Optimization of Carbon Dioxide Sequestration in Brine Aquifers

You might also like

Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 27

Optimization of carbon dioxide sequestration in brine aquifers

David Cameron & Lou Durlofsky Smart Fields Seminar April 15, 2009

Optimizing CO2 sequestration is similar to optimizing oil production


Reservoir Data

Set Well Controls

Update Model

Optimize Well Settings


April 15, 2009

SFC

CO2 sequestration process

Flue gas is separated to get pure CO2 stream


CO2 is injected in brine aquifer as supercritical fluid A long equilibration phase (~1000 yrs) follows injection CO2 is trapped in different ways

April 15, 2009

SFC

Time scales for trapping mechanisms

(from Benson, 2007)

April 15, 2009

SFC

Potential goal: minimize mobile CO2


Definition of mobile CO2 Proportion of CO2 above residual saturation after equilibration
Controlling parameters Well configurations Model parameters

April 15, 2009

SFC

Contents of this talk

Description of the reservoir model


Formulation of optimization case study Optimization methods

Optimization results from case study


Summary and conclusions

April 15, 2009

SFC

Reservoir model features


Full reservoir model (to scale) Base case parameters

Dimensions 10 km x 10 km x 250 m
Grid 25 x 25 x 20 Permeability 0.1 10000 md (M = 100) Porosity 0.15 to 0.35

Kv/Kh 0.1 Inner section (not to scale)


Porosity

Max inj. per well 50 000 STB/d (super critical) Total injection 5 Mta (500 MW coal plant) Injection time 30 yrs

Equilibration time 1 000 yrs


900 ft

April 15, 2009

SFC

These figures were obtained using the preprocess file


Permeability 1000 (md)

TO SCALE

NOT TO SCALE

900 800 700 600 500

250 m

400 300 200 100 0

250 m

April 15, 2009

SFC

Simulation features

Used Stanfords General Purpose Research Simulator (GPRS)


Two phase (water and CO2) compositional

Peng Robinson EOS


Hysteresis (Carlson method) No mineral trapping

April 15, 2009

SFC

Case study: 6 fixed wells Optimize using rate control


Well configuration Control parameters: The fraction of total injection rate at each well (vectors over time): R1, , R6

I1 (7,9)

I2 (13,9)

I3 (19,9)

Bounds:

Constraint:

R i 0.001,0.4
I4 I5 (7,17) (13,17) I6 (19,17)

R
i 1

Possible objective functions: 1) Mobile CO2 2) CO2 under cap rock


SFC 10

April 15, 2009

Base case: equal injection rates


Base case phase distribution
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 mobile residual aqueous Top Layer

proportion of injected CO2

200

400 600 time (yrs)

800

1000

April 15, 2009

SFC

11

Base case: equal injection rates


Base case phase distribution
0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 mobile residual aqueous Top Layer

proportion of injected CO2

200

400 600 time (yrs)

800

1000

April 15, 2009

SFC

12

Base case: saturation and dissolution snap shots


cross section

Base case CO2 saturation


t = 15.5 yrs
1 0.8 0.6 0.4 0.2 0

Base case dissolved CO2 (kg)


x 10 5 4 3
7

t = 30.0 yrs

t = 15.5 yrs

t = 30.0 yrs

t = 750.3 yrs

t = 1000.0 yrs

2 1 0

t = 750.3 yrs

t = 1000.0 yrs

1 0.5 0 0.5 01

1 0.5 0 0.5 01

April 15, 2009

SFC

13

General Pattern Search (GPS) optimization method


ADVANTAGES

Gradient free Easy and robust

Parallelizable

DISADVANTAGES

Local method
Slow
14

April 15, 2009

SFC

Mesh Adaptive Direct Search (MADS) optimization method

Generates random positive basis set


Samples an area more densely than GPS

April 15, 2009

SFC

15

Pattern search (N+1) polling

Fewer function evaluations per iteration Less likely to find global optimum
GPS (N+1) is biased

April 15, 2009

SFC

16

Minimization of mobile gas phase with 6 unknowns using different techniques


0.153 0.152 0.151 0.15

GPS GPS Np1 MADS MADS Np1

Objective

0.149 0.148 0.147 0.146 0.145 0.144 0 50 100 150


17

Function evaluations
April 15, 2009

SFC

Different methods converge to different solutions


Base Case
0.25

GPS

GPS (Np1)

MADS

injection fraction

0.2 0.15 0.1 0.05 0 well 1 well 3 well 5

April 15, 2009

SFC

18

Minimization of mobile CO2 phase using more unknowns


0.16 0.15 0.14 0.13 0.12 0.11 0.1 0

6 unknowns 12 unknowns 18 unknowns

Mobile CO2

100

200

300

400

500

Function evaluations
April 15, 2009

SFC

19

Time updates tend toward bang-bang solutions


6 par
0.45 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 well 1 well 3 well 5

12 par

18 par

Injection fraction

April 15, 2009

SFC

20

Minimizing CO2 phase in the top layer


0.22 0.21 0.2 0.19

Objective

0.18 0.17 0.16 0.15 0.14 0.13 0.12 0 50 100 150 200 mobile (6 par) mobile (12 par) top layer (6 par) top layer (12 par)

Function evaluations
April 15, 2009

SFC

21

Optimization with respect to mobile and top layer CO2 give similar results
mobile 6 par
0.45 0.4

top 6 par

mobile 12 par

top 12 par

Injection fraction

0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 well 1 well 3 well 5

April 15, 2009

SFC

22

Base case versus optimal solution


0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0

Base case phase balance


mobile residual aqueous 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 200 400 600 time (yrs) 800 1000 0 0

Optimized phase balance (18 par, mobile CO2)


mobile residual aqueous

proportion of injected CO2

proportion of injected CO2

200

400 600 time (yrs)

800

1000

April 15, 2009

SFC

23

Cross sections for dissolution and saturation cross section


Base case dissolution (1000 yrs)
t = 1000.0 yrs x 10 5
7

Optimized dissolution (1000 yrs)


t = 1000.0 yrs x 10 5
7

Base case saturation (1000 yrs)


1 0.5 0 0.5 1 t = 1000.0 yrs 0.2 0.15

1 0.5 0 0.5 1

Optimized saturation (1000 yrs)


t = 1000.0 yrs 0.2 0.15

0.1

0.1

0.05

0.05

April 15, 2009


1

SFC
1

24

Summary and conclusions


Summary

Formulated CO2 sequestration as an optimization problem Designed a realistic synthetic model for CO2 injection Optimal solutions reduced mobile CO2 by 30 % Solutions exhibit bang-bang behavior

Conclusions

Bang-bang injection increases dissolved and residually trapped CO2

April 15, 2009

SFC

25

Future directions

Test using more parameters


Test more optimization methods Use position of wells as additional optimization variables Incorporate closed-loop optimization

April 15, 2009

SFC

26

Acknowledgements
Huanquan Pan Yaqing Fan David Echeverria for modeling and GPRS help for modeling and GPRS help for optimization and coding help

April 15, 2009

SFC

27

You might also like