Small-Scale EOR Pilot in the Eastern Eagle Ford Boosts Production

The Lower Eagle Ford in the northern Eagle Ford extension, north of the San Marcos Arch, is similar in age and time of deposition to the Eagle Ford in south Texas. Across this basin, we see higher siliceous content and lower carbonate contents when compared to the Eagle Ford deposition south of the Arch (Vallabhaneni et al. 2016). The subject well is on the northern end of this extension, located in Madison County, Texas, USA. Well logs, core, and production results show that the reservoir quality across the basin is quite comparable. The target reservoir in Madison County is a siliceous mudstone that has a total organic carbon content in the range of 3 to 8% as observed in cores. Average reservoir pressure gradients place it in the overpressured category at 0.71 psi/ft with a bottomhole static temperature of ~235°F. Oil produced ranges from 39° to 42° API with solution gas/oil ratios from 500 to 700 scf/bbl. Long-term water production is low, settling at ~20% of total fluid production within a few months of the start of production.

Development in the Lower Eagle Ford has accelerated in the past few years as large public exploration and production operations made their entrance into the subbasin. Before the advent of horizontal drilling, there were many penetrations in vertical wells, but there were very few completions due to the characteristically low permeability, and those that were attempted were uneconomic. As operators came to this area attracted by the potential of horizontal development of the tight conventional Woodbine sands, which lie 400 to 600 ft true vertical depth above the Eagle Ford, they also began to experiment with the Eagle Ford shale. The progression of technique for completion has been the primary driver in the improving results from the reservoir over time, a very similar pattern to the one we have seen in every major horizontal shale basin in the US. The standard progression of longer laterals, large proppant to small, gelled fluids to slick, increasing sand volumes, and decreasing stage spacing has generally led to higher initial production and estimated ultimate recoveries in the search for higher returns.

Miscible gasflooding is a proven viable process that enhances oil recovery in conventional and low-permeability reservoirs (Rao 2001; Gozalpour et al. 2005; Sohrabi et al. 2008; Kovscek et al. 2008; Arshad et al. 2009; Wang and Gu 2011; Anand et al. 2019; Sie and Nguyen 2020a, 2020b; Liu et al. 2022). Several laboratory investigations and field tests of the efficiency of different gas injection strategies for EOR in shale formations have been reported in the literature over the past few years (Hawthorne et al. 2014; Sheng 2015; Hoffman and John 2016; Alfarge et al. 2017, 2018; Tovar et al. 2018a; Hamdi et al. 2018, 2021; Hoffman 2018; Jia et al. 2019; Carlsen et al. 2020a, 2020b; Huang et al. 2022). The hydrocarbon gas huff ‘n’ puff process was found to be the most promising strategy based on both reservoir simulation and field-scale pilot testing (Alfarge et al. 2018; Hoffman and John 2016; Hoffman 2018). However, the use of field gas, as in shale huff ‘n’ puff, has not been well explored because, unlike CO₂ which is another commonly used huff ‘n’ puff, field gas is a mixture of hydrocarbon components that have different interactions with organic matter–rich shales, and thus a laboratory evaluation of cyclic injection of field gas in shales would be difficult. It is commonly observed that the composition of natural gas has been oversimplified in most previous laboratory huff ‘n’ puff studies (Li et al. 2017a; Hawthorne et al. 2017).

An experimental study of huff ‘n’ puff processes by Hawthorne et al. (2017) showed that both CO₂ and ethane could recover nearly all of the oil initially in Bakken shale cores. The authors of that study also concluded that the use of methane above its minimum miscibility pressure with the crude oil reduced oil recovery by only 10%. It has also been well established that enriching methane with natural gas liquids in huff ‘n’ puff could significantly reduce the minimum miscibility pressure and enhance oil extraction (Tran et al. 2021). Song et al. (2021) conducted huff ‘n’ puff experiments on Upper Devonian Duvernay core samples and observed that adding propane to the injected mixture of methane and ethane resulted in an increase in oil recovery. However, the oil recovery factor does not monotonically increase with the injection pressure. The experimental work by Thomas et al. (2020a, 2020b) showed that the increase of recovery factor for pure methane cyclic injection became marginal as the huff pressure was raised above an optimal value. The compositional complexity of crude oil could also strongly impact the performance of a gas huff ‘n’ puff process. For example, the oil recovery in shale samples with methane cyclic injection could be as high as 80% of gas condensate but reduce by more than 30% for a volatile oil (Thomas et al. 2020a). The impact of the initial solution gas before gas injection has also been found important (Mahzari et al. 2021)

The industry has long understood that shale reservoirs have characteristically low primary recoveries (US Energy Information Administration 2016). This was confirmed by the petrophysical work done during primary development. As the companies in the area searched for ways to improve the overall economics of shale development, this area was identified as a good candidate to test the effectiveness of huff ‘n’ puff recovery. Due to the low development density in the subject area, many of the well-to-well interference problems experienced in overdeveloped reservoirs could be avoided. Another significant factor in the decision to implement the project was the availability of relatively low-cost rich injection gas that was controlled by the operator of the well. To determine in which well to first implement the process, the available wells were screened, and wells that had low-intensity initial completions or that were producing more than 100 barrels of oil per day (BOPD) were eliminated. Injecting into a well producing more than 100 BOPD was an economic risk the company was not willing to take, as the outcome of the experiment was unknown and it was feared that the process may damage the wellbore or reservoir. In screening for a suitable example well, it was recognized that the size of the primary estimated ultimate recovery would probably be indicative of the potential recoveries that could be expected from an EOR process. In the final selection, the Montana 1 H (42-313-31220) was the oldest well that had a full size fracturing job applied to it, had not demonstrated any well interference with nearby wells during stimulation, and was not near any known faults. At the time the first injection began, the subject well had been producing for 4.3 years and had a cumulative production of 152,000 bbl of crude oil, 47 million ft³, and 48,000 bbl of water. The minimum breakdown pressure observed while stimulating this well was 8,087 psi as observed through the instantaneous shut-in pressures after each stimulation. Measurements taken after the well primary production began indicate a reservoir pressure of approximately 6,174 psi.

Once a decision was made to implement the pilot, experiments were designed to understand the fluid-fluid interactions that would guide many of the design and sizing decisions for the pilot. Once the physical interactions were understood, engineering design sized and scoped appropriate equipment for the pilot. It was important that the equipment selected would use the known parameters and allow field experiments to be designed that would allow for quick learning cycles and modification. The remainder of this paper will serve to describe the process of experimental design both at the laboratory and field level and to highlight some significant results and observations.

The injection gas used for all laboratory tests in this work consists of 66.20 mol% methane, 18.65 mol% ethane, 10.48 mol% propane, 3.55% n-butane, and less than 1 mol% n-pentane. A core plug sample (2 in. long and 1 in. in radius) taken from a shale formation was used in the gas huff ‘n’ puff experiments. The permeability of this core plug is expected to be very close to 0.05 md. This permeability was obtained from a permeability test (i.e., the rock-eval pyrolysis) using a twin core plug taken from the same well at nearly the sample depth. It is uncertain that this measured permeability includes fracture permeability. The core porosity of 4.98% was determined based on the bulk volume of the core plug and the amount of dead crude oil saturated in the core plug before the field gas extraction experiment. The specific gravity of the dead crude oil used in this work is 0.8167 at 60°F. This oil sample was taken from a shale oil formation producing at a very low gas/oil ratio. It is noted that the use of dead crude oil in the huff ‘n’ puff experiments in this work would not precisely capture the interaction of actual crude oil at very low gas/oil ratio in the reservoir with the injection gas. The composition of the dead crude oil sample was measured by standard condensate analysis (GPA-2103) and is presented in the  Appendix.

Fig. 1a shows a schematic of the experimental setup used to determine the minimum capillary pressure (MCP). The improvements on this design were based on the work by Hawthorne et al. (2016). Three capillary tubes (World Precision Instruments) with inner diameters of 1.12, 0.84, and 0.68 mm, respectively, were used in the experiment. The capillary tubes and a piston were placed into a custom-designed high-pressure sapphire visualization cell. The crude oil with the desired volume was transferred into the cell while the piston was at the bottom of the cell and the cell was heated up to the reservoir temperature (235°F). The synthetic field gas was then injected into the cell until the system reached a target pressure. The hydrocarbon liquid column heights in the capillary tubes (e.g., Figs. 1b and 1c) were monitored and recorded by a camera. The MCP is an estimate of the minimum pressure above which the liquid imbibition into the capillary tubes is completely inhibited because the interfacial tension between the gas and the oil at pressures above the MCP becomes so small that the capillary pressure could be negligible (De Gennes et al. 2004). Thus, the effect of system pressure on the oil extraction during a huff ‘n’ puff process could be better evaluated without a significant effect of capillary pressure (Sie and Nguyen 2020c).

The residual oil in the reservoir core was removed using the gas mixture and the following procedure. The core was placed in a high-pressure cell in the oven, and then the system was vacuumed for an hour. After the system reached 235°F, the gas was injected into the cell until the pressure of the system reached 4,500 psia. This pressure was selected based on the MCP as discussed in the Results and Discussion section. The core was soaked for 3 days at this pressure to remove the residual oil and then placed into a glass vial at atmospheric pressure for another 12 hours to remove the residual gas. Next, the clean core was saturated with the dead crude oil and soaked in the synthetic field gas in the high-pressure cell at 235°F and 4,500 psia. The same gas mixture was used to flush the cell at 4,500 psia after a 24-hour soaking period to remove the produced oil from the cell. This soak ‘n’ flush process was repeated until no further oil production was observed.

The beginning points of the facility design included consideration of both the pressure required to stimulate the reservoir as well as an acceptable flow rate of gas into the reservoir. Due to the cyclic nature of this process, a mobile equipment design was specified. Above average gas pressures required for the process demanded an in-depth review of equipment and procedures to ensure safety. The details of each of these points and the impact on the facility design are discussed.

The pressure required to stimulate the reservoir needed to be above the MCP of approximately 4,500 psia (refer to Results and Discussion), but significantly below the breakdown pressure of 8,087 psig. These pressures are substantially above conventional field gas pressures (1,200 psig) and rapidly demanded a thorough search for unconventional compression equipment as traditional field gas compressors are limited to approximately 1,200 psig. Compressors achieving higher pressures have historically been almost exclusively purpose-designed and -fabricated large stationary compressors (Jacobs 2019). However, a relatively new compressor design capable of developing pressures of up to 5,500 psig for a relatively new artificial lift method, high-pressure gas lift , entered the market and offered promise. The compressor is a critical piece of equipment for the project, influencing a number of important factors in the success of this project.

Gas flow rates to be injected into the reservoir were bound by two factors: On the low end is gas compressor throughput, which was approximately 1.5 MMscf/D, and on the high end is gas availability from neighboring wells, which was limited to approximately 4 MMscf/D. To minimize the time to pressurize the reservoir, two compressors were installed in parallel. Acknowledging this project as a pilot and the potential for learning and needed changes, the idea of scalability (both up and down) was highly desirable.

The desire existed in this pilot to try the gas EOR process across a number of various wells in the field. In lieu of building an expensive and long gas gathering and/or distribution system, utilizing mobile equipment was deemed necessary. Skidded equipment (compressors, production separators, flow meters, etc.) must be movable by flatbed trucks with rolling tailboards and winches as opposed to requiring cranes. Connection of the compressor process gas system to the wellhead was done with 1502 flow iron, which provided the practical benefit of eliminating the need for welding and hydrotesting operations during installation. Standard seals were replaced with Viton seals to raise the temperature limit of the flow iron to over 300°F. In keeping with the rapid mobilization/demobilization theme, utility connections, such as scrubber dump lines were made via flexible high-pressure hydraulic hoses. In the end, a modular and easily transportable equipment approach was taken and became valuable to the economic viability of the project.

A process hazard analysis or hazard and operability analysis is insightful in any new project; however, given the above average pressures here, these analyses were deemed imperative and revealed some necessary safeguards. Because lower pressure equipment previously existed, in a move of extreme caution, blinds were installed on the valving in the production tree wing to isolate the production facility from high pressures during injection phases.

Given the high pressures combined with the high Btu value of the gas, hydrate formation was also a concern. Referencing the adjacent hydrate formation chart (Fig. 2), it can readily be seen that as gas pressures and gas specific gravities increase, so does the temperature at which hydrates can form. In process pressures of 4,000 psia and gas specific gravities of 0.8, hydrates can form in excess of 75°F. Thus, hydrate formation is conceivable at even springtime and falltime ambient temperatures. Elevating gas temperatures and insulating pipes to maintain elevated temperatures in gas lines becomes critical for reliable and safe operations. In cold weather months, use of methanol as an antifreeze agent is a further aid and added insurance.

The mobility of and, specifically, the rapid deployment/redeployment capabilities of the compressor skid are key. A self-contained compressor package that houses oil tanks and fuel supply systems on-skid reduces set time and costs. Requirements for off-skid utilities/services (i.e., instrument air supply, electricity, etc.) should also be eliminated. A truly mobile package should only require three off-skid connections: (i) inlet gas, (ii) gas discharge, and (iii) scrubber drains.

As mentioned above in the Facility Safety section and further discussed elsewhere (Elmer and Elmer 2017), elevated gas temperatures are needed to help manage hydrates. The compression process inherently generates heat, which can be useful. Most compression packages contain oversized gas cooling, negating the positive impact the compression process can contribute to managing hydrates. In EOR applications, gas injection temperatures tend only to be limited by component limitations, which in many cases is the wellhead, for which a 250°F limit is quite typical. As such, little to no process gas cooling in the compression process is oftentimes possible and is the case with the compression equipment selected for this project.

A related matter is dealing with hydrates on the compression skid itself, specifically in the compressor utility systems. Scrubber dump systems are notorious for creating hydrates during the dump process and benefit from heat tracing and insulation. Further, the fuel gas system, now that it is on-skid and with its large pressure drops (and consequently cooling effects brought on from the Joule-Thompson effect), also warrants added heat from heat tracing. This heat tracing is easily provided by heat exchangers and use of coolant from the engine jacket water system.

Fig. 3 shows the results of the capillary rise experiment. The MCP was 4,429.76 ± 138.89 psia, which was estimated by extrapolating the linear regression lines to the pressure at which the oil column height is zero (Hawthorne et al. 2016). When the field gas becomes more miscible with the crude oil, the interfacial tension between them approaches zero (Rao and Lee 2003; Hawthorne et al. 2017). Thus, the MCP is expected to be close to the minimum miscibility pressure. It is noted that the minimum miscibility pressure is often determined by either the slimtube experiment or using an equation of state, which is very different from the capillary raise method for the MCP measurement described above. It is commonly used to evaluate the efficiency of multiple-contact miscibility processes in conventional gasflooding. Based on this result, an oil extraction experiment was conducted at 235°F and 4,500 psia. This experiment confirmed that more than 80% of the initial dead oil in place could be recovered at this reservoir condition. It was also observed that the preponderance of oil recovery occurred in the first two days of extraction.

During the huff stage, the injection gas in the fracture invades the matrix due to a large pressure difference between the fracture and the matrix and thus creates gas-oil mixing zones in the matrix. Thermodynamic equilibration redistributes the phase and composition in the mixing zone (Anand et al. 2019; Tang et al. 2020). During the soaking stage, thermodynamic equilibration redistributes the components at the fluid interface in the mixing zone according to the local composition. The injection gas components dissolve into the crude oil-rich phase and migrate deeper into the matrix by concentration-driven molecular diffusion. Meanwhile, the crude oil components are brought to the vapor phase and diffuse from the matrix into the fracture (Guo et al. 2009; Tharanivasan et al. 2006). During the pressure depletion (puff) stage, as the pressure in the fracture reduces to the puff pressure, the resulting pressure gradient induces the convective flow. Also, increasing gas expansion as pressure continues to drop could carry more crude oil from the matrix to the fracture (Mahzari et al. 2019, 2020). When the huff ‘n’ puff process is operated at high interfacial tension (e.g., below the MCP), the lower huff pressure provides less driving forces for the mass transfer between the oil and the injection gas (Tang et al. 2020; Sie et al. 2018, 2019; Anand et al. 2019). As the interfacial tension is reduced at elevated huff pressure (e.g., above the MCP), this mass transfer is greatly enhanced.

The pilot was designed with a staged approach. Keeping the early injection cycles small minimized the impact of reservoir containment loss and the impact to nearby producers should injection gas break through unexpectedly. The early cycles were kept small as the following series of questions was answered experimentally: What will the pressure response to injection look like? Is there an appreciable fluid composition change between injection and production? How much of the injected gas will be recovered? How fast will oil recovery happen?Table 1 summarizes the design parameters for five gas injection cycles and the associated production responses. Note that the incremental oil recovery was estimated based on comparing the oil rate at the end of each cycle with the oil rate projected by a decline curve analysis had the primary production continued.

The first injection cycles did demonstrate that injection gas was contained and recoverable in large enough quantities to allow the volumes on subsequent cycles to be increased. It was apparent that the injection pressure would have to be raised if large volumes were to be used. The containment was indicated by the consistent rise in injection pressure as injection continued during each cycle. Additionally, the injection well is directly underneath a parallel horizontal well in a sand formation ~500 ft true vertical depth above the injection well and no increase in gas production was observed from this or any surrounding wells.

The facility was operated at its maximum injection capacity throughout the pilot. In the beginning of each injection, the available gas rate from the field limited the injection rate; in the middle of each period, the experiment was limited by the horsepower available for compression; and toward the end, the experiment was limited by the maximum allowable operating pressure of the surface facilities. The injection behavior across all five cycles has been remarkably stable and repeatable (Fig. 4). With the initial 20 MMscf of gas pressurizing up the wellbore and near-wellbore region before establishing a relatively stable injection pressure, much larger volumes of gas have been forced into the far-field regions. Some spread between Cycles 4 and 5 can be observed in the injection pressures, though these are largely due to differences in the daily injection rate at various times. While a high injection rate was planned to be held constant until maximum pressures were achieved, this was not always possible, though operational improvements were made with each cycle that improved injection throughput and equipment run times. Basic design improvements such as adding a slug catcher to catch any liquids coming into the system off the pipeline and strategically placing methanol injection throughout the dump piping were very helpful.

The first three cycles were pressure limited by the maximum allowable operating pressure of the system at the time, with the last two cycles being volume limited after the improvement in the system maximum allowable operating pressure. Once an injection cycle was ended, the wellhead and flow lines had their high-pressure sides blinded off and the production side blinds removed. The well was then flowed at a constant gas rate until produced liquids hit. Due to the constraints of the existing production facility, the maximum sustainable flowback rate was ~1 MMscf/D, during which time a line heater was used along with methanol and an extra medium pressure separator to minimize hydrate formation. Once liquids began to hit, a fixed choke was used to allow pressure to slowly continue to bleed off. The transition from rate limited flow to fixed choke flow and finally to starting the gas lift system when tubing pressures got below 200 psig allowed a gradual bottomhole pressure drawdown to be achieved. Initial flowback during each withdrawal (puff) cycle was 100% injection gas, with liquids gradually increasing in volume as the pressures decreased. Oil samples showed that while the early oils made were quite light, these oils were being dropped out of the gas stream due to the extreme pressure drop from the wellhead pressure to the facility pressures early in the flowback. The API gravity and color change in the produced fluids can be observed in Fig. 5. Once the oil gravity stabilized during each withdrawal period, the gravity and quality of the oil produced were indistinguishable from oil produced before the test.

Gas composition during the injection and production cycle was monitored and can be seen in Fig. 6. The injection gas for the project is a stream of wet produced gas from nearby wells. The Btu content is slightly lower than that of the native gas from the subject well. It has been established that in some cyclic EOR processes, the light ends of the reservoir oil may vaporize into the high-pressure gas stream, leaving behind a heavier and heavier fraction of the oil with every cycle (Sie and Nguyen 2020c, 2022). Based on the sampling of the produced gas stream throughout this pilot test, there is no indication that the gas composition materially changed by being cycled into and out of the well as a working fluid. There is also no indication that any significant fraction of native gas was mixing into the injection gas. This would indicate that the oil being extracted was dead oil without a significant solution gas component.

Both the injection and production responses were stable (Fig. 7), and when compared on a time scale, the production profiles exhibited the characteristics of oil production from an injection gas-crude oil mixing zone. During the huff phase, an injection gas-crude oil mixing zone is generated in the shale matrix, and an induced fracture network referred to as a stimulated reservoir volume is formed. The injection gas components partition from the injection gas into the crude oil driven by thermodynamic phase behavior and continue to penetrate deeper into the stimulated reservoir volume. The peak oil production rate and production rate decline are strongly influenced by the oil content and the extent and volume of the mixing zone, which has been shown to increase with system pressure (Vallabhaneni et al. 2016; Li et al. 2017b; Sie and Nguyen 2022). This is consistent with the liquid production profiles for the five injection cycles (Fig. 7). The volume, rate, and decline of the oil production response appear to be proportional to the injection volumes, with the higher sustained peaks correlating with the larger later injection volumes. The days to peak oil production did continue to get longer for each subsequent cycle as the mixing zone continued to propagate deeper into the stimulated reservoir volume. Based on the core experiments that showed high first-cycle recovery, the delayed peak through cycles could indicate that each cycle is bringing oil incrementally farther from the wellbore. Further investigation could yield useful insights into the reservoir and fluid characteristics and allow quantitative conclusions to be drawn from future experiments. Some attempts have been made to identify oil production mechanisms from the fracture network and the matrix by inverse modeling of these data.

Initial gas production rate was limited by the ability of the surface facility to heat and process the gas. The solvent extraction process that characterizes huff ‘n’ puff in this implementation is dependent on pressure to maintain the desired mass transfer between the injection gas and the oil in the mixing zone. Thus, slower flowback rates were selected to continuously promote the far-field mass transfer and to reduce the effect of competing gas flow on the relative permeability of oil due to excessive pressure depletion. An example of the implementation of this strategy can be seen in Fig. 8, where the initial choke at the surface was adjusted to try and maintain a constant gas rate during the initial flowback period while the flowback was primarily gas. As the fluid fraction of the flowback increased, the gas rate began to fall precipitously due to the rapidly increasing hydrostatic pressure from liquids entering the tubing. Once this rapid fall was observed and confirmed by increasing fluid rates, the choke was held constant until tubing pressure declined to the point that gas lift was needed. This conservative approach seeks to minimize that rate of drawdown in the well while allowing liquids to be produced at reasonable rates.

A primary economic driver is the recovery of the injected gas as it has an economic value that will be lost if the gas is not recovered. It is hard to be certain what the final recovery of any of the cycles would be because no production phase has yet been run until all the gas was recovered or until the well stopped producing gas. From Fig. 9, we can see that when the production phase ended for the next cycle to begin, there was still meaningful gas recovery taking place. If the recovery profiles were projected out to approximate a final gas recovery, it would be in the range of 70 to 80% of the injected volume.

It should also be noted that during Cycle 5 of this project, a gas measurement error caused the underreporting of produced gas from approximately Day 150 until Day 275. The appropriate gas volumes would be a curve projected through the gap caused by the bad data and would positively affect both Figs. 8 and 9. For accuracy, no massaging of the data is done to account for this error in the data presented in this paper.

When comparing the gas flowback from cycle to cycle (Fig. 9b), the slope is a mirror image of the injection plots (Fig. 4), where a slope change can clearly be observed as flow transitions from near-wellbore to far-field or matrix-dominated flow regimes. The amount of gas recovered before liquid production started was also larger with each cycle. Whether this is a function of solely the larger volumes of injection gas each cycle placed or it is also a function of fluids having to migrate from deeper in the formation in each subsequent cycle is difficult to ascertain but is an important area of interest that deserves further study.

The overall effects of the implementation of the huff ‘n’ puff process on this well can be seen in Fig. 10. The cumulative baseline projection, obtained from a decline curve analysis using a reserve estimation software, is shown and makes clear that this trial produced incremental oil over the baseline curve. When trying to decide how long to flow a cycle, these data iarehelpful because they highlight the lost production during injection and how that is made up for during the production phase. The short initial cycles prove not to be overly effective at changing the trajectory of the cumulative production plot. It is because the production time is not very long compared to the injection time during which no oil is produced. The larger longer cycles prove to be the most effective at recovering incremental volumes over the baseline projection. Future cycles will use the profiles created here to optimize the size and timing of the cycles to maximize the recovery and net present value.

These data represent the first well in the northern Eagle Ford extension to have cyclic gas injection implemented; this is one of the few single well examples of huff ‘n’ puff implementation. Preparations for the pilot were focused on several areas, including (1) gaining practical insights that could be used in real time to optimize the process, (2) creating high-frequency learning loops that incorporated the real-time results from the laboratory, and (3) conducting field experiments and driving design and implementation changes. This process of continuous incorporation of learnings allowed several step changes that yielded a result that displayed incrementally better results with each cycle of implementation.

Laboratory experiments were designed to provide understanding and insight into the nature and thresholds of the fluid-fluid interaction as well as the fluid-rock interactions. The MCP for the crude oilfield gas system was about 4,500 psia at the reservoir temperature. An oil recovery test on a shale sample indicates that more than 80% of the initial dead oil in place could be recovered by field gas injection at MCP.

The results of the field experiments answered some basic questions about the containment of injected gas within the zone and allowed for increasing volumes to be utilized. Both on injection and withdrawal, the system displayed remarkable consistency and repeatability in its behavior. The injection pressure response was shown to be predictable based on the volume injected and indicative of the flow regime that the injectant front was propagating through. Likewise, the withdrawal phases demonstrated that under similar flowback procedures, similar flow profiles can be achieved even with incrementally larger volumes of injected gas. The recovery of very high API oil rapidly abated and reverted to normal reservoir quality fluids as the pressures declined during each withdrawal cycle. A critical observation was that good quantities of gas were recovered each cycle and that the composition of that gas was not substantially changed during the injection and withdrawal processes.

The implementation of the huff ‘n’ puff process in this well was not only a technical success in that it produced incrementally more oil than was predicted based on its decline curve, but it was also an economic success from its early stages. Lessons learned in this project have already allowed for several other low-cost tests and allowed the reduction of capital implementation cost to be reduced from approximately USD 250,000 to less than approximately USD 100,000. One of the lessons that enabled this success was designing a system that is flexible, modular, and can be easily adapted as a project progresses.

The authors acknowledge MD America Energy LLC for the support of this study. Special thanks are expressed to Chao-yu Sie, a former PhD student at The University of Texas at Austin, for conducting the laboratory experiments.

This paper (SPE 209429) was accepted for presentation at the SPE Improved Oil Recovery Conference, Virtual, 25–29 April 2022, and revised for publication. Original manuscript received for review 2 May 2022. Revised manuscript received for review 21 January 2023. Paper peer approved 23 January 2023.