Advancements and Operational Insights in the Bakken Shale: An Integrated Analysis of Drilling, Completion, and Artificial Lift Practices

Abstract

This chapter provides an in-depth analysis of the Bakken Petroleum System (BPS) in the Williston Basin, focusing on Improved Oil Recovery (IOR) techniques. It explores the significant advancements in drilling, completion designs, and artificial lift methods that have markedly boosted oil recovery in this prime unconventional resource basin. The chapter traces the history of oil production in the Williston Basin, highlighting the transformative impact of horizontal drilling and multistage fracturing. It delves into advanced drilling operations, emphasizing the role of high-performance motors, geosteering, and real-time downhole data in enhancing drilling efficiency. Additionally, the chapter examines the evolution of well-completion strategies, from traditional to innovative horizontal completions, and assesses their effectiveness through data analytics, numerical modeling, and field studies. The vital role of artificial lift systems in combating rapid production decline in shale formations is analyzed, comparing the efficacy of ESPs, Sucker Rod Pumps, and Gas Lifts. The interconnectivity between operational aspects is discussed, providing a unified view of how integrated strategies and technological advancements drive optimized oil recovery in the Bakken formation. This study aims to offer insights and strategic guidance for industry stakeholders, particularly concerning IOR in unconventional oil resources.

1. Introduction

The Williston Basin holds the distinction of being among the earliest consistent producers of shale oil resources in the United States. Oil was first discovered in the basin in 1936, and the region became a major oil province in the 1950s with the discovery of large fields in North Dakota. Production initially peaked in 1986, but significant increases in production began in the early 2000s due to the application of horizontal drilling techniques, especially in the Bakken Formation [1]. A myriad of successful drilling and completion methodologies took root and were developed within the context of the Williston Basin, and these have since been adopted and implemented nationwide. In its nascent production phase, it constituted the highest production yield from unconventional resource basins in the United States. As of June 2023, the Williston Basin continues to be a significant player in the United States’ oil production landscape, ranking as the second-largest producing region in the country, surpassed only by the Permian Basin [2].

In 2010, the Energy & Environmental Research Center (EERC) executed an in-depth, multidisciplinary research program to rigorously examine the key elements influencing successful oil production in the BPS, North Dakota. This study emphasized four primary areas: geology, geochemistry, geomechanics, and drilling/completion engineering. Initial results underscored the efficacy of horizontal drilling in conjunction with multistage fracturing, the role of geological variables in determining hydrocarbon production rates, and increased production in Mountrail County, linked to heightened concentrations of organic carbon and thicker shale deposits. Emphasizing optimal completion practices involving horizontal drilling and fracture stimulation is fundamental to liberating reservoir fluids held tightly within the formation. Notably, horizontal drilling of the Bakken’s middle member combined with multistage fracturing has surpassed the performance of all prior Bakken wells in North Dakota. On the contrary, multilateral wells, despite their reduced per-foot drilling costs, seem to present no significant advantage in terms of production. The study revealed that in the Williston basin, wells in Mountrail County, one of the largest producers, predominantly used heavy, oil-based mud for drilling, whereas Dunn County wells employed water-based brine. These drilling fluids can cause harm to horizontal wellbores and impact their wettability. Although water-based fluids facilitate quicker and more economical drilling, they lead to increased pipe wear and come with density limitations. Conversely, oil-based fluids lessen drag and pipe wear and provide more flexibility in density. Oil-based drilling muds are crucial for the installation of oil-based swell packers used in multistage fracturing [3].

Based on data from May 2023, the top producing operators in North Dakota in terms of oil and gas production have shown some changes. Continental Resources, Inc. remains a dominant player, producing 6,254,994 barrels of oil and 18,699,327 MCF of gas from 2179 active wells. They’re followed by Marathon Oil Company and Hess Bakken Investments II, LLC, which produced 2,550,434 and 2,298,305 barrels of oil, respectively. Burlington Resources Oil & Gas Company LP has made a notable jump in its contributions, producing 1,991,140 barrels of oil. On the gas production side, Hess Bakken Investments II, LLC remains a significant contributor with 9,638,090 MCF, while Whiting Oil And Gas Corporation produced 7,409,455 MCF. In terms of active wells, Continental Resources, Inc., Whiting Oil And Gas Corporation, and Hess Bakken Investments II, LLC continue to be major operators with 2179, 1786, and 1423 wells respectively. The wells with the highest oil production have shifted. The ‘GT USA 11-18H’ operated by Marathon Oil Company in Dunn County stands out with 65,539 barrels of oil. Following closely are the ‘Dennis Fiu’ series of wells, particularly ‘Dennis Fiu 4-8H’ operated by Continental Resources, Inc. in Dunn County producing 52,828 barrels. As for gas production, the ‘Sorenson Federal 153-96-9-4-14H’ well operated by Ovintiv USA Inc. in McKenzie County dominates, generating a whopping 146,161 MCF of gas. Interestingly, many of the top producing wells are located in Dunn County, with Continental Resources, Inc. being a particularly dominant operator in the region [4].

The drilling statistics report from the North Dakota Industrial Commission from 1951 to 2021 indicates a substantial growth in total footage over the years, peaking in 2014 at 44,941,118. However, a decline occurred afterward, likely attributed to the impact of the 2015 oil price collapse followed by the COVID-19 pandemic. Nevertheless, there was a slight recovery observed in 2021, with the total footage reaching 14,129,194 compared to 12,914,547 ft. in 2020. These statistics highlight the consistent drilling activity in North Dakota over the years, resulting in a combined footage of 384,882,036 [5].

Figure 1 illustrates the initial oil production, initial water cut (WC), initial Gas Oil Ratio (GOR), and wells drilled per year for the BPS. The average values per year are reported. The graph shows a steady increase in initial oil production (24 h), an increase in initial GOR and initial water cut. These observations are attributed to the development of several technologies that permitted the increase in the performance of these wells. The initial water cut trend is a result of an increase in hydraulic fracturing treatment volumes described by Miller et al. [6]. The increase in GOR trend is attributed to the tighter well and fracture spacing explained by Acuña [7]. The implemented fracture spacing results in smaller rock fragments, thus less pressure support. This will cause the pressure in these fragments to decrease rapidly below bubble point pressure resulting in higher GOR. When offset wells are drilled next to existing wells the rock can already be below bubble point pressure resulting in higher GOR. We note the increase in the number of child wells reported by Latrach et al. [8]. The well GOR per location during different periods is illustrated in Figure 2, the figure shows a relatively low initial GOR at early development (2005–2010). After this period, well density increased at the center of the basin with higher GOR. Wells that are further from the highest well density area had a lower initial GOR (Figure 3).

Figure 1.

Oil production in the Bakken and Permian regions (in thousands of barrels per day). Data source: U.S. Energy Information Administration [2].

Figure 2.

Initial GOR per well location. (a) Wells drilled between 2005 and 2010, (b) wells drilled between 2010 and 2015, (c) wells drilled between 2015 and 2023.

Figure 3.

Average initial oil production, initial GOR, initial water cut, and wells count in the BPS.

The performance change in Figure 2 is not representative of the whole basin as there some are companies that perform better than others in terms of initial production. In this study, only the top 6 operators are studied. The top 6 are selected based on the number of wells they own in the Bakken. Figure 4 illustrates the evolution of initial production parameters for these operators. Even though all operators showed an increase in initial production over time, they do not perform the same. Figure 5 illustrates the distribution of well locations per operator. Even though operators have wells within reasonable proximity of one another. These wells did not perform the same; thus the well’s initial production is likely a result of operators’ approach for development.

Figure 4.

Average initial oil production, initial GOR, initial water cut and wells count per operator in the BPS.

Figure 5.

Wells location per operator.

This research endeavor represents an ambitious attempt to unravel the multiple changing variables that characterize oil production within the BPS. It aims to shed light on the divergent performance of different companies, track the evolution of their outputs over time, and contextualize these within the broader trajectory of oil production within the BPS. The primary focus of this paper is to perform a thorough literature review and data analytics study to interpret observable trends in three key areas: drilling practices, artificial lift mechanisms, and completion design.

The methodology of this study is twofold. Firstly, the performance of each operator is meticulously examined, thereby enabling a detailed comparison between the players in the field. Our focus is particularly geared towards those producers who have demonstrated a consistent increase in production and have an extensive array of wells that underpin their operational success. Secondly, through a comprehensive review of academic articles and industry publications associated with each operator, we gain insights into their specific practices, technological preferences, and operational strategies.

2. Key insights into the Bakken formation’s role in oil production

The Bakken Formation, spanning across parts of North Dakota, Montana, Saskatchewan, and Manitoba, is a significant oil and natural gas reserve in the United States [9]. Dating back to the Late Devonian to Early Mississippian periods, this formation within the Williston Basin is characterized by layers of shale, siltstone, and sandstone [10]. Originally considered marginal due to low permeability, technological advancements in horizontal drilling and hydrofracturing have transformed it into a major oil and natural gas producer [11]. By 2012, these developments elevated North Dakota to the second-largest oil producer in the U.S., behind Texas [12]. The USGS estimates substantial undiscovered resources in the Bakken Formation, making it a key player in North America’s energy sector [13].

3. Exploring the background of oil and gas drilling in the Williston Basin

3.1 General well design and challenges in the bakken formation

The Djurisic et al. [14] study describes a three-phase well drilling design in the Bakken region:

• Surface Interval: A 13–1/2 inch hole is drilled to 2100 feet using a rock bit. It involves cementing 9–5/8 inch casings with a rotary assembly. Challenges include managing winter water and maintaining hole inclination.

• Intermediate Interval: Drilling reaches the kickoff point with 8–3/4 inch PDC bits, then curves towards the Middle Bakken horizon. The section uses oil-based mud and a 7-inch casing. Main challenges are PDC bit wear and geological considerations.

• Production Interval: A lateral hole is drilled to about 20,000 feet with 6-inch PDC bits. The stage uses brine mud and possibly a 4–1/2-inch liner. Challenges here include steering accuracy, managing BHA walk, torque, tool reliability, and handling casing wear and well control.

3.2 Drilling operations and technologies

In North Dakota’s horizontal drilling operations, conventional mud motors and Measurement While Drilling/Logging While Drilling (MWD/LWD) systems, along with onsite geological analysis, are frequently used for drilling curve and lateral sections. Positive displacement mud motors with Adjustable Kick-Off angles are used in drilling the curve. The MWD system measures the earth’s gravity and magnetic field, and wellbore inclination with survey stations crucial for tracking the wellbore’s position. These systems need sufficient non-magnetic spacing and proper magnetic corrections for accuracy. The temperature sensor is another vital component, as most sensors are temperature calibrated. The placement of LWD gamma ray modules varies across service providers [15].

3.3 Key studies and their findings

This collection of research papers provides a detailed insight into the advancements in drilling technologies and efficiencies in the Bakken and Williston Basin Petroleum Systems, focusing on several key technical aspects:

Djurisic et al. [14] demonstrated a significant 50% increase in the Rate of Penetration (ROP) over 18 months, attributing this improvement to advancements in drilling technology and techniques, including the use of high-performance motors with pre-contoured stators and real-time downhole drilling data analysis. This enabled a more profound understanding of critical drilling parameters such as weight on bit and torque. Pearson et al. [16] discussed how technological advances have led to a substantial boost in oil production, even amidst falling oil prices and economic challenges. They highlighted how strategic changes in completion performance and well design, such as transitioning from Generation 1 to Generation 4 and then Generation 5 designs, significantly improved drilling returns. Lolon [17] focused on the success of horizontal well completions in the Bakken formation, noting a threefold increase in production rates compared to vertical wells. The study emphasized the effectiveness of longer laterals, staged treatments, and cleaner fluids, along with the importance of fracture design and proppant selection. Johnson and Courrege [18] reported on the increased oil and gas production in the Bakken Shale using openhole packer and sleeve completions. This technology proved beneficial in maximizing exposure to hydrocarbon reserves and reducing environmental impacts.

Southcott and Harper [19] detailed WPX Energy’s significant reduction in geosteering errors through the advanced processing of 3D seismic data. This technological advancement led to more accurate mapping and lower well costs. Brandt et al. [20] presented an engineering-based evaluation of the energy intensity and net energy return (NER) in oil production. They highlighted the increasing energy cost associated with drilling, underlining a declining trend in NERs. Gutierrez et al. [21] introduced a geosteering methodology using azimuthally sensitive gamma-ray sensors for precise well positioning, leading to enhanced production rates.

Lim et al. [22] emphasized the integration of production engineering with drilling strategies, using Torque & Drag (T&D) modeling to ensure quality wellbores and facilitate successful liner installation. Veazey [23] reported on Whiting Petroleum Corp.’s strategic expansion in the Williston Basin to deepen its drilling inventory, while Halliburton [24] introduced the Cerebro intelligent bit technology, enhancing drilling efficiency through high-frequency motion mapping and vibration measurement. Jerrard et al. [25] stressed the importance of real-time observation in cement operations, suggesting adjustments in slurry design based on bottom hole temperature observations. Halliburton [26] discussed a customized water-based fluid system for drilling, focusing on wellbore stability and contamination prevention, while Halliburton [27] introduced NitroForce®, a high-torque motor technology that significantly reduced well time and increased drilling efficiency. Gundersen et al. [28] explored the implementation of Wet Shoetrack Completions, a cost-efficient technique for unconventional oil and gas extraction, and Schmidt et al. [29] examined well-to-well interactions, suggesting that sacrificial fracs in child wells could optimize production. Lastly, Ouadi et al. [30, 31] introduced the innovative fishbone drilling method, which increases hydrocarbon recovery and reduces environmental impact by featuring a central wellbore with multiple branching laterals.

3.4 Overcoming obstacles in North Dakota’s drilling industry

Reddy and Pitcher’s [32] case study highlighted the complexities encountered in geosteering a horizontal well in the Middle Bakken Dolomite. The operator initially faced difficulties with the wellbore straying from the desired zone, necessitating multiple open-hole sidetracks. To mitigate these challenges, they deployed a comprehensive geosteering service that included an azimuthal resistivity tool, dedicated software, and specialist personnel, leading to more accurate well positioning and reduced need for costly unplanned sidetracks. Bassarath and Maranuk’s [33] research focused on the challenges of drilling long lateral wells in the Bakken shale, introducing the Targeted Bit Speed (TBS) technology. This technology significantly improved 3D directional control, akin to a rotary steerable system, and was instrumental in reducing sliding time, increasing the Rate of Penetration (ROP), and drilling smoother wellbores, thus enhancing the overall efficiency of shale drilling operations.

McCormick et al.’s [34] study proposed the use of expandable liner hangers (ELHs) for the completion of long horizontal wells in shale plays, notably in the Bakken Shale. ELH systems, which combine the liner hanger, packer, and tieback receptacle into one unit, simplify operations and have demonstrated excellent performance in managing large loads, thereby increasing reliability, safety, and profitability in shale operations. The study by Parayno et al. [35] addressed the drilling challenges in the Bakken shale and suggested Managed Pressure Drilling (MPD) as an effective solution. MPD, which allows for instant changes to wellbore pressure without regular mud weight adjustments, led to improved drilling efficiency, better hole cleaning, and fewer issues related to downhole pressures. Chinander’s [36] study, along with Tobben’s [37] report, discussed the fluctuating state of oil production in the Bakken region, peaking in late 2019 and then declining in 2020 due to the COVID-19 crisis. These studies raised concerns about the future productivity of the Bakken Shale formation, especially given the increased gas production from mature wells, which is negatively affecting crude output. Additionally, the Western Dakota Energy Association [38] highlighted the potential advantages of extending underground drilling laterals from two miles to three miles, particularly in less productive areas of the Bakken. This approach could lead to fewer overall wells needed and a shift towards enhanced oil recovery in more extensively drilled areas. Lastly, Jean’s [39] article in the Williston Herald reported on North Dakota’s recent peak in rig counts, which is being stifled by severe labor shortages and infrastructure limitations. Despite rising oil prices, the state faces challenges in expanding its labor pool and accommodating the increasing gas-to-oil ratio with adequate infrastructure.

4. Hydraulic fracturing and completion design

This section attempts to summarize the most relevant research and publications related to hydraulic fracturing and completion design in the Bakken Petroleum System. It tracks the latest technologies and their applications. The section is organized into three subsections.

The first subsection is statistical studies. It summarizes the statistical analysis conducted by top operators to understand production efficiency related to different completion designs. It also reports some statistical findings related to refracturing, parent-child wells interference, drilling directions, and stress estimation using Diagnostic Fracture Injection Tests (DFITs).

The second subsection is field pilot studies. It summarizes testing that was conducted in the Bakken. Operators tried to understand the effect of completion design in the Bakken Petroleum System by testing different ideas in the field. In their attempt they used several diagnostic approaches and techniques to understand the completion design and technologies. They also report some of the challenges and how they overcame them.

The third and last subsection is numerical modeling studies. This subsection reports modeling efforts that were conducted to history match and pilot field studies. These efforts leverage physics-based modeling to understand the behavior of the reservoir and draw conclusions about the efficiency of hydraulic fracturing and completion design in the Bakken.

4.1 Statistical studies

Griffin et al. [40] conducted a comparative statistical analysis of the performance of different operators in the Bakken. They found out that operators with advanced completions (Table 1) performed better than their competitors in terms of averaged cumulative production and EUR. The performance is notably different in higher water cut wells. The authors state that the use of ceramic proppant results in better production; however, we believe that the uplift of production is related to the use of higher injection volumes because, comparing operators B and E, the injection and proppant volumes are the same; however, the proppant types were different with higher proppant cost for operator E and lower EUR and cumulative production. They also noted that cumulative water cut represented a very good proxy for reservoir quality (lower water cut indicates better reservoir quality). Their study was limited to a well basis, and not to a drilling and spacing unit (DSU). Note that in unconventional plays development, the fractures from these wells create a connected system that imposes an interdependency of well performance.

Operator	Wells #	Liner	Stages #	Comp. Type	Fluid	Prop. Lbs/ft	Vol. Bbls/ft	Sand %	Ceramic %	Cost MM$	year Cum	EUR MBOE
A&C	45	SP	35	PNP	SW	396	25.1	0	100	4.2	195	1050
B	144	SP	34	PNP	Hyb.	395	7.9	38	62	3	140	800
E	56	Cmt	25	PNP	Gel	353	7.8	0	100	3	124	700
L	157	SP	30	PNP	Gel	288	5.7	68	31	2.3	105	600
Y&T	68	SP	28	BS	Gel	300	6.1	83	15	2.3	107	600
R	76	SP	25	PNP	Gel	264	6.5	100	0	1.9	87	500

Table 1.

Averaged values for completion properties and performance for different operators for 30% water cut [40].

Lolon et al. [41] built a multivariate statistical machine-learning model. To evaluate the impact of different fracture designs on well performance. They used data from 6800 wells from the North Dakota Industrial Commission and Frac Focus. They used root mean square error in cross-validation to measure the performance of their models. They reported that water-cut is inversely proportional to cumulative oil production. They also noted that water-cut can be used as a proxy for reservoir quality. The TVD was a proxy for oil maturity. Higher treatment rates, volume and proppant mass had a positive effect on production (Figure 6). Note that these effects are not absolute as multiple parameters change from one geographic area to another one. The paper outline sensitivity analysis for these different parameters using the machine learning models to quantify the effect of changing one parameter at a time to support the previous conclusions.

Figure 6.

Correlation between reservoir and completion parameters to 180 days oil cumulative production normalized per lateral length modified from [41].

Figure 6 illustrates the Pearson Coefficient of Correlation for various factors impacting oil recovery in the Bakken Petroleum System, revealing both positive and negative relationships. Higher water cuts and increased stage spacing show negative correlations, suggesting that excessive water production can impede oil flow and wider fracturing stages might lead to a less efficient fracture network, respectively. This indicates that optimizing water management and stage spacing could be crucial for enhancing oil recovery. In contrast, a positive correlation with treatment volume and proppant mass emphasizes the benefits of sufficient hydraulic fracturing inputs in creating expansive, conductive fracture networks that improve hydrocarbon migration. Interestingly, the organic content as measured by TOC shows a contrasting effect; while higher TOC in the upper Bakken correlates negatively with oil recovery, suggesting complexities in geological variability or non-productive organic matter, the positive correlation with TOC in the lower Bakken implies a region richer in hydrocarbon-generating organic matter. The analysis underscores the nuanced interplay between geological, operational, and technical factors in optimizing oil recovery in unconventional shale plays.

Pearson et al. [42] emphasized the importance of near-wellbore conductivity. They supported their idea with the fact that most of the pressure drop occurs near the wellbore; thus higher proppant conductivity is required. They ran a series of laboratory tests and simulations in the BPS to support their idea. The tests showed a time-dependent proppant conductivity under the same stress condition. They conducted a numerical simulation to history match well production in the Bakken. The model was then used to sensitize on using different proppant combinations of ceramic and brown sand with ratios varying from 0 to 100% of the total injected proppant. The time of injecting ceramic proppant was also changed between lead and tail for different concentrations. The economics were also evaluated for these cases showing a significant incremental for using a 5% lead, 5% tail ceramic proppant. The latter design only increases the completion cost by 8% but results in an incremental oil recovery of 8%. Their study showed the potential production uplift of having higher conductivity in the near wellbore region.

Taghavinejad et al. [43] conducted a Rate Transient Analysis study on 73 wells of the Bakken and Three Forks Formations to quantify the effect of Fracture Driven Interaction (i.e., frac-hit). Their work divided the BPS wells into 3 generations, summarized in Table 2. Note that the pump rates contradict other papers, especially for Gen 2 and Gen 3 actual rates and sand types. In their work, they concluded that the Fracture Driven Interaction (FDI) is distance and production time-dependent. FDI resulted in 50% losses in EUR, up to 80% incremental water for first Gen wells. Increased proppant and fluid volume increased production only up to a certain value above which the increased volumes did not have a significant effect. These results can be area specific where other responses have been noted in the following studies.

Metric	Gen 1 (Parent)	Gen 2 (Infill)	Gen 3 (Infill)
Completion Date	2009–2012	2012–2015	2015–2016
Formation	Three Forks (TF)	70% TF, 30% Bakken	45% TF, 55% Bakken
Completion Type	OH SS	OH SS, PNP	PNP
Number of Stages	10–26	17–30	26–30
Cluster per stage	1	1–5	5
Stage Spacing (ft)	390–970	319–593	310–460
Sand Concentration (lbs/ft)	85–225	160–625	195–280
Sand Type	20/40 and 20/70	20/40 and 20/70	30/30 and 40/70
Fluid Volume (bbl/ft)	3–5	4–9	4–8
Fluid Type	Slickwater/ XL tail-in	Slickwater/ XL tail-in	Slickwater/ XL tail-in
Pump Rate (BMP)	46	46	46

Table 2.

Evolution of hydraulic fracture job design in the BPS [43].

Cozby and Sharma [44] conducted a statistical study on the impact of FDI on the production of parent and child wells. Their study covered several plays across the United States, including the Bakken play. Previous to their study, there were several papers written on the same topic, including [6, 45]. The results from these studies showed that on average, BPS’ offset wells overperform BPS’ parent wells. They also found that wells in the BPS that had been drilled and stimulated before 2017 resulted in incremental production due to FDI, whereas wells that were drilled after 2017 resulted in underperformance in the parent wells. This was attributed to the evolution of completion size. Wells drilled before 2017 had a smaller completion size resulting in an under-stimulated volume of the reservoir accessed once the child well was drilled and fractured. They reported that the two main factors that affect the parent well response are well spacing and the Cumulative production volume before the stimulation of the child well (Figure 7). The reader is referred to Gupta et al. [46] for a complete review on FDI.

Figure 7.

Bakken wells spacing (primary and offset) and cumulative produced volume relationship to the first 12 month production percent change [44].

The heatmap presented in the Figure 7 provides a visual representation of the relationship between the volume of Barrels of Oil Equivalent (BOE) produced by parent wells and the spacing between drilling sites, along with the corresponding change in production over a 12-month period. The left panel of the heatmap indicates that as the spacing between wells increases, there is a trend towards greater initial production volumes, denoted by the transition from cooler to warmer colors. This suggests that wider well spacing may lead to less competition for resources between wells and potentially enhance individual well performance. The right panel, which shows the data count, reflects the distribution of data points across different spacings and production changes, indicating the most commonly encountered scenarios. Together, these heatmaps provide a dual perspective, linking well spacing to production efficacy and highlighting areas where data are most abundant, thereby offering strategic insights for drilling and development planning in the context of Improved Oil Recovery (IOR) efforts in shale formations.

Other than FDI, refracturing is used to access the under-stimulated reservoir volume. Lantz et al. [47] reported the results from several refracturing treatments in the Bakken for 14 wells (cemented horizontal liner wells). During the refracturing operations, they noticed that the stress gradient dropped from 0.73 psi/ft. to 0.66 psi/ft. due to depletion (more details on stress changes can be found in [48]). The refracturing treatments were pumped at 50 bbl/min with 4 lb./gal sand concentration. They reported that the EUR estimate increased by more than 30%. The reader is referred to for more details about Refracturing candidate selection and treatment design [49].

Dalkhaa et al. [50] reviewed 272 refracs that were conducted in the BPS. They used production rate changes, peak oil, change in GOR, and incremental EUR as metrics for the evaluation process. They found that open hole completions in the Bakken are under-stimulated compared to cased hole wells. They reported an incremental EUR of 340 MSTB in open hole refrac’ed. wells, and 175 MSTB in cased hole wells. They also noted that the decrease in GOR is an indicator of accessing new stimulated reservoir volume. The decrease was quantified for 90-days averaged daily production where open hole wells GOR decreased by 10% and cased hole wells did not decrease. They also conducted an economic analysis of refracs potential revenues and identified 400 potential wells as candidates for refracs. Following the study, Schmidt et al. [29] reported that 184 wells have been refrac’ed. between 2017 and 2021 with a decreasing trend from the highest levels in 2017.

Rostami et al. [51] conducted a detailed statistical analysis on the drilling direction of 7000 wells on well cumulative production and normalized production per lateral. The direction was then compared to the orientation of the least principal stress. They reported that wells drilled in the direction of the least principal stress were more economical compared to wells drilled in other directions. However, at most north or south leases, operators may find themselves with limited acreage, which results in drilling in the direction of maximum horizontal stress. These wells were more expensive to drill with higher cumulative production.

McClure et al. [52] reported results from a statistical 62 DFITs study to compare DFITs interpretation approaches. They report that a high percentage of Bakken DFITs does not show a clear closure which makes it challenging to estimate stress using the compliance method. This is related to the higher permeability of the play compared to other unconventional plays. The permeability ranges were sometimes outside the applicability of the compliance DFIT interpretation method. Dohmen et al. [53] reported results from 100 DFITs interpretations. The stress gradient was reported as 0.72–0.73 psi/ft. in the MB and 0.71–0.79 psi/ft. for TF. The pore pressure gradient was reported as 0.61–0.70 psi/ft. for MB and 0.72–0.73 for TF.

4.2 Pilot studies

Weddle et al. [54] evaluated the use of particulate diverters and increased number of perforation clusters per stage. In their attempt, they investigated the efficiency of their approach using radioactive tracers (RA). They investigated results from 27 wells that adapted the paper. The completion had 15 clusters per stage with 2 shots each at 0–180-degree phasing. The pumping rate was 80 bpm. The production results were compared based on 180 days of cumulative production. Their results reported that well production increased by 240% for high cluster efficiency (i.e., uniformity) compared to nearby wells. They also noted that the use of particulate diverters improved cluster efficiency. They note that the practitioner needs to account for settling velocity to determine the maximum perforation number. Their approach is limited to near wellbore uniformity (i.e. limited to measurable radioactive tracer distance).

Limited entry was first introduced by Murphy and Juch [55] and Lagrone and Rasmussen [56]. The method was mainly used to ensure equal flow rates between different perforations at different breakdown pressures. The approach is then adapted in unconventional plays to overcome the following issues [57]:

• Stress variability along the lateral length (90% of laterals are in the 750 range)

• Near-wellbore friction variability from cluster to another (P50 of 625 psi from step down tests)

• Stress shadow between cluster and from previous stages (formation elastic properties dependent)

• Fracture propagation variability in different clusters and elastic properties anisotropy

• Perforation friction changes due to erosion and perforation diameter range (approximately 500 psi)

Note that the values in brackets are values measured for the Bakken [57]. The pressure drop through perforation can be calculated as follows [58]:

$$\Delta P_p=\frac{0.2369\times Q^2\times \rho }{N_p^2\times D_p^2\times C_d^2}$$E1

Where $\Delta P_p$ is the perforation pressure (pressure drop across perforation (psi)). $Q$ is the flow rate in bpm. $\rho$ is the fracturing fluid density (lb/gal). $N_p$ is the number of open perforations. $D_p$ is the perforation diameter (in). $C_d$ is the coefficient of discharge.

Weddle et al. [57] incorporated the use of the Extreme Limited Entry (XLE) in their designs to drive the number of clusters from 11 clusters per stage to 15 cluster per stage for the rate of 80 bpm. The pressure drop at perforations was designed at 2000 psi. They measured their success using data from fiber optic, radioactive tracers and production data. For a 9500 ft. lateral, they have drastically reduced the total number of stages from 50 to 27. The pilot project resulted in an average incremental production of 10%. The authors stress the importance of accounting for plug shifting that can cause leakage, thus, lower performance. The proppant transport and settling velocity estimations limit the maximum number of perforation clusters. The perforation orientation was mentioned, as some orientations provided better production compared to others. For more details on the perforation design, the reader can consult the following articles [59, 60, 61, 62].

Lorwongngam et al. [63] investigated the design of XLE in the Williston Basin through a pilot field trial. They used pump pressure and step rate tests to ensure they attained high pressure. Radioactive tracer and fiber Distributed Acoustic Sensing/Distributed Temperature Sensing (DAS/DTS) and downhole camera were used to measure the uniformity of fluid distribution. They found out that the efficiency when using XLE design was independent of geological targets except for some cases. Using XLE design, the operator increased the cluster number to 15 per stage. The design was achieved using a limited entry pressure drop of 1500 psi. The design implements 2–3 perforations per cluster. In their field trial, they faced plug leakage issues that were identified using RA tracer data. Leakage problems were detrimental to the success of fracturing fluid uniform distribution.

Lorwongngam et al. [64] Improved the previous design by conducting field trials (XLE 2.0). They investigated the optimum shots per cluster, the minimum rate per cluster, and the maximum possible cluster number per stage. Their study was conducted for three years based on the Uniformity Index. A suite of fracture diagnostics was used to evaluate the eXtreme Limited Entry (XLE) design, including radioactive tracer, fiber optic, pressure step-down test, downhole camera, perforation acoustic imaging tools, lateral bottom hole gauges, in-well and offset fiber optic, production data. The results showed that using one shot per cluster can increase the cluster number to 20 per stage with 5 BPM for 4–1/2″ liner, and 6 BPM for a 5–1/2 liner. The pressure was 1900 psi. This reduces the total number of stages, thus, the rig time, leading to lower completion cost by 12%. Table 3 reports the evolution of hydraulic fracturing design evolution.

Design Type	Cluster per Stage (CPS)	Shot per Foot (SPF)	Estimated Cluster Spacing (ft)	Total Clusters/well	Total Stages per Well
Baseline Original Design	3–6	3–6	35	280	36–42
XLE 1.0 Design of Experiment	6–15	2–3	33	300	30–32
XLE 2.0 Design of Experiment	8–20	1	30 >	300 <	25–27

Table 3.

Hydraulic fracturing design evolution [64].

Lorwongngam et al. [64] also noted that plug leakage and inefficient isolation can lead to lower cluster efficiency, thus, lower production. The authors also tried diverters in their trial; however, the XLE design was proven to give better production. They also suggest avoiding two perforation wireline runs.

Dohmen et al. [65] introduced the concept of Microseismic Depletion Delineation (MDD) for the first time. This technique was first introduced in the Bakken to delineate the depleted area from a producing parent well. The concept of the idea relies on the stress changes due to depletion. As the stress changes, the effective stress Mohr circle gets closer to the Coulomb failure envelope. Thus, fracture orientation in the critically stressed orientation will slip due to a change in pore pressure. Thus, the authors injected fluid in a depleted well and monitored the microseismic events resulting from the shear failure of critically stressed fractures to map the drained area. This concept was then used to plan infill drilling while accounting for depleted zones. The reader is referred to for more details on the approach [53, 66].

Cipolla et al. [67] reported the results of a field pilot of 6 laterals in which one is an observation lateral that was not stimulated drilled in the TF formation. The wells were equipped with SWPM and fiber optic to record FDI events and map far-field drainage. One MB well was equipped with fiber optic to measure the stage level uniformity and FDI occurrence from the stimulation of other wells. Another MB well was equipped with microseismic geophones to estimate the fracture geometry. Various sets of measurement and diagnostics techniques were used. The aim of these techniques is reported in Table 4.

	Comp. Efficiency	Stim. Efficiency	Diversion	Frac- Geometry	Frac Morphology	Prod. Logging	Drainage Mapping	Well/FM Connectivity
Fiber Optic
Microseismic
Pressure Gauges
Oil/Water Tracers
Proppant Tracers
Camera
Fingerprinting
	Completion Design and Frac Characterization					Well Spacing and Frac Design
	Development Optimization

Table 4.

Diagnostics technics to record FDI events [67]. Red for high confidence measurement, and grey for lower confidence measurer.

Using their diagnostics, they reported fracture lengths ranging from 1900 ft. to 3800 ft. pumping the same volumes. They attributed this change to the effect of stress shadowing. They also analyzed FDI vs. fracture length data and Fracture length vs. injected volume to build empirical correlations and calibrate the fracture toughness. The uniformity index was used to calibrate the fluid distribution between clusters of the same stage. The pressure in the TF offset well was recorded during completion and production at several intervals of the lateral. The lateral pressure was used to assess the depletion of different fracture stage designs that intersect the observation wells at different pressure measurement locations. Some designs had better drainage than others. However, all designs of wells drilled in the MB drained from the TF formation at 450 ft. away from the primary well. Three fracturing designs were mainly evaluated to improve the production within one well. The third design had better connectivity to the reservoir for the longest period (note that the connectivity declined with time). The uniformity index from fiber data was highest at 14 clusters/stage and diminished with increasing cluster number. The uniformity index did not improve even with pod diverters (it got worse). This work resulted in a new concept called Augmented Drainage Development (ADD) [68]. ADD is an unstimulated well drilled in the middle of stimulated wells. This well drains from the induced fractures from offset wells to increase production. On the same project, Miranda et al. [69] reported an analysis of interference tests between wells in the MB and the TF formations using Chow Pressure Group Technique (CPG) [70]. They reported that the fracture system created in the DSU behaves as an interconnected system. The communication varies depending on the distance between wells and fracture design, this interference degrades with time. A MB well can drain up to 450 ft. from the lateral towards TF formation. These type of tests can be used to determine optimum well spacing and drainage performance.

4.3 Numerical modeling studies

Cipolla et al. [71] built a well-constrained model that was calibrated using MDD, microseismic, production logging tool, bottom hole gauge, RA tracers, and Image Logs geomechanics modeling and production data history matching to understand the impact of parent well depletion on the offset well fracture geometry. Their numerical study showed an effective fracture half-length of 900 ft., and a primary drainage of 400 ft. to both sides of the well. Wells drilled in the Bakken produced from TF and vice versa. A severe asymmetry in the offset well’s fractures resulted from the depleted area near the primary well. In their approach, they used a fracture conductivity of 60md-ft for the propped fracture part and 4md-ft as a conductivity for the unpropped fracture part.

Lorwongngam et al. [72] modeled a pilot pad in the Bakken. To constrain their model, they used DFIT for stress and pore pressure estimation, microseismic for hydraulic fracturing geometry, MDD for drained area mapping, and RA tracer to quantify the uniformity index and the fracture propped half length of the fracture treatment. Oil and water tracers were used to quantify the connectivity between wells and bottom hole gauges and production data to history match the reservoir performance. They found that the effective fracture half-length is approximately 700–900 ft. for MB wells and 450 ft. for TF wells. The unpropped region contributes to the flow of hydrocarbons from the reservoir. Fracture conductivity ranges from 30md-ft in low stress to 9 md-ft for high stress propped fractures and 0.2 md-ft for unpropped fracture conductivity for high stress. The conductivity of the fracture in LBS is low, which limits the conductivity between wells in the TF and in the Bakken. This study’s application was to increase well spacing and reduce the total number of wells per DSU.

Cipolla et al. [73] and Fowler et al. [74] conducted a numerical study to calibrate stress changes due to depletion (i.e., Biot coefficient). They used MDD to map the drained area, Microseismic to monitor fracture geometry, tiltmeter to measure fracture height, and time-lapse DFIT to measure stress and pore pressure changes due to depletion. They drilled two vertical wells next to a parent horizontal well that was depleted for 10 years. The DFITs were performed in the vertical wells to reduce the complexity of the problem. They reported that the connectivity distance between wells exceeds 1000 ft. The fracture from the vertical well grew asymmetrically towards the parent well. Used a Biot coefficient of 0.34 to match the MDD and FDI response. They compared two modeling approaches that resulted in similar results. First is the use of DFN modeling, and the second one using pre-existing planer fractures. The reader is referred to McClure et al. [75] for more details on planer fracture modeling.

Merzoug and Rasouli [76] built another model for the same previous case study using a hydraulic fracturing modeling approach to estimate fracture geometry. In their work, they found a different Biot coefficient of 0.7. They reported that the use of pre-existing fractures modeling approach changes the depletion zone and pressure distribution significantly. Their values align with results from laboratory work reported by Ling et al. [77] and Ma and Zoback [78]. Their study used proxy modeling to optimize the fracture geometry of offset wells. They reported that offset wells with wider spacing (1320 ft), tighter cluster spacing (26 ft), higher cluster numbers (20), and higher injection volumes (1900 bbl/cluster) resulted in better Net Present Values (NPV).

McClure et al. [79] reported results from a collaborative modeling study on the FDI effect in all major shale plays in North America. They analyzed three Bakken datasets as part of their study. The results showed that wells in the MB drain from the TF formation and vice versa. The FDI had a neutral effect on Bakken. The Bakken petroleum system is composed of a thin prolific zone. Thus, numerical modeling studies show that lateral fracture growth was important for better production in the Bakken, whereas vertical growth was not beneficial. The fracture spacing and well spacing were the dominating parameters in the production from the Bakken, with more production at tighter spacing. However, the decision depends on the interest of the operator in terms of optimizing net present value or discounted return on investment.

5. Artificial lift activity in the Bakken

According to Lane and Chokshi [80], the decline rates in shale plays, when production is not choked, can reach a staggering 60 to 80% within the first year. In the Bakken formation specifically, the average annual decline rate is as high as 69%. This significant drop is closely tied to the hydraulic fracturing process, which is essential for stimulating the low permeability formations characteristic of shale plays.

After the initial flowback of hydraulic fracturing fluid, a phase of high production rates typically follows as the fracture and surrounding rock start to drain. However, due to the low permeability, the migration of hydrocarbons to fractures is limited, leading to a swift drop in production rates [81, 82].

This rapid decline necessitates the implementation of artificial lift systems early in the well’s lifespan, often just a few months after production begins. In some shale plays, artificial lift systems are even installed during the well commissioning phase, despite the expectation of natural flow during the initial months [81, 82].

Lolon et al. [41] analyzed the data for several wells completed in the Bakken/Three Forks formations and observed that artificial lift is required after only a few months of production due to the high decline rate. Ganpule et al. [83] arrived at a similar conclusion when they conducted a study examining the impact of completion practices on recoverable reserves in the Bakken/Three Forks formations. They found that the decision to switch to artificial lift systems was largely dependent on individual operator strategies, which were influenced by factors such as production targets and operational expenses. The study noted that some operators opted to transition to artificial lift systems relatively early, within 2 to 3 months of initial production, while others delayed this switch until 4 to 6 months into production. In some exceptional cases, high-performing wells did not require artificial lift systems until 18 to 24 months after the first production. This variability underscores the importance of operator-specific strategies in optimizing well performance and recoverable reserves.

5.1 Types of artificial lift systems

Based on a study by Eisner et al. [84], there are around 2 million operational oil wells globally, with over half - more than 1 million - utilizing some form of artificial lift. Sucker-rod pumps are the most common type of artificial lift, being used in over 750,000 wells. In the United States alone, approximately 350,000 wells employ sucker-rod pumps for lift.

A significant 80% of all U.S. oil wells are classified as stripper wells, producing less than 10 barrels per day, often with some water cut. Most of these stripper wells rely on sucker-rod pumps for lift. Looking at the non-stripper wells, which produce higher volumes, 27% use rod pumping, while 52% use gas lift. The remaining wells use other methods of lifting, including Electric Submersible Pumps (ESPs), hydraulic pumps, and other techniques.

In a more recent study, Pankaj et al. [85] provided a general breakdown of the artificial lift systems currently in use in U.S. onshore wells. They found that approximately 40% of wells starting with an artificial lift system use gas lift, while 36% employ ESPs. Rod lift is used in 13% of cases, plunger lift in 7%, and jet pumps in 4%.

In the Bakken formation, Patron et al. [86] stated that only ESPs and jet pumps (JP) can be implemented during the high-flow period. As production declines, a common practice is to move to sucker rod pumps (SRP), as a lower flow rate artificial lift method that allows a smooth transition with their ability to pump the required volumes from deep installations.

Britvar and Williams [87] reported that as of November 2016, Oasis Petroleum was operating on a substantial land area of 540,000 acres, with a primary focus on the Bakken and Three Forks formations. Most of the wells in operation were utilizing rod pumps for production. In addition, 30 wells were employing gas lift methods, while 35 wells were using ESPs for production. Britvar and Williams [87] also provided insights into the typical life cycle of artificial lift systems in the Williston Basin. Historically, operators have primarily used pumping units in this region. A typical well would employ a 912-pumping unit, which is moved to the location during the initial flowback, with rods being run after the well is loaded up. However, as completion designs evolved to move more fluid, some operators transitioned to long-stroke pumping units capable of significantly increasing fluid movement. The evolution of completion design in the basin necessitated the exploration of high-capacity lift systems to match improved well productivity. Artificial lift mechanisms such as ESPs and gas lift were identified as capable of moving considerably larger volumes at operating depths exceeding 10,000’ TVD. The successful application of ESPs in Bakken and Three Forks wells has notably enhanced early life production rates, positively impacting total well economic returns.

In their study, Patron et al. [86] implemented an innovative artificial lift selection workflow to examine an unconventional well situated in the Three Forks formation. The process began with a pre-screening phase, where they evaluated the seven primary types of artificial lift systems based on both technical and economic criteria. Their analysis concluded that only Electric Submersible Pumps (ESP), Jet Pumps (JP), and Sucker Rod Pumps (SRP) were viable options for use throughout the well’s lifespan. Interestingly, Gas Lift, a commonly used artificial lift system, was not included in their selection. This exclusion was primarily due to the substantial capital investment required to upgrade existing installations to support a gas lift injection infrastructure. The outcomes of their analysis are comprehensively presented in Table 5.

Artificial Lift Transitional	Artificial Lift Stage Late	Production Stage
Flowrate	700 B/D	200 B/D
Electrical Submersible Pump (ESP)	Applicable	Not Applicable
Jet Pump (JP)	Applicable	Not Applicable
Sucker Rod Pump (SRP)	Not Applicable	Applicable
Gas Lift (GL)	Not Applicable	Not Applicable
Progressive Cavity Pump (PCP)	Not Applicable	Not Applicable
Electrical Submersible Progressive Cavity Pump (ESPCP)	Not Applicable	Not Applicable
Plunger Lift (PL)	Not Applicable	Not Applicable

Table 5.

Results of artificial lift selection strategy for a well in the three forks formation [86].

5.2 Field studies

In their study, Clark et al. [88] utilized production data from the North Dakota Industrial Commission to construct a database tracking production from all wells drilled and completed in the Bakken and Three Forks formations. They developed type curves based on rod pump artificial lift for two groups of wells, depending on the operating area.

Interestingly, they observed that wells with jet pumps tended to deviate positively from the type curve, indicating increased production. In fact, jet-pumped wells with the same completion strategy produced 11% more oil in the first year than rod-pumped wells.

The study further analyzed individual wells using hyperbolic decline to a terminal decline rate of 6%. The results showed that for the 10 wells with at least 12 months on a jet pump, the average increased production above the type curve for the first 12 months was 15.4 MBO. This resulted in an average incremental income of $1045 M per well, assuming 100% Working Interest, 78% Net Revenue Interest, $100/Barrel of Oil, and a differential of $13/Barrel of Oil.

Clark et al. [88] concluded that jet pumps can be an effective method for lifting Bakken and Three Forks wells in the Williston Basin, yielding superior production and economic results compared to rod-pumped wells. This is particularly true in higher water cut environments. They also noted that production results can be further optimized through automation controlling pump discharge to handle well slugging while optimizing drawdown.

Nickell and Treiberg [89] presented two case studies of Bakken wells that employ rod pumps, demonstrating the significant impact of optimizing a well’s downtime on its operation. They utilized an autonomous control algorithm to modulate the idle time of the wells, aiming to reduce failures or increase production.

In the first case study, a well experiencing fluid pound in the Bakken started with a downtime of 30 minutes, which the autonomous optimization algorithm increased to 100 minutes. This adjustment reduced the number of cycles per day from 30 to 8, without affecting production. This reduction in incomplete fillage pump strokes is expected to decrease failures on the downhole equipment, thereby increasing the well’s runtime. The second case study also involved a Bakken well experiencing pump-off conditions with fluid pound. The well began with a downtime of 50 minutes, which the autonomous downtime optimization increased to 300 minutes. This increase led to a decrease in cycles per day from 25 to 5, again without reducing production or runtime. This adjustment is expected to reduce the number of fluid pound strokes by 36,500 per year, improving the overall health of the well and reducing failures.

Bestgen et al. [90] discussed the challenges faced by oil-producing wells in the Williston Basin (Bakken and Three Forks formations) that utilize Electrical Submersible Pump (ESP) lift. These wells encounter severe scale and corrosion issues due to high TDS brines and high shear and high-temperature operating conditions. The authors then presented the results of a field trial that compared the new combination product with a conventional scale/corrosion inhibitor. The new product provided a 73% increase in run time, and when used in conjunction with the preconditioning corrosion inhibitor treatment, the run time showed a 100% increase. The paper concludes that any extension in ESP run time gained through mechanical or chemical means offers improved well production, lower operating expenses, and improved profitability to the producer.

Orji et al. [91] conducted a study on the optimization of unconventional wells operated by Hess in the Bakken that are using sucker rod lift. Their research revealed that the optimization process necessitates a delicate balance between reservoir inflow, gas separation (pump efficiency), and structural integrity parameters. They utilized a dimensionless pressure and dimensionless time (pDtD) approach, which proved effective in capturing complex physical phenomena associated with unconventional wells, such as transient, multi-stage fractures, and superposition effects. This approach was found to be simpler and faster to calibrate compared to traditional numerical simulator history matching exercises, making it particularly useful for scenarios involving a high number of wells.

The authors clarified that their approach is intended to complement, not replace, classical reservoir engineering work by addressing short-term operational challenges. They also highlighted the time-dependent nature of unconventional wells, indicating that the modeled production potential is a moving target, and thus, optimization recommendations should be implemented as swiftly and safely as possible. The study also underscored the importance of ensuring the structural integrity and gearbox loading of the pumping unit, the adequacy of communication and control systems on the pumping unit, and the accuracy of the metering and allocation method on a given site for successful optimization.

Rankin et al. [92] performed a case study analysis to evaluate improvements and profitability achieved with superior completions in the Bakken/Three Forks horizontal wells. A side observation from this study is that wells completed with multiple Plug and Perf stages flowed naturally for longer periods compared to most wells completed with fewer frac sleeves (3 out of 4 wells), which required the installation of pumping units earlier in their life cycle.

Wells F, G, and H, which utilized Plug & Perf methods with ceramic proppant, show higher initial production rates but also experience a steep decline. This rapid decline necessitates the earlier installation of pumping units as indicated by the proximity of the stars to the y-axis. In contrast, Wells I, J, K, and L employed Frac Sleeves with white sand proppant, demonstrating lower initial production rates but a more gradual decline. Notably, the cumulative production for these wells remains competitive with that of the higher-performing wells over time, illustrating the trade-off between initial production vigor and long-term decline rates.

According to a study by Yuan et al. [93], the impact of well trajectory on the production performance of ESP-lifted shale wells was investigated. The study utilized a representative set of field data to model the shale well with an ESP installed. The wells under study had similar reservoir characteristics but varied in their well trajectories, including smooth and highly sinuous toe-up and toe-down trajectories. A transient dynamic multiphase simulator was used to integrate ESP performance and wellbore models over the life of a shale well. The study found that well trajectory does not significantly impact the production performance of an ESP-lifted well, even under conditions of slugging flow. The research also highlighted the importance of using a transient wellbore simulator to capture the continuous, long-period transient process inherent in the production of unconventional shale wells. The study concluded the workflow described could be used to evaluate selected ESP performance and timing for installation and withdrawal for a particular well over its whole life cycle.

6. Application of machine learning in artificial lift

Pennel and Hsiung [94] presented a workflow to diagnose artificial lift problems using machine learning. They trained their models using production data for over 1 year from operating wells in the Bakken. Their objective was to predict the operational problems of sucker rod pumps and gas lift systems including tubing and pump failure along with the suboptimal performance of sucker rod pumps and gas injection. The data consisted of time series sensor and controller data for 800 wells, reviewed by experts in each lift type to identify periods of failure and sub-optimal performance.

Freeman et al. [95] embarked on a study to optimize unconventional wells in the Bakken using an Internet of Things (IoT) device with high-performance computational capabilities. The researchers developed an IoT device capable of real-time analysis and higher-order mathematics, which was connected to a cloud-based analytics software platform. This technology was deployed on 50 representative wells in the Bakken, with the device connected to the legacy rod pump controller via Modbus connection. The results from the IoT device showed immediate differences in key downhole parameters when compared to the traditional rod pump controller.

The study found that the higher-accuracy physics-based inputs fed into machine learning algorithms, which dynamically classified wells into key operating states of under-pumping, over-pumping, and dialed-in. The results showed that for wells that were under-pumping, Equinor was able to increase oil production by up to 33%. For wells that were over-pumping, Equinor was able to decrease the number of strokes by 11% and increase pump efficiency by 14%. The study concluded that the vision of autonomous well operations is possible to implement, and the investment in modern optimization technology provides lasting, repeatable value through many operational parameters.

Table 6, as derived from the study by Freeman et al. [95], presents a concise summary of the production efficiency gains achieved through the implementation of an optimization strategy on unconventional wells in the Bakken formation. The metrics depicted in the table provide a quantitative comparison of well performance before and after the application of the optimization techniques.

Metric	Average
Total production before (bpd)	1842
Total production after (bpd)	2447
Production uplift	605
Production increase on under-pumping wells (%)	32.8%

Table 6.

Summary statistics showing efficiency gains without impacting production (after Freeman et al. [95]).

Specifically, the table shows that the average total oil production before optimization was 1842 barrels per day (bpd). After optimization, the production increased to an average of 2447 bpd, indicating a significant uplift of 605 bpd. This demonstrates a clear enhancement in well productivity attributable to the optimization process.

Moreover, a notable metric of improvement is the average production increase of 32.8% on wells that were initially under-pumping. This statistic underscores the potential for production gains by addressing inefficiencies in the pumping process and optimizing operational parameters. The table effectively highlights the tangible benefits of applying targeted optimization techniques in terms of increased oil production and improved well performance, reinforcing the value of such interventions in the context of IOR strategies.

7. Discussion

The drilling operations within the Williston Basin have yielded valuable insights that can guide the industry towards more effective and efficient extraction methods. The implementation of advanced drilling technologies, such as high-performance motors and real-time downhole drilling data, has significantly enhanced drilling efficiency and oil production. Moreover, the accuracy of geological placement through precise geosteering techniques has proven instrumental in minimizing errors and maximizing production.

Horizontal well completions, with their higher production rates compared to traditional vertical completions, have clearly established themselves as the way forward. When paired with effective drilling practices such as using cemented liner wells with diversion, longer laterals, and cleaner fluids, substantial production improvements have been achieved.

Attention to the vibration issue is critical, and further research into vibration measurement and mitigation is a pressing need for enhancing operational productivity. Likewise, minimizing sliding during operations through motor-driven rotary steerable systems enhances hole quality and offers better directional control.

The role of infrastructure, especially in terms of accommodating produced gas and oil, emerged as a central factor affecting drilling operations. Similarly, the availability of skilled labor directly impacts the productivity of drilling sites. As high-yielding core areas near exhaustion, identifying potential new drilling locations has become a crucial step in maintaining production levels.

Adapting to innovative practices like the use of wet shoetrack completions can usher in cost savings and operational optimization. Furthermore, the adoption of advanced well completion techniques, such as expandable liner hangers and MPD, are critical in overcoming the challenges associated with long lateral drilling and completion of horizontal wells.

Given that drilling operations are energy-intensive, strategies aimed at improving energy efficiency are necessary to mitigate the environmental impact. Lastly, the ability to cope with industry fluctuations due to economic or other crises is integral to maintaining operational continuity and productivity.

The evolution of hydraulic fracturing design was the result of a combination of science-oriented trial-and-error projects. Operators started with small hydraulic fracturing designs using cross-linked gel. The trend then changed to increasing fluid volumes and mass. The optimization went from a well-basis improvement into a DSU optimization where the recoveries are quantified on a well basis. The increased number of infill wells caused concern for operators because of overlapping drainage areas. However, In Bakken, a trend was noted. Old wells treated with lower volumes showed an uplift in production, whereas the newer wells were not influenced by FDI. Some operators attempted to reduce the number of stages. This resulted in a decrease in costs while keeping production at higher levels. This reduction in the number of stages is associated with an increase in the number of perforation clusters. This was made possible using high limited entry pressures and High Viscosity Friction Reducer fluids with a minimum rate of 5–6 bpm per cluster. Other operators keep the high number of stages with tighter well spacing. The two approaches are reported in Appendix A for two different operators. The field testing and numerical modeling approaches showed that the wells drilled in the MB drain from the TF formations and vice versa. The wells drilled in the direction of the least principal stress were more optimum in recovery because they create the largest surface area. Refrac operations were executed in several wells in the Bakken. However, the success rate was limited. The fracture geometry was estimated to propagate upward towards the Lodgepole and downward towards the lower TF. The fractures drain 450 ft. (measured) and a total drainage of 700–900 ft. (modeled) away from the lateral. The estimated stress gradient is around 0.7 psi/ft. with a Biot coefficient of about 0.34–0.7.

The rapid decline in oil production in shale plays like the Bakken formation requires the implementation of artificial lift systems early in the well’s lifespan. Often, artificial lift systems are installed just months after production begins, despite the expectation of natural flow. The timing of this switch can vary greatly depending on individual operator strategies, influenced by production targets and operational expenses. Regarding the types of artificial lift systems used, global studies have shown that sucker-rod pumps are the most common type, followed by gas lift in high volume wells. Electric Submersible Pumps (ESPs) and jet pumps are typically implemented during the high-flow period in the Bakken formation. As production declines, a common practice is to transition to sucker rod pumps.

Artificial lift systems and their implementation strategies have significantly evolved over time to keep up with changing well productivity. For instance, high-capacity lift systems such as ESPs and gas lift have been identified as capable of moving considerably larger volumes, and have significantly enhanced early life production rates, thus positively impacting total well economic returns. The optimization of unconventional wells necessitates a delicate balance between reservoir inflow, gas separation, and structural integrity parameters. Techniques like dimensionless pressure and dimensionless time approach can capture complex physical phenomena associated with unconventional wells and offer simple, swift solutions to address short-term operational challenges.

Field studies on the Bakken and Three Forks formations have shown interesting results. Jet-pumped wells tend to deviate positively from the type curve, indicating increased ultimate production. Autonomously controlled rod-pump wells can significantly reduce failures or increase production by optimizing downtime. Furthermore, ESP-lifted wells can greatly increase their runtime by combating severe scale and corrosion issues through innovative chemical treatments.

Machine learning has been increasingly applied to diagnose artificial lift problems and optimize unconventional wells. Machine learning models trained on production data can predict operational issues like tubing and pump failures, improving the overall health of the well and reducing failures. In parallel, the advent of Internet of Things (IoT) devices in the oil and gas industry allows real-time analysis and higher-order mathematics, which can classify wells into key operating states of under-pumping, over-pumping, and dialed-in, and optimize their operational parameters accordingly.

8. Conclusion

This chapter summarizes the best practices performed by operators in the Bakken Petroleum System. It reports the recent findings and operations for completing wells in the unconventional play. It can serve as a roadmap for people starting to work in the Bakken or a lesson learned paper for people around the world to continue from current best practices in the Bakken at the time the paper was published. The Drilling operations had accelerated dramatically bringing the cost to very low levels. The hydraulic fracturing design was changes several time to improve well productivity and different concepts about fracture geometry have been changed through time. The artificial lift application approaches have been tested in the Bakken and had different performance under different conditions. The optimization process changed from a well basis into a DSU basis. The Bakken evolution was made possible thanks to field trial, since projects supported with numerical simulation and statistical studies. Learning from previous practices is what drove the completion evolution and it will continue in the future.

Appendix

	Well #	Frac-Vol. (bbls)	Sand Lb (lb)	#Stages	Max Inj Rate (bpm)	Max Treatment Pressure	lat length	Well Spacing (ft)	Cum Oil	Cum Water	Cum Gas	IP Oil	IP Water	IP G as	Date	Bench
Operator M	37,778	152,832	8,492,476	40	108.8	9551	9968	700	326,894	192,890	564,961	1833	1006	2346	7/7/2021	MB
	37,777	159,275	8,507,435	40	110.8	9358	9890	450	367,194	311,225	690,061	2188	1757	2591	7/7/2021	TF Bench 1
	37,774	153,367	8,499,679	40	109.7	9883	9663	450	333,552	165,385	587,024	2391	1048	2944	6/15/2021	MB
	37,773	157,379	8,502,985	40	111	9548	9635	450	384,685	288,834	641,572	2488	1910	3034	6/15/2021	TF Bench 1
	37,772	155,647	8,489,220	40	111.5	9868	9940	450	270,302	310,827	601,335	1145	1486	1421	6/15/2021	TF Bench 1
	37,787	156,965	8,477,568	40	106	9595	9544	450	338,912	329,079	597,432	1351	1657	1504	8/6/2021	TF Bench 1
	37,786	147,605	8,332,270	40	69.3	9554	9625	450	326,935	188,470	556,636	1498	880	1811	9/2/2021	MB
	37,740	200,161	12,479,510	46	92.4	9474	11,290	500	406,094	332,738	318,517	1615	1488	1089	8/25/2021	MB
Operator H	37,979	261,486	16,003,347	20	123	8889	10,918	660	299,084	372,992	475,557	1289	1833	1531	5/7/2021	MB
	37,978	261,212	15,993,640	20	124	9250	10,903	660	327,738	383,660	578,005	1354	1970	2368	4/28/2021	MB
	37,128	185,939	9,733,347	26	79	9074	9896	660	142,013	220,781	287,803	833	1518	1455	7/3/2021	MB
	37,129	171,741	10,030,421	24	91	9261	9921	660	160,466	239,811	344,453	908	1564	1204	7/10/2021	MB
	37,130	171,568	10,027,942	24	91	8902	9912	660	172,154	208,757	385,018	835	689	866	7/7/2021	MB
	37,706	223,681	10,018,586	24	93	9410	9962	450	186,170	342,117	460,124	689	1852	1343	9/27/2021	TF Bench 1
	38,310	316,661	15,034,262	26	90.9	8851	9893	1100	233,524	278,600	482,183	1079	865	1616	11/7/2021	MB
	36,520	203,602	9,964,933	24	91	9527	9928	450	226,118	253,462	447,761	891	1224	1471	10/8/2021	TF Bench 1

Nomenclature

ADD	Augmented Drainage Development
AL	Artificial Lift
BPS	Bakken Petroleum System
BS	Ball and Sleeve
Cmt	Cemented
Comp.	Completion
CPG	Chow Pressure Group Technique
DAS/DTS	Distributed Acoustic Sensing/Distributed Temperature Sensing
DFIT	Diagnostic Fracture Injection Test
DSU	Drilling and Spacing Unit
ESP	Electrical Submersible Pump
ESPCP	Electrical Submersible Progressive Cavity Pump
FDI	Fracture Driven Interaction
Gel	XL Gel
GL	Gas Lift
GOR	Gas Oil Ratio
HFE	High-Frequency Extender
Hyb.	Hybrid Slickwater/Gel
IOR	Improved Oil Recovery
IoT	Internet of Things
JP	Jet Pump
MD	Measured Depth
MDD	Microseismic Depletion Delineation
MWD/LWD	Measurement While Drilling/Logging While Drilling
NER	Net Energy Return
NPV	Net Present Value
OH	Open Hole
PCP	Progressive Cavity Pump
PL	Plunger Lift
PNP	Plug and Perf
Prop.	Proppant volume/ length
RA	Radioactive Tracer Analysis
ROP	Rate of Penetration
SP	Swell Packer
SRP	Sucker Rod Pump
SS	Sliding Sleeve
SW	Slick Water
TF	Three Forks
TVD	True Vertical Depth
Vol.	Volumes
WC	Water Cut
XLE	Extreme Limited Entry