0702

RESERVOIR SIMULATION AND INTERPRETATION OF THE

RECOPOL ECBM PILOT IN POLAND

W.F.C. van Wageningen and J.G. Maas

Shell International Exploration & Production

Kesslerpark 1, 2288 GS Rijswijk, The Netherlands

ABSTRACT

The RECOPOL project is a joint industry project (JIP) and the RECOPOL site is located in the west central Upper

Silesian basin in the South of Poland near the Czech border. The pilot area consists of a small fault-block, which

has a triangular shape. The deposits in the block dip 12 degrees to the north and consist of alternating layers of

sandstone, clay and coal. The main objective of the RECOPOL project is to demonstrate that CO2 injection in coal

is a feasible option under European conditions and that CO2 storage in coal layers is a safe and permanent

solution.

The RECOPOL pilot was simulated with Shell's proprietary simulator MoReS and the results were compared

to the existing field data. The reservoir simulations have been conducted to obtain a better understanding of the

field behavior in the RECOPOL pilot. Buoyancy proves to be important for the transport of methane in a coal-bed.

Because of the high cleat permeability and because vertical cleat permeability is likely of the same order as

horizontal cleat permeability, the gas and water segregate in the cleat system due to buoyancy. A small grid in the

vertical direction was necessary to model this effect. Because gas accumulates at the top of the coal layer, gas

may escape to surface if there is no sealing cap-rock. Therefore, cap-rock integrity is important for both CBM and

ECBM.

The RECOPOL pilot shows that CO2 injection enhances the production of methane in two ways: (1) CO

2

enhances de-sorption of methane and (2) The injected CO2 pushes methane towards the producer (CO

2 drive). In

RECOPOL, the CO2 adsorption was very slow, which likely enhanced the drive effect. An unexpected early

breakthrough of CO2 was also observed in the RECOPOL pilot, which was likely caused by CO

2 overshooting the

water in the cleats. Thus, it is likely hat the lower part of the coal seam never came into contact with the CO2. As a

result, there was less CO2 sequestration and enhanced methane recovery than anticipated. The RECOPOL pilot

helped to identify two important mechanisms: slow matrix diffusion and phase segregation in the cleats. These

mechanisms are relevant for both CBM and ECBM.

INTRODUCTION

Today, coalbed methane (CBM) is a mature technology. In 2005, the annual CBM production in the US was 1.7

Tscf and the proved reserves were 20 Tscf (source: EIA-DOE). The CBM process can be enhanced by injecting

CO2 into the coalbed, so called enhanced coalbed methane (ECBM). ECBM has a double goal: the enhanced

production of methane and the storage of CO2 in coal beds. Contrary to CBM, ECBM is far from mature as is

pointed out by White [1] in his review on the status of ECBM technology.

Worldwide a few pilots on ECBM have been conducted. In 1996, Burlington Resources started the first large-

scale (4 injectors, 7 producers) CO2 -ECBM pilot in the San Juan basin, [2-4]. The pilot was a technical success.

The methane recovery was enhanced and at the same time CO2 was sequestered. However, without additional

credits for CO2 storage the economics are poor. Outside the US only a few other ECBM pilots have been

conducted. The Alberta Research Council (ARC) has conducted a single well (huff and puff) ECBM micro-pilot with

RESERVOIR SIMULATION AND INTERPRETATION OF THE RECOPOL ECBM PILOT IN POLAND 2

flue gas and CO2 in Alberta, Canada. ARC also has completed a similar pilot in the Shanxi province in China

together with the Chinese State company CUCBM, [5]. The Japanese company JCOAL has an active (two-wells)

ECBM pilot in Hokkaido Japan. The study presented here will present the modeling and interpretation of the

RECOPOL (two-wells) ECBM pilot in Poland, [6].

The RECOPOL project is a joint industry project (JIP) funded by the European Union. The RECOPOL pilot is

lead by TNO and takes place in the upper Silesian basin in Poland. The main objective of the RECOPOL project is

to demonstrate that CO2 injection in coal is a feasible option under European conditions and that CO

2 storage in

coal layers is a safe and permanent solution. Although CO2 injection ended June 2005, the pilot is still on going.

Currently, the focus is on monitoring and verification.

Shell has joined the project in 2002 as an end-user and has participated actively in the planning

of the field activities and all project meetings. The successful frac-job in April 2005 was designed and supervised

by Shell experts and resulted in the first significant CO2 injection of the project, which turned the pilot into a

success. Today a new consortium has been formed and the RECOPOL pilot continues under a different name:

MOVECBM.

GEOLOGY AND WELL LOCATIONS

The RECOPOL test site is located in the west central Upper Silesian basin in the South of Poland (Fig. 1) near the

Czech border and falls under the concession area of a Silesian mine. The pilot area consists of a small fault-block,

which has a triangular shape. The deposits in the block dip 12o to the north and consist of alternating layers of

sandstone, clay and coal. A detailed description of the geology can be found in [6]. A small summary of some key

features is given below.

The Upper Silesian coal basin is bound by the Carpathian fore-deep and is structurally complex compared to

commercial CBM basins in the US. Faults cut into the underlying coal seams and have destroyed the lateral

continuity of the coal seams. Like most other European basins, the Upper Silesian basin underwent several

subsequent burial and uplift phases. Because the coal seams were buried deeper in the past than at present

times, permeability is relatively low, in the range of 0.5 to 2 mD.

On the test site, there are two old wells (MS-1 and MS-4), which were used for a CBM production pilot in

the period 1995-1997. The distance between MS-1 and MS-4 is about 375m, where MS-4 is located up dip from

MS-1. Initially, both MS-1 and MS-4 were considered for the ECBM pilot. However, reservoir simulations

conducted by TNO, [6] indicated that the distance between MS-1 and MS-4 was too large to achieve breakthrough

of CO2 within the project lifetime. Therefore, a new well (MS-3) was drilled between MS-1 and MS-4. During the

pilot CO2 was injected in the new well MS-3 and production took place in MS-4. The MS-1 well was not used during

the project.

Several coal layers having thicknesses in the range of 1 to 3m are located in the Upper Carboniferous (Fig. 2),

which are covered by Miocene shale. It was planned to inject CO2 in three coal layers (364, 401 and 405), but

there were indications that CO2 only entered into the top layer 364. Six layers (357-510) were completed in the old

production wells MS-1 and MS-4, whereas the new injection well MS-3 was completed in three layers (364, 401

and 405). The bottom layers 501 and 510 of MS-4 were plugged off with a bridge plug. Table 1 shows an overview

of the completions, coal seam thickness and depth of each well. Note that initially the completions of the newly

drilled well MS-3 were not fracced, which resulted in poor injectivity. A successful frac-job of MS-3 was completed

in April 2005, which resolved the injectivity problems.

It can be observed (Tab. 1) that the producer MS-4 was only fracced in three layers (364, 405 and 510). A

frac-job is usually necessary to connect the well to the cleat system and proppants are often used to keep the

cleats open. Several layers of MS-4 were not fracced so it can be assumed that these layers did not contribute to

the production. Furthermore, well tests in MS-3 had indicated that layer 405 has almost no permeability (~1 µD).

Layer 357 and 401 were not fracced and the lower layers (501 and 510) were plugged off. This makes it likely that

the only significant contribution to CBM production came from layer 364. The low production rates (water and gas)

observed in the RECOPOL pilot support this assumption.

THEORY: TRANSPORT, EOS AND SORPTION MODEL

The basic transport mechanisms of ECBM considered in our modeling are:

• Bulk transport of the gases through the natural fracture or cleat system (Darcy flow) including buoyancy

effects.

W.F.C. VAN WAGENINGEN AND J.G. MAAS 3

• Cleat to matrix transport by diffusion and vice versa.

• Exchange of methane and CO2 inside the coal matrix (adsorption/desorption)

The main part of the flow will take place through the vertical (unstressed) cleats. The horizontal cleats are mainly

closed due to the overburden pressure. The vertical and horizontal cleat permeability are likely of the same order,

because the cleats are connected and extend in the vertical direction. The permeability of a cleat can be estimated

by considering the flow through a cleat assuming it is similar to the flow between two parallel plates. Note that in

reality the flow through a cleat system deviates from the (ideal) flow through two parallel plates due to obstacles or

other imperfections of the cleat system, [7]. The solution of this flow is the well-known Poiseuille profile and can be

adequately modeled with the Darcy equation.

The permeability of the coal matrix is very small, which limits Darcy flow from matrix to cleat and vice versa.

The main transport mechanism between matrix and cleat is therefore diffusion. We model the diffusion with Fick's

law:

(1)

where Φm is the mass flow-rate, D is the diffusion coefficient, ∂c/∂x the concentration gradient and A the cross

sectional area through which the mass transfer takes place. In this study the diffusion coefficient sets the transport

rates of the different gases between matrix and cleat. In reality, the diffusion process between matrix and cleat is

very complex. Furthermore, the water content has a large influence on the diffusion rate, especially at the smallest

pore scale. In the literature, more detailed (bidisperse diffusion) models based on the work of Ruckenstein [8] are

currently used, e.g. [9]. We believe that currently there are too many unknowns to justify a more detailed diffusion

model. More accurate experimental data is necessary to have a solid basis for the more advanced models. Hence,

a simple empirical approach is our preferred option and used in this study.

We used the Shell Modified and Improved Redlich-Kwong equation of state (SMIRK-EOS), [10].The SMIRK-

EOS is a two parameter Redlich-Kwong type EOS, in which both parameters, a and b, are a function of

temperature T. The SMIRK-EOS is equal to

(2)

where R is the gas constant and v the specific volume.

The sorption process is modeled by means of the (extended) Langmuir equation, which describes the

amount of adsorbed gas of each component, Gcmp

as a function of pressure p and composition:

(3)

where VL,cmp

and pL are the Langmuir volume and pressure, y

cmp is the mol fraction of a component in the gas phase,

and wa and w

we are the ash and moisture content, respectively. In practice, both V

L and p

L are used to fit

experimental sorption data to the Langmuir isotherm. The Langmuir isotherm is modeled in Shell's proprietary

reservoir simulator, MoReS [11], via chemical reaction modeling, where it is assumed that adsorption is in

equilibrium with desorption for a given p and T. The chemical equilibrium is effectively equal to Eq. 3.

Fort the simulation of the RECOPOL pilot no swelling model was used. Instead, field data and the outcome of

reservoir simulations were compared and analyzed for evidence of swelling (e.g. change in the injection pressure

during the CO2 injection).

FIELD OPERATIONS

A graphic overview of the historic injection and production data is shown in the figures 4 and 5, respectively. The

total amount of injected CO2 was 320 10

3 m

3, the cumulative water production was 580 m

3, the total produced CH4

was 16.8 103 m

3 and the back-produced CO

2 was 22.6 10

3 m

3. It follows that 93% of the injected CO

2 was stored in

the coal seam and 7% was back-produced. The main events of the pilot are summarized below:

• Begin CBM operation (day 0; 28 May 2004)

• Begin CO2 injection (day 70; 6 Aug 2004)

RESERVOIR SIMULATION AND INTERPRETATION OF THE RECOPOL ECBM PILOT IN POLAND 4

• Observed increase in methane production (day 90; 26 Aug 2004)

• Pump frozen (day 257-263; 8-14 Feb 2005)

• Fall off test injector (day 263-278; 14 Feb - 1 Mar 2005)

• Observed decrease in methane production (day 272; 23 Feb 2005)

• First (unsuccessful) Frac-jobs with brine only (day 278-279; 2-3 Mar 2005)

• Injection resumed but at very low rate (day 279; 2 Mar 2005)

• Successful Frac-Job with proppant (day 327; 19 Apr 2005)

• Dramatic increase in CO2 injection (day 328; 20 Apr 2005)

• Big increase in methane production (day 334; 26 Apr 2005)

• Significant CO2 breakthrough in producer (day 336; 28 Apr 2005)

• Pump broken (day 352; 14 May 2005)

RESERVOIR MODEL

The reservoir is located at a depth of 1000m and dipping 12~degrees north. The reservoir model represents a

triangular area of 1.35 km2 (333 acre) and a single coal seam (364) having a thickness of about 3m is modeled.

The GIIP of the model area equals 9.1 Mm3 (0.32 Bscf). A dual-porosity dual-permeability model has been used for

this study. We assume that the matrix is in Langmuir equilibrium and that there is segregated flow in the cleat

system and that the relative permeability depends linearly on the saturation (Sat=0, Kr=0; Sat=1, Kr=1). Transport

between matrix and cleats is dominated by diffusion.

The permeability was determined by laboratory experiments, well tests and history match. All methods indicate

an average effective permeability of the reservoir in the range of 1 to 2 mD, except for the bottom coal layer (405),

which has a much lower permeability of about 1 µD. It is therefore not likely that CO2 entered into this layer. The

cleat porosity is around 0.5% and the cleat spacing is 0.025m. Together with a permeability of 1.3 mD, these

values gave a best match for the water production and CO2 injection/breakthrough.

The permeability is the effective permeability of both cleats and matrix. The cleat porosity was in the order of

0.5% so the cleat permeability equals 1.3/0.005 =260mD. Due to the large density difference (ρH2O

/ρCH4

~ 15), the

relatively low radial pressure gradient and the large cleat permeability, it is likely that the desorbed CH4 and water

will segregate in the cleats. As a consequence, CH4 will accumulate near the top of the coal seam, which brings

about the importance of a sealing cap-rock for CBM in order to prevent the flow of methane to surface. The

segregation of gas and water was confirmed with simulations with a radial flow model (Fig. 6).

To accurately describe the phase segregation, the 3 meter thick coal-seam was modeled with 10 grid blocks in

the vertical direction, which were refined (factor 1.5) towards the top of the seam (Tab. 2). The full field model has

a [7x5x10] grid with local grid refinement around the wells (Fig. 7). If the grid block of the well would be taken too

large, the matrix near the well would depressurize by a too large extent. This would result in more desorption of

methane and hence production would be over-predicted. This issue is addressed in MoReS via local grid

refinement near the wells (Tab. 2) and (Fig. 7). The radial model was also used to determine the correct level of

grid refinement in the X and Y direction. For that, the radial model was compared to the full field model, which has

larger grid blocks in the horizontal plane. There was good agreement between the two models indicating that the

grid refinement in the X-Y direction was sufficient (Fig. 8).

It is known that CO2 dissolves in water. The solubility proved to be small. At the most 3% of the injected CO

2

dissolves into the brine. To keep the model simple, this effect was neglected in this study.

DIFFUSION AND SORPTION PARAMETERS

Busch et al [12] have investigated the diffusion process on coal particles. They have indicated that there are two

sorption processes in coal, a fast and a slow process. The results were obtained by fitting the sorption data with

two first order reactions representing the fast and slow sorption process. Van Krevelen [13] describes the diffusion

processes in coal at different scales. The fast and slow diffusion process can be linked to the different pore scales

of coal.

The measurements of [12] show that for large particles (>0.3mm) about 65% of the CH4 sorption is fast and

35% is slow. Smaller particles (<0.3mm) show a trend of increasing percentage of fast sorption with decreasing

particle size. Particles larger than 0.3 mm all show about the same small percentage of fast sorption indicating that

large particles could represent the coal matrix. The observation that there are two diffusion processes is in line with

W.F.C. VAN WAGENINGEN AND J.G. MAAS 5

the observed behavior in the desorption analysis on cores to determine the gas content. After 100 days only 60%

of the gas was de-sorbed. The other 40% remained trapped in the coal matrix.

It appears that at least 35% of the gas in the RECOPOL is diffusing at a very slow rate and as a result will not

contribute much to the production. This effect is incorporated in the simulator by reducing the Langmuir volumes.

In order to obtain a good history match the Langmuir parameters had to reduced by a factor of 6.5, which can

indicate that the amount of slow diffusion is larger under field conditions.

There are also other explanations possible for the low gas rates observed in the field such as a higher ash and

moisture content or reduced thickness of the coal seam due to shale/sandstones. In other words there are different

ways to explain the field data. However, the main principle remains the same: there are only a limited amount of

sorption sites available or accessible for methane and carbon dioxide.

In the reservoir simulations, the diffusion coefficient was set to 10-9 m

2/s, which represents a fast diffusion

process. The effect of the slow diffusion process was not modeled; by adjusting the Langmuir parameters the

amount of coal accessible by fast diffusion has been reduced to 15%. The physics behind this is that part of the

coal is not accessible, because diffusion into the coal matrix is very slow. The adsorption/desorption process is

described by the Langmuir relation (Eq. 3). The coal parameters as obtained in the lab are: pL (CH

4) = 2490 kPa,

pL (CO

2) = 2300 kPa, V

L (CH

4) = 17 sm

3/ton and V

L (CO

2) = 17 sm

3/ton. The average ash and moisture content are

25% and 5%, respectively. The plot of the gas content as function of pressure is shown in figure 9. The vertical line

at 100 bar indicates the field gas content as measured from cores and cuttings.

RESULTS AND DISCUSSION

The pilot's aim was to prove the feasibility of CO2 sequestration in coal. It was not aimed at economical production

of methane, which indeed was marginal. The main reasons for the low methane production rates are:

• the adsorption of CO2 and desorption of CH

4 are slow in Silesia coal

• the gas content of the coal is low

• the coal seam is thin and has a relatively high ash content

Reservoir simulations have been conducted to get a better understanding of the field behavior in the RECOPOL

pilot. We used a simple geological model considering only one coal seam. To accurately simulate the segregation

of the phases in the cleats, the model has a relatively dense grid in the vertical direction that was refined toward

the top of the coal seam. The grid blocks around the wells were also refined in the x and direction in order to get a

more accurate pressure distribution and sorption behavior around the wells. The level of grid refinement was

based upon a 2D radial model, which was also used to determine the CBM baseline. The CBM baselines of the full

field and 2D radial models are in good agreement with each other (Fig. 8). This shows that the level of grid-

refinement is adequate.

The CBM baseline was established as follows. First, the water rate of the model was matched by varying the

permeability and cleat porosity. Subsequently, the Langmuir parameters were decreased with a constant factor to

obtain the CBM baseline. The main differences between the final model and field were:

• The CBM baseline of the model is somewhat too high

• The decline of the water-production of the model is somewhat too slow

No further adjustments have been made to the model to get a better match. This keeps the model simple. Note

that the aim was not to get a perfect match, but to understand main features of the field behavior.

The base (CBM) model (Fig. 8) was used to study the effect of CO2 injection (Fig. 10). The CO

2 injection of the

model was set to the field rate (all other parameters were left unchanged) and its response on the production was

studied. The response of the methane production to the CO2 injection was similar to the field (Fig. 10). Shortly after

the injection, the methane production started to increase although to a lesser extent than the field rate. The field

reached a higher ECBM plateau than the model, but we obtained the correct response. Previous models not

considering the phase segregation all failed to predict the early enhancement of methane due to CO2 injection.

This demonstrates that phase segregation most likely takes place in the coal seam and that small grid cells near

top of coal layer are necessary to describe this effect.

The new model shows that early enhancement of CH4 can be the result of CO

2 flooding. The injected CO

2

pushes the methane that has accumulated at the top of the coal seam toward the producer. Note that the break-

through of the CO2 occurred too early in the model (Fig. 11), which was likely caused by the over-prediction of

dispersion of CO2. In a grid-block, CO

2 and CH

4 mix instantly, while in reality there will be a delay due to diffusion.

We summarize the main differences between the field data and simulation below:

RESERVOIR SIMULATION AND INTERPRETATION OF THE RECOPOL ECBM PILOT IN POLAND 6

• The enhancement of methane at the start of the CO2 injection is under-predicted leading to a lower

(simulated) ECBM plateau.

• The break-down of the pump was not modeled

• The decline in methane production due to the decrease in injection rate in the period just before the frac-

job is too short in the simulation result. This can be related to the breakdown of the pump, which was not

modeled.

• The break-through of CO2 is too fast and the amount of CO

2 production after the frac-job is under-

predicted

The main shortcomings of the current model are believed to be:

• There is too much adsorption of CO2 to the coal

• The dispersion of CO2 is too fast (numerical effect)

These issues are currently investigated and addressed in a new model, which will contain a better description of

the diffusion processes at different scales (micro-pore and macro-pore diffusion). The influence of the amount

adsorption of CO2 was studied by investigating the sensitivity of the Langmuir parameters of CO

2 (Fig. 12-13). The

Langmuir parameters of CO2 were changed and it was found that when the amount of adsorption of CO

2 was

decreased, the enhancement of methane production was closer to the field rate. However, the decline of the

methane rate before the frac-job was still too slow. The post-frac CO2 production was closer to the field rate, but it

was still lower. The fact that the simulated rates were closer to the historic rates, can be an indication that less

adsorption of CO2 took place in the field test.

Although there is no direct evidence of coal swelling in the RECOPOL pilot, it appeared that the cleat

permeability reduced after the post-frac CO2 injection. The pressure decline of the field was much slower than the

pressure decline of the simulation (Fig. 14, day 375 - day 550). This indicates reduced cleat permeability due to

coal swelling. Moreover, after a shut-in period of 4 months the injector was still over-pressured (THP ~50 bar).

In the next phase of the pilot (MOVECBM), it is planned to produce back the (former) injector. If there was

indeed a strong reduction in permeability, it is expected that this effect can also be observed in the amount of

produced water. We also hope to establish the amount of CO2 that is physically adsorbed to the coal and the

amount remaining in the cleats.

CONCLUSION

Buoyancy, which is often neglected in (E)CBM simulations, proves to be important for the transport of methane in a

coal-bed. Because of the high cleat permeability and since vertical permeability is most likely of the same order as

horizontal permeability, the gas and water segregate in the cleat system due to buoyancy. The gas accumulates at

the top of the coal layer, which brings about that the cap-rock integrity is important for both CBM and ECBM.

The RECOPOL pilot showed that CO2 injection enhances the production of methane. We found that methane

is not only enhanced by the sorption mechanism, but also due to CO2 flooding. The injected CO

2 pushes the

methane that has accumulated near the top of the coal layer toward the producer. Due to slow diffusion into the

coal matrix, there was less adsorption of CO2 , which likely enhanced the flooding effect. Furthermore, the CO

2

breakthrough occurred much faster than anticipated earlier.

Two likely causes for the early breakthrough observed in both pilot and reservoir simulations are that the bulk

of the CO2 overshoots the water (Fig. 15) and that due to slow diffusion there is less CO

2 adsorption to the coal

matrix. We expected that more CO2 will adsorb to the coal and more CH

4 will desorb, if diffusion is faster. This will

lead to a later breakthrough of CO2. Faster diffusion is not expected to have any effect on the overshooting of the

CO2. When the matrix is saturated, CO

2 will still migrate to the top of the coal seam. The only way to prevent the

overshooting of CO2 is by lowering the water level in the coal seam.

Because the CO2 overshoots the water in the cleats, it is likely that in RECOPOL a mayor part of the coal

seam never got into contact with the CO2. As a result, there will be less CO

2 sequestration and enhanced methane

recovery. Furthermore, it indicates the importance of injector location and design of an ECBM operation. Ideally,

the CO2 should be injected in the bottom part of the coal seam to maximize exposure of the coal to CO

2. Multi-

laterals may be used to achieve this.

The RECOPOL pilot helped to identify two important mechanisms (diffusion and phase segregation in the

cleats) relevant for both CBM and ECBM. This novel insight can help find ways to enhance diffusion into coal

matrix (e.g. dewatering prior to injection) and to optimize the location of injectors and producers (e.g. using multi-

laterals).

W.F.C. VAN WAGENINGEN AND J.G. MAAS 7

ACKNOWLEDGEMENTS

The authors would like to acknowledge Henk Pagnier and Frank van Bergen from TNO for their efforts in making

the RECOPOL project a success.

REFERENCES

1. White, C.M ,et al., 2005: "Sequestration of Carbon Dioxide in Coal with Enhanced Coal bed Methane

Recovery-A Review", Energy & Fuels, V.19, No. 3.

2. Stevens, S.H., Spector, D. and Riemer, P., 1998: "Enhanced Coal bed Methane Recovery Using CO2

Injection: Worldwide Resource and CO2 Sequestration Potential", SPE-48881.

3. Reeves, S.R. et al., 2004: "The Tiffany Unit N2 – ECBM Pilot: A Reservoir Modeling Study",

Topical Report DOE, DE-FC26-0NT40924.

4. Reeves, S.R. et al., 2002: Selected Field Practices for ECBM Recovery and CO2 Sequestration in Coals

based on Experience Gained at the Allison and Tiffany Units, San Juan Basin,Topical Report DOE, DE-FC26-

00NT40924

5. Wong, S., Law D. and Gunter B.,2005: "Enhanced Coal-Bed Methane Test at South Qinshui Basin, China",

Greenhouse Issues, V. 78.

6. Pagnier, H., et al., 2006: "Reduction of CO2 emission by means of CO

2 storage in coal seams in the Silesian

Coal Basin of Poland", TNO, RECOPOL Final Report.

7. LeGrain, P.H. 2006: "Etude de l'influence de la rugosite sur l'ecoulement de fluide dans les fissures

rocheuses, PhD-Thesis, Faculte Polytechnique de Mons.

8. Ruckenstein, A.S., et al., 1971: "Sorption by solids with bidisperse pore structures",Chem. Eng. Sc., V. 26.

9. Shi, J.G. and Durucan S. 2003: "A bidisperse pore diffusion model for methane displacement desorption in

coal by CO2 injection", Fuel, V. 82.

10. Drexhage, J.J. and Welsenes, A.J., 1990: "Physical properties of pure compounds. Parameters for the

SMIRK equation of state", Shell Internal report.

11. Por, G.J., Boerrigter, P., Maas, J.G. and De Vries, A., 1989: "A Fractured Reservoir Simulator Capable of

Modeling Block-Block Interaction", SPE-19807.

12. Busch, A. et al., 2004: "Methane and carbon dioxide adsorption-diffusion experiments on coal: upscaling

and modeling", Int. J. Coal Geol., V. 60.

13. Van Krevelen, D.W., 1993: Coal: Amsterdam, Elsevier Science Publishers B.V., ISBN 0-444-89586-8.

TABLES

Table 1: Coal layers and completion types

Table 2: Grid block sizes in X,Y and Z direction

RESERVOIR SIMULATION AND INTERPRETATION OF THE RECOPOL ECBM PILOT IN POLAND 8

FIGURES

Figure 1: Location of RECOPOL Pilot (source: TNO)

Figure 2: Coal layers and wells (source: TNO)

W.F.C. VAN WAGENINGEN AND J.G. MAAS 9

face cleat

butt cleat

top view

Coal matrix

horizontal cleat

vertical cleat

inaccessible matrix

(slow diffusion)

fast diffusion layer

Figure 3: Cleats in Coal matrix

Injection

Production

Injection

Figure 4:Historic CO2 injection and production data

RESERVOIR SIMULATION AND INTERPRETATION OF THE RECOPOL ECBM PILOT IN POLAND 10

CH4

H2O

Figure 5: Historic CH4 and water production data

Radial Model Gas Saturation

S=0.05

S=0.9

Figure 6: Gas saturation (Radial model ) on log scale

W.F.C. VAN WAGENINGEN AND J.G. MAAS 11

Top View Side View

Grid in between wells (Top View)

Figure 7: Grid [7x5x10] with local grid refinement near wells

CH4 (field)

CH4 (2D sim)

CH4 (3D sim)

H2O (field)

H2O (sim)

Figure 8: Comparison of 3D model and 2D radial model, water and CH4 production (CBM base case)

RESERVOIR SIMULATION AND INTERPRETATION OF THE RECOPOL ECBM PILOT IN POLAND 12

CO2

CH4

CH4 (field)

Figure 9: Adsorption characteristics for RECOPOL coal (DAF). The vertical line at 100 bar indicates the expected gas

content of the field.

H2O (field)

CH4 (CBM base case)

CH4 (field)

H2O (sim)

CH4 (ECBM base case)

Figure 10: Effect of CO2 injection on produced CH4 (ECBM base case)

W.F.C. VAN WAGENINGEN AND J.G. MAAS 13

CO2 (Field)

CO2 (sim)

Figure 11: Back-produced CO2 (ECBM base case)

CH4 (ECBM base case)

CH4 (CBM base case)

CH4 (ECBM VL(CO2) * 0.1)

CH4 (field)

H2O (field)

H2O (sim)

Figure 12: Water and Methane production (ECBM: VL (CO2) * 0.1)

RESERVOIR SIMULATION AND INTERPRETATION OF THE RECOPOL ECBM PILOT IN POLAND 14

CO2 (ECBM VL(CO2) * 0.1)

CO2 (ECBM base case)

CO2 (Field)

Figure 13: Back-produced CO2 (ECBM: VL (CO2) * 0.1)

BHP (sim)

BHP (field)

Figure 14: Injection pressure: comparison of simulated and field BHP

W.F.C. VAN WAGENINGEN AND J.G. MAAS 15

Figure 15: ECBM base case; Gas saturation and CO2 mole fraction in cleats (matrix is not shown) at day 375 (end of

the CO2 injection), Dark Gray = 0 and White = 1