DEVONIAN GAS TECHNOLOGY-CONCLUSION STUDY DETERMINES BETTER WELL COMPLETION PRACTICES

Jim J. Venditto, D.E. McMechan Halliburton Services Duncan, Okla.

G. SimpsonHalliburton Logging Services Houston

Paul V. Hyde, L.L. Friend, R.E. Schindler Columbia Natural Resources Inc. Charleston, W.Va.

The 9-month Halliburton Services and Columbia Natural Resources (CNR) study showed a number of ways for improving gas production from Devonian shale and other "tight gas" formations.

This final part of a series of five articles (started in OGJ, Oct. 5, 1992, pp. 70-76) discusses these conclusions and the supporting evidence (see box for highlights).

PERFORATION ORIENTATION

In the West Virginia study well, the effects of perforation direction on fracture initiation were illustrated by breakdown pressure differences and fluid distribution. This was determined with a prototype directional gamma ray logging (DGRL) tool that measures azimuthal distribution of the tracer.

The perforation (4,857-4,860 ft) aligned with the least principal stress broke down at 3,238 psi, while the perforations (4,850-4,853 ft) aligned perpendicular to the least principal stress broke down at 2,731 psi.

Fig. 1 shows the bottom hole pressures of the two breakdowns superimposed. The curves indicate that 500 psi additional treating pressure was required to break down the misoriented perforation.

With the perforation aligned with the fracture direction of the formation, breakdown pressure was 2,731 psi compared to 3,238 psi with perforations out of phase.

In this example, minimum stress was constant across both sets of perforations.

The major difference in required breakdown pressure between perforations having the same stress indicated that greater differences in breakdown pressures may occur when fracturing long intervals with few perforations.

Because the stress varies by several hundred psi over the typical fracture treatment interval, perforations that are pointing in wrong directions in a higher stress location may never break down during a fracture treatment.

This problem cannot be predicted without having both stress and perforation direction information, and performing a bailout at the beginning of the pad or during the breakdown is the only way to ensure that all perforations are treated.

Tracer logs run after the fracture treatment indicated that the fractures probably did not join into a single fracture over the entire treated interval. Based on a series of after fracture tracer logs, some perforations did not have a fracture.

CASED HOLE LOGS

The casing was perforated with tubing-conveyed guns at 2 shots/ft, phased 180. The bottom perforations (4,857-4,860 ft) were perpendicular to the anticipated direction of fracturing of N67E, while the upper perforations (4,850-4,853 ft) were oriented parallel to the direction of fracturing.

The direction was determined by:

• Studying oriented cores
• Running a circumferential acoustical scanning (CAS) tool log
• Placing an experimental extensometer in the open hole during a microfrac treatment. The extensometer is a joint development of Halliburton and Total Compagnie Franais des Ptroles.

The purpose of this perforating configuration was to determine effects of perforation orientation on hydraulic fracturing operations.

Bottom hole pressure measurements made during the breakdown treatment indicated that the upper set of perforations broke down at approximately 500 psi less than the lower perforation interval (Fig. 1). Measurements made by a prototype DGRL tool also indicated that the upper perforations (in phase with the frac direction) were more efficient.

In addition to the oriented perforations, more perforations were added with a casing gun from 4,853 to 4,857 ft before the minifrac test.

The spectral gamma ray (SGR) logging tool measured the extent of the radioactive tracers from the tool. The SGR log also indicated the vertical extension of the fracture.

Tests in the interval between 4,850 and 4,860 ft showed that the fracture was contained.

During the breakdown treatment, a radioactive isotope tagged the acid used in the stimulation. The SGR-computed log indicated that the breakdown test obtained a fracture from 4,846 to 4,877 ft.

During the minifrac, 17,000 gal of 90% nitrogen-foamed fluid were tagged with 0.27 millicurie/1,000 gal radioactive tracer. The SGR-computed log indicated the fracture height to be from 4,832 to 4,894 ft. The upper extension was slightly higher than expected based on borehole stress information computed from the full wave-form sonic (FWS) log.

The DGRL tool is a prototype designed to measure the downhole azimuthal distribution of radioactive traces. The tool was attached to the bottom of the SGR logging tool.

Conclusions from the DGRL data indicated that the fracture at the lower perforations had reoriented around the outside surface of the casing, then propagated in the anticipated fracture direction. This reorientation accounted for the higher gamma counts recorded during the dynamic passes with the tool because more area was filled with radioactive tracer.

In the upper set of perforations, the fracture was found to have propagated in the direction of the anticipated fracture direction, which was N67E. The lower gamma ray activity recorded during dynamic passes was explained by the tracer material located in the narrow fracture opened during the breakdown test.

The DGRL verified the basic ENE direction of the fracture and that orientated perforating improved the fracture treatment.

Note, a nondirectional gamma ray logging tool would show a higher signal in the misoriented perforation, thus wrongly indicating a better fracture in improperly oriented perforations.

MINIFRAC

The minifrac test consisted of a single pump-in, shut-in test of 20,000 gal of 90 quality foam pumped at 30 bbl/min. The pressure decline was monitored for 2 hr.

At the end of 2 hr, the fracture had not fully closed, but the temperature survey was run to avoid losing fracture height information.

Although the exact fluid loss coefficient could not be determined, a conservative value was estimated by assuming that the fracture closed in 2 hr. Sufficient information was obtained for the fracture design simulators.

Fluid loss was so low that pad size design was dictated by the need to create sufficient fracture width for proppant placement rather than emphasizing fluid loss. Thus, temperature surveys were more valuable than a more refined fluid-loss value.

On a large scale, the stiffness of the formation perpendicular to the anisotropy should be lower than the stiffness of the matrix. This is a result of both the matrix being compressed and closure of the natural fractures.

With higher matrix Young's modulus, wider fractures may occur than predicted. The minifrac was simulated using a 3-D fracture simulator. Modifications were made to achieve a reasonable match in height growth between the model and the observed height from the radioactive tracer.

The Young's modulus was decreased from 5 x 106 to 5 x 105. This appeared to be a reasonable procedure based on measurements from the fracture extensometer and the existence of anisotropy in the formation (Fig. 2F).

In Fig. 2, the 3-D computer simulation showed that predicted geometry of the minifrac was in reasonable agreement with post-minifrac, radioactive-tracer log measured fracture boundaries at 4,832-4,894 ft.

The extensometer measured more well bore deformation perpendicular to the fracture direction than in the fracture direction before breakdown occurred. This indicated less stiffness in the formation perpendicular to the fracture.

The fracture would be more easily contained by stress contrasts and would be shorter than anticipated from designs based on a higher Young's modulus.

Determination of Young's modulus in an anisotropic formation may be made by:

• Running a directional, full-wave sonic log
• Directional plugging of an oriented core
• Running a downhole extensometer.

Directionally drilled horizontal plugs taken from oriented cores showed a variation in Young's modulus that corresponded with the direction of maximum horizontal stress. The highest Young's modulus was parallel to maximum stress and the lowest was perpendicular to maximum stress.

Young's modulus measurements made by laboratory examination on nonoriented cores, or nondirectional full-wave sonic logs will most probably be too high. In this case, computer design simulation will predict a greater fracture length than will be achieved.

Fluid loss is not generally a problem in the Devonian shale, and the use of minifracs is not necessary as a routine part of completion procedures. Based on the response of the first short perforated interval, where treating pressure increased with each injection period, it appeared a new fracture may have been initiated from the same interval on each cycle. Also, multiple fractures were observed in the microfrac tests.

Procedures that require repeated pump-in/flowback cycles should be conducted with caution in the Devonian shale.

In the study well, a pressure build-up test was conducted after the minifrac treatment to evaluate the fracture length from the minifrac and to aid in designing the fracture length and conductivity.

After the minifrac, the well was cleaned up for about 7 hr and shut in for 1 week. After the shut-in, the well flowed 1 week at a constant rate (23 Mcfd) before being shut in for the buildup test.

Build-up data were acquired first with a downhole electronic memory (DEM) gauge for 135 hr, then with a bourdon tube gauge for 280 hr.

Preliminary analysis of the pseudopressure data was performed by matching along the infinite-conductivity, vertical-fracture type curve. The end of the pseudopressure data was matched with the end of the half-slope period on the type curve. This resulted in a maximum value of permeability, a minimum effective formation permeability of 0.00048 md, and a minimum fracture half-length of 61 ft.

INTERFERENCE TEST

Following the build-up test, a fracturing treatment was performed with DEM gauges placed in two offset wells. The interference test was designed to determine whether the wells were in hydraulic communication with each other and to what degree.

Fracture orientation could be estimated if interference was observed in one well and not the other. No pressure response due to interference was observed in either observation well.

FRACTURING

A three-stage fracturing treatment, on Apr. 19, 1991, was conducted down the casing by perforating the lower interval, breaking down the lower interval with acid, and fracturing with nitrogen foam of 90 to 75 quality at sand concentrations of 0.5-4 ppg. Foam quality decreased with increasing sand concentration.

About 50,000 lb of 20/40 sand were placed in each of the lower two intervals and 40,000 lb in the upper interval.

In the standard treatment following the treatment of the lower interval, a ball was dropped to seal a frac baffle between zones. The next interval was perforated, broken down, and fractured. This procedure was again repeated for the upper interval.

The well was then flowed back. To control fluid placement in each interval, the perforations were limited to about 20 at an injection rate of 50 bbl/min.

On this well, the lower interval frac job was shut in at 11:22 a.m., the middle interval at 2:25 p.m., and the upper interval at 5:18 p.m. Flowback started at about 6:30 p.m.

In the lower zone, two fractures were indicated by the postfrac tracer log. Six unconnected fractures were indicated from the tracer log with little extension up or down from the perforations for five of the six indicated fractures in the middle zone.

Stress log data were not available for the top interval. The tracer log, however, showed no fluid entry for the lower two perforations, 3,460 and 3,491 ft.

The major advantage of the procedure used in April is the cost savings from fracturing all the intervals in 1 day; however, the primary disadvantage is the long shut-in time for all but the last interval.

Fracture closure time may be longer than the static stability of the foam. This may cause proppant settling to the bottom of the fracture with little or no proppant at some or all of the perforations.

Because of the foam flush, expansion of foam in the well bore as the pressure declines after shutdown could overdisplace the sand slurry.

Fig. 3 illustrates a four-step sequence of fluid flow during overdisplacement caused by expanding gases in foamed fluid.

Because of the balls and baffles, the volume of the well bore changed during the postfrac pressure decline. There was no simple way to calculate the correct flush volume.

This potential problem has not generally been addressed in foam fracturing because flowback before closure was assumed part of foam fracs. The problem was assumed to be severe only at low pressures with a large well bore volume.

Based on the information obtained from the study well, a modified three-stage fracturing procedure with flowback and cleanup performed after each stage was subsequently conducted on an offset well.

In Stage 1, the postfrac SGR log indicated that over the interval logged the fracture heights were 4,823-4,630, and 4,600-4,550 ft.

Proppant was tagged with iridium. Variations in the borehole stress measurement appeared to contain the upper interval, but no correlation could be determined in the lower interval.

The SGR log run on Stage 2 indicated only low concentrations of antimony. This could have been caused by overdisplacement of the slurry containing the tagged sand (antimony) during the treatment, or that the perforations were more aligned with fracture orientation.

Iridium also appeared in the near-well bore region of the frac. This probably occurred during the flowback process. It appeared that the heights of the fractured zones were possibly contained by variations in the borehole stress profile.

Based on the SGR log, Stage 3 appeared to be the most effective of all the stages except that the bottom two perforations were not treated.

Proppant in Stage 3 was tagged with iridium. Higher levels of radioactive tracer present in this zone indicated that the quick turnaround time on this, the last of three stages, left more proppant near the well bore (Fig. 4). To identify the producing intervals, a production log run, on July 17, 1992, showed gas flow primarily from the perforated interval, 3,900-4,310 ft.

This interval contributes about 75% of the total gas production and about 10 bw/d.

The interval from 3,100 to 3,500 ft produces about 25% of the total gas production with minimal water production.

Copyright 1993 Oil & Gas Journal. All Rights Reserved.