with the differences in stimulation characteristics. Well #1’s stimulation, a single-point injection method using less rate and less volume, most likely induced less fracture complexity. This resulted in fewer microseismic events and a narrower event distribution. In contrast, the multiple-entry point stimulations of Wells #4, #2, and #3 had more rate and volume and, most likely, induced greater fracture complexity, based on the microseismic responses. These stimulations resulted in more microseismic events with event distributions, for some stages, extending across the neighbouring wellbores.
Figure 4 shows the cumulative microseismic response for the three stimulations. Overall, the three stimulations induced two distinct distributions of events. In the region between the two distributions, there was an azimuthal trend of very few events. This may indicate a geologic feature running northeast to southwest across the area.
Figure 4. Cumulative microseismic response from all four wells. Microseismic event symbols are coloured by stimulated well and are scaled by the relative event size.
This study is not definitive. Additional study, similar to this, is needed for definitive conclusions. It posed the question of best completion technique but its answer was not uniquely determined. What was determined is a partial answer. Based on Figure 2, the cumulative production graph, the coiled-tubing actuated completion appears to be the most effective completion. However, the production of the other wells could have been adversely affected by other factors besides their completion techniques. These other factors include, but are not limited to stimulation order; pressure depletion while Well #1 was producing; the possible geologic feature suggested by the microseismic response; or undetected geologic variability or heterogeneities between the wells.
Although the microseismic response from a hydraulic fracture stimulation is not a proxy for production, the microseismic responses from these wells showed the effects of the different stimulation techniques. However, the microseismic responses may have also been affected by the previously mentioned factors: stimulation order; pressure depletion; and geologic heterogeneity. More than three stimulations are required to determine the best completion technique.
Adapted by David Bizley
Read the article online at: https://www.energyglobal.com/upstream/special-reports/17072015/testing-in-the-fields/