Design and History Matching of Waterflood/Miscible CO2

1
Design and History Matching of Waterflood/Miscible
CO2 Flood Model of a Mature Field The Wellman
Unit, West Texas
by Jose Rojas Master of Science Candidate
Chair of Advisory Committee Dr. David
Schechter Committee Members Dr. Duane McVay and
Dr. Luc Ikelle
2
Content

• Research Objectives
• Review of Geology
• Historical Reservoir Performance
• OOIP and Water Influx (Material Balance)
• Simulation Model
• Model Calibration History Matching
• Results Primary Depletion
• Waterflooding
• CO2 Injection
• Conclusions and Recommendations

3
Objectives
Revise and integrate data from the reservoir
description to develop a full field,
three-dimensional black oil simulation model to
reproduce via history matching, the historical
performance of the reservoir under primary,
secondary and tertiary stages of depletion
Secondly, develop a calibrated model that can be
used to evaluate, design and plan future
reservoir management decisions.
4
Review Of Geology
Location

• Terry county, TX, along the
• Horseshoe Atoll reef complex that
• developed in North Midland during
• Pennsylvanian and early Permian
• time

5
Review Of Geology
Field is considered geologically unique, because
it comprises two types of reef construction

• Built in shallow muddy(turbid) water
• Encroaching shales at the flank
• Smaller cone-shape structure
• Oil bearing

• Built in deep clear water
• Large mound shape structure
• Strong depositional dip
• Water bearing

6
Review Of Geology
Reef on Reef Depositional Model

• Wolfcamp deposited on top of the prominent Cisco
  Reef
• Curved layers at the bottom, more horizontal in
  upper structure

7
Review Of Geology
Isopach Structure Map
8
Historical Reservoir Performance
Primary Depletion (1950 1979)
Cum. Oil 41.8 MMSTB RF 34.6
9
Historical Reservoir Performance
10
Historical Reservoir Performance
CO2 Injection (1983 1995)
11
Chronological Stages of Depletion
12
OOIP and Water Influx Material Balance
13
OOIP and Water Influx Material Balance
Hurst and Van Everdigen

• Estimate and validate previous OOIP assessments

• Estimate water influx rate prior waterflooding

Results
Final Aquifer Properties

• OOIP (N) aprox 125 MMSTB
• We10 approximately 8.0 MMRB

H, Feet 68
K, md 25
?, Fraction 0.9
Ro/Re 2
Ro, Feet 3000
Angle (f1) 360
14
Simulation Model
Full field, 3-D black oil simulation Imex CMG
Grid System

• Use of flexible grids corner point,
• non - orthogonal geometry.

• 27 x 27 gridblocks I,J direction

• K, direction subdivided in 23 layers
• based on porosity correlations
• (geological description)

• Total 16,767 gridblocks

15
Simulation Model
3D Structure Development
16
Simulation Model Input Data
Production data

• Over 45 years of monthly cumulative oil, gas
  and water production from 47
• wells was converted into daily rate schedules
  for simulation

• Model initially constrained by oil rates and
  water/CO2 injection rates

Pressure data

• Pressure measurements reveal good communication
  within the reservoir
• Use of BHP corrected and averaged to a common
  mid-perforation
• Static BHP seemed to be representative of the
  average reservoir pressure

17
Simulation Model Input Data
Isopach Maps

• Use of isopach maps resulted from geological
  and petrophysical study in 1994

• Geological and stratigraphic correlation (Core
  vs Log data)
• Quantify major rock properties
• Lateral and areal continuity

• 60 geological contoured maps from gross
  thickness, porosity and NTGR were
• digitized
• Interpolation between contour allows model to be
  populated

18
Simulation Model Input Data
Permeability

• Use previous estimates from correlations between
  open-hole logs and core
• measurements

K 10(0.167 Core porosity 0.537)
Water Saturation
19
Simulation Model Input Data
Fluid Properties

• Use PVT properties contained in previous lab and
  reservoir studies
• Bubble point 1248 1300 psig
• Rs, 400-500 SCF/STB
• Oil Gravity, 43 API
• OFVF, 1.30 RB/STB
• Oil Viscosity, 0.4 cp
• Black oil fluid type

Relative Permeability

• Special core analysis for core well No. 7-6
  included measurements on only
• two samples with a low non-representative
  permeability

• Use functions derived from Honarpours
  correlation (past studies)

20
Simulation Model Input Data
Capillary Pressure Data

• Only 4 samples, K gt 1 md
• (Special core analysis)

21
Model Calibration History Matching

• Objective Validate the model adjusting the
  reservoir description until
• dynamic model match the historical production
  and pressures

• Weight and rank properties by level of
  uncertainty (quality,
• source, amount, availability of data)

Historical Responses to be Matched

• Fieldwide average reservoir pressure
• Fieldwide production rates
• Fieldwide GOR and Water cut
• Arrival times
• Individual responses (lesser degree)

22
Results Primary Depletion
First simulation runs

• No aquifer modeled

• Poor pressure response

• Need of external energy recognized

23
Results Primary Depletion
Preliminary runs
Aquifer Calibration

• Carter and Tracy Analytic
• MB case too strong (top)
• Aquifer size (?,h), trans. (K,h)
• Reference datum adjusted
• Influx 20 greater, best case

24
Results Primary Depletion
Preliminary runs

• Water arrival time and cumulative
• did not match

• Highest corresponds to MB

• Poorest corresponds no aquifer

• Poorest corresponds no aquifer

• Best pressure match sharp gas
• increase

25
Results Primary Depletion
26
Results Primary Depletion
Diagnosis
27
Results Primary Depletion
Final Results
28
Results Primary Depletion
Final Results
29
Results Waterflooding
Injector Location

• 4 producers converted
• to water injectors (1979)

• Located at the flank
• forming a perimeter belt

• Injection below and above
• OWOC _at_ - 6,680 ft

• Model primarily constrained
• by historical injection rate
• schedule

30
Results Waterflooding
Initial runs
Fluid production needs to be controlled !
31
Results Waterflooding
Diagnosis
Model Pressure Map
32
Results Waterflooding
Model Calibration for Final Pressure Match
33
Results Waterflooding
Model fluid match
34
Results Waterflooding
WOC Movement
35
Results CO2 Injection
Miscible Displacement
Highlights

• Modification of the black oil sim.

• Pseudo-miscible option with
• no chase gas

• Based on the Todd and Long-
• staff theory

• Modifies physical properties
• and flow characteristics of
• the miscible fluids

• Requires definition of new param.

• CO2 PVT prop., MMP, ?o(P)

36
Results CO2 Injection
Initial runs
What is happening ?

• Abnormal increase in reservoir pressure

• VRR greater than 1, correlates with sharp
  pressure increase
• VRR decreased (1992) correlating with decrease
  in pressure

37
Results CO2 Injection
Initial runs
Solvent Rate
Water Rate

• Model is not able to reproduce rapid water rate
  increase (1986)

• In 1986, insufficient water and solvent
  production results in
• a dramatic increase in reservoir pressure

38
Results CO2 Injection
Most Uncertain Parameters influencing Production
of Fluids

• Relative Permeability
• Functions
• end points, shape, crit. Sat
• New set for middle reef
• Account for ESPs

• Vertical Transmissibility
• Aquifer/reservoir
• vertical arrays
• local refinements
• Kv / Kh gt 1

• Aquifer Properties
• ?, h,k

Model Calibration

• Re-interpretation
• Completion intervals
• Plug-downs
• Include wells high
• on the struct.

Local Absolute (K) Lesser degree

Negative Skin Stimulations - Acidizing
Uniform Mod. Fluid PVT
K.H Term Prod / Inj Index
39
Results CO2 Injection
Diagnosis / adjustments
40
Results CO2 Injection
Sensitivity runs

• Good pressure match primary. Lost
• during waterflooding, poor at CO2 Inj.

• Excess of H20 (waterflooding)

• Overall insufficient water and solvent
• production (tertiary), causing over-
• pressurization.

41
Results CO2 Injection
Final match

• Daily oil rate primary constraint
• expanded to daily total liquid rate
• (oil water)

• Match is preserved (primary, H2O Inj.)

• Water and H2O breakthrough matched

• Oil match sacrificed to match pressure

42
Results CO2 Injection
Final match
43
Chronological Stages of Depletion
44
Results CO2 Injection
Chronological Oil Saturation Distribution
a) Primary depletion
b) Waterflooding
c) CO2 miscible flooding Oil saturation
considered overestimated due to the
excess of oil production
45
(No Transcript)
46
Conclusions

• Original fluids in place (according to
  simulation)
• Oil 127.1 MMSTB
• Water 139.0 MMSTB
• Gas 54.3 BSCF
• Model OOIP, proved to be in close agreement
  not only with past estimations
• but also with the analytical solution of the
  material balance technique
• previously presented.

2. Cumulative water influx (8 MSTB) was
estimated from application of the
material balance theory and correlates quite
well with water influx obtained in the
best case being 8.5 MMSTB (first 10 years).

1. The natural aquifer greatly influenced
   production of fluids and consequentially, the
   predicted average reservoir pressure.

4. The initial set of aquifer parameters was
derived analytically by the Hurst and Van
Everdigen theory and finally tuned by
sensitivity analysis
47
Conclusions Cont.
5. The Carter and Tracy (analytic) method
resulted as the best alternative to model the
Cisco aquifer over the Fetkovich (analytic) and
the numerical aquifer method.

• The Cisco aquifer provided energy and supplied
  water that encroached
• uniformly advancing the WOC 208 ft
  (prior to waterflooding) and an additional 210
  feet (prior to CO2 injection) being in excellent
  agreement with field observations.

• The use of a flexible grid system, honored the
  characteristic structure
• of the cone-shaped double anticline. The
  distorted grid blocks
• allowed a good representation of Wellman
  Unit geological features.

• Historical water production and breakthrough
  times were identified as
• one of the most difficult parameters to
  match and one that greatly influenced the
  behavior of the predicted reservoir pressure
  response.

48
Conclusions Cont.

• A complete pressure match was achieved
  through primary depletion,
• waterflooding and CO2 injection, however the
  match on liquid production
• was compromised in order to tune the final
  pressure match.

• The results of this work provide the
  foundation for future research into
• this hydraulically complicated reservoir

49
Recommendations for Future Work.

• More research is recommended on the geology
  of the field with the aim
• of simplifying the total number of
  gridblocks, specifically the number of
• layers (23) by the use of some of the
  upscaling methods in the literature.

• Consider the use of pseudo-functions during
  simplification of the
• existing model to increase the accuracy when
  modeling the production of
• fluids.

• Place additional effort to update the current
  model by incorporating
• production and injection data from 1995 to
  the present time, thereby it
• can be used to assist future reservoir
  management decisions.

50
(No Transcript)
51
OOIP and Water Influx Material Balance
Havlena and Odeh

• Validate existence and influence of external
  energy (aquifer)
• Use performance data and fluid properties prior
  waterflooding

(1)
H O re-arranged the general equation (1) into
linear form. (final WE)

• Assumption No initial gas cap (m0)

• Plot Qp vs. DBt should result straight line

52
OOIP and Water Influx Material Balance
Hurst and Van Everdigen

• Estimate and validate previous OOIP assessments

• Estimate water influx rate prior waterflooding

• Water influx (We) at some arbitrary T
• Series of discrete pressure steps

Havlena and Odeh (M.B.E)

• m 0
• Pore and cw neglected (usually low)

(2)
FNEo We
53
OOIP and Water Influx Material Balance

• From (2) a plot F/Eo vs. We/Eo should be
  linear with intercept F/Eo N
• when We/Eo 0 and with unit slope
• 2, 2.5 and 3.5 dimensionless radio reD
  re/ro were assumed to obtain
• dimensionless function WD
• Iterations to reD and aquifer properties until
  satisfy aquifer fitting theory HO

54
OOIP and Water Influx Material Balance

• Best fit with reD 2
• OOIP (N) aprox 125 MMSTB
• We10 7.3 MMRB