a history matching approach to estimate caprock morphology...
TRANSCRIPT
A History Matching Approach to Estimate
Caprock Morphology Parameters for CO2
Storage in Saline Aquifers
Masoud Ahmadinia
Supervisors:
Seyed Shariatipour
Odd Andersen
ABOUT ME
Education
• Ph.D. Flow Measurement and Fluid Mechanic, Coventry University
• M.Sc. Petroleum Engineering, Polytechnic University of Turin
• M.Sc. Environmental and Land Planning Engineering, Polytechnic University of Milan
• B.Sc. Petroleum Engineering, Shiraz University
Work Experience
• TNO. UTRECHT, NETHERLANDS (Internship)
• SINTEF. OSLO, NORWAY (Research placement)
HISTORY MATCHING - OBJECTIVES
• Improve and validate the reservoir simulation model
• Better understanding of reservoir processes
• Improve the reservoir description and data acquisition program
• Identify unusual operating conditions
HISTORY MATCHING - METHODS
Manual
• Run simulation for
historical period
• Compare results to
actual field data
• Adjust simulation input to
improve match
• Selection of input data
based on knowledge and
experience
Automatic
• Minimizes the objective function;
i.e., difference between observed
reservoir performance and
simulation results
HISTORY MATCHING – COMMON PARAMETERS
• Aquifer size
• Vertical permeability barriers
• Flow capacity, kHh
• kV/kH ratio
• Pore volume
• Relative permeability
CO2 PLUME DEVELOPMENT @ SLEIPNER
• World's first industrial-scale GCS project.
• Time-lapse seismic monitoring data are available from 1996 to 2010.
Chen Zhu et al. 2015
CURRENT STUDY
• The numerical simulation is performed on a synthetic model with a
specific slope and rugosity pattern.
𝑧 𝑥, 𝑦 = A[sin 𝜔1𝑥 + sin ሿ𝜔1𝑦 + 𝑨𝒙 sin 𝜔2𝑥 +𝑨𝒚 sin 𝜔2𝑦 + 𝑥 tan 𝑺𝒙 + 𝑦 tan 𝑺𝒚
CURRENT STUDY
𝑧 𝑥, 𝑦 = A[sin 𝜔1𝑥 + sin ሿ𝜔1𝑦 + 𝑨𝒙 sin 𝜔2𝑥 +𝑨𝒚 sin 𝜔2𝑦 + 𝑥 tan 𝑺𝒙 + 𝑦 tan 𝑺𝒚
Observed Initial guess Calibrated
CURRENT STUDY
• The results of the plume shape is recorded and subsequently
reinterpreted as "observed" data.
• A new synthetic model is generated, without knowledge of the first.
• This model could be regenerated to change rugosity magnitude and
slope directions.
MODEL PARAMETERS
Parameter Value
Reservoir dimensions (NX×NY×NZ) 101×101×4
Reservoir size (km) (LX×LY×LZ) 15×15×0.1
Cell size (m)DX×DY 148.5
DZ 25
Rock compressibility (1/bars) 4.35E-5
Water density (kg/m3) 1000
CO2 density (kg/m3) 745.6
Residual water saturation (Srw) 0.27
Residual CO2 saturation (Src) 0.20
Permeability (mD) 5
Porosity 0.2
Simulation period (years) 1010 (100 × 0.1 years + 100 × 10 years)
Number of time steps 200
Water viscosity (Pascal-second) 8.0E-4
CO2 viscosity at 150 bar (Pascal-second) 6.4E-5
VERTICAL EQUILIBRIUM
• Large disparity in lateral and vertical scales, plus differences in density
between the supercritical CO2 plume and the resident brine vertical fluid
segregation will be almost instantaneous compared with the up-dip
migration.
• The flow of a thin CO2 plume in 3D can be approximated in terms of its
thickness to obtain a 2D simulation model
INITIAL GUESSES & SEARCH RANGE FOR EACH OF THE PARAMETERS
Ax
(meter)
Ay
(meter)
Sx
(meter)
Sy
(meter)
Limits for scenario a 0-10 0-10 0-5 0-5
Limits for scenario b 0-20 0-20 0-10 0-10
Observed 9 7 2 3
Case# of
iterations
Initial guess
Ax
(meter)
Ay
(meter)
Sx
(meter)
Sy
(meter)
1a 12
0 0 0 0b 17
2a 6
5 5 1 2b 11
3a 7
15 12 0 0b 14
4a 8
15 12 6 7b 7
5a 9
20 20 10 10b 9
• Simulations are performed on the
new synthetic model, and coupled
with a nonlinear optimization
routine, where the investigated
parameters (slope, rugosity
magnitude) are varied, with a view
to match the original "observed"
data.
OPTIMIZED VALUES FOR ALL THE CASES
case 1a 1b 2a 2b 3a 3b 4a 4b 5a 5b
Error (%) 16.41 1.01 1.37 1.78 5.33 1.81 0.49 0.45 0.54 0.36
RESULTS & CONCLUSION
• The results were able to reproduce the impact of caprock
topography on the plume evolution.
• Uncertainties in caprock slope and rugosity may impact the
simulation outcome.
• Setting tighter bounds for the parameter ranges did not result
in a better match, which can be explained by the presence of
local optima.
CURRENT WORK
A comprehensive sensitivity analysis on the impact following parameters on
the plume shape for Sleipner model.
Parameters:
• Density (5 variables)
• Temperature (5 variables)
• Injection rate (5 variables)
• Porosity (20 perturbations)
• Permeability (20 perturbations)
• Caprock morphology (20 perturbations)
53 x 203 = 1,000,000 Simulations
Thank You