Professional Documents
Culture Documents
SPE 69690 SAGD Performance Optimization Through Numerical Simulations: Methodology and Field Case Example
SPE 69690 SAGD Performance Optimization Through Numerical Simulations: Methodology and Field Case Example
It is shown that an optimized steam chamber development is Stable gravity displacement is particularly important to reach
obtained by adjusting the steam injection rate to the potential a favorable energy balance. In SAGD, the heated oil remains
of the reservoir (fluids and geology) and by monitoring the always in contact with the heated region, as it gets drained
production rate during the process/operations to keep the along the sidewalls of the steam chamber [9]. Thus, energy
steam chamber as large as possible but away enough from the losses from heated oil, which has not been produced, are
production well to prevent any steam breakthrough. The minimized. The energy balance is commonly estimated by the
results are in good agreement compared with Butler's Steam Oil Ratio (SOR), which represents the volume of steam
analytical model (oil rate and steam chamber shape). A very necessary to produce one volume unity of oil. On the
good history match is obtained in the field case example. The economic standpoint, SAGD is thus very interesting because
2 P. EGERMANN, G. RENARD, E. DELAMAIDE SPE 69690
shape degrades to fit the low rate required (smaller chamber Field Case application
due to unproduced heated fluids, lower slope of its walls). In that part, the previous methodology is illustrated by an
Simulations with optimization application on a SAGD field case with real data [16].
The base case is now simulated with the optimization Reservoir parameters of the Celtic field
procedure described in the previous section. The opening of Following the 1996 success of two Single Well SAGD wells
the wells to the nominal regime is established progressively in in the Celtic field, Saskatchewan, Mobil Oil Canada piloted a
70 days. Dual Well SAGD well pair in 1997 [16]. The well pair was
Qualitatively, the observed steam chamber shape is more located in the Sparky-General Petroleum « channel » sand
triangular (Figure 10) and closer to what can be expected with a net pay thickness of about 20 m and a dead oil
from Butler's model. The frontier of the chamber remains viscosity of about 15,000 cP. Table 3 lists key reservoir
between the two wells, which means that there are no parameters that have been confirmed from the SAGD pilot.
unproduced heated fluids. The four pictures of Figure 11 The reservoir is quite homogeneous without any shale
enable to follow the steam chamber development from the barriers. The oil is heavy, however light enough to allow a
rising period to the lateral extension regime. The impact of start-up using and Huff & Puff type procedure. In the present
the regulation is particularly sensitive on the BHT since study, a preheating phase has been simulated through heat
temperature is our control parameter. Figure 12 shows that conduction of the two wells.
DT is maintained at around 25°C all along the simulation.
Compared to the non-optimized simulation, the general From the information given in the literature, it is clear that a
evolution of flow rates is the same except for the water rate. A Black-Oil history match simulation can be done. Oil viscosity
decrease is observed for the longer times. Hence, monotony of versus temperature has been derived using the Svrcek and
the water and oil rates is the same when the process is Mehrotra correlation. At steam temperature, i.e. 240° C, oil
optimized. This behavior can be attributed to the non-storage viscosity is close to 3 cP. A good drainage can therefore be
of unproduced fluids at the basis of the steam chamber. When anticipated.
retention time of heated fluids into the steam chamber The injection-production history shown on Figure 17 has been
increases, heat has to be provided to maintain their derived from Fig. 11 of the original paper. In a first step it is
temperature. Practically, this is achieved by the condensation simulated straightforward without implementation of any well
of a fraction of the injected steam. This lost steam is produced monitoring option.
rapidly and is not used to develop the chamber. It makes the Prior analysis of field data
bottom of the steam chamber going up which increases the From the data given in the article, it is possible to anticipate
part of injected steam used to maintain the temperature. This what will be the shape of the steam chamber :
phenomenon coupled with gravity segregation of oil and Ø A temperature differential of about 60° C between the
water leads to a gradual increase of the water rate. injection and production wells is indicated. It suggests a
Compared to the first simulation, oil rate is multiplied by a relatively wide distance between the steam chamber and
factor two (Figure 13). The steam injection rate follows the the producer.
same evolution as the produced water rate, which shows that Ø Production curves of Figure 17 are similar to those
the injected water is entirely recycled by the producer after obtained when simulating without adapted production
steam condensation. The phenomenon of condensation can be schedule. Instead of the same monotony, oil and water
observed by plotting the water saturation profiles (Figure 14). rates do not vary in the same way at long times. This
In the virgin area, Sw is equal to Swi (0.15). At the border of expresses a degradation of the Water Oil Ratio.
the chamber, Sw gets through a maximum around 0.45.
All these indications out into evidence a better development of Therefore, an accumulation of hot fluids is predictable at the
the steam chamber in terms of energy balance as showed by basis of the steam chamber, just above the production well.
Figure 15. A significant decrease of the cumulative SOR is Results of the history match
obtained in the optimized case. Due to the monitoring control Figure 18 shows that the match is pretty good between field
and the adapted injection rate, heated fluids do not stay observed flow rates at the producer (oil and water) and
unproduced in the bottom part of the chamber. That helps simulated ones. The temperatures at the two wells and
lateral development of the chamber, decreases the WOR and corresponding temperature differential are plotted in Figure
heat losses and favors the SOR. 19. Values are similar to what has been observed on the field,
60° C. The shape of the steam chamber shown on Figure 20
Comparison with the complete model from Butler is provided confirms the conclusions that have already been drawn
on Figure 16. Details on the modelling approach can be found concerning the accumulation of hot fluids at the basis of the
in reference 1. The best fit was obtained for a rising steam chamber. We emphasise that these results have been
parameter equal to 2.6 and oil permeability equal to 0.7 obtained without any particular adjustment of the simulated
Darcy. data.
SPE 69690 SAGD PERFORMANCE OPTIMIZATION THROUGH NUMERICAL SIMULATIONS: METHODOLOGY AND FIELD CASE EXAMPLE 5
Vertical permeability D 1
Porosity % 35 Mobile fluids
Viscosity at Ti cP 5000
Table 1: Reservoir properties
Underburden
Temperature °C Oil viscosity cP
27 5000 Legend: Producer
Injector
66 242
121 23
177 6.4
232 2.9 Figure 1: SAGD principle
282 1.8
316 1.4
1.E+04
Table 2: Viscosity evolution with temperature
1.E+03
Depth m 490
Net pay m 20
Porosity % 33 Viscosity cP 1.E+02
Oil saturation % 80
Gravity API 12° 1.E+01
Initial Reservoir Pressure KPa 2,900
Initial reservoir Temperature °C 24°C @ 490 m
1.E+00
Horizontal Permeability D 7 (4)
Vertical Permeability D 1
Oil Viscosity @ 24° C cP 12 500 1.E-01
Bottom Water Thickness Normally zero 0 200 400 600
Temperature °c
Table 3: Typical reservoir parameters (Celtic field)
Producer
0.55 0.55 0.5 0.45
Table 5: Water saturation scale Figure 3: Synthetic case: gridding of the reservoir
SPE 69690 SAGD PERFORMANCE OPTIMIZATION THROUGH NUMERICAL SIMULATIONS: METHODOLOGY AND FIELD CASE EXAMPLE 7
300
250
Temperature °c
Inj 200
Prod T Prod
150 T Inj
10 days 50 days DT
100
50
0
0 200 400 600 800 1000
Time days
100 days 150 days
Figure 4: Synthetic case: temperature profiles during th Figure 7: Synthetic case: non-optimized BHT
preheating period
80
500 days
70
60
rate m3/d
50
40
30
Qoil surface
20
Qwater surface 1000 days
10 Qsteam surface
0
0 200 400 600 800 1000
Time days
Figure 8: Synthetic case: gas saturation profiles
Figure 5: Synthetic case: non-optimized rates
3.0
500 days
Cum SOR
2.5
Inst SOR
2.0
SOR
1.0
1000 days
0.5
0.0
0 200 400 600 800 1000
Time days
Figure 6: Synthetic case: non-optimized SOR Figure 9: Synthetic case: temperature profiles
8 P. EGERMANN, G. RENARD, E. DELAMAIDE SPE 69690
200
500 days
150
Rates m3/d
Triangular shape
100
Qoil surface
50 Qwater surface
1000 days Qsteam surface
0
0 200 400 600 800 1000
Time days
Figure 10: Synthetic case: Sg profiles (optimized)
Figure 13: Synthetic case: Rates (optimized)
500 days
Condensation front
200 days Rising period 250 days
1000 days
Figure 11: Synthetic case: Temperature (optimized) Figure 14: Synthetic case: Sw profiles (optimized)
300 2.5
250 2.0
Temperature °c
200 T Prod
1.5
SOR
T Inj
150
DT 1.0 Optimized
100
Non optimized
0.5
50
0 0.0
0 200 400 600 800 1000 0 200 400 600 800 1000 1200
Figure 12: Synthetic case: BHT (optimized) Figure 15: Synthetic case: comparison of SORs
SPE 69690 SAGD PERFORMANCE OPTIMIZATION THROUGH NUMERICAL SIMULATIONS: METHODOLOGY AND FIELD CASE EXAMPLE 9
300
Oilrate bottom conditions m3/d
160
250
140
Temperature °c
120 200
100
150 Tprod
80 Tinj
60 Simulation 100 DT
40 Butler 50
20
0 0
0 200 400 600 800 1000 0 200 400 600
Time days of production Time days
Figure 16: Synthetic case: comparison with Butler's model Figure 19: BHT from history match
210 days
550 days
Figure 17: Celtic injection-production history Figure 20: Steam saturation at two different times
400
200 Field oil rate Field water rate
350 Optimized water rate Optimized oil rate
175
300
150
Rates m3/d
125 250
Rate m3/d
100 200
75 150
Field oil rate
50 Field water rate 100
Simulated water rate Incremental production
25 Simulated oil rate 50
0 0
0 200 400 600 0 100 200 300 400 500 600
Time days Time days
300 300
200 200
150 Tprod 150
100 Tinj Simulation
DT 100
Butler
50 50
0 0
0 200 400 600 0 200 400 600
Time days Time days
Figure 22: BHT after optimization Figure 25: Comparison with Butler's model
2.5
2.0
Cumulative SOR
1.5
1.0
Optimized case
0.5 Base case
0.0
0 200 400 600
Time days
210 days
550 days