Case studies can be instructive in the evaluation of other coalbed methane (CBM) development opportunities. The San Juan basin, located in New Mexico and Colorado in the southwestern U.S. (Figure 1), is the most prolific CBM basin in the world. It produces more than 2.5 Bscf/D from coals of the Cretaceous Fruitland formation, which is estimated to contain 43 to 49 Tscf of CBM in place. For a long time, the Fruitland formation coals were recognized only as a source of gas for adjacent sandstones. In the 1970s, after years of encountering gas kicks in these coals, operators recognized that the coal seams themselves were capable of commercial gas rates. CBM development benefited greatly from drilling and log data compiled from previous wells targeting the deeper sandstones and an extensive pipeline infrastructure that was built to transport conventional gas. These components, along with a U.S. federal tax credit and the development of new technologies such as openhole-cavity completions, fueled a drilling boom that resulted in more than 3,000 producing CBM wells by the end of 1992. The thickest Fruitland coals occur in a northwest/southeast trending belt located in the northeastern third of the basin. Total coal thickness in this belt locally exceeds 100 ft and individual coal seams can be more than 30 ft thick. The coals originated in peat swamps located landward (southwest) of northwest/southeast trending shoreline sandstones of the underlying Pictured Cliffs formation. The location of the thickest coals (Figure 1) coincides with the occurrence of overpressuring, high gas content, high coal rank, and high permeabilities in the San Juan fairway ("fairway"). The overpressuring is artesian in origin and is caused by water recharge of the coals through outcrops along the northern margin of the basin. This generates high vertical pressure gradients, ranging from 0.44 to 0.63 psi/ft, which allow a large amount of gas to be sorbed to the coal. Coal gas in the San Juan basin can contain up to 9.4% CO2 and 13.5% C2 . Chemical analyses suggest that thermogenic gases have been augmented by migrated thermogenic and secondary biogenic gas sources, resulting in gas contents ranging up to 700 ft 3 /ton. Coal rank in the fairway ranges from medium- to low-volatile bituminous and roughly coincides with those portions of the basin that were most deeply buried. Coals in the fairway typically have low ash and high vitrinite contents, resulting in large gas storage capacities and excellent permeabilities of 10 md from well-developed cleat systems.
A useful first step in the characterization of any new coal area is to compare its characteristics with those of successful CBM projects. Table 2 summarizes the characteristics of several successful projects in the US and includes parameters related to reservoir properties, gas production, gas resources, and economics. The table shows that successful projects have many similarities, including high permeabilities and high gas resource concentration; however, the table does not include aspects such as government incentives or high-value markets, which could elevate a marginal project to commercial status.
The process of drilling and completing coalbed methane (CBM) wells is similar to wells in conventional reservoirs. Coring, however, can pose special challenges. The first step in creating a drilling program for a CBM well involves gathering information about existing wells in a given area. After these data are gathered and analyzed, a preliminary drilling and completion prognosis can be drafted with the input of field operations personnel. An important aspect in drilling frontier or appraisal wells is to keep the drilling procedures relatively simple.
Unlike conventional reservoirs, coal seams are the source, trap, and reservoir for coalbed methane (CBM). A comparison of the two reservoir types shows profound differences in reservoir properties, storage mechanisms, flow mechanisms, and production profiles. Organic matter constitutes more than 50% of coal by weight and more than 70% by volume. Type refers to the variety of organic constituents.
This case study helped an operator in the Powder River Basin approach an optimized completion design. The operator used geomechanical measurements, hydraulic fracture modeling, and fracture diagnostics on two horizontal wells. The two wells are near a previously-completed, producing well (i.e., “parent” well).
While drilling the two horizontal wells, the operator acquired geomechanics data. This method, called drill bit geomechanics, measured the variability along the laterals. These data produced geomechanically-informed perforation and stage placements to minimize the differences in minimum horizontal stress across each stage. Additionally, the operator engineered the perforation sizes, which increased perforation friction to overcome the measured variability. The authors used the near-wellbore geomechanics data, along with other data, in a hydraulic fracture simulator. In general, standard hydraulic fracture simulators assume constant mechanical properties in each geologic layer. Compared to this standard practice, adding measured geomechanics data can more accurately predict which perforation clusters may be stimulated. To test two different fluid systems, the operator designed a “hybrid” (i.e., combination of slickwater and crosslinked gel) treatment for Well 1 and a slickwater treatment for Well 2. Fracture diagnostics reported their effectiveness. Diagnostics included: 1) proppant tracers to evaluate the perforation efficiency, 2) oil-soluble fluid tracers to quantify by-stage production contribution, and 3) water-soluble fluid tracers to assess inter-well communication. Also, the operator had used proppant tracers on the parent well, providing a baseline for results comparison.
Compared to the parent well, the two study wells showed 15-22% higher perforation efficiency. This suggests the engineered design changes created more even proppant distributions. Understanding the geomechanical variability, the operator recognized the engineering required to overcome it. The oil-soluble tracer, although affected by the parent well's depletion profile, showed higher perforation efficiency can increase oil production. Between the two study wells, Well 1 had higher perforation efficiency than Well 2 and it slightly out-produced Well 2. This suggested the hybrid design was likely the more effective design. The hydraulic fracture simulator with near-wellbore geomechanics data predicted perforation efficiency similar to that measured by the proppant tracer. Across both wells’ traced stages, the predicted efficiency and measured efficiency were within 3%. The measurements validated the modeling method.
This paper describes a method of improving completion designs through 1) geomechanics data measured while drilling, 2) modeled perforation cluster efficiency, 3) a measurement of proppant placement effectiveness, and 4) an estimate of stage-by-stage production. For the Powder River Basin operator, this method informed decisions about the next completion design iterations. Operators in any unconventional basin could apply this workflow to approach an optimized completion.
The completion design process for most horizontal wells in shale reservoirs has become a statistical evaluation process, rather than an engineering-based process. Our paper presents an alternative approach using an engineering approach to define the reservoir properties and the effectiveness of the fracture treatments. We then use these results in an economic analysis that allows the engineer to be predictive with respect to how capital is spent in the completion process.
This paper presents a methodology for both the evaluation of the reservoir and the design of the well completion where the engineer can make economic decisions and determine the change in the return on investment as a function of the change in capital expenditure. The engineer can then be able to “optimize” the completion and fracture treatment designs based on Net Present Value, Return on Investment or any other economic parameter desired. We use a rate transient analysis approach to estimate reservoir and fracture properties. We present case histories in the paper, and the interpretation of the production analyses of these case histories yields information about the formation permeability and the effective lengths and number of hydraulic fractures created during the completion process.
With knowledge of the reservoir and fracture properties in hand, the engineer can then determine the “optimum” completion design for future wells. This understanding can be achieved much quicker and for much less money than the cost to drill the number of wells necessary to make statistical analysis meaningful. The results of the case histories indicate that many completion designs are not in the “optimum” range. Too much capital is being spent increasing stage count when it should be going to increased effective length. The focus on early-time production has ignored the effect that more fractures has on ultimate recovery.
The results and conclusions in this paper will run contrary to much of the direction most unconventional completion designs have been evolving over the past 5 to 10 years. A much greater emphasis on achieving increased effective lengths will be demonstrated and that increased stage count can prove detrimental to economic success over the well's life. Processes in the paper will also prove valuable for smaller operators that do not have a large well counts that are usually required to achieve a meaningful statistical evaluation.
Objectives/Scope: Understanding that capillary forces will act to limit petroleum fluid saturations in water-wet fine-grained rocks, including organic rich source rocks, dates back at least to Hubbert (1953). Likewise, Philippi (1965) noted relationships identifying sorption in/on organic matter as a significant storage mechanism in organic-rich rocks. Contrast these early insights with current unconventional reservoir evaluation, where we observe a disconnect between in situ (core exhumed to surface) measured total water saturations vs. the produced cumulative water volumes from a given stimulated rock volume. Water-free production in gas shales, from gas-wet organic matrix pores, created an early impression that unconventional plays don't produce water. So, in more liquid-rich plays, water cuts were initially under-appreciated: e.g. >80% in the Wolfcamp (stock-tank basis). If measured Sw is so low (core-based calibration), where is the water coming from; or is there an alternative method to more accurately relate in situ to produced water and petroleum production?
Methods/Procedures/Process: Adapting organic sorption models from the 80's, we can split total hydrocarbon volatiles into sorbed and, by difference, non-sorbed (fluid phase) yields. Converting to volumes and adding back dissolved gas using a formation volume factor (FVF) we can estimate the bulk volume fluid phase. This new approach then yields observations regarding remaining water-filled pore volume versus sorbed and non-sorbed hydrocarbon volume explaining the high water cuts in the Permian Basin stratigraphy; and additionally may indicate sweet spots in pore systems in different parts of the rock compared to alternatively derived saturations.
Results/Observations/Conclusions: The final piece of the puzzle comes from basin modeling of petroleum charging in the 90's. Some scientists applied conventional reservoir relative permeability to fine-grained rocks, but new research predicted that progressively finer grained rocks with higher irreducible water should be able to flow oil at progressively higher Sw: at 100nD, both oil and water should flow at Sw > 80%. Lower petroleum phase saturations and adjusted relative permeability curves may better explain observed production behaviors and profoundly alter our view of recovery factor and stimulated rock volume.
Applications/Significance/Novelty: The method offers an alternate and independent method to Dean-Stark-based core / SWC saturation analysis and its pitfalls. Saturation patterns after removal of immobile sorbed oil are different to those derived using the Dean-Stark based method, implying sweet spots / landing zones can be further optimized even in maturing shale plays. Lower oil-in-place – representing only the potentially mobile fluid phase petroleum – means that fracture stimulation has a higher recovery factor than previously thought, with profound effects on the infill volumes / opportunities for future field developments and therefore ultimately for US – and global – oil supply projections.
Interdisciplinary Components: Cross-over technology from organic geochemistry to petrophysics to reservoir engineering.
A fundamental component of a real-time drilling analytics system is automatic rig state detection. High frequency time series data (typically one data point per second) from multiple sensors on a drilling rig is processed and labeled with drilling states including: slide drilling, rotate drilling, pick up, in slips, and others. With labeled time-series data, the real-time system can derive operational KPIs (key performance indicators) with extremely high resolution, e.g., a statistical summary of rotary versus slide drilling time for the rig supervisor and drilling engineer to analyze efficiency. Later, such information can be leveraged to develop algorithms to detect abnormal drilling events and drive closed loop control.
A workflow was developed to clean and fill in any missing data. A rules-based model was then applied to classify the data into seventeen rig states. For the state “drilling”, a sub-classification was made to label rotate drilling and slide drilling. However, it is difficult to categorize “slide drilling” solely based on surface RPM due to top drive oscillation. In order to achieve acceptable accuracies, three machine learning models to classify “rotate drilling” and “slide drilling” were evaluated: Random Forest, Convolutional Neural Network (CNN), and a hybrid Convolutional Neural Network / Recurrent Neural Network (CNN/RNN).
Machine learning models were built for two basins, one model each, to accommodate different drilling styles. For the Delaware Basin, 10 wells with 9 million rows of data were chosen, and for the DJ Basin, 12 wells with 2 million rows of data were chosen. A legacy, rules-based algorithm was applied to label each row as rotate or slide drilling, and the misclassified records were manually corrected. The machine learning models were found to be far superior to rules-based models. For the wells in the training set, the accuracies of our rules-based models were 70% and 90% respectively, while the accuracies of our machine learning models were over 99%. The CNN model was proven to be the best model, excelling with high accuracy, short computation time, and scalability for big data applications.
The data cleaning, preprocessing, and machine learning algorithm has been deployed in Anadarko's Real-Time Drilling (RTD) ecosystem (Cao et al., 2018, 2019), which consists of four layers: a data source, analytics, data storage, and UI layer. KPIs, directional statistics, and engineering models are calculated in real-time and visualized through a web-based UI. This system can be accessed by any member of the drilling operations team. The system is regularly used to evaluate, compare, and optimize well performance. Future plans include pushing analytical models to the rig site with edge computing to facilitate drilling guidance and levels of automation. To our knowledge, this is the first time that a deep learning model has been used to analyze drilling time series data in a production real-time system.
The Finn-Shurley field produces petroleum from the Upper Cretaceous Turner Sandstone of the Powder River Basin. The Turner is a member of the Carlile and is overlain by the Sage Breaks and underlain by the Pool Creek members of the Carlile. The Turner is interpreted to be a shallow marine shelf sandstone deposited along the eastern side of the Western Interior Cretaceous Seaway. Sand-shelf-bar orientation across the field is roughly east-west. Trapping occurs where sandstone beds get shalier up-dip. The field is located along the shallow east margin of the Powder River Basin south of the Clareton lineament.
Three to four coarsening upward cycles are present in the Turner in the field. Most of the production in Finn-Shurely comes from the lower two cycles. Each cycle consists of burrowed to bioturbated, heterolithic mudstones and sandstones coarsening upwards into fine-grained laminated to burrowed sandstones. Trace fossil present fall into the shelf Cruziana ichnofacies. The sandstones are largely litharenites. Porosities range from 11-17% and permeabilities range from 0.06 to 0.5 md. Source rock analysis of the Turner shales indicate Ro values averaging 0.63 and Tmax values of 433°C. Source beds for the oil and gas in the Turner are thought to be the Mowry and Niobrara formations. The low thermal maturity suggests lateral migration of oil into the stratigraphic trap.
The field extends over an area roughly circular in shape of ~65 square miles. Productive depths across the field are 4450 to 5700 ft. First production is reported as 1965 and cumulative production from ~750 vertical wells is 23.6 MMBO and 38.9 BCFG. Cumulative gas oil ratio is 1688 cu ft gas per barrel oil. Average production per well is approximately 31.5 MBO and 52 MMCFG. Horizontal drilling activity in the field area has recently commenced. Although the production is fair to marginal, the field provides an excellent example of trapping style as well as a depositional model for Turner Sandstone elsewhere in the Powder River Basin. Recent drilling in the deeper overpressured parts of the Powder River Basin has encountered excellent production from the Turner (> 1,000 bbls oil equivalent per well).
Finn-Shurley Field is part of a continuous accumulation within the Turner Sandstone in the Powder River Basin. Distinct oil-water contacts are not present in the field area. The accumulation is underpressured and regarded as unconventional.
Core measurements of fluid wetting characteristics frequently indicate oil-wet conditions for organic-rich unconventional reservoirs. To define fluid wetting differences between inorganic and organic porosity components using commonly available logging suites is a petrophysical challenge, but is necessary to correctly define hydrocarbon saturation and hence hydrocarbon volumes.
The calculation requires specific input with respect to thermal maturity, either vitrinite reflectance (Ro) or level or organic maturity (LOM). Also, input of the conversion of TOC from weight to volume percent is required.
By comparing inorganic porosity and organic porosity with resistivity (Pickett plots) it can be shown that the inorganic porosity is mostly water-wet, with Archie saturation exponents typically 2 or less, whereas the organic porosity is usually oil-wet, with Archie saturation exponents frequently 3 or more. Also, it is often observed that the Archie cementation exponent for organic porosity is much less than 2, suggesting linear flow paths.
Application of the different values of Archie exponents for the inorganic and organic porosity elements allows quantification of water saturation and therefore provides input for oil-in-place calculations.
These findings are significant because traditional petrophysical analysis does not consider the marls to have any reserve potential.
Of important significance is the ability to quantify the resource potential of the organic porosity fraction, which standard petrophysical analysis does not provide. It can be shown that significant volumes of mobile hydrocarbons are associated with organic porosity, which is routinely not calculated.