Simulation of Steam Flooding at West Coalinga Field

1
Simulation of Steam Flooding at West Coalinga
Field

• Lekan Fawumi, Scott Brame, and Ron Falta
• Clemson University
• School of Environment

2
Objective

• Evaluate the effects of different representations
  of interwell permeability on steam flood behavior

3
Outline

• Introduction to steam flooding
• Numerical simulation of steam flooding
• West Coalinga model area and permeability
  distributions
• Steam flood simulations using facies tract,
  facies group, and facies fractal representations

4
Steam Flooding in Heavy Oil Reservoirs

• The main benefit comes from a large reduction in
  the oil viscosity with increased temperature
• Large pressure gradients also help mobilize oil
• Lower interfacial tension and solvent bank
  effects may also help, but are secondary

Viscosity of West Coalinga Crude Oil Chevron
5
Numerical Simulation of Steam Flooding Physical
Processes

• A field steam flood simulator must include at a
  minimum
• a mass balance on water and oil
• an energy balance
• three-phase flow of gas, water, and oil phases
• heat transfer by convection and conduction with
  phase change effects
• capability for three-dimensional flow in
  anisotropic heterogeneous media

6
PDE for water component
7
PDE for Oil component (pseudo-component)
8
PDE for Multiphase Heat Transfer
An energy balance gives
9
Lawrence Berkeley Laboratory TOUGH2 codes
http//www-esd.lbl.gov/TOUGH2/

• Publicly available 3-D multiphase heat and
  compositional flow codes for heterogeneous porous
  and fractured systems
• Developed over a 20 year period, originally for
  geothermal reservoir modeling
• Codes are distributed by (with FORTRAN source
  code) DOE Energy Science and Technology Software
  Center http//www.osti.gov/estsc/
  estsc_at_adonis.osti.gov . The cost to
  organizations with DOE affiliations is 670,
  while the cost for private US companies is 2260.
• A new graphical users interface (developed with
  DOE funding) is available from Thunderhead
  Engineering, Inc. http//www.thunderheadeng.com/p
  etrasim/

10
T2VOC version of TOUGH2

• Special version of TOUGH2 developed for
  environmental steam flood applications Falta et
  al., 1995
• Code considers 3 phase flow of 3 mass components
  air, water, and an organic chemical (which may be
  oil)
• Full heat transfer and thermodynamics are
  included
• Problem may involve 3-D flow in heterogeneous,
  anisotropic porous or fractured systems.
• A new multicomponent hydrocarbon version called
  TMVOC was just released by LBNL in May.

11
Computational effort for steam flood simulation
compared to single-phase isothermal flow

• Increased number of simultaneous equations -- 3X
• Newton-Raphson linearization at each time-step -
  5 iterations per time-step -- 5X
• Smaller time-steps due to N-R convergence
  difficulties -- 5-10X
• Ill-conditioned, stiff matrices at each N-R
  iteration of each time-step -- 2-5X
• Net result A steam flood simulation takes at
  least 150 -500 times more computational effort
  than a single-phase flow simulation with the same
  resolution

12
Steam flood modeling resolution compared to a
single-phase flow simulation
Single-phase
Multiphase
Gridblock resolution (same volume)
Modeled Volume (same resolution)
13
Estimated relationship between number of
gridblocks and simulation time (2Ghz cpu)
106 5x105 0
16 cpu
4 cpu
Number of gridblocks
1 cpu
0 5
10
Simulation time, days
14
Standard repeated 5-spot pattern
Lines of symmetry
injectors
producers
Basic Element Of symmetry, 1/8 of five spot
15
(No Transcript)
16
3
0
0
8
0
0
8
-
1
3
0
0
6
0
0
2
3
9
2
2
9
2
3
9
W
2
2
9
W
3
0
0
4
0
0
2
2
8
-
2
B
2
3
8
3
0
0
2
0
0
2
3
8
W
2
2
8
W
2
2
8
3
0
0
0
0
0
2
3
8
A
8
-
2
2
9
9
8
0
0
1
2
8
B
1
2
8
1
1
8
B
2
9
9
6
0
0
Northing
2
3
7
2
2
7
2
9
9
4
0
0
8
-
3
2
3
7
W
2
9
9
2
0
0
1
2
7
1
2
7
B
2
9
9
0
0
0
2
3
6
1
1
8
A
8
-
4
2
9
8
8
0
0
2
3
6
W
0
0
2
9
8
6
1
5
8
7
5
0
0
1
5
8
7
7
0
0
1
5
8
7
9
0
0
1
5
8
8
1
0
0
1
5
8
8
3
0
0
1
5
8
8
5
0
0
E
a
s
t
i
n
g
17
Well 118A
Complete well log showing facies tracts, facies
groups, and bounding surfaces. Logs such as this
were compared to well 118A to characterize the
location of bounding surfaces and facies groups.
18
Table 2.2 Characteristics of Facies Tracts within
the Temblor Formation  
 
Table 2.4 Characteristics of the facies Groups
from Bridges (2001).  
63
Facies Tracts Used in Model
             
   
19
Facies Tract Model
20
Well 118A
Complete well log showing facies tracts, facies
groups, and bounding surfaces. Logs such as this
were compared to well 118A to characterize the
location of bounding surfaces and facies groups.
21
Table 2.4 Characteristics of the facies Groups
from Bridges (2001).  
Facies Groups Used in Model
   
22
Facies Group Model
23
Facies Fractal Model

• A 3-D fractal distributions of k are generated
  using the properties of each facies group on a
  fine grid
• Based on the location in the coarser simulation
  grid, the facies group type is known, so the
  appropriate fractal k values are extracted,
  preserving the facies group structure in the
  model
• The fine grid fractal k values are upscaled to
  the simulation grid using an arithmetic mean for
  the horizontal permeability, and a harmonic mean
  for the vertical permeability. This upscaling
  can have a large effect on the final k values
  used in the simulation!

24
Facies Fractal Permeabilities
25
Facies Fractal Model
26
Comments on water phase relative permeability and
initial oil saturaton

• Our choice of the water phase relative
  permeability curve was based on a fit of data
  from a core from Chevron
• The initial oil saturation in the model was
  interpolated from Chevron values derived from the
  well logs
• HOWEVER these values resulted in simulations
  where the water to oil ratio was off by a factor
  of 10 or more compared to field values!
• To better match the field values, we reduced the
  water relative permeability endpoint from .56 to
  .15, and
• We increased the oil saturations everywhere by
  20 (with an upper limit of 70 oil)

27
Initial and final oil-water relative
permeabilities
28
Estimated Oil Saturations at the Start of Steam
Flooding
29
Facies Tract Temperatures at 5 years
30
Facies Tract Oil Saturations at 5 years
31
Facies Group Temperatures at 5 years
32
Facies Group Oil Saturations at 5 years
33
Facies Fractal Temperatures at 5 years
34
Facies Fractal Oil Saturations at 5 years
35
(No Transcript)
36
(No Transcript)
37
Conclusions

• The three permeability representations predict
  similar oil and water production from the field.
  The facies group model arguably provided the best
  match of the oil production rate
• Only a single realization of the facies fractal
  model was simulated. A Monte Carlo simulation
  approach would be needed to see the true effect
  of the facies fractal permeability representation
• Upscaling the fine grid fractal values to the
  simulation grid scale presents some important and
  unresolved issues. This could be a useful area
  for future theoretical research
• The over-prediction of water rates may be due to
  the choice of boundary conditions.
• The rate of water production is sensitive to the
  shape of the water relative permeability curve.
  The applicability of measured core values in
  field scale simulation seems questionable.