Characterization and Simulation of the Panoma Field (Wolfcampian);

a Tight, Thin-Bedded Carbonate Reservoir System Southwest Kansas

Martin K. Dubois1, Alan P. Byrnes1, Shane C. Seals2, Randy Offenberger2, Louis P. Goldstein2, Geoffrey C. Bohling1, John H. Doveton1, and Timothy R. Carr1

1Kansas Geological Survey, 2 Pioneer Natural Resources USA, Inc.

Acknowledge support from Hugoton Asset Management Project industry partners:Pioneer Natural Resources USA, Inc.Anadarko Petroleum CorporationBP America Production CompanyConocoPhillips CompanyCimarex Energy Co.E.O.G. Resources Inc.OXY USA, Inc.W. B. Osborn

Presentation Outline

Overview, unique problems and lithofacies (Marty)

Petrophysical properties and relationships (Alan)

3D cellular model (Shane)

Initial simulations (Randy)

What’s next

Setting

Giant stratigraphic trap(s)Kansas Panoma 2.8 TCF Kansas Hugoton 24 TCF

Kansas Hugoton EUR 35-38 TCF(Olson, etal, 1996)Panoma EUR ???

Challenges and Key Points

Challenges:Automate & upscaleData volume (5200 sq miles, 2600 producers, 10,000+ wells)

Direct measurements of Sw by logs is problematic (must use property-based OGIP)

Facies representation in model is critical

Free water level varies and not documented Automate Automate

Volumetric OGIPAutomate

Some key points:1. Thinly layered reservoir, moderate to low-

crossflow between zones (pressure data indicates differential depletion)

2. Matrix properties drive the system and thin high perm layers may control flow

And one more challenge:Material balance GIP is problematic due to lack good pressure data by zone.

SIP for commingled production is available, but this represents the lowest possible SIP of the most permeable of the commingled zones.

Why Model These Mature Reservoirs?Panoma SIP vs Cum Gas

(1968-2002)

0

50

100

150

200

1 2 3Cum Gas (TCF)

Avg

. SIP

(psi

g)

(Discovered 1958)

SIP is actually pressure of highest perm zone in commingled production

Hugoton SIP vs Cum Gas(1968-2002)

0

100

200

300

400

10 15 20 25 30

Cum Gas (TCF)

Avg

SIP

(psi

g)Goals:

1. Functional comprehensive geologic and engineering models for simulation and reservoir management

2. Resolve zonal differential depletion questions

3. Resolve question of continuity between two reservoir systems that are regulated separately

Seven Sequences, Eight Lithofacies

Gas production from upper seven marine-nonmarine sequences of Council Grove

• Depth (top) 2400-2800’• 300 feet below Chase

~ 50% Marine, ~ 50% Nonmarine

Lithofacies DistributionCouncil Grove, Panoma Field

26%

23%12%4%

8%7%16%

4%

NM Silt & Sd

NM Shly Silt

Mar Shale & Silt

Mudstone

Wackestone

Dolomite

Packestone

Grnst & PA Baf

(For A1 - B5)20% are bulk of pay

(Keystone wells)

Council Grove

Lithofacies

Nonmarine Shaly Siltstone

Core Slab

Cm

0.5 mm

T h i n S e c t i o n Photomicrograph

4.6%0.000024 md

Close-up Core Slab

0.5 mm

Thin Section Photomicrograph

M-CG Oncoid-PeloidGrainstone

Cm

21.2%32.3 md

Phyloid Algal Bafflestone

20.6%1141 md

Core Slab

Silty WackestoneCore Slab3.4%

0.0024 md

0.5 mm

Thin Section Photomicrograph

0.5 mm

Thin Section Photomicrograph

Close-up Core Slab

Dolomite13.9%1.1 md

Cm

Lithofacies classes and their depositional environment

Facies Stacking Patterns• Migrating facies belts

response to rapid glacial-eustatic SL fluctuation

• Facies vertically stacked in predictable manner

• Sequences bounded by exposure surfaces

• Facies log response predictable

• M-NM and Rel-Pos Geologic Constraining Variables helpful

Predict lithofacies at well scale with Neural networks• Select e-log predictor variables and

develop geologic constraining variables

• Train N-Nets on core lithofacies• Run N-Net models on 500 wells

and output facies curves in LAS format (automated, batch process)

• Import lithofacies curves files into geologic applications

Measuring error in test set predictions Core Lithofacies 6-8 and Predicted Lithofacies 6-8(Used PE when available)

0

50

100

150

2 6 10 14 18 22 26 30

X-Plot Log Porosity %

Freq

uenc

y

Core LithPred Lith

Porosity Distribution in Pay Lithofacies

Predicted Lithofacies “Scorecard” (Counts)

(Nnet mod1, no RelPos, w/ PE)

Effectiveness MetricsAbsolute accuracy

Accuracy within one facies

Proportional representation

Porosity (and perm) distribution in facies populations

Panoma Stratigraphic X-Sec

Neural network predicted lithofacies, 5 log variables, 1 geologic constraining variable

Property-based OGIPOGIP (property-based) is good early test of facies predicition, k-phi-Sw transforms and Phi correction.

OGIP = f (Sw,P, T, Z, Phi)

Sw = f (facies unique properties, Phi, FWL)

Automated; generate OGIP well curve in LAS format

Vol OGIPa & Cum Production Keystone Wells, Council Grove

0

1

2

3

4

ALEXANDER

BEATYKIM

ZEYNEWBY

LUKE

GU4SHANKLESHRIM

PLINSTUART

BC

FG

as

CumGasVol OGIPa

Using "best guess" FWL based on anecdotal data (perforations, tests, lowest producingperfs) and Pippin (1985). Cum gas is ~ 80% ERU.

Modified after Pippin (1985)

Generalized Field X-Section

Panoma

Hugoton

Gas-Water

Variable Free Water Level

Estimating free water level

Ratio OGIP50 / Cum Gas“Fault” map overlay

Ratio is relative to FWL

FWL can be back-calculated by est. OGIPmb and solving for FWL.

Related to minor faulting?

Cum Gas by Section 2002

Porosity HistogramNM Clastics

0%

5%

10%

15%

20%

25%

2 4 6 8 10 12 14 16 18 20 22 24Porosity (%)

Perc

ent o

f Pop

ulat

ions

NM Silt & Sand L1NM Shaly Silt L2

Porosity HistogramLimestones

0%

5%

10%

15%

20%

25%

30%

2 4 6 8 10 12 14 16 18 20 22 24Porosity (%)

Perc

ent o

f Pop

ulat

ions

Mdst/Mdst-Wkst L4Wkst/Wkst-Pkst L5Pkst/Pkst-Grnst L7Grst/Grst-Baff L8

Density log calibration

Porosity

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 100Wetting Phase Saturation at 400' Above Free Water (%)

Wet

ting

Phas

e Sa

tura

tion

at H

eigh

t Abo

ve

Free

Wat

er L

evel

(%)

80 ft 2.18200 ft 1.39300 ft 1.15400 ft 1.00500 ft 0.91600 ft 0.84800 ft 0.74

Packstone Bafflestone

Permeability

logki=0.0588 logka3-0.187 logka2+1.154 logka-0.159logki=0.0588 logka3-0.187 logka2+1.154 logka-0.159

Mudstones-Bafflestones

0.00001

0.0001

0.001

0.01

0.1

1

10

100

0 2 4 6 8 10 12 14 16 18 20 22In situ Porosity (%)

In s

itu K

l. Pe

rmea

bilit

y (m

d)

Allmudstonemud-wackestonewackestonewacke-packstonepackstonepack-grainstonegrainstonebafflestone8

Mudstone-Wackestone Wackestone0.0001

0.001

0.01

0.1

1

0.0 0.2 0.4 0.6 0.8 1.0

Water Saturation (fraction)

Rel

ativ

e Pe

rmea

bilit

y (fr

actio

n)

w -10 mdg-10 mdw -1 mdg-1 mdg-0.1 mdw -0.1 mdg-0.01 mdw -0.01 mdg-0.001 mdw -0.001 md

Capillary Pressure300’ Above Free Water

Capillary Pressure Curves Pkst/Pkst-Grainstone(Porosity = 4-18%)

10

100

1000

0 10 20 30 40 50 60 70 80 90 100Water Saturation (%)

Gas

-Brin

e H

eigh

t Abo

ve F

ree

Wat

er (f

t)

Porosity=4%Porosity=6%Porosity=8%Porosity=10%Porosity=12%Porosity=14%Porosity=16%Porosity=18%

Gas and Water Relative Permeability

0.001

0.01

0.1

1

0 20 40 60 80 100Water Saturation (%)

Rel

ativ

e Pe

rmea

bilit

y (fr

actio

n)

8.08 md1.09 md0.72 md0.201md0.188md2907': 8 08 md

0.0001

0.001

0.01

0.1

1

0.0 0.2 0.4 0.6 0.8 1.0

Water Saturation (fraction)

Rel

ativ

e Pe

rmea

bilit

y (fr

actio

n)

w -10 mdg-10 mdw -1 mdg-1 mdg-0.1 mdw -0.1 mdg-0.01 mdw -0.01 mdg-0.001 mdw -0.001 md

Building the Structural Framework

Building the Structural FrameworkDefine grid increment and area of

interestConstruct top horizon for Council

Grove (A1_SH) top Create isochore for each zone, and

hang isochores from top horizonGenerate layers (define cell thickness)

Import wells with tops and logs11,367 total wells, 527 of which

contain log data10,840 wells with tops only (no logs)527 wells with tops, predicted facies

curves, “probability” curves, and porosity curves (Facies from two Nnet models, 352 with PE and 175 without PE)

Model Architecture

SH LM "Dummy" TotalA1 23 41 12 76B1 19 16 12 47B2 12 15 12 39B3 20 15 12 47B4 17 18 12 47B5 8 34 12 54C 28 61 12 101

Layers per ModelCells in modelXY = 1000 X 1000 feet

5,200 square mile model

7 Models (one per cycle)

Average model 8.6 million

Maximum 15 million (C cycle)

Minimum 5.7 million (B2 cycle)

Architecture QC

Structural Cross Section Grid

Panoma Facies Modeling

Facies ModelUp-scaled predicted facies to fit layeringBiased facies trends based on what we

know about the geology of the systemPopulated cells in between wells

(Sequential Gaussian Indicator)

Facies “Biasing”Non-biased

A1-LM Grnst-PA)

2:1 Trend Biased(NE-SW)

A1-LM Grnst-PA)

Heavy Trend Biased(NE-SW)

A1-LM Grnst-PA)

Facies Distribution Biased

B1-LM (Grnst-PAbaf)

Panoma Petrophysical Modeling

.0001

100101.1

.01.001

Petrophysical Models

Up-scaled porosity curve to fit layering and generated porosity model

Used perm facies transforms and porosity values in cells to generate permeability model

LithPhiNDUpscalePhiND Perm

Permeability in Facies 8, A1_LM

PermeabilityA1_LM

327 Layers

Upscale to Dynamic Model

Initial Simulation; single well

0.4 0.8 1.2 1.6 2.00

60

120

180

240

300ALEXANDER D2 - PANOMA

BHP/ZFlowing

0.4 0.8 1.2 1.6 2.00

60

120

180

240

300ALEXANDER D2 - PANOMA

Gas Cum (BCF)

BHP/ZFlowing

Pres

sure

We have just begun the transitionfrom a static model to the simulator,Beginning at the single well level.

Example from one of the key wells

1.00.75

0.75

1.0

Alexander D21.6 BCF cum.

10^2

10^3

10^4

10^5

10

1975

198 3

1 99 1

201 5

200 7

1 999

Mon

thly

Pro

d.

Initial dynamic models

Parameters• 640 Acre Section

• Cell Size: 390’ X 415’

Layers:

• Upscale from well – 6

• Geomodel – 41 (from 327)

• Well Location: Center

• 0.6’ X 315’ Fracture

• 100 Year Run

• Swc 30 %

• BHPi 260 psia

Three Runs1. Upscaled from well

2. GeoModel, Rate Specified

3. Geomodel, Pressure Specified

A1_LM

B1_LM

B2_LM

B3_LM

B4_LM

B5_LM

GR Facies

Core facies at single well

Perm, Layer 1

0.2 7.6Range (md)

Upscaled permeability from static model(map view layer 1)

Initial Simulation Results• Upscaled stochastic model performed better than one with upscaled well data alone

• Rate specified decline provided better match than pressure specified

Model (by Rate)

Well Model

2.6 BCF

1.4 BCF

Cum. Gas vs. Time

Summary; What’s next?

1. Many obstacles overcome by effort and automation.

2. Upscaling to more manageable model size for largerscale simulations (9, 81 wells).

3. Devise methodology to simulate on even larger scales.

4. On to the Chase (Hugoton) and into OK Panhandle