Dollarhide Devonian CO₂ Flood: Project Performance Review 10 Years Later

The Doilarhide Devonian Unit was formed in 1959. It is located in the Doilarhide Field, Andrews County, Texas, 70 miles west of Midland, Texas, (fig. 1 & 2) Discovered in June 1945 by Magnolia Petroleum and Humble Oil & Refining Co., the Devonian is one of 4 productive intervals within the Doilarhide Field. Subsequent to primary depletion (1947-61) and a successful waterflood (1962-85), a patented tertiary CO₂-hybrid WAG miscible process was engineered to maximize recovery. The Doilarhide Devonian Field is one of two Devonian miscible CO₂ floods currently operating in West Texas and produces a 40 API light crude oil at a depth of 7800 ft. Production peaked in the vicinity of 8000 BOPD under primary, 9000 BOPD under waterflood and is currently at 3080 BOPD and climbing under miscible CO₂ injection. Original oil in place is estimated at 145.8 million bbls of oil. The estimated primary and secondary recoveries from the Unit total 62.6 MMBBLS or 43% of the OOIP. Approximately 19.5 MMBBLS or 13.4% of this volume are attributable to primary recovery. Ultimate tertiary recovery is expected to be 20.4 MMBO with an additional 7.3 MMBO captured through infill drilling. Primary and secondary performances are discussed in greater detail by Poole (1).

This CO2 tertiary project was implemented over the course of 10 years (Fig. 3). Phase 1 was the first of 5 phases to be completed in this field in 1985-86. Phase 2 then followed in 1987-88 completing tertiary development over an area called the southern fault block. Although simulation studies (2,3) predicted recoveries in the order of 14% under tertiary development, mixed results were obtained from CO₂ injection in this block. There was no question however that additional oil was being captured through miscible displacement. By May 1988, production in phases 1 and 2 had increased to 1672 bbls/d up from 924 bbls/d in August 1985. This prompted the continuation of the development into the Phase 3 area, located in the northern fault block. By the end of 1990, the entire Phase was under CO₂ injection and showing earlier than anticipated response in several wells. Finally, 1993 and 1994 saw the addition of 15 infill producers and 24 well conversions to CO2 injectors to bring the total well count to 101 active producers and 80 injectors. The field was originally drilled on 40 acre spacing and is currently producing on 20 acres after an extensive infill drilling program. Performance to date has been less than anticipated and current efforts are being made to alleviate operational difficulties causing this shortcoming.

This paper presents an update of the flood performance and a cursory evaluation of its behavior over the past 10 years of existence. A comparison of the different phases will be provided to emphasize the influence of the geology and operation strategies on sweep efficiency and response to CO₂ flooding. Modifications, to maximize sweep efficiency, flood response time and improve project economics, were made to the later development phases based on observed behavior in Phases 1, 2, and 3. The findings obtained from these are currently being used to optimize the overall production strategy for this tertiary recovery project.

The Doilarhide Devonian reservoir is productive from the Thirtyone Formation. Sailer et al. (4) divided the Thirtyone Formation at Doilarhide field into 5 main lithologies (fig. 4). These facies may be simplified into two main reservoir facies separated by a tight limestone. They will be referred to in this paper as the upper and lower porosity of the Unitized interval. The Thirtyone formation has a conformable contact with the underlying Silurian Wristen Formation and an unconformable contact with the overlying strata. Regionally, the Upper Devonian Woodford Shale is above this unconformity, however, locally the Woodford Shale has been removed by erosion.(5) This has occurred at Dollarhide on the crest of the anticline.

The Dollarhide Devonian field is a faulted, asymmetric anticline subdivided into four major fault blocks by a series of transverse faults which act as flow barriers. Located within the Dollarhide Devonian Unit are three of the four blocks, the fourth being located in North Dollarhide. The top of the Devonian structure is truncated by the Pennsylvanian unconformity which removed the Mississipian Limestone, Mississipian Woodford shale, and the Upper Devonian reservoir.(6). The middle and southern fault blocks show areas where all Devonian has been eroded altogether.(fig. 5)

The northern fault block, which houses phases 3 and 4, possesses the highest reservoir quality. It is characterized by good reservoir continuity and homogeneity both in the Upper and Lower porosity intervals. Faulting in this block is relatively minor and mostly concentrated on the edges. This fault block shows the best Lower porosity pay and has the best recorded recoveries under waterflood. The southern fault block, which includes the Phase 1 and 2 areas is slightly inferior in quality to the northern fault block. This is mainly due to the elaborate system of small faults throughout the pay sections. Although continuous and of good quality, the Upper and Lower porosity pays are highly variable which accounts for the lower recoveries under waterflood. Lastly, the middle fault block, which consists of Phase 5 and small portions of phases 2 and 4, is highly discontinuous and variable in nature. Lower reservoir quality and heavy faulting have resulted in lower than anticipated recoveries. The interpretation of a 3-D seismic survey conducted in 1988 revealed considerably more faulting than originally thought and provided great insight into problem well behavior and infill drilling location selection.

The Lower porosity interval accounts for 83% of the total oil in place and is therefore the main target for tertiary recovery in this reservoir. The Upper porosity shows permabilities and porosities and net pay averaging 12 mD, 8.5 percent and 28 ft as compared to the Lower porosity�s 9 mD, 17 percent, and 48 ft.

The original tertiary flood project was to be implemented in four phases over an 11 year period. This was done to reduce early peak CO₂ demand, maximize the utilization of recycled CO₂, minimize manpower requirements, and provide flexibility for acceleration or deceleration of project development depending on market conditions. More importantly, however, this would provide enough time to evaluate early Phase 1 behavior and adjust the subsequent phases accordingly to maximize recovery. The southern fault block was selected for the Phase 1 area because of its secondary performance, its geological characteristics, and its remaining oil in place. This Phase was to have a 25% hydrocarbon pore volume slug of pure CO₂ injected followed by chase water, all at a reservoir pressure of 3600 Psia, well above the 1600 Psia minimum miscibility pressure. A WAG process would be instituted only if premature breakthrough became a problem. CO₂ was trucked in for an 8 month period during which the necessary CO2 transportation pipelines construction and supply contracts were finalized.

Additional computer modeling of the Phase 1 area showed that WAG injection provided the benefit of greater mobility control at the expense of rate acceleration.(2) Optimization of the model showed that a large slug of CO₂ followed by WAG injection could capitalize on the benefits of both mechanisms. Using this process, a 14% incremental recovery could be obtained, however, the slug size had to be increased to 30%. This process was subsequently patented by UNOCAL under the name HYBRID-WAG process where a 9% pore volume of CO₂ was to be injected followed by the remaining 21% at a 1:1 WAG ratio.

Phases 1 and 2 were developed according to this design, -seeing 8.9% and 9.0% HCPV of CO₂ injection prior to going to WAG. The Phase 3 area was put on CO₂ injection in May 1990 and saw a 6.6% HGPV of CO₂ injected prior to going to WAG. It was felt that the higher permeability of the northern fault block would precipitate breakthrough and therefore require mobility control at an earlier stage. Phases 4 and 5, located in the middle fault block of the Dollarhide Field, were developed in 1993-94 through a series of infill wells and conversions. CO₂ injection commenced in November 1993 in both phases.

Tertiary response, for Phase 1, was first experienced at the 47-9 well after only 9 months, considerably sooner than the anticipated 18 months predicted in the original computer simulation study. This was mainly due to the irregularity of the pattern as the 47-26 injector was drilled on 10 acre spacing. Other wells in the area, such as 47-6 and 47-7 responded in 22 and 21 months respectively. On average, response time for wells in the Phase 1 area was 35 months and production peaked at 350% of the pre-flood rate. CO₂ production was first detected 17 months after first injection and only in minimal quantities. Peak CO₂/oil ratios were recorded in November 1993 at 3.7 MCF/BBL and have since decreased to 1.8 MCF/BBL.

When examined individually, a number of wells outperformed the Phase average response. More specifically, wells 46-6, 47-8, and 48-4 showed production increases of 475%, 500%, and 910% respectively. This is explained by the high reservoir quality of the area and above average injectivity of the supporting injectors. Also, the WAG process, once believed to enhance conformance and recovery has been shown to impair well productivity and oil cuts.(7) Several patterns have shown decreases in oil production within 1 month after CO₂ injection was reestablished. Once the pattern was put back on CO₂ injection, a period of 2 to 3 months would be required to regain the rate loss. The fact that cycling, within the Phase 1, 2 and 3 areas had been scheduled in such a manner that at least 2 injectors were on CO₂ injection and two on water injection for each pattern, obscured the adverse injectivity effects caused by water. This conclusion became obvious once continuous CO₂ injection was re-established in mid- 1994.(7)

Primary and secondary recoveries for Phase 1 were estimated from production decline at 15.8% and 25.7% respectively, giving a secondary to primary recovery ratio of 1.63. The decline rate under waterflood was estimated to be 11.9 %. Under CO₂ flood operations, the production rate, after peaking at 1240 BOPD, declined at 6.9% until 1994. An internal study showed that, in certain areas of the reservoir, CO₂ injection had been predominantly going into the Upper porosity where only 18.2% of the OOIP is located. (8) As of August 1994, Phase 1 had seen on average in excess of 24% of the CO₂ going into the Upper porosity and some individual patterns posted numbers as high as 84%. Pattern optimization and profile modification have, over the past two years, helped eliminate the decline, redirect injection, and turned the trend around to see an increase in the oil rate (fig. 6). At this time, tertiary recovery for this Phase is projected to be 16.3% using exponential decline. To date a total of 34.8 BCF equating to a 23.1% HCPV of CO₂ has been injected and only 0.3 BCF produced over the course of 10 years. CO₂ breakthrough and vertical migration have not been a problem. As anticipated, watercut in this Phase has decreased to 70.8% from 84.8% prior to initiation of CO₂ injection.

Tertiary development of the Phase 2 area commenced as the bottom fell out of the oil market. With a less than attractive perspective on the horizon, UNOCAL chose to delay the project until some firming of the WTI price. As of July 1988, CO₂ was being injected at a rate of 5.5 MMCFD into 12 wells and development of the southern fault block was complete. When taking into consideration the poorer reservoir quality of Phase 2, this rate compares favorably with the 15 MMCFD injection in the 21 injectors of Phase 1. First response to CO₂ injection in the Phase 2 area was recorded at the 44-7D well some 7 months later. On average, however, production response in the Phase 2 area averaged 34 months. CO₂ breakthrough occurred 18 months after start of injection and only in minimal quantities. Maximum recorded CO₂/oil ratios were measured in December of 1991 at 8.47 MCF/BBL and have since settled at 2.3 MCF/BBL.

A detailed examination of the individual well behavior reveals some interesting factors. 40-35 and 40-36, which hold the best Upper porosity in the middle fault block, have shown the greatest production increase of the Phase 2 wells coming in at 880% and 710% increases respectively. This is explained by the relatively higher permeability of the Upper porosity in relation to the Lower porosity. This, in combination with the fact that over 55% of the CO₂ injection within the pattern has gone into the Upper porosity explains the high increase of short duration. Unlike phase 1, the wells of Phase 2 were only slightly affected by the WAG cycles. The relatively limited occurrences of good Lower porosity development in this Phase minimized the influence of the loss of injectivity caused by the reduced relative permeabilities encountered following a WAG cycle.

Ultimate primary and secondary recoveries for this Phase were estimated at 13.1% and 22.2% respectively for a secondary to primary ratio of 1.69. A decline rate of 19.8 % under waterflood was slowed considerably once miscible operations were initiated (fig. 7). Peak rate for the Phase 2 area was recorded in August of 1993, at 546 BOPD, after the drilling of the 20 acre infill 40-38D. On a 2.7% tertiary decline from 1989 until the drilling of 40-38, phase 2 production has stabilized 350 BOPD and is currently on an incline. Much like in the Phase 1 area, on an injected pore volume basis, the Upper porosity's 13.7% is substantially more CO₂ than the Lower porosity's 9%. For the Phase 2 area, 24.2% of the OOIP is located in the Upper porosity and it has received 34% of the CO₂ injection to date. As in Phase 1, profile modifications and pattern optimization are being used to maximize and possibly accelerate production. Under current operations, tertiary recovery is expected to be 16.8% using exponential decline. To date, a total of 14.2 BCF of CO₂ (17.4% HCPV) has been injected and only 0.3 BCF of it produced since 1988. CO₂ breakthrough has once again not been a problem and continuous CO₂ injection has been implemented throughout the Phase. Consequently, watercuts in Phase 2 have dropped from 84.4% down to 73.4% currently.

Phase 3 development followed immediately after Phase 2 completion. With a portion of the infill drilling program initiated in 1986, the phase was infill drilled to 20 acres in 1989 and 1990. A total of 20 producers was drilled and 18 wells converted to CO₂ injection. Oil production prior to CO₂ injection was 590 BOPD and currently averages 1100+ BOPD. The first indication of tertiary response was recorded at the 13-1 and 14-1 wells within 3 months as compared to the phase average value of 27 months. This was highly unusual and after closer examination it became obvious that the fault system associated with the area surrounding the two wells was responsible for the rapid response. Tertiary peak oil rate to pre-CO₂ production rate for this phase calculates to 1.61. Currently, phase 3 is seeing the highest injection rate of all phases at over 18 MMCFD. In order to prevent CO₂ migration to outside acreage, water injection was maintained at the northern and western edges of the Unit. CO₂ production was first recorded in minimal volumes 13 months after first injection. CO₂/oil ratios peaked at 6.0 MCF/BBL in February 1994 and are currently at 3.9 MCF/BBL.

On a well by well basis, 10-106, 15-5, and 15-4 really stand out. They show respective production increases of 470%, 690%, and 390% as compared to a phase individual well average Peak to pre-flood rate of 250%. This anomaly is due mainly to the fact that adjacent injection wells are either open hole completions or have 3 or more high shot density injectors (6+ SPF) within the pattern. Once again, this phase is still in the preliminary stages of CO₂ flooding and several wells are inclining at the current time. Of interest is the occurrence of a breakdown in the CO₂ injection system in December of 1994 for a period of 3 weeks. Production in the phase 3 area, which had been on a steady incline for over 3 years, was almost immediately affected resulting in an estimated 200+ BOPD production loss. CO₂ injection was restored shortly after this failure but production rates lagged in resuming their incline. Production has only recently recovered and now tracks the pre-WAG incline rate.(fig 8)

Decline predicted primary and secondary recoveries for phase 3 calculate to be 12.2% and 39.9% respectively for a secondary to primary ratio of 3.27. The high secondary recovery in this phase is attributable to superior reservoir quality and the absence of minor faulting present in all other phases. A decline rate of 9.4% under waterflood compares favorably with phases 1 and 2. Under miscible flood operations, the oil production is on an increasing trend and currently averages over 1100 BOPD. As for the other two phases, phase 3 has seen a substantially greater pore volume of upper porosity CO₂ flooded. Historical injection profiles show a cumulative 39.4% of the injection going into the upper zone. In certain wells having both upper and lower pay present, up to 65% of the injection has gone into the upper section. This problem is being alleviated by an aggressive profile modification workover program. This is doubly important as over 84% of the OOIP of phase 3 is located in the lower pay interval. Predicted tertiary recovery for this phase is currently estimated at 16.0% using exponential decline. To date, a total of 27.5 BCF (13.9% PV) of CO₂ has been injected into phase 3 and 3.8 BCF produced. CO₂ production has escalated in two of the 29 producers but has stabilized at 1.1 MMCFD and 0.75 MMCFD. Watercuts have decreased from 91% in 1989 to below 80% currently.

Phases 4 and 5 were developed simultaneously in 1993 and 1994 with the drilling of 15 infill producers and conversion of 27 injectors. Currently, phase 4 has 14 producers and 13 active injectors as compared to phase 5's 14 producers and 11 injectors. Encompassing an area of poorer reservoir quality, these two phases were anticipated to have lackluster responses to tertiary miscible CO2 flooding. Oil production prior to tertiary development was 189 BOPD and 267 BOPD for phases 4 and 5 respectively. These figures have increased to what is today 595 BOPD and 276 BOPD. Several operational problems have plagued these two phases, from severe scale formation to injection string leaks and poor downhole conditions. Injection started in the third quarter of 1993 and CO₂ volumes total 3.8 BCF and 3.1 BCF for phases 4 and 5 respectively. At this stage, CO2 production has been measured only in very small volumes, (fig. 9,10)

Only six wells have demonstrated significant response to date and all are part of the phase 4 area. Strangely enough, the responses have been significant ranging from a 125% increase in 27-1 to a 950% increase in the 26-12 well. This is not surprising when one considers that the well 26-4, which is the common injector to the wells showing high responses, initially had the Unit's highest IP rate. On average, however, phases 4 and 5 have had increases of 144% and 66% respectively. Limited profile information is available at this time to make an accurate evaluation of the distribution of injected CO₂ and zonal displacement. Unlike the first three phases, phases 4 & 5 have remained on continuous injection and will continue in this manner until breakthrough problems are encountered.

Ultimate primary and secondary recoveries calculated to 12.4% & 27.6% and 12.9% & 22.1% for phases 4 and 5 respectively giving secondary to primary ratios of 2.23 and 1.71. Both phases showed strong secondary decline rates coming in at 11.5% and 14.0%. Phase 4 is currently producing at its highest level since flood initiation (595 BOPD) and 4 wells are responsible for 90% of the phase's production. Phase 5 peaked in early 1994 during the infill drilling phase at 397 BOPD and is currently holding steady at 275 BOPD, slightly above the pre-flood rate of 265 BOPD. Watercuts have seen a significant drop in phase 4 going from 89+ % in 1993 to less than 70% currently; Phase 5 has yet to see a decrease.

The major challenge faced by UNOCAL in optimizing this flood was to increase the displacement efficiency, accelerate the reservoir depletion, and increasing project economic viability without sacrificing recovery. Injection profile analysis confirmed that considerable improvement in displacement efficiency could be achieved by redirecting the majority of the injection into the lower porosity zone. This, combined with higher injection rates would offset pattern withdrawals which had historically been below unity. For this purpose, a proposal was put forward to increase perforation density in the injection wells. This was based on the belief that the magnitude of the response to CO₂ flooding was directly correctable to the cumulative shot density of the pattern injectors. New information – is currently being gathered to allow presentation of this hypothesis in a subsequent SPE paper. To date, however, production gains in excess of 600 BOPD have been achieved by using this approach and additional increases are expected as the work is completed.

In order to have an effective CO₂ flood, a CO₂-Hydrocarbon miscible solvent bank has to be formed and maintained to maximize displacement. The introduction of water through the WAG process hinders this mechanism and severely reduces displacement efficiency. Similarly, an adverse effect of the WAG process is the resulting oil trapping from the introduction of water. Laboratory verified that simultaneous injection of solvent and water into waterflooded cores results in trapping of both oil and solvent. Experiments using long Berea cores demonstrated that WAG ratios between 1 and 3 severely reduced oil recovery.(9,10,ll,12,13,14). These conditions closely resemble that of the Devonian Formation at Dollarhide and confirm what has been witnessed in the field. This was again reinforced by Huang & Holm in their Dollarhide Devonian core flooding tests where recoveries under continuous CO₂ injection were 12 to 16% greater than under WAG. (15) Hadlow states that on average, CO₂ projects will show a 20% loss in injectivity in the WAG process, hence slowing recovery. (16)

In all phases, water production and watercut increased prior to the oil response indicating a more efficient and uniform displacement front due to reduced capillary forces at the pore throat level. This was more prominent in areas where watercuts had reached 90%+ prior to CO₂ injection. Beliveau & Payne(17) have documented similar behavior in Canada.

Long before the implementation of the CO₂ miscible flood, scaling has been a major concern at Dollarhide. During the waterflood operations, produced waters from the Clearfork, Devonian, Fusselman and Ellenburger were mixed and reinjected. This, in combination with the introduction of CO₂, has contributed to considerably worsen the scale occurrences in the Devonian. In order to minimize this problem, several modifications to the water handling facilities were made resulting in the elimination of 98+% of the fines and particulate from the injection stream. The resulting effect on scale formation has yet to be noticed, however, the amount of fill present in the injectors has noticeably decreased.

Of interest is the progression of scale with various stages of the CO₂ flood. In the early stages of the flood, the majority of the scale recovered from producers and injector conversions is calcium sulfate. This stage is felt to be limited to an average of two cleanouts. The second stage shows more iron sulfide concentration with traces of calcium sulfate and calcium carbonate and approximately lasts from 4 to 12 cleanouts. The majority of the scale recovered from the third stage is calcium carbonate and appears to occur for 4 to 6 cleanouts. Asphaltenes come next for a very short period, usually 2 to 3 cleanouts over a period of 12 to 18 months. Finally, once all these stages have been attained, few, if any at all, scale problems are encountered again. It is still too early at this point to tell if scale occurrences have been totally eliminated as few wells have passed the asphasltene stage, but results are encouraging. Proximity to faults and fractures seem to play a key role in the amount of scale encountered in.producers.

The original design for the Dollarhide Devonian CO₂ Miscible Flood called for a project CO₂ utilization factor of 6. Currently, the five phases show higher utilization rates as they are still in the early stages of the flood. Figure 11 shows the various utilization rates as a function of time for all five phases. Phase 1, which started in 1985 shows a steady decrease in the instantaneous utilization factor until mid-1994. Phase 2 has behaved in a similar fashion since commencing injection in 1988. Phase 3 was on a steady increase until 1993 and has since leveled at the 14 MCF/BBL level. Phases 4 and 5 being at the early stages of the flood, are still expected to rise over the next several years before a downward trend is seen. Currently, the instantaneous factors average 14 MCF/BBL for 4 out of the 5 phases and is decreasing on a yearly basis. At this time, projections for the full-life project utilization factor are between 6 and 8 MMCF/BBL.

The WAG process is detrimental to productivity and recovery at Dollarhide.

Tertiary response for the Dollarhide Devonian CO₂ flood is expected to average 250% of the pre-flood rates.

Maximizing CO2 injection by increasing shot density in the Lower porosity interval can significantly increase well productivity.

The high variability of the reservoir quality and the faulted nature of this reservoir make individual pattern monitoring essential in the optimization of this tertiary flood.

Scaling occurrences are highly dependent on fluid throughput, flood maturity and proximity to faults and fracture systems.

The author wishes to thank UNOCAL and the other coowners of the Dollarhide Devonian Unit for permission to publish this paper.