SPE 71090

Advanced Reservoir Modeling In Desorption-Controlled Reservoirs Scott Reeves, Advanced Resources International, Houston, Texas; Larry Pekot, Advanced Resources International, Arlington, Virginia

Copyright 2001, Society of Petroleum Engineers Inc. permeability, etc. Analysis of core and other data suggests that This paper was prepared for presentation at the SPE Rocky Mountain Petroleum Technology another porosity and permeability system is required to Conference held in Keystone, Colorado, 21–23 May 2001. account for this effect, specifically within the matrix blocks This paper was selected for presentation by an SPE Program Committee following review of themselves. An advanced, triple-porosity/dual-permeability information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to model has therefore been developed, in which gas desorbs correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at from the internal matrix block surfaces, migrates via SPE meetings are subject to publication review by Editorial Committees of the Society of conventional Darcy flow through micro-permeability matrix, Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is and into the cleat system where it then flows to the wellbore. prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous Water can also be stored both within the matrix blocks and in acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. the cleat system. In essence, this model requires that desorbed Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. gas must work its way through the matrix before reaching the cleat system, and must establish a relative permeability to gas Abstract within the matrix block before it can do so. This geometry is Reservoir models typically utilized for desorption-controlled similar to conventional dual-porosity models, with the reservoirs such as coals and gas shales possess dual- addition of an adsorbed gas component. porosity/single-permeability characteristics. In this case dual- porosity means that two in-situ locations exist that can be used Comparisons of this new model versus the historical for gas storage, adsorbed within the matrix and in the free modeling approach confirm that the new model predicts lower form in the cleat system. Single-permeability, which refers to gas and higher water production rates, consistent with field the cleat system, is the only permeability network that gas or evidence. Further, more accurate production forecasts can be water must flow through to reach the wellbore. While this achieved using measured well test information (for the cleat approach to modeling coals and shales has become accepted permeability), low cleat porosities (which are known to exist), practice, experience has shown that the models can frequently and lab-derived porosity and permeability data for the matrix be in gross error when forecasting well or field performance block properties. This paper presents the historical accuracy based on limited reservoir and/or production data; gas problem with reservoir simulation in desorption-controlled production is usually over-predicted and water production reservoirs, the practical theory behind the new model, under-predicted. The implications for economic decision- comparisons between the new and conventional models, and making in an exploration mode are obvious, and there are some example applications. many examples of projects that have suffered from this very problem. Further, reservoir parameters derived from history- Introduction matching, when historical gas production does exist, are In the early 1980’s, when producing gas from coalbeds in the commonly found to be inconsistent with measured U.S. was in its infancy, one of the greatest hurdles for resource permeability and gas sorption/content data. While there has development was having the tool(s) to accurately predict long been considerable effort focused on improved data collection term well and field performance so that investment decisions procedures, such as well testing and gas content measurement could be made. It was in response to this technology obstacle for example, these problems persist. that the staff of Advanced Resources International (ARI) developed the COMET coalbed reservoir simulator (and later While performing reservoir studies in the Antrim a desktop version – COMETPC) with funding from the Gas shale and low-rank coal plays throughout the world, it became Research Institute (now the Gas Technology Institute, or GTI), clear that the accepted assumption of gas desorbing directly which became the de-facto industry-standard tool for coalbed from the coal matrix into the cleat system is not entirely valid. methane (CBM) modeling. Since its earliest versions, the In practice, gas production occurs much later than the models model has been upgraded to a fully 3-D version (COMETPC- predict, and cannot be adequately explained though the normal 3D, funded by a industry consortium in 1989-90). At its core, parameters of sorption time, permeability, relative 2 SCOTT REEVES; LARRY PEKOT SPE 71090 the COMET model is based on the Warren & Root sugar-cube incorporating it either in the adsorbed state within the matrix model to replicate the dual-porosity nature of coals 1 (i.e., gas block, or as free gas in the natural fractures17. Essentially, if adsorbed on the internal surfaces of the coal in the matrix that “extra” gas was placed in the natural fractures, it was blocks, and a cleat system through which two fluid phases - produced too quickly, and if it was placed in the adsorbed free gas and water - flow in the reservoir). state, it was produced too slowly. In either case, to achieve material balance the initial saturation conditions in either the A common schematic for illustrating the nature of natural fractures or the matrix blocks had to be modified from fluid flow in coals is shown in Figure 1, and the model actual measured (and known) values, and the producing rates representation is shown in Figure 2. The processes of would still be in error. To solve this problem, the traditional desorption and diffusion (Figure 1 (a) and (b)) are lumped dual-porosity representation was modified to incorporate a together as one of the sugar-cubes and gas release, described third porosity (gas storage) system within the matrix block to by parameters such as diffusion coefficient and sorption time, provide needed free gas (and water in some cases) storage occurs directly into the cleats. Details on the reservoir capacity. Fundamentally, rather than lumping together mechanics incorporated into the model are provided in the desorption and diffusion (Figure 1 (a) and (b)), they were references2,3. This representation of coalbed methane decoupled such that each could be discretely modeled. This reservoirs has become widely accepted, and now numerous enabled material balance in each porosity system to be models have been developed with a similar premise and are honored, improved well performance forecasting, and the commercially available in the marketplace. COMETPC-3D, direct use of lab and well test data in the model. The following and presumably other models, have been used for a wide sections describe the new model, COMET2, how it performs variety of purposes, including coalbed methane reserve relative to the previous versions, and the implications. estimation4,5, resource assessments6, development of exploration strategy7, development of field spacing rules8, and Description of Triple-Porosity/Dual-Permeability evaluation of best completion, stimulation and operating Model practices9,10,11. COMETPC-3D has also been used extensively for coalbed reservoir characterization, most notably as a Model Description. A conceptual illustration of the new centerpiece of GTI’s extensive R&D programs in the San Juan reservoir model is provided in Figure 3. The mechanics of gas and Warrior basins in the late 1980’s and early desorption are similar to prior versions of COMET (i.e., are 1990’s12,13,14,15,16, which played a significant role in facilitating governed by a Langmuir isotherm and sorption time), as are the growth of U.S. CBM production. the traditional physics of two-phase flow in the cleats. The difference lies in the addition of an entirely new porosity However, mounting evidence suggested that the dual- system within the matrix blocks. This system is characterized, porosity replication being utilized in the various CBM models in terms of reservoir parameters, by values for porosity and was not matching actual field performance without significant permeability, relative permeability (like the cleats, this micro- adjustments to measured reservoir properties. Specifically, porosity is also a two-phase system), capillary pressure, gas results of core analysis could not be directly utilized in the and water saturations, as well as other common reservoir models – permeability estimates were too low, and porosity engineering parameters. The net effect of this system is estimates were too high, to be representative of the cleat severalfold. Firstly, as alluded to above, it allows the material system (core studies tend to provide matrix properties, and do balance to be honored, and fluids to be properly allocated to not capture the larger-scale reservoir features such as open their actual physical locations within the reservoir. Secondly, cleats and natural fractures). In addition, well test permeability with additional water within the matrix block porosity, more estimates, when directly input into a reservoir model, tended water is produced during early times, consistent with field to yield predicted gas rates that were too high and water rates evidence (without having to artificially increase cleat porosity that were too low. This presented a significant problem in to unrealistic values). Third, since a gas saturation must be coalbed methane exploration, since the gas rate predictions established within the matrix blocks before gas can flow into upon which investment decisions were made were typically the cleats, the early time gas production is delayed, again too high. As a result, many international coalbed methane bringing the predictions into better alignment with field projects proceeded when perhaps they should not have, giving evidence. international CBM a “bad investment” stigma. As a practical matter, by decoupling the intra-matrix The prediction problem became an issue that had to properties from cleat properties, better utility of reservoir be addressed when ARI was performing reservoir characterization measurements can be achieved. For example, characterization R&D in the mid-1990’s on behalf of GTI in core measurements of porosity, permeability, relative another desorption-controlled gas play, the Antrim Shale in permeability, and capillary pressure, previously unusable, can the Michigan Basin. Research there suggested a that a now be input as matrix block properties. In addition, well test significant component of the gas resource existed in the free permeability estimates, which typically had to be revised state within the matrix blocks themselves, and its effect on gas downward to match actual well performance, presumably due producing rates could not be adequately modeled by to the need to use a hybrid permeability value in the model SPE 71090 ADVANCED RESERVOIR MODELING IN DESORPTIO N-CONTROLLED RESERVOIRS 3

(matrix and cleat permeability combined), can now be used The remaining three runs attempt to replicate the with more confidence as being directly representative the bulk 3f/2k model using variations of cleat permeability and coal permeability (cleats and secondary natural fractures porosity. Run 3 varies permeability only to achieve a combined). reasonable match for cumulative gas production. Permeability had to be reduced to 0.4 md, over an order of magnitude Comparative Simulations. To illustrate the difference in well lower, to achieve this result, with considerable differences in performance predictions based on the historical dual- water production still existing also (by over an order of porosity/single-permeability (2f/1k) approach and the new magnitude). Run 4 attempts to match cumulative water triple-porosity/dual-permeability (3f/2k) approach of production by adjusting only porosity; a rough match was COMET2, a series of comparative simulation runs were achieved with 8 %, a higher value that is typically used in performed. Specifically, five runs were performed; the first 2f/1k simulations. In this case cumulative gas production is being a COMET2 simulation using a complete 3f/2k reservoir over predicted by a factor of almost 3. description, based on the assumption that this is the data that would be available from laboratory measurements of gas The final simulation, Run 5, attempts to match both content, sorption properties, matrix flow properties, and cleat the cumulative gas and water production from Run 1 by permeability estimates derived from well tests. The next series varying both cleat permeability and porosity. The best match of runs uses the 2f/1k coal description, and attempt to was achieved with a permeability of 2.0 md and a porosity of replicate the results of the first run using various combinations 13%, both very different values from those in the 3f/ 2k of reservoir parameters. Table 1 provides the relevant input model, as well as unrealistic in terms of porosity. One can parameters for the five simulations. All other reservoir clearly appreciate the potential for inaccurate predictions parameters remained constant for all of the cases. (specifically overstating gas production and understating water production) using laboratory and field reservoir property Before discussing the results of these model runs, measurements in conjunction with the 2f/1k simulation several points of clarification are needed with respect to how model. certain differences in the model were handled to ensure the comparisons were made on as consistent a basis as possible. Note that runs 2,3,4 and 5 had no manipulation of First, the initial water saturations in the cleat and matrix relative permeability in an attempt to replicate the 3f/2k case. porosity systems were set to 100% for each case. Second, the In a typical history matching exercise, adjustment to relative relative permeability relationships adopted for the cleats, the permeability curves might be heavily relied upon. Thus, a matrix, and composite (for the 2f/1k runs) are shown in history match with a 2f/1k simulator can usually be created Figure 4. In the 3f/2k model, straight-line curves were used without the simulation engineer realizing that 2f/1k for the cleats, similar to what would typically be used for simulation is a more simple representation of in-situ natural fractures, and curves consistent with laboratory conditions than a 3f/2k model. measurements utilized for the matrix. The composite curves utilized for the 2f/1k model are consistent with those derived from history-matching actual CBM well performances. The Enhanced Coalbed Methane Recovery And Carbon selection of these particular curves was based on a literature Sequestration Applications review 18,19. Third, the sorption time (tao) utilized for the 3f/2k model was only a fraction of that used for the 2f/ 1k Reservoir Mechanics. Another new and unique feature of the model. This was because in a 2f/1k model tao accounts not COMET2 model is its ability to simulate enhanced coalbed only for desorption from the coal surface, but also diffusion methane recovery processes (i.e., nitrogen or carbon dioxide through the matrix, which the 3f/2k model accounts for flooding), and CO2 sequestration in coalseams, both of which 20,21,22,23 separately. are receiving an increasing level of worldwide interest . The mechanism by which CO2 (or N2) can enhance the The results from the simulations, presented in terms coalbed methane recovery process, and/or CO2 is sequestered, of gas and water production versus time, are provided in is a complex mix of physical and chemical interactions that Figures 5 and 6, as well as Table 2. The base case, Run 1, must achieve equilibrium simultaneously in the sorbed state using the 3f/ 2k model, yields an early-time gas rate less than and in the gaseous state. Coal has the capacity to hold all other cases, and a high early-time water rate, generally considerably more CO2 than either CH4 or N2 in the adsorbed consistent with field results. The first 2f/ 1k simulation, Run state (in an approximate ratio of 4:2:1). As a result, in the 2, simply uses the same cleat properties (k=5md, f =1%). A presence of multiple gases (e.g., CO2, CH4 and N2), the stark contrast in results can be seen – gas production increases amount of each in the adsorbed state would be in (by a factor of 4-5) and water production decreases (by a approximately these proportions. However, since any injected factor of 6-7) – and demonstrates the dramatic difference in gas for ECBM is unlikely to be of exactly that composition, a results between the two model types. partial-pressure disequilibrium will be created in the gaseous phase (i.e., in the coal cleat system). Adsorption/desorption of 4 SCOTT REEVES; LARRY PEKOT SPE 71090 individual components will therefore occur until the gases in Obviously, this is attractive since power plant flue gas is both the sorbed and gaseous states are each in equilibrium, and comprised mostly of these two components. are in equilibrium with each other. Water production increases with either N2 or CO2 As an example, consider ECBM recovery via N2 injection. The higher water response with N2 is surmised to be injection. Under certain conditions, the equilibrium ratio of a result of its lesser compressibility and higher viscosity than CH4 to N2 in the adsorbed state is 2:1, but is 1:3 in the gaseous methane. CO2 is quickly adsorbed by the coal matrix, which state. As N2 is injected, it flushes the gaseous methane from releases methane to the fracture system. Hence, it occupies a the cleats, creating a near 100% N2 saturation. The partial minimal portion of the in-situ pore space. The additional pressure of methane in the gaseous cleat-system phase is methane released from the CO2-flooded areas sweeps reduced to 'zero', a disequilibrium condition in a system additional water to the producer. containing both methane and nitrogen. As a result, methane desorbs and is drawn (or 'pulled') into the gaseous phase to Field Tests. There are currently two field sites where CO2 or achieve partial-pressure equilibrium. This is why the N2- N2 injection is being performed for ECBM purposes, and in ECBM recovery process is referred to as methane stripping. general support the modeling findings described above. The On the other hand, as CO2 is injected, it becomes first field site is the Tiffany Unit, operated by BP in the preferentially adsorbed onto the coal, displacing methane. northern San Juan basin (Figure 11). Amoco began to However, there is no 'pull' on the methane into the cleat investigate ECBM techniques in the late 1980s, primarily via system, rather it is 'pushed' from the matrix by the CO2. laboratory experiments, which involved injecting a gas, or mixture of gases such as N2, CO2, or flue gas, to improve COMET2 models these processes based on the theory recovery. Building on the success of laboratory and pilot tests, of an extended Langmuir isotherm24. With the extended the company moved forward with the first and largest full Langmuir isotherm, the concentration of each gas component scale N2-ECBM commercial pilot known as the Tiffany Unit. can be directly calculated from its partial pressure. Only the After nine years of primary production, nitrogen injection was Langmuir constants from pure gas sorption are used, and no commenced in January 1998 into 12 wells, with production binary constants are needed. This makes the technique very from 34. Production from the Tiffany Unit is shown in Figure simple and quite easy to use in mathematical formulations. 12. Similar to the modeling results, an immediate impact on There is a body of laboratory and theoretical work that gas production was observed with N2 injection, as well as with supports the validity and utility of the extended Langmuir its intermittent cessation. Early N2 breakthrough was also isotherm approach for this application25,26,27, although it is less observed. However, an increase in water production is not precise than some equation–of-state approaches that account apparent. Further field evaluation is planned to study this for gas mixture interactions on a molecular level. response further.

Modelling Results. To illustrate the differences in how N2 The second site, the Allis on Unit, operated by and CO2 injection affect CBM recovery, a series of COMET2 Burlington Resources, is the world's first experimental (pure) simulations were performed. In the model, a quarter 5-spot CO2-ECBM recovery pilot. The field is also located within the pattern was assumed which, after a period of primary northern portion of the San Juan basin (Figure 11). The pilot production of 5 years, a series of cases were run where either comprises of four CO2-injection wells and nine methane N2 or CO2 was injected into the center well. The assumed production wells. Formerly, these wells had been produced reservoir parameters for the model are similar to those in using conventional pressure-depletion methods for over five reference 23, and reflect a San Juan Basin setting. The results years. During 1995, Burlington drilled four injection wells and are provided in Figures 7-10, and Table 3. The model results initiated CO2 injection. The production response is shown in indicate a immediate and significant gas production Figure 13. In this case, an immediate water production enhancement with N2 injection. However, N2 breakthrough increase was observed, but an immediate gas production occurs fairly quickly and becomes a high percentage of total increase is not readily apparent. However a series of events, production. Hence any enhanced recovery benefit gained by such as the shutting-in and re-opening of wells, re- N2 injection must be balanced against higher gas treatment completions, and gathering system pressure reductions, have costs. Injection of CO2 also results in an immediate gas complicated understanding the field response. Further and production response, albeit less so than with N2. In the case of more detailed reservoir studies are needed (and planned) to CO2, however, no significant breakthrough of CO2 is predicted understand the pilot response and reconcile any potential over the 20-year simulation (the model assumes a differences with model predictions. homogeneous reservoir for all cases, absent of potential reservoir pathways for early breakthrough). These behaviors Since these two sites represent a highly unique have an important impact on an integrated ECBM recovery opportunity to gain insights into the reservoir mechanics of and CO2 sequestration project; to achieve the desired low net- CO2 sequestration and ECBM recovery, the U.S. Department cost for CO2 sequestration, an injection gas consisting of both of Energy has selected them for detailed reservoir N2 and CO2, in optimised proportions, is the likely outcome. characterization and modeling study by ARI using the SPE 71090 ADVANCED RESERVOIR MODELING IN DESORPTIO N-CONTROLLED RESERVOIRS 5

COMET2 model28, 29. This project, just now underway, will Pressure, and Coal Rank on Coalbed Methane Exploration provide additional insights into the processes of ECBM Strategy”, SPE 21490, SPE Gas Technology Symposium, Houston, (January 23-25, 1991). recovery and CO2 sequestration mechanisms, as well as an improved predictive capability for future project screening and 8. Young, G.B.C.; Paul, G.W.; McElhiney, J.E.; McBane, R.A., “A design. Future publications will present the results of that Parametric Analysis of Fruitland Coalbed Methane Reservoir work. Producibility”, SPE 24903, 67th SPE Annual Techincal Conference and Exhibition, Washington, (October 4-7, 1992). Conclusions This paper presents a new model for reservoir simulation in 9. Young, G.B.C.; Paul, G.W.; Saulsberry, J.L.; Schraufnagel, desorption-controlled reservoirs. The primary advancement R.A., “A Simulation-Based Analysis of Multiseam Coalbed th made is that rather than lump desorption and diffusion Well Completions”, SPE 26628, 68 SPE Annual Technical together within the matrix blocks, they are decoupled and each Conference and Exhibition, Houston, (October 3-6, 1993). described and modeled explicitly. The new model is hence 10. Young, G.B.C.; Kelso, B.S.; Paul, G.W., “Understanding Cavity described as triple-porosity/dual-permeability, as distinguished Well Performance”, SPE 28579, 69th SPE Annual Technical from dual-porosity/single-permability models commonly Conference and Exhibition, New Orleans, (September 25-28, employed for coalbed methane simulation. The new model 1994). allows for more direct utilization of laboratory and field- derived reservoir properties, and improved allocation and 11. Young, Genevieve B.C.; Paul, George W., “A Simulation-Based material balance of reservoir fluids. The new model provides Approach To Evaluating Coalbed Methane Reservoir very different predictions than those generated from dual- Producibility and Operating Practices”, Case Studies in Coalbed porosity/single-permeability models, but ones that are believed Methane Reservoir Simulation, International Coalbed Methane more consistent with both actual field performance and Extraction Conference, London (January 24-25, 1994). measured reservoir properties. The new model can also 12. Young, G.B.C.; McElhiney, J.E.; Paul, G.W.; McBane, R.A.; simulate enhanced recovery of coalbed methane via nitrogen “An Analysis of Fruitland Coalbed Methane Production, Cedar or carbon dioxide injection, as well as carbon dioxide Hill Field, Northern San Juan Basin”, SPE 22913, 66th SPE sequestration in coals. Hence a significant step forward in Annual Technical Conference and Exhibition, Dallas, (October modeling capability in desorption-controlled reservoirs is 6-9, 1991). believed to have been achieved. 13. Young, G.B.C.; Paul, G.W.; McBane, R.A., “Cedar Hill and References Tiffany: Case Studies in Coalbed Methane Reservoir 1. Warren, J.E. and Root, P.J.: “The Behavior of Naturally Simulation”, Symposium on Coalbed Methane Research and Fractured Resrvoirs,” SPEJ (Sept. 1963) 245-255; Trans., Development in Australia, Townsville (November 19-21, 1992) AIME, 228. 14. Advanced Resources International, Inc., “Characterization of 2. Sawyer, W.K.; Paul, G.W.; Schraufnagel, R. A., “Development Fruitland Coal Through Reservoir Simulation”, GRI Topical and Application of a 3D Coalbed Simulator”, CIM/SPE 90-119, Report GRI-92/0461.1, November 1993. International CIM/SPE Technical Meeting, Calgary, (June 10- 13, 1989). 15. Young, Genevieve B.C.; Paul, George W.; Saulsberry, Jerry L.; Schraufnagel, Richard A., “Characterization of Coalbed 3. Paul, G.W.; Sawyer, W.K.; Dean, R.H., “Validation of 3D Reservoirs at the Rock Creek Project Site, Alabama”, Paper No. Coalbed Simulators”, SPE 20733, 65th SPE Annual Technical 9380, International Coalbed Methane Symposium, Tuscaloosa Conference and Exhibition, New Orleans (September 23-26, (May 17-21, 1993). 1990). 16. Advanced Resources International, Inc., “Reservoir 4. Zuber, M.D.; Kuuskraa, V.A., “A Reservoir Simulator-Based Characterization of Mary Lee and Black Creek Coals at the Methodology for Calculating Reserves of Coalbed Methane Rock Creek Field Laboratory, Black Warrior Basin”, GRI Wells”, Paper No. 8952, Coalbed Methane Symposium, Topical Report GRI-93/0179, August 1993. Tuscaloosa (April 17-20, 1989). 17. Kuuskraa, V.A.; Wicks, D.E.; Thurber, J.L., “Geologic and 5. McElhiney, John E.; Koenig, Robert A.; Schraufnagel, Richard Reservoir Mchanisms Controlling Gas Recovery From the th A.: “Evaluation of Coalbed-Methane Reserves Involves Antrim Shale”, SPE 24883, 67 SPE Annual Technical Different Techniques”, Oil & Gas Journal, (October 30, 1989). Conference and Exhibition, Washington, (October 4-7, 1992).

6. Advanced Resources International, Inc., “Reservoir Simulation 18. Young, Genevieve B.C., “Coal Reservoir Characteristics From Study of U.S. Coalbed Methane” Technical Report for the Simulation of the Cedar Hill Field San Juan Basin”, Methane USGS in support of the 1995 National Assessment of U.S. Oil From Coal Seams Technology, (July 1992). and Gas Resources, 1995. 19. Meaney, Kevin; Paterson, Lincoln, “Relative Permeability in 7. Reeves, S.R.; Decker, A.D., “A Reservoir Simulation Coal”, SPE 36986, 1996 SPE Asia Pacific Oil & Gas Investigation Into the Interaction of In-Stiu Stress, Pore Conference, Australia, (October 28-31, 1996). 6 SCOTT REEVES; LARRY PEKOT SPE 71090

20. Puri, R.; Yee, D., “Enhanced Coalbed Methane Recovery”, SPE 20732, 65th SPE Annual Technical Conference and Exhibition, New Orleans, (September 23-26, 1990).

21. Stevenson, M.D.; Pinczewski, U.; Downey, R.A., “Economic Evaluation of Nitrogen Injection for Coalseam Gas Recovery”, SPE 26199, SPE Gas Technology Symposium, Calgary, (June 28-30, 1993).

22. Stevens, Scott H.; Spector, Denis; Riemer, Pierce, “Enhanced Coalbed-Methane Recovery By Use Of CO2“, SPE 48881, JPT, (October 1999).

23. Stevens, Scott H.; Schoeling, Lanny; Pekot, Larry; “CO2 Injection for Enhanced Coalbed Methane Recovery: Project Screening and Design”, Paper No. 9934, International Coalbed Methane Symposium; Tuscaloosa, (May 3-7, 1999).

24. Advanced Resources International, Inc.; COMET2 User’s Guide, March 1998

25. Arri, L.E.; Yee, Dan; Morgan, W.D.; Jeansonne, M.W., “Modeling Coalbed Methane Production With Binary Gas Sorption”, SPE 24363, SPE Rocky Mountain Regional Meeting, Casper, (May 18-21, 1992).

26. Harpalani, S.; Pariti, U.M., “Study of Coal Sorption Isotherms Using A Multicomponent Gas Mixture”, Paper No. 9356, Proceedings of the 1993 International Coalbed Methane Symposium, Tuscaloosa, (May 17-21, 1993).

27. Hall, F.E.; Chunhe, Zhou; Gasem, K.A.M.; Robinson, R.L. Jr., “Adsorption of Pure Methane, Nitrogen and Carbon Dioxide and Their Binary Mixtures on Wet Fruitland Coal”, SPE 29194, SPE Eastern Regional Conference & Exhibition, Charleston, (November 8-10, 1994).

28. Reeves, S.R.; Schoeling, L., “Geological Sequestration of CO2 in Coalseams: Reservoir Mechanisms Field Performance, and Economics”, Greenhouse Gas Technology Conference, Cairns (August 2000).

29. Reeves, S.R.; Stevens, S.H., “CO2 Sequestration”, World Coal (December2000).

SPE 71090 ADVANCED RESERVOIR MODELING IN DESORPTIO N-CONTROLLED RESERVOIRS 7

Table 1 - Variable Parameters in 3f/2K versus 2f/1K Comparative Simulations

CLEATS MATRIX Porosity (%) Permeability (md) Tao (days) Porosity (%) Permeability (md) Run Model 1 3f/2k 1 5 0.1 8 0.01 2 2f/1k 1 5 10 n/a n/a 3 2f/1k 1 0.4 10 n/a n/a 4 2f/1k 8 5 10 n/a n/a 5 2f/1k 13 2 10 n/a n/a

Table 2 – Cumulative Production Results, Comparative Simulation Study

CUMULATIVE PRODUCTION (20 YEAR) Run Gas Water (MMcf) (Mbbls) 1 398 994 2 1,784 158 3 396 87 4 1,156 948 5 411 993

Table 3 – Incremental Methane Recoveries from N2/CO2 Injection (Quarter 5-Spot Pattern)

NITROGEN CARBON DIOXIDE Base Case @ 250 Mcfd @500 Mcfd @ 250 Mcfd @500 Mcfd Total Recovery (MMcf) 1,171 2,312 2,933 1,542 2,147 Incremental Recovery (MMcf) n/a 1,141 1,762 371 976

8 SCOTT REEVES; LARRY PEKOT SPE 71090

Figure1: Processes in the Reservoir Transport of Coalbed Methane Gas

Figure 2: Warren & Root Reservoir Model Idealization

SPE 71090 ADVANCED RESERVOIR MODELING IN DESORPTIO N-CONTROLLED RESERVOIRS 9

Micro-porosity System Gas Adsorbed on Coal

Coal Cleats

Figure 3: Comet2 Representation of Triple-Porosity/Dual-Permeability System

1

0.9

Composite 0.8

0.7

0.6 Cleats

0.5

0.4

0.3 Relative Permeability [fraction]

0.2

Matrix 0.1

0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Water Saturation [fraction]

Figure 4: Relative Permeability Curves 10 SCOTT REEVES; LARRY PEKOT SPE 71090

1000

2phi/1k Base Case, Run #2

2phi/1k Por/Water Match, Run #4

100 Gas Rate, Mscfd 2hi/1k Perm/Por Match, Run #5 10

2phi/1k Perm/Gas Match, Run #3

3phi/2k Base Case, Run #1

1 0 30 60 90 120 150 180 210 240 Month Figure 5: Gas Production Forecasts from Comparative Simulations

1000

3phi/2k Base Case, Run #1

2phi/1k Por/Water Match, Run #4

100

2phi/1k Perm/Por Match, Run #5 Water Rate, Bpd

10

2phi/1k Base Case, Run #2

2phi/1k Perm/Gas Match, Run #3

1 0 30 60 90 120 150 180 210 240 Month

Figure 6: Water Production Forecasts from Comparative Simulations SPE 71090 ADVANCED RESERVOIR MODELING IN DESORPTIO N-CONTROLLED RESERVOIRS 11

1000 70.00% N2 @ 500 Mcfd

60.00% N2 @ 250 Mcfd

50.00% Net Methane

40.00%

100

30.00%

Gas Rate, Mscfd Base Case Nitrogen Content

20.00%

Nitrogen Content 10.00%

10 0.00% 0 30 60 90 120 150 180 210 240 Month

Figure 7: Gas Production Response to N2 Flooding

1000

N2 @ 250 Mcfd

N2 @ 500 Mcfd

100 Water Rate, Bpd

10

Base Case

1 0 30 60 90 120 150 180 210 240 Month

Figure 8: Water Production Response to N2 Flooding

12 SCOTT REEVES; LARRY PEKOT SPE 71090

1000

CO2 @ 500 Mcfd

CO2 @ 250 Mcfd

100 Gas Rate, Mscfd Base Case

10 0 30 60 90 120 150 180 210 240 Month

Figure 9: Gas Production Response to CO2 Flooding

1000

CO2 @ 250 Mcfd

CO2 @ 500 Mcfd 100 Water Rate, Bpd

10

Base Case

1 0 30 60 90 120 150 180 210 240 Month

Figure 10: Water Production Response to CO2 Flooding

SPE 71090 ADVANCED RESERVOIR MODELING IN DESORPTIO N-CONTROLLED RESERVOIRS 13

L A P L AT A C O . A R CH U L ET A C O .

D u r an g o P a g o s a F l o r i d a R i v e r S p r i n g s P l a n t

T i f f a n y U n i t San Juan CO A L S i t e Basin Outline F A CO L OR A DO I R NE W ME X I C O W A D u l c e Y A l l i s o n U n i t

A z t e c

F a r m i n g t o n B l o o m f i e l d

K A O 9 91 0 0 .C D R Figure 11: Location of N2/CO2-ECBM Pilots, San Juan Basin

Tiffany Unit Production

100000 Nitrogen Injection Suspended

Gas Production Water Production Nitrogen Injection 10000

1000

Rate, Mcfd or Bpd 100

10

Water Measurement Discrepency

1 May-90 Sep-91 Jan-93 Jun-94 Oct-95 Mar-97 Jul-98 Dec-99 Apr-01 Sep-02

Figure 12: Tiffany Unit N2 – ECBM Performance

14 SCOTT REEVES; LARRY PEKOT SPE 71090

Allison Unit Production

10000000

CO 2 Injection Resumed (Jul, 1996) CO2 Injection Suspended, Five Wells Re-Opened (Nov, 1995)

1000000 Begin CO 2 Injection, Five Wells Shut-In (May, 1995)

Line Pressures Reduced 100000

10000

Monthly Gas, CO2 or Water, Mcf Bbls Gas 1000 Water CO2 Injection

Five Wells Reworked

100 Aug-87 Dec-88 May-90 Sep-91 Jan-93 Jun-94 Oct-95 Mar-97 Jul-98 Dec-99 Apr-01 Sep-02

Figure 13: Allison Unit CO2-ECBM Performance