1. Integration of reservoir simulation with time-lapse seismic modeling
Ying Zou, Laurence R. Bentley, and Laurence R. Lines
Good afternoon,
My talk is on the integration of reservoir simulation with time-lapse seismic modeling. My co-authors are Dr Bentley and Dr Lines.

2. Objectives Time-lapse seismic survey processing and interpretation
Synthetic
seismic modelling
Reservoir simulation
First I want to talk little about the objectives of this work. This work is a part of on going combined study of seismic survey analysis, reservoir simulation and seismic modeling for the Pikes Peak heavy oil field. The ideal process is to run seismic processing and reservoir simulation independently first and then convert the output of reservoir simulation to synthetic seismic section. After comparing synthetic seismic to real seismic, the reservoir model is updated and the reservoir simulation is rerun. Then this process is repeated until the synthetic differences match the observed differences. In this way, the reservoir model is better constrained than by doing production history matching alone.
We reprocessed two time-lapse seismic lines and it is on the CD of this year’s report. In last year’s report we have shown the synthetic time-lapse seismic modeling based on estimated heat zones since we did not have engineering data to run reservoir simulation. Now we have received all the required engineering data for reservoir simulation from Husky and we have run a history matching on a partial reservoir for Pikes Peak field. This talk will focus on the conversion of reservoir simulation to seismic modeling.
Some of the things I am going to highlight in my presentation are on the next slide.

3. OUTLINE Theory and methodology
Geology and geophysical background of Pikes Peak heavy oil field
Reservoir production activity and reservoir model
Reservoir simulation results
Synthetic time-lapse seismic sections
Conclusions
Future work
Acknowledgements
First, is the theory and methodology that I used to convert engineering data to geophysical model,
Second, geology and geophysical information is must know background.
Then we go to the main part of this talk. A general information of the production activity and reservoir model followed by the reservoir simulation results.
Then Synthetic seismic sections with some discussion.
At last, conclusions and future work.

4. Theory and methodology
Ko, Kg, Kw, and rf from Batzle and Wang (1992)
average of: Kf=1/(So/Ko+Sg/Kg+Sw/Kw)
Kf= (SoKo+SgKg+SwKw)
Gassmann equation:
Ku = Kd + (1- Kd/Ks)2 / [ f/Kf + (1-f)/Kf –Kd/Ks2]
m Kd no change with fluid content
Batzle and Wang’s relationships are used to calculate fluid bulk moduli and densities. We chose the combination of harmonic and arithmetic average for fluid bulk modulus. We assume shear and dry bulk moduli do not change with fluid content and they were derived from well logs. Then we use Gassmann’s equation to calculate saturated bulk modulus. The detailed procedure is presented in 2001 CREWES report. Now let us look it the geology and geophysical background of Pikes Peak field.

5. Geology and geophysics of Pikes Peak heavy oil field
Waseca F~
So~80%
This is a diagram of typical Pwave, Swave, density, and Gamma ray logs. The producing reservoir is in the Lower Cretaceous Waseca Formation. It is about 500 meters below the surface. The reservoir’s porosity is around 0.32 to 0.36 and with 80% heavy oil saturation. The Waseca formation has been divided into a homogeneous, well-sorted predominantly quartz lower unit and sand-shale interbeded upper unit.
Steam drive technology has been applied to enhance recovery by reducing the effective viscosity. Next slide shows the locations of the production and injection wells and the locations of the time-lapse seismic lines.
Seam drive technology

6. Reservoir production activity
North
Two seismic lines almost overlap each other marked in blue line. The honey comb shape is 7 point steam drive pattern. The wells around the two seismic lines are mostly cyclic steam stimulation wells. We marked the 21 wells that are included in the reservoir model. The next slide is the geometry of the reservoir model.

7. Reservoir Model North 1D2-6 280mX3000m Grid size: 20mX20m
Top two layers: interbeded
Bottom layer: clean sand
CSS from 1D2-6
Steam days
History matching:
This is the partial 3D reservoir model around the seismic line that is shown as a yellow line. It is 280 m wide 3000 m long.The grid elements are 20 m X 20 m horizontally and vary in thickness. The top two layers are the interbedded sand and shale zone and the bottom layer is the clean sand zone.
Cyclic Steam Stimulation started in southern part of the reservoir in 1983 at well 1D2-6. Average steam injection duration was 10 to 30 days followed by few day’s soak and 5 to 10 month production. We have run a history matching based on the injection and production history from Jan 1981 to Aug. 2003
The next slide is the preliminary history matching results for well 1D2-6.
North

8. Reservoir simulation results Reservoir history matching
The overall history matching is not too bad. The cumulative liquid production from history file is in red and the same parameter from the simulation is in blue. The mismatching started around 1985, when well bottom hole pressure is lower than minimum bottom hole pressure constrain. The liquid rate from history file which is in green matched well with the liquid rate from simulation which is in pink. The other wells have more or less the same level of matching.
Now let us look at the temperature and pressure change with time in 3D.

9. Reservoir simulation results 3D Temperature change
From (20 years)
This is a short movie which starts from year 1981 and end at year 2000 when the second seismic survey was acquired. The direction of the temperature change is the same as the direction of production activity progress. Temperature increase is due to steam injection and temperature moves about 5 m to 8m per year.
Now let us look it the pressure change in time on a 3D view.
North

10. Reservoir simulation results 3D Pressure change
From (10 years)
This pressure change movie only run from year 1981 to year 1991 when the first seismic was shot. The pressure change spreads much faster than temperature. Now let’s have a look.
The area of interest for us is the middle profile of the model which corresponds to seismic survey profile. On the rest of the slides, we will look into details of oil saturation, temperature, gas saturation, and pressure distribution on this profile for 3 time steps.
North

11. Reservoir simulation results Reservoir oil saturation change
Jan 1981
Feb 1991
March 2000
This is the oil saturation at starting time 1981, the first seismic survey time Feb. 1991, and the second survey time March In 1991 the temperature change is still around well 1D2-6 because the production out of this profile had not have any impact on this profile yet. In year 2000, more production activities had started and the reservoir was heated on this profile. Oil flew easier due to viscosity decrease. Temperature change is shown on next slide.

12. Reservoir simulation results Reservoir temperature change
Jan 1981
Feb 1991
March 2000
On these temperature profiles, we can see the temperature change corresponds to oil saturation quite well. The highest temperature is 250 centigrade which is steam temperature. Now let’s look it gas saturation on these three dates.

13. Reservoir simulation results Reservoir gas saturation change
Jan 1981
Feb 1991
Gas saturation is appear on the top layer first and it seems appear before temperature change. For example, in 1991 on last slide, the temperature change only appear three grids away right from well 1D2-6, but on the gas saturation profile, gas appeared 12 to 13 grids away right from this well. This is because gas saturation is subject to pressure change. The pressure decrease during production spread very faster and the gas came out of the oil.
Next slide is the pressure distribution at the three dates.
March 2000
Combination of Methane and water vapor

14. Reservoir simulation results Reservoir pressure change
Jan 1981
Feb 1991
March 2000 N S
This slide future shows the pressure front spread very fast. In 1991, almost every element in the model had pressure changed.
So far, we have seen the reservoir simulation results, that are fluid saturation, temperature, and pressure change with time. Using our rock physics procedure, we obtained velocity and density distribution for the reservoir at the three time steps. Please look at the next slide for seismic model building.

15. Synthetic seismic sections Model construction
Reservoir top
Reservoir bottom S N
1D2-6
D15-6
Top Devonian
Top Cambrian
To build velocity and density model for seismic modeling, we used well logs from well 1D2-6 and D15-6 for the earth model above the reservoir. Inside the reservoir are the calculated velocities and densities from reservoir simulation output. Since we did not have well log penetrated Devonian we just give some estimated velocity and density values for the earth model underneath the reservoir.
Now let’s look the synthetic seismic section from this model.

16. Synthetic seismic sections synthetic seismic section in 1981
We did zero offset FDM modeling and post-stack migration for three models correspond to time 1981, Feb 1991, and March 2000.
This is the synthetic section for 1981 before production. The reservoir top and bottom are marked here.

17. Synthetic seismic sections synthetic seismic section in 1991
This is the synthetic seismic section for Feb 1991 after 10 years production. There is some character change at the south end.

18. Synthetic seismic sections synthetic seismic section in 2000
This is the synthetic seismic section for March 2000 after 20 years production. Character change area is larger than on 1991 section. To further look into the impact of reservoir change on seismic sections, we should look at the difference sections.

19. Synthetic seismic sections Difference between 1991 and 1981
Gas saturation
Temperature N S
This is the difference section between 1991 and We also put temperature and gas saturation plots in the reservoir on the top. We found that both temperature and gas saturation change correspond to seismic change. However, gas saturation corresponds to seismic change more directly. Thin gas saturation causes bright spot and thick gas zone causes bright spot and observable time delay. This can be seen on next slide too.

20. Synthetic seismic sections Difference between 2000 and 1991
Temperature
Gas saturation N S
This is the difference section between 2000 to If there is gas saturation in all three layers in the reservoir model, there are obvious time-delay on seismic difference section. The thin gas layer only causes bright spot.
Due to time reason we have not upgrade our reservoir model yet, but I would like to put real seismic difference section with this synthetic difference section together to show you the preliminary match.

21. 1D2-6
D15-6
D15-6
1D2-6
The up section is the difference section of seismic survey 2000 to The lower section is the synthetic seismic difference section. We can see the banding effect on both sections which is due to temperature and gas saturation lateral change.

22. Gas saturation is the dominant factor for seismic change
Conclusions
Gas saturation is the dominant factor for seismic change
Thicker gas zone corresponds to more time delay on seismic section
High temperature regions correspond to regions of seismic difference energy
Pressure propagation is much faster than temperature
Now come to our conclusions.

23. Future work
Include the moduli dependence on effective stress (reservoir pressure)
Compare seismic survey results to synthetic results
Modify reservoir model to improve the match between the synthetic and real seismic sections
In future we will add the effective pressure influence on moduli to our rock physics procedure and compare the synthetic seismic sections with real seismic sections, modify reservoir model, and go back and forth to get a possible realistic model.

24. Acknowledgements CREWES sponsors
Husky Oil, Larry Mewhort, Don Anderson
CMG, Dennis Coombe, Peter Ho
I want to think CREWES sponsors for their constant support. A special thank to Husky for contribute their data to us. We really appreciate the input from Larry and Don at Husky and we also want to thank Dennis and Peter at CMG for their help on reservoir simulation.