Exploring the Future Potential of AI-Enabled Smartphone Processors
Fawag snorre pilot_revisit
1. Mobility Control Using Foam
Snorre/WFB FAWAG Pilot Revisit
Arif Ali Khan (NTNU), Ying Guo (Total E&P Norge),
Dag Wessel-Berg (SINTEF), Jon Kleppe (NTNU), Dennis Coombe (CMG)
2. Content
• Introduction
– Project motivation and scope
• Snorre WFB pilot simulation – a revisit
– Updated reservoir model
– Porting from ECLIPSE to STARS and model validation
– Full-field model vs sector model
– Simulation benchmarking
• Simulation of FAWAG pilot
– Sensitivity study with respect of foam parameters
• Summary, conclusions and recommendations
3. Introduction
• Project motivation
– Start out as a master thesis
– New Snorre reservoir model established in 2005
– Evaluate STARS as field simulator for FAWAG
• Project scope and ambition
– Transfer the most recent reservoir model for Snorre WBF pilot to
STARS simulator
– Re-evaluate the foam effect on the new Snorre WBF model
– Establish a feasible workflow for FAWAG simulation
– Recommend potential improvement for STARS simulator
4. A feasible workflow for FAWAG simulation
• Assumptions
– Most of the FAWAG project will likely be implemented after a significant
period of water or gas flooding, and the oil saturation in the flooded
areas is around Sorw or Sorg
– The reservoir model is well established
• Pre FAWAG simulation
– Full-field simulation before FAWAG using Eclipse
– Establish a sector model for the pilot region and validate using Eclipse
• FAWAG simulation (field + sector)
– Port the models from Eclipse to STARS
• Use Petrel to export the grids, rock property, faults etc to STARS + manual
adjustment of the data file for STARS
• Use Eclipse simulated saturation and pressure map before the FAWAG start
to initialize the STARS simulation
– History-match the FAWAG data or predict the field behaviour using
STARS
5. Snorre WFB pilot - history
• Pilot time: November 1999 –
2001
• Wells involved:
– P32 (inj)/P39 (prod), 1550 m
– P32 (inj)/P42 (prod), 1450 m
• 2 FAWAG cycles after gas
injection and WAG operation
for about 3 years
• Total surfactant injected:
~140 tonn
– Slug 1: 15262 SM3 (0.49
wt%) for 9,5 days followed
by 100 days gas injection
– Slug 2: 31733 SM3 (0.2
wt%) for 20.3 days followed
by more gas injection
• Ref: SPE 75157 by A.
Skauge et.al.
8. From Eclipse to STARS
STARS
WFB – Full Field (2005 –
Statoil)
ECLIPSE
WFB – sector (This work)
9. Establish the sector model for FAWAG using
Frontsim
Water injection
Gas injection
Effect from nearby wells
Selected sector
10. SPE75157 vs this work
Geological Differences
• Stochastic perm / porosity
• Faults
PVT
• Composional - 4 Component
• Black oil - 5 Component
Rock Compressibility
Foam
• Part never matches with old
parameters versus new
model
New Eclipse Type Well
Section
• New well Index
• Deviated
• Trans and skin included
Absolute
Permeability
Porosity
FAULTS
11. SPE75157 vs this work
Lay
er
#
division
s
K
(md)
Ф Layer
height
(m)
Layer
height
(mD*M)
4.2 14 700
2 480
1 120
610
6.2
14.0
12.2
S1 5 3500 0.259
S2 7 400 0.236
S3 5
1
80
90
0.225
0.191
P- 39
Pdummy
P-32
25x21x20 = 10 500 grid blocks
P-32
P- 42
P- 39
Lay
er
#
division
s
K
(md)
Ф Avg
layer
height
(m)
Layer
height
(mD*M)
1.69 -
-
S3 8 - - 4.06 -
S4 6 - - 4.06 -
2.77
S1 4 - -
S2 16 - -Stochastic
values
29 x 26 x 34 = 25 636 grid blocks
12. Model porting from Eclipse to STARS
• Grid data – Use Petrel to export files with STARS format
(trivial)
– Grid coordinates and faults
– Permeability and porosity values for each blocks
– Transmissibility multipliers for each blocks
– Pressure + Saturation values for each blocks
• Fluid data – Manual entering to STARS input files (trivial)
• Well data - Convert Eclipse schedule files using with
CGM/IMEX to STARTS input files + 70% manual input
file adjustment (tedious)
• Add foam in STARS (trivial)
Arif – please review and
change to match what
you have done.
13. FAWAG pilot simulation workflow
Eclipse
Field model
S + P profile
in restart files
STARS
Field model
Field simulation
results including
foam effect
STARS
Field model
Eclipse
Field model
S + P profile
in restart files
Field simulation
results including
foam effect
WAG period FAWAG period
19. Simulation benchmarking
• Full field model – Snorre
– No of grid blocks: XYZ= 43 x 146 x 34 => 213 452 grid blocks
– No of wells: 8 up to the end of FAWAG pilot
– Total simulation run time
• Eclipse* up to FAWAG: ~ 20 hours
• STARS** incl FAWAG ~ 18 hours
• STARS** for FAWAG only: ~ 6 hours
• Sector model
– No of grid blocks: XYZ= 29 x 26 x 34 => 25 636 grid blocks (~ 8 time
smaller than full field model)
– No of wells: 3 up to the end of FAWAG pilot
– Total simulation run time
• STARS** for FAWAG only: ~ 2 hours
* Unix / Solaris – 32 bits
** High-End PC (Pentium4, dual processor, 3GHz/1.5GB RAM), 32 bits
20. Foam simulation – phenomenological approach
• The physical phenomenon included in the sensitivity
studies
1. Necessary surfactant concentration in order for foam to grow
(fsurfmin)
2. Foam dry out when there is not sufficient water (Swmin)
3. Foam killed by surplus of oil (Sof)
4. Foam collapse due to viscous forces (Nc) – criticall capillary
number
21. Foam Model in STARS:
( ) FMskk w
o
rg
f
rg ×=
)1
1
654321 ⋅⋅⋅⋅⋅⋅⋅⋅⋅+
=
FFFFFFMRF
FM
• FM Dimensionless Interpolation Factor
FM = 1 (no foam)
FM = 0 (strong foam)
• MRF Reference mobility reduction factor
• F1 Surfactant concentration term
• F2 Water saturation term
• F3 Oil Saturation (foam killer)
• F4 Gas Velocity term
• F5 Capillary Number term
• F6 Critical Capillary Number term
35. Conclusions and recommendations
• The simulation exercises on the foam WFB FAWAG pilot
demonstrate the feasibility STARS as a tool for field scale studies for
FAWAG.
• More understanding and tuning of the reservoir simulation for
FAWAG is needed in order to obtain trust-while results.
• Acquisition of lab data must be designed closely related to the
planed field operation
• Refined grid analysis should be done in the close-well areas, but
STARS must be made more robust for this.
• Hysteresis for WAG period should be included in the simulations
• MRF as dependent on the FAWAG cycles, pressures, etc (input
table?)
• Oil saturation plays important role for foam behaviour, specially in
the saturation ranges where foam is likely to be generated or killed
and more sophisticated modelling is desired to capture the possible
phenomena, e.g. Sw and its distribution, wettability, oil composition.