HISTORY MATCHING A CO2 HUFF ‘N’ PUFF EOR PROJECT WITH CMG’S GEM AND CMOST

University ofNorth Dakota

Grand Forks

Charles D. Gorecki, * Damion J. Knudsen, Guoxian Liu, Jordan M. Bremer, James A. Sorensen, Edward N. Steadman, and John A. Harju

For more information, contact: Charles D. Gorecki, cgorecki@undeerc.org, (701) 777-5255; James A. Sorensen, jsorensen@undeerc.org, (701) 777-5287; Edward N. Steadman, esteadman@undeerc.org, (701) 777-5279; John A Harju, jharju@undeerc.org, (701) 777-5157

EERCEnergy & Environmental Research Center®

Putting Research into Practice

A CO2 huff ‘n’ puff (HNP) enhanced oil recovery (EOR) project was carried out in the

E. Goetz 1 well located in the Northwest McGregor field of Williams County, North Dakota,USA. The HNP is one of the Plains CO

2 Reduction (PCOR) Partnership Phase II field

validation tests where CO2 was injected into a fractured carbonate reservoir for the dual

purpose of CO2 EOR and CO

2 storage. Because of the high level of uncertainty in many of

the reservoir properties, part of this research explored methods of uncertainty identification and sensitivity analysis of geological properties by history-matching production and injection data with time-lapse seismic and wireline technologies.

Initially, homogeneous cases were produced for the purpose of determining approximate fracture and matrix properties. This method employed experimental design and sensitivity analysis using Computer Modeling Group’s GEM and CMOST software packages to determine the most probable sets of reservoir properties. Based on the results of the homogeneous model analysis, heterogeneous models were developed using geostatistical methods to generate different stochastic realizations of heterogeneous geological and fluid properties for history matching. As with the homogeneous cases, the heterogeneous models were fed into an uncertainty and sensitivity analysis to determine the heterogeneous fracture and matrix property “heavy hitters,” reducing the number of simulation factors that most affect injectivity and production while reducing history-matching simulation runs. In the end, a small number of heterogeneous cases were generated, which can be used to predict the incremental oil recovery potential and long-term CO

2 fate.

The PCOR Partnership is a collaborative program assessing regional CO2 storage

opportunities. Its primary sponsor is the U.S. Department of Energy National Energy Technology Laboratory, with additional support from its more than 80 partners.

DUAL POROSITY AND PERMEABILITY MODELING WORKFLOW

y = 0.0004e31.7230x y = 0.0014e35.2912x y = 0.0038e34.3836x y = 0.0003e11.8095x

0.0001

0.001

0.01

0.1

1

10

0.001 0.01 0.1 1

Kw

ate

r (m

D)

Effective Porosity (vol/vol)

Effective Porosity vs. Kwater NW McGregorMicro-Facies

Mudstone Packstone Grainstone Anhydrite

Samples from 8076’

kf = kd2/km

kf = fracture permeability (mD) kd = average permeability from DST (mD) km = core permeability (mD)

ke = km + Φf*kf

ke = effective permeability ~ DST permeability (mD)Φf = fracture porosity

kf = 84.4 x 105 x W3/ZΦf = W/Z x 100

Z = fracture spacing (cm)W = fracture width (cm)

MAT

RIX

(M)

FRA

CTU

RE (F

)

HISTORY MATCHING AND MONITORING, VERIFICATION AND ACCOUNTING SUMMARY

The primary goals of this PCOR Partnership pilot project included determining the short- and long-term fate of a small volume of injected CO

2 (440 tons) and

determining if a CO2 HNP in this type of reservoir could be effective. In order to

determine the fate of the injected CO2 and the potential incremental oil

production, time-lapse RST and VSP were used along with detailed static and numerical modeling. The VSP was an excellent tool for acquiring modeling parameters such as property semivariograms, near wellbore horizon vertical deviations, and production of covariables for petrophysical simulation. However,because of the highly fractured nature of the reservoir and the extremely low pore volume of the fractures, which contained most of the CO

2, there is a high

level of uncertainty as to whether or not the injected CO2 could be detected by

the repeat VSP survey. The baseline and repeat RST logs were found to be an excellent tool for determining the vertical extent of the injected CO

2 near the

wellbore, and the modeling and history-matching activities give good support for the overall extent of the injected CO

2 and will lead to better predictions of

future incremental oil recovery.

Selected Sources(1) Lucia, F. J. , 2007, Carbonate reservoir characterization, 2d Ed.: Springer

Publishing.(2) Djebbar, T., and Donaldson, E., 2004, Petrophysics, theory and practice of measuring reservoir rock and fluid transport properties, 2d Ed.: Gulf Publishing.

AcknowledgmentsThe authors would like to thank Eagle Operating, Computer Modeling Group,

Schlumberger, North Dakota Geological Survey, Praxair, and all PCOR Partnership partners for their support.

ABSTRACT

Step 1: A multimineral petrophysical analysis (MMPA) was com­pleted after well log normalization and synthetic curve creation for missing curves using neural networks.

Steps 2F and 3F: The fracture intensity was quantified with core, and the other fracture properties were populated using MMPA and pressure buildup analysis.

ship between microfacies, effective porosity, and permeability to water. This graph, along with the QEMSCAN analysis in Step 3M, helps verify the petrophysical analysis.

vertical seismic profiles to acquire spatial variability in porosity and absolute acoustic impedance (AAI).

porosity was used to acquire a better horizontal correlation for short- and long-range spatial variability.

Step 7M: Reservoir properties were populated using inverted VSP as a covariable for SGS. The smaller upscaled injection grid is shown in the center of the larger geostatistical grid.

Step 8M: Gas volume in the matrix was estimated by numerical simulation.

Step 8F: Gas volume in fractures was estimated by numerical simulation.

Step 7F: The DFN was then upscaled and populated with volumetric properties such as fracture porosity and perme­ability for input into the simulation grid.

Step 6F: Based on the volumetric grid of fracture intensity, DFN modeling was performed to produce discrete fractures.

Step 5F: Using synthetic logs of fracture intensity along with sequential Gaussian simulation (SGS), a fracture intensity volumetric grid was produced which was the input for discrete fracture network (DFN) modeling.

Step 4F: Fracture widths, spacing, permeabilities, and porosities were approximated using the above forumla.

which parameters most affect injection and production history matching. s GEM and CMOST. After hundreds of simulation runs and adjusting the heavy-hitter variables, a decent

match was obtained. More simulations will be run to adjust the final few months of history to form a better match for future incremental recovery estimates, as well as to predict long-term CO2 fate.

Step 2M and 3M: The MMPA was completed for all wells in the study area and results were verified with QEMSCAN analysis. The sample above was from approximately 8076’ measured depth.

Energy & Environmental Research Center, University of North Dakota, 15 North 23rd Street, Stop 9018, Grand Forks, North Dakota, 58202-9018

NW McGregor Oil Field

E. Goetz 1(NDIC 3392)

Range

Sem

ivar

ian

ce

Sill

SW_MATRIX(0.6, 0.99)

Step 4M: This graph helps illustrate the relation- Step 5M: Seismic inversions were performed on Step 6M: A semivariogram analysis of the inverted VSP

NETGROSS_MATRIX(0.22, 0.36) PERMIJ_FRACTURE(5.3378, 8.418)

POR_FRACTURE(0.004356, 0.006823)

SW_FRACTURE(0.6, 0.99)ROCKDEN_FRACTURE(164.41, 169.07)

DIJFRAC_CON(0.1992, 1.0361)

DKFRAC_CON(0.957, 3.877)PERMK_MATRIX(0.00429, 0.06233)

PERMK_FRACTURE(1.3878, 2.1887)

SWCON_MATRIX(0.31123, 0.74772)

Pa

ram

ete

rs

95th Percentile

(T1) (T2) (T3) (T1)

PERMIJ_MATRIX(0.00865, 0.04732) SWCRIT_MATRIX(0.31123, 0.74772)

SOIRW_MATRIX(0.01463, 0.08345)

POR_MATRIX(0.05089, 0.09429)ROCKDEN_MATRIX(164.41, 169.07)

SORW_MATRIX(0.01463, 0.08345)

0 5E+07 1E+08 1.5E+08 2E+08 (T2)Cumulative Gas (cubic feet)

SW_MATRIX(0.6, 0.99)PERMK_MATRIX(0.00429, 0.06233)

SWCRIT_MATRIX(0.31123, 0.74772)

SOIRW_MATRIX(0.01463, 0.08345)POR_MATRIX(0.05089, 0.09429)

ROCKDEN_FRACTURE(164.41, 169.07)

SORW_MATRIX(0.01463, 0.08345)PERMK_FRACTURE(1.3878, 2.1887)SWCON_MATRIX(0.31123, 0.74772)

POR_FRACTURE(0.004356, 0.006823) PERMIJ_MATRIX(0.00865, 0.04732)

ROCKDEN_MATRIX(164.41, 169.07)

SW_FRACTURE(0.6, 0.99)

Pa

ram

ete

rs

95th Percentile (T3)

PERMIJ_FRACTURE(5.3378, 8.418)

NETGROSS_MATRIX(0.22, 0.36) DIJFRAC_CON(0.1992, 1.0361)

DKFRAC_CON(0.957, 3.877)

Cumulative Oil (reservoir barrels)

0 100 200 300 400 500 600 700

Step 9: Uncertainty analysis was performed using CMG’s CMOST to estimate Steps 10 and 11: Time-lapse reservoir saturation tool log and production were matched using CMG’