Larry W. Lake
The University of Texas at Austin
Larry_Lake@mail.utexas.edu
Outline
• Introduction
• The Model
• Validation
• Updates of Technology
Prior and Current Work
Past
• Belkis Refunjol
Present
• Jorge S’Antana Pizarro (Petrobros) • Tom Edgar, ChE
• Introduction
• The Model
• Validation
• Updates of technology
CRM Continuity Equation
Continuity: Production Rate:
i(t) q(t)
c t Vp
dp
dt
i(t) q(t)
q(t) J p pwf
Ordinary Differential Equation:
dq(t) 1 1 dpwf
q(t) i(t) J
dt dt
c t Vp
J
t t0 t t0 t t0
pwf,t pwf,0
( ) ( ) ( )
q(t) q(t0 )e
I(t) 1 e c t Vp
1 e
t t0
Signal Response
Connectivity
ττijij==61days day
ffijfijij===100%
65%
0%
Capacitance-Resistance Model (CRMT)
ctV p
I(t) q(t)
J
Time constant
qk qk 1e
t
1 e
t
I
k
Capacitance-Resistance Model (CRMP)
ctV p
j
qj(t) J j
I5 I4
Time constant
f5j f4j I3
I6
f6j
j I2
f3j
I1 f2j
f1j
np
f11
f ij 1 f12 f13
Drainage volume
j 1 around a producer
Inter-well connectivity or gain
t t ni
f ij I ik
j j
q jk q j k 1e 1 e
i 1
Capacitance-Resistance Model (CRMIP)
ctV p
qj(t)
ij
J ij
Ii(t)
Time constant
ij
np fij
f
j 1
ij 1
Inter-well connectivity or gain
ni
t t
ij
q jk qij k 1e
ij
1 e f ij I ik
i 1
Mature West Texas Waterflood
• Introduction
• The Model
• Validation
• Updates to technology
Validation
How do we scientifically validate
geoscience hypotheses?
Remember:
Characteristics of a reservoir can be inferred
from analyzing production and injection data
only
Recognizing testable hypotheses can be subtle and
requires practice. To do it, ask “how would one test this
hypothesis”.
• Introduction
• The Model
• Validation
• Updates to Technology
Updates to Technology
discontinuous changes
500
• Wrong water 0
0 10 20 30 40 50 60 70 80 90
Time, month
production data
• Unknown operational
changes
Data Issues Remedies
• Restart CRM
• Outlier detection
Warm start
Gainfit Remove small
Gainfit #1
gains
Calculate residuals
Gainfit #2 Gainfit #3
and replace outliers
Calculate residuals
Fracfit #1 Fracfit #2
and remove outliers
~2.5 hrs
computation
Reservoir
model
time
Warm start
Gainfit Remove small
Gainfit #1
gains
Calculate residuals
Gainfit #2 Gainfit #3
and replace outliers
Calculate residuals
Fracfit #1 Fracfit #2
and remove outliers
Warm start
Gainfit Remove small
Gainfit #1
gains
Calculate residuals
Gainfit #2 Gainfit #3
and replace outliers
Calculate residuals
Fracfit #1 Fracfit #2
and remove outliers
<10 min
Reservoir
computation
model time
•For gain>0.2
0.85 - 1
0.75 - 0.85 I19 I21 P29 I26
P89
0.65 - 0.75 P9 I98
P4
I99 P84
I96
I113
I75 P110
150 30 150 10 30
5
5
180 0 180 0
270 270
10
150 30
5
180 0
210 330
240 300
270
Updates to Technology
OUTCOME
Reservoir property estimates (volume, pressure, etc.)
Future production forecast
Support production planning (number of wells, production
rate schedule, producing life)
ICR Model
•Assumptions
•No gas •Well
•Well 2
•No volatile oil 1
•No aquifer •Reservoi •Reservoi
r2
r1
•Well
3
•Reservoir 3
c
V
tpi
q (
c
V)P P
(
0
)i N tp
i wi
fi pi
Ji i i i
• Summary
• Future Work
• Constraints
nj
f
j 1
ij 1 for all i (6)
Equation 7 ensures that injected water does not adversely affects the
reservoir production.
fij Iik J j j
t / j t / j k
qˆ jk qˆ j ( k 1)e 1 e
wf wf
N
i
p, j
t i 1
•where
• NP,j is the cumulative total liquid
production from a producer j
• CWIi is the cumulative water injection
into a injector i •* Reservoir barrel
(rb)
Independent variables used for the CRMP Independent variables used for the ICR
3000 3.00E+06 model
Water injection rates (rb/day)
2500
produced (rb)
I -Np4
1500 2 W1
0.00E+00
I 50 70 90 W2
1000 3 W3
-1.00E+06
W5
500
-2.00E+06
0
50 60 70 80 90 100 -3.00E+06
Time periods (month) Time periods (month)
•g2(x1,x2
) •g3(x1,x2
)
•g1(x1,x2
)
•g4(x1,x2
)
f(x1,x2)
•Global minimum
•Global minimum x2 •With constraints
•without constraints x1
•2480 ft
•(Same example as in Sayarpour, 2008)
fi2
0.5 fi2_linear_Np
0.4 fi1_linear_Np 0.10
0.3
0.2 0.05
0.1
0.0 0.00
I01 I02 I03 I04 I05 I01 I02 I03 I04 I05
P03-fi3 P04-fi4
0.20 0.8
0.18 0.7
0.16
0.6
0.14
0.12 0.5
fi3
fi4
0.10 0.4
fi3_nonlinear fi4_nonlinear
0.08 0.3
fi3_linear_Np f4_linear_Np
0.06
0.2
0.04
0.02 0.1
0.00 0.0
I01 I02 I03 I04 I05 I01 I02 I03 I04 I05
0.6
15
R2
0.5
0.4
10
0.3
5 0.2
0.1
0 0.0
P1 P2 P3 P4 P1 P2 P3 P4
Np (rb)
Np (rb)
2.00E+06 2.50E+05
2.00E+05
1.50E+06
1.50E+05
1.00E+06
1.00E+05
Data Data
5.00E+05 5.00E+04
Linear_Npk Linear_Npk
0.00E+00 0.00E+00
50 60 70 80 90 100 50 60 70 80 90 100
Time period (month) Time period (month)
P03 P04
8.00E+05 3.00E+06
7.00E+05
2.50E+06
6.00E+05
2.00E+06
5.00E+05
Np (rb)
Np (rb)
4.00E+05 1.50E+06
3.00E+05
1.00E+06
2.00E+05
Data 5.00E+05 Data
1.00E+05
Linear_Npk Linear_Npk
0.00E+00 0.00E+00
50 60 70 80 90 100 50 60 70 80 90 100
Time period (month) Time period (month)
Time Constants
30
Tau_CRMP Tau_ICR
25
Taus (day)
20
15
10
0
P1 P2 P3 P4
fij
0.3 R² = 0.3681
388 0.2401 0.2314
0.2
426 0.2068 0.2010
0.1
426 0.2751 0.2510
442 0.1270 0.1418 0.0
0 200 400 600 800 1000 1200
493 0.0444 0.1646
Distance between well-pair (ft)
510 0.3100 0.4077
527 0.2742 0.2643 Gains vs Distance (ICR Model)
0.6
543 0.3502 0.3413 fij_ICR
576 0.1253 0.1250 0.5 Linear (fij_ICR)
650 0.2059 0.2369 0.4 y = -0.0003x + 0.401
677 0.1246 0.1226 R² = 0.2487
fij
0.3
745 0.0861 0.1439 0.2
806 0.2895 0.2504
0.1
815 0.2756 0.2323
860 0.2457 0.1878 0.0
0 200 400 600 800 1000 1200
959 0.1687 0.1722 Distance between well-pair (ft)
The results show that as the distance between interwell-pairs increases, the well
connectivity tends to decrease.
fij = -0.0003(ft-1)*dij (ft)+ 0.402 for 110 ft < dij < 959 ft
•Application of the ICR •57
Synfield-4: q(t) vs time
-0.2
Surface subsidence (ft)
-0.8
-1
-1.2
Jim Honefenger (P.E. Moseley & Associates, Inc.) The Gardner Group63
Validation
Characteristics of a reservoir can be inferred from
analyzing production and injection data only
Field Injectant Independent Data Agree
CRMT
mod elÊparameters j , fij CRMP
,f CRMIP
ij ij
fittingÊwindow int ervalÊwithÊnoÊexternalÊchanges
Basic Definitions
ScatterÊof ÊproducerÊrateÊaboutÊmod el
R 2Pr oducer 1
ScatterÊof ÊproducerÊrateÊaboutÊmean
R 2j fittingÊint erval
qöj t qj t;mod elÊparameters
t Last 2
tt First
1
qöj t qj
t Last 2
tt First
Steady-State Connectivity Map
80
60
40
Better CO2
Performance
20
Producer
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Interwell Connectivity
Two Equally Viable Solutions
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
After 10 days
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
After 30 days
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
After 90 days
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
After 180 days
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
After 365 days
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
After 2 years
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
After 4 years
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Transient Interwell Connectivity
4 years <<
80
60
40
20
Produce r
Water Injector
Carbon Dioxide Injector 0 1,000 ft
0
0 20 40 60 80 100
Mature West Texas Waterflood
-1.5
0 50 100 150 200 250
Producer
ProducerNumber
Number
Oil Rate vs. Cumulative Oil Production
•Confidence band
for regression
line
•103,042 •109,252
•39,956 bbl
•~
$2,800,000
Producer 184 – Good Fit
200
Historic Total Production
180
Modeled Total Production
160
y R2 = 0.961
Bbl/ a 140
d err = 0.146
day / 120
l
100
b
b 80
60
40
20
0 20 40 60 80 100 120
Month
Month
Producer 127 – Good Fit
600
Historic Total Production
Modeled Total Production outliers
500
y
a 400 R2 = 0.696
Bbl/ d
err = 0.037
day /
300
l
b
b 200
100
0
0 20 40 60 80 100 120
Month Month
Producer 74 – Poor Fit
180
160
140
y
a 120
Bbl/ d
day / 100
l
80
b
b R2 = -1.03
60
err = 0.143
40
20
Historic Total Production
Modeled Total Production
0
0 20 40 60 80 100 120
Month
Month
Producer 201 – Poor Fit
1200
Historic Total Production
1000 Modeled Total Production
y
a 800
Bbl/ d R2 = 0.793
200
0
0 20 40 60 80 100 120
Month Month
728000
Gain Map
Injector
726000
Producer
724000
722000
720000
ft
P210
718000
P103
716000
I 58
714000
712000
475000 480000 485000 490000 495000 500000
ft
Producer 210 (large distance)
600
Historic Total Production
Modeled Total Production
500
Bbl/ R 2 0.882
y err 0.093
daya 400
d
/
300
l
b
b 200
100
0
0 20 40 60 80 100 120
Month
Producer 103 (skipped over)
150
y
a 100
d R 2 0.635
Bbl/
/ err 0.110
l
day
b
b
50
0
0 20 40 60 80 100 120
Month
t
c
e Lost Injection
j 1
n
0.9
I
0.8
Np f
1 fij o 0.7
j1
0.6
n
o 0.5
i 0.4
t
c 0.3
a 0.2
r
F 0.1
0
0 20 40 60 80 100 120 140
Injector Number
Injector Number
Time Constants
100000
Time Constant (days) .......
1000
100
10
1
11
21
30
42
56
75
87
1
101
111
123
133
142
153
163
175
184
195
205
216
225
234
Producer Number
Reservoir A
CRM Fit – Total Field
4
x 10
6
Bbl/
5.5 R2 = 0.956
day
5
y
a
d 4.5
/
l
b 4
b
3.5
Historic Injection
3
Historic Total Production
Modeled Total Production
2.5
0 20 40 60 80 100 120
Month
Month
Future Injection
Real-Time Optimization
MPD & and Risk Management
Contracting
Blowouts;
Strategies
Inevitable VOI; Impact Drlg Safety;
(lump sum v
Dis- of Estimates Offshore FUTURE:
cost plus?)
appointment & Methods Spills Life Cycle
Assessments
Uncertainty
Updating Capital
Allocation w/
Valuing Price Simple Model
Uncertain
Forecasts Development
Arrivals
Portfolio
Real Options
Optimization
Optimal Injection and Predicted Oil Production for
x 10
the Field
4
5
bbl/day
3 Historic
Optimal
2
0 20 40 60 80 100 120 140 160 180 200
Month
3000
2500
bbl/day
2000
1500
Historic Oil Production
1000 Predicted Oil Production
Extrapolated Oil Production
500
0 20 40 60 80 100 120 140 160 180 200
Month
e
i
F Injection Shares
l 3.5
a Historic Share
t 3 Predicted Share
Percent ofo
Total T 2.5
f 2
o
t 1.5
n
e 1
c
r 0.5
e
P 0
0 20 40 60 80 100 120 140
Injector
InjectorNumber
Number
i
O
Production Shares
d 2.5
l Historic Share
e Predicted Share
i
Percent of
2
F P195
Total P112
f 1.5
o
t 1
n
e
c 0.5
r
e
P 0
0 50 100 150 200 250
Producer
Producer Number
Number
Retrodiction