You are on page 1of 9

COUPLING THE RESERVOIR SIMULATOR POWERS WITH THE SURFACE FACILITIES NETWORK SIMULATOR PIPESOFT

Tareq M. Al-Shaalan, Ali H. Dogru and Larry S. Fung

Al-Shaalan graduated from the University of WisconsinMadison in 1997 with a PhD in mechanical engineering. He is a developer working on Saudi Aramcos massively parallel simulator POWERS in Saudi Aramcos Technology Development Division. He worked with Lab Research and Development for two years and has published five technical papers on combustion science technology with the National Association of Corrosion Engineers (NACE) and the Society of Petroleum Engineers (SPE). Dogru is general supervisor of Saudi Aramcos Technology Development Division, in which he supervises 20 engineers and scientists involved with reservoir and production technologies. His division develops and implements new technologies for reservoir and production engineering with an emphasis on multiphase flow meters, downhole separation and downhole monitoring. Dogru has an MSc from the Technical University of Istanbul and a PhD from the University of Texas in Austin (UT). Prior to joining Saudi Aramco, he worked for Mobil Research & Development Company and Core Labs Inc., both in Dallas. He held academic and teaching positions at UT in mechanical engineering, at the California Institute of Technology in chemical engineering and at the Norwegian Institute of Technology in petroleum engineering. He has 25 years of experience in the oil and gas industry and has published 25 technical papers for the SPE and other scientific journals.

SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002 13

Fung supervises Reservoir Technology in Saudi Aramcos Technology Development Division. He is a lead developer for the massively parallel simulator POWERS. Prior to joining Saudi Aramco he was with Epic Consulting Services Ltd. Prior to that he spent 11 years at the Computer Modeling Group, where he developed features for its commercial simulators. He has published more than 20 technical papers on reservoir simulation techniques such as adaptive implicit method, dual porosity dual/permeability modeling and control volume finite element method, as well as coupled petroleum geomechanical modeling. Fung holds BS and MS degrees from the University of Alberta in Canada. He has 20 years of experience in the oil and gas industry.

INTRODUCTION
Super computers and large data storage systems make the modeling and simulation of complex problems more viable. The latest technology in parallel computers allows the simulation of hydrocarbon movement in giant reservoirs coupled with the surface network to be accomplished in a practical time frame. Reservoir simulation can be done accurately without the surface network if the well flow rates are predicted correctly. However, it is very difficult to predict the correct well flow rates without knowledge of the capacity of the surface network. The wells constitute the inner boundaries of the reservoir. They are connected together on the surface through a network of pipes (fig.1). The deliverability of the wells and the reservoir depends on the capacity of the surface network, in addition to reservoir pressure and reservoir productivity, or injectivity. The flow rates of the wells may change with time. Due to the nature of the dynamic flow in the surface network, any change in the flow rate of one of the wells will affect the flow rates of other wells. Shutting down one of the wells could increase the deliverability of the other wells in the same surface network. Increasing the flow rate of one of the wells could cause backflow in the other wells.

ABSTRACT
Two simulators play an important role in navigating Saudi Aramcos reservoirs: the Parallel Oil, Water and Gas Simulator (POWERS) is an in-house black-oil simulator; PIPESOFT is a commercial product that simulates steady state three-phase flows of oil, water and gas in pipes and pipe networks. These simulators are coupled at the sand face via in-house software. This procedure eliminates the need to generate hydraulic flow tables. With this coupling, fluid flow is modeled from the reservoir to the gas-oil separation plant (GOSP). As a result, the uncertainty of wellhead pressure required by the flow-table approach is no longer needed. As the coupled simulation proceeds, PIPESOFT receives continuous data for reservoir pressure, productivity index, water cut and gas-oil ratio from POWERS. This process mimics the field process better and results in more realistic rate predictions than the conventional flow-table approach. Since it provides a more rigorous solution, the predicted rates can be substantially different than those of the flow-table method. The coupled simulator was tested on two offshore oil fields. The first example comprised 20 years of prediction involving 60 wells. For comparison purposes, simulation runs were carried out with and without coupling. The results of the coupled model showed that the use of flow tables with constant wellhead pressure could over-predict or underpredict actual well performance. Decoupling under-predicted the field target oil rates by 20 percent. The second offshore field, with a complicated surface network (250 wells), was tested for 40 years of prediction. The coupled simulation was stable and added only 10-13 percent overhead to the total computer time.

Well Interior node GOSP Linkage


Fig. 1. An example of a surface network.

14 SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002

To improve the accuracy of the well flow rates, the reservoir simulator needs to be coupled with the surface network simulator. Therefore, the fluid movement can be modeled all the way from the reservoir to the plant (e.g., injection plant or GOSP). The coupling provides an efficient tool for better future planning and development opportunities for both the reservoir and the surface network. Coupling reservoir simulators with surface network simulators goes back to Dempsey J. R. (1971). More development and research on the coupling, or full-field model, have been done since then. Wallace and Van Spronsen (1983) and Stoisits, R. F. et al. (1992) implemented the full-field model in their studies of the simulation of the reservoir coupled with the surface network. All the studies agree that the coupling gives a more realistic result for future development and optimization. There are many ways to couple a reservoir simulator with a surface network simulator (Trick, M.D. (1998) and Breaux, E.J. et al. (1985)). The coupling can be done at the following locations: Wellhead with inflow performance relationship; Reservoir level with inflow performance relationship; Sand face; and Trap level with inflow performance relationship. Trick and Breaux et al., discuss these methods in detail. Hepguler et al. (1997) integrated the famous reservoir simulator Eclipse with the surface network simulator Netopt. The data shared between the two simulators is transferred through a parallel virtual machine (PVM). Trick coupled Eclipse with another surface network simulator, FORGAS. PVM is employed to transfer the data between the two simulators. This paper discusses the integration of the reservoir simulator POWERS, (Dogru, A.H., et al. 1999) with the surface network simulator PIPESOFT.

R E S E RV O I R S I M U L AT O R : P O W E R S
POWERS is Saudi Aramcos in-house black-oil simulator. It utilizes the latest technology in parallel computing to simulate a multi-million cell model in a reasonable amount of time. The governing equations used to simulate the reservoir are integrated in time using either implicit or IMPES methods. POWERS well models apply many constraints on bottom hole pressure that will be used to flow the wells. The bottom hole pressures executed by POWERS are: Bottom hole pressure corresponding to maximum well flow rate (BHPQ); Bottom hole pressure corresponding to bubble point pressure for producer (BHPbp); Bottom hole pressure corresponding to fracture pressure for injector (BHPfr);

Bottom hole pressure corresponding to maximum draw-down pressure for producer (BHPDDP); Bottom hole pressure corresponding to maximum build-up pressure for injector (BHPBUP); and Bottom hole pressure corresponding to minimum wellhead pressure (BHPWHP). For producer wells, the operating bottom hole pressure (BHPprod) that POWERS utilizes in its calculations is: (1) BHPprod = max (BHPQ, BHPmin, BHPDDP, BHPWHP) For injector wells, the operating bottom hole pressure (BHPinj) is: (2) BHPinj = min (BHPQ, BHPmax, BHPBUP, BHPWHP) The operating bottom hole pressures calculated from equations 1 and 2 will be manipulated further to produce the target flow rate for a given group of wells. Like other reservoir simulators, POWERS knows the capacity of the surface network through the wellhead pressure (WHP). The bottom hole pressure BHPWHP is calculated from a given WHP and flow table generated from a simulator for the surface network. A constant WHP is assumed during the reservoir simulation. However, in real time, WHP pressure is dynamic. The time characteristic of the pipe flow is within a minute. The time characteristic of the reservoir flow is within a day. Since the time characteristic of the pipe flow is very small, any small change in the flow rate of one of the wells will be felt instantaneously by the other wells connected to the same surface network. As a result, WHP at one well is affected by the flow rates of others wells in the same network. It assumes constant WHP is reasonable for high productive, or injective wells. For low productive, or injective wells, the assumption of a constant WHP pressure may fail. Another drawback of using wellhead pressure is the error of interpolation associated with the flow tables. Flow tables are generated for different flow rates, well pressures (P), water cuts (WC), gas-oil ratios (GOR) and WHPs. The interpolation is typically employed to determine the well flow rate for a given P, WHP, WC and GOR. The accuracy of the interpolation depends on the method used in the interpolations. The error of interpolation is generally high for multiphase flows (Schiozer, D.J. and Aziz, K. 1994). To reduce the uncertainties in calculating WHP and to improve the prediction of the reservoir simulation, POWERS needs to be linked with another simulator for the surface network. The coupling eliminates the need for the flow table and WHP. The well flow rate is calculated directly from the surface network simulator.

S U R FA C E N E T W O R K S I M U L AT O R : P I P E S O F T
On the surface, the wells are connected together through a network of pipes (fig. 1). The surface network consists of

SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002 15

nodes and linkages. It could be a gathering network, a distributive network or a combination of both. A linkage consists of pipes, tubes, traps, valves, pumps, compressors, separators, chock valves and other mechanical devices. The nodes are either boundary nodes or interior nodes. The boundary nodes are wells, sources and sinks. The interior nodes are gathering points for the linkages. The water injection plant is an example of a source node. The GOSP is an example of a sink. Detailed information on the nodes and linkages can be found in any manual of a surface network simulator. PIPESOFT is a commercial software that simulates steady-state three-phase flow in the surface network. The governing equations required to solve for the surface network are complex and nonlinear. Therefore, PIPESOFT uses the Newton-Raphson method to solve the equations iteratively. If the flow is multiphase and the surface network is complex, the solution may not converge. Low well pressure may also cause the solution not to converge. The simulation of a surface network requires either pressure or flow rate to be specified at the boundary node. For the case of multiphase flow simulation, WC and GOR are specified at the source, or well node. The productivity or injectivity index (PI) has to be given when the boundary node is a well. Once the boundary values are given, PIPESOFT calculates, for example, temperature, pressure drops and the flow rate in each branch of the network. The output of PIPESOFT is valuable for optimizing the surface network. When POWERS is coupled with PIPESOFT, the flow rates or the pressures at the wells are valuable for reservoir simulation. The values P, WC, GOR and PI for each well are changing with time. Most surface network simulators model the reservoir as a big tank to mimic P as a function of time. However, for a heterogeneous reservoir, or tight reservoir, this assumption results in a high error. The big tank model, also, is not able to reproduce WC, GOR and PI as a function of time. To accurately model the surface network and get the time data for P, WC, GOR and PI, PIPESOFT has to be linked with a reservoir simulator.

COUPLING POWERS WITH PIPESOFT


A full-field model is essential to optimize production of a reservoir and to design for the optimum surface network. A reservoir simulator coupled with a surface network simulator gives a more rigorous solution than when both run as stand-alones. There has been work already done to link the reservoir simulator Eclipse with the surface network simulators, FLOGAS (Trick, M.D., 1998) and Netopt (Hepguler, G. et al. 1997). In this paper, POWERS and PIPESOFT are linked together.

PIPESOFT runs on IBM and POWERS runs on CM5, IBM or SGI. For these simulators to be linked together and work as one, an interface (POPSI) is developed to transfer data between the two simulators. As mentioned in the introduction, there are many ways to couple a reservoir simulator with a surface network simulator. To eliminate the need for the flow table, both POWERS and PIPESOFT are coupled at the sand phase (Trick, M.D., 1998; Hepguler, G. et al. 1997). The algorithm of the interface POPSI is given in fig. 2. At the beginning of time integration, POWERS calls POPSI to send P, WC, GOR and PI for each well to PIPESOFT. To minimize the overhead of data transfer, POPSI puts all the data in one buffer and sends it to the machine where PIPESOFT runs. The data is transferred through a Unix socket. Then POPSI extracts the data for PIPESOFT and waits for PIPESOFT to finish its calculations. When the calculations have converged, POPSI extracts the well flow rates and sends them to POWERS. In case the solution does not converge, POPSI uses the old well flow rates and sends them to POWERS as a temporary solution for the current time step. The good initial estimations for the unknown pressures and flow rates in the surface network are necessary for the solution of PIPESOFT to converge. POPSI employs the following estimates to help the solution converge. The nodal flow rates of the previous time step are used as initial estimates. The nodal pressures, other than well pressures, of the previous time step are utilized as initial estimates. The wells that are closed at the previous time step are kept closed all the time. When POWERS receives the well flow rates from PIPESOFT (Qpipesoft), it calculates the bottom hole pressure, BHPpipesoft from Qpipesoft. (3) BHPpipesoft = P - Qpipesoft / PI For producer wells, the operating bottom hole pressure (BHPprod) is calculated from (4) BHPprod = max (BHPQ, BHPmin, BHPDDP, BHPpipesoft) For injector wells, the operating bottom hole pressure (BHPinj) is calculated from (5) BHPinj = min (BHPQ, BHPmax, BHPBUP, BHPpipesoft) Note that when POWERS is coupled with the surface network, there is no need for BHPWHP (compare equations 1, 2, 4 and 5). The pressures BHPprod and BHPinj calculated from equations 4 and 5 will be manipulated further to meet the target flow rate for a given group of wells.
Example Application

The coupled simulators were tested on two Saudi Aramco offshore oil fields, A and B. Field A is modeled with the

16 SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002

Start of time

POWERS calls POPSI to send pressure, productivity index, WC and GOR to PIPESOFT

POPSI extracts the data for PIPESOFT

POPSI runs PIPESOFT

NO Calculations converge POPSI restores old flow rates

YES

POPSI extracts well flow rates and sends them to POWERS

POWERS compares the flow rates from PIPESOFT with other constraints to choose the optimum well flow rates

1,330 psi and a maximum injection pressure of 7,500 psi. The field oil target rate was set to 120 MBPD from 1999 to 2000. Then it was raised to 140 MBPD until 2001. After that it was reduced back to 120 MBPD to 2020. There were no drilling or workovers allowed to keep the target oil rate. When there was over production in the oil rates, the wells with WC less than 10 percent were scaled down. Any producer well with an oil rate of less than 250 BPD was shut in. Field B was modeled with the grid points of 70, 168 and 8 in the x, y and z directions, respectively. Two-phase flow (oil and water) was assumed in the reservoir. The simulation of the surface network considers the evolution of the gas. The history period covered 1950 to 1998. The prediction period was from 1998 to 2040. During the prediction period, the field was operated to its potential rate to test the stability of the coupling. The two simulators were coupled at the beginning of the prediction period. There were 250 producing wells coupled with the surface network. The injectors were not part of the surface network. Three cases were considered: Case A1 field A without coupling; case A2 field A with coupling; and case B1 field B with coupling. For case A1 (uncoupled case), the reservoir simulation was accomplished using flow tables and wellhead pressures of 300 psi for producer wells.

R E S U LT S A N D D I S C U S S I O N
The effect of coupling on the field prediction is demonstrated through case A1 and case A2. The field oil rates for case A1 and case A2 are plotted in fig. 3. The oil target rate for case A1 can be maintained until 2003. Then it starts to decline due to low reservoir pressure. However, with the coupling (case A2), the oil target rate can be sustained for an additional four years before it starts to decline. The number of active wells for both cases is plotted in fig. 4. The number of active wells for the two cases agrees very well until year 2003, when case A1 predicts a decline in reservoir deliverability. In 2003, the flow table of case A1 starts to shut in some of the producing wells due to backflow resulting from high WC and low reservoir pressure. On the other hand, due to surface network calculations, the surface network has enough capacity to flow these closed wells for four more years. The use of flow tables and wellhead pressures of 300 psi underestimates the oil potential rate for field A (fig. 5). The decline in field potential is due to a loss in the reservoir pressure and a high WC. The accurate field potential cannot be accomplished without the coupling. The field potential is influenced by the capacity of the surface network in addition to the reservoir pressure. The larger the surface network the higher the field potential. Installing

POWERS does the time integrations

Fig. 2. The algorithm (POPSI) for the interface that couples POWERS with PIPESOFT.

grid points 84, 128 and 128 in the x, y and z directions, respectively. Two-phase flow (oil and water) was assumed in the reservoir (the reservoir is kept above its bubble point pressure). Even though the reservoir pressure is maintained above the bubble point pressure during the reservoir simulation, the simulation of the surface network considers the evolution of the gas. The history match for field A covers the period from 1970 to 1999. On the other hand, the field prediction period covers 1999 to the year 2020. The coupling of the two simulators started at the beginning of the prediction period. There were 60 producing wells coupled with the surface network. The injector wells were not part of the surface network. Field A was simulated with a bubble point pressure of

SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002 17

160 140 Case A2 Oil Rate (MSTBD) 120 Rate (MSTBD) 100 80 60 40 20 0 1995 2000 2005 2010 Year 2015 2020 Case A1

250 200 150 100 Case A1 50 0 1995 2000 2005 2010 Year 2015 2020

Case A2

Fig. 3. The field oil rate for case A1 and case A2

Fig. 5. The field oil potential rate for case A1 and case A2

60 50 Well Counts 40 30 20 10 0 1995 2000 2005 Year 2010 2015 2020 WHP (psi) Case A1 Case A2

300 250 200 150 100 50 0 1995 2000 2005 Year


Fig. 6. Wellhead pressure for well W1 and case A2

2010

2015

2020

Fig. 4. The number of producers wells that are flowing for case A1 and case A2

pumps or compressors in the surface network increases the field potential. Chocked valves in the surface network reduce the reservoir potential. If the surface network goes through a high elevation, some of the wells cannot flow. Five wells are presented in this paper to demonstrate the effects of the coupling on well performance. These wells are given the names W1, W2, W3, W4 and W5. The rest of the wells have more or less the same behavior. For case A1, the flow table and the wellhead pressure of 300 psi are used to estimate the capacity of the surface network. However, coupling (case A2) proves that the assumption of the wellhead pressure of 300 psi underestimates the capacity of the well W1 (fig. 6). From the calculations of the surface network, the wellhead pressure is very close to 300 psi at the start of the prediction period. Then it starts to decline to 200 psi at the end of the simulation. Therefore, the assumption of a wellhead pressure of 300 psi is valid only at the beginning of the prediction peri-

od. The wellhead pressures for all producing wells have the same behavior as W1. The use of the flow table (case A1) may under-predict or over-predict well performance. For W2, case A1 underestimates the oil rates by 600 BPD at 2000 (fig.7). At the end of the simulation, both case A1 and case A2 predict roughly the same oil rate. On the other hand, case A1 overestimates the oil rate for W3 by 2,000 BPD at 2000 (fig. 8). Case A1 and case A2 predict roughly the same oil rates for W4 through all the periods of the simulation. The difference in the oil rate of W4 between the two cases is less than 50 BPD (fig. 9). Case A1 shuts in the well W5 (fig. 10) due to high water cut. The water cut for W5 is 92 percent. However, the surface network calculations show that the surface network still has enough capacity for W5 to flow to the network. Fig.7 and fig. 8 show that when the prediction starts in year 1999, case A1 predicts a sudden change in the oil rates

18 SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002

1.2 1.0

1.2 1.0 Oil Rate (MSTBD) 0.8 0.6 0.4 0.2 0.0
2000 2005 Year 2010 2015 2020

Oil Rate (MSTBD)

0.8 0.6 0.4 0.2 0.0 1995 Case A1

Case A2

Case A2

Case A1

1995

2000

2005 2010 Year

2015

2020

Fig. 7. Oil rate of W2 for case A1 and case A2

Fig. 9. Oil rate of W4 for case A1 and case A2

6.0 5.0 Oil Rate (MSTBD) Case A1 Case A2 Oil Rate (MSTBD) 4.0 3.0 2.0 1.0 0.0 1995 2000 2005 2010 Year 2015 2020

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.0 1995 2000 2005 2010 Year 2015 2020 Case A1 Case A2

Fig. 8. Oil rate of W3 for case A1 and case A2

Fig. 10. Oil rate of W5 for case A1 and case A2. Note that the well is closed for case A1.

for W2 and W3. It reduces the oil rate for W2 sharply to 200 BPD and it increases the oil rate for W3 sharply to 5,800 BPD. Case A2, however, makes a smooth transition from the history simulation to the prediction simulation. PIPESOFT utilizes the iterative Newton-Raphson method to solve for the non-linear equations of the surface network; due to the high non-linearity of the equations, the solution may require lots of iterations. Fig. 11 plots the number of iterations for the surface network for case A2. Initially, the number of iterations is very high, 52 iterations, because the initial estimates for the unknowns are far from the solution. The closer the initial estimates are to the solution, the faster the solution will converge. For the reservoir simulation, the well rate and pressure at the previous time step are close to the solution of the current time step. Using the solution at the previous time step as the initial estimate decreases the number of iterations greatly from 52 iterations per time step to five iterations per time step.

Sometimes the number of iterations gets as low as two iterations per time step. The average number of iterations per time step is 6.7. One reason a reservoir engineer may prefer to use the flow table over the coupled model is the increase in CPU time for the coupled model. But it is proven in this study that this is not always the case. In this study, POWERS runs on CM5 and PIPESOFT runs on IBM or SP2. The total time of simulation for case A1 and case A2 is 21.6 minutes and 23.7 minutes, respectively. The calculations of PIPESOFT add only 10 percent extra time to the total time of the reservoir simulation. This 10 percent is very acceptable considering the accuracy resulting from the coupling. The stability of the coupling is also tested on case B1 where field B flows to its oil potential rate and it has 240 wells coupled with the surface network. For this case, PIPESOFT adds 13 percent extra time to the total time of the reservoir simulation. The calculations of PIPESOFT are

SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002 19

60 50 Number of Iterations 40 30 20 10 0 1995 2000 2005 Year 2010 2015 2020 Number of Iterations

300 250 200 150 100 50 0 1990 2000 2010 2020 Year
Fig. 12. The number of iterations of PIPESOFT for case B1

2030

2040

2050

Fig. 11. The number of iterations of PIPESOFT for case A2

stable all the time. Fig. 12 plots the number of iterations for the surface network to solve. The highest number of iterations per time step is 300 and it happens at the beginning of the prediction period. The lowest number of iterations per time step is four. The average number of iterations per time step is 11.

Subscript

CONCLUSIONS
The parallel reservoir simulator POWERS was coupled with PIPESOFT to improve the accuracy of reservoir deliverability predictions. The coupling removed the uncertainty in the wellhead pressures for the reservoir simulator and the reservoir pressures for PIPESOFT. With this coupling, the fluid flow was modeled all the way from the reservoir to the GOSP. The coupled simulator was demonstrated in two oil fields for prediction times of 20 years and 40 years. The results showed that the use of constant wellhead pressures and the flow tables could over-predict or under-predict the well deliverability. The numerical experiments indicated that coupling was computationally stable and added only 10 to 13 percent overhead to the reservoir simulation time.

bp BUP DDP fr inj l o prod

Bubble point pressure Build-up pressure Draw-down pressure Fracture pressure Injector Liquid Oil Producers

ACKNOWLEDGMENTS
The authors acknowledge Rick Pawlas for providing the field data and K.M. Rafique for the surface network data.

REFERENCES
Breaux, E.J. et al. 1985. Application of a Reservoir Simulator Interfaced with a Surface Facilities Network: A Case History, SPE 11479, Society of Petroleum Engineering Journal, June. Breaux, E.J. et al. 2000. Linking Reservoir and Surface Simulators: How to Improve the Coupled Solutions, SPE 65159 presented at SPE European Petroleum Conference, Praise, France, Oct. 24-25. Dempsey, J.R. et al. 1971/ An Efficient Method for Evaluating Gas Field Gathering System Design, JPT, P.1067-1073, September. Dogru, A.H., et al. 1999. A Massively Parallel Reservoir Simulator for Large Scale Reservoir Simulation, SPE paper 51886 presented at the 1999 SPE Reservoir Simulation Symposium, Houston, Texas, Feb. 14-17.

N O M E N C L AT U R E
BHP GOR GOSP P PI Q WC WHP Bottom Hole Pressure Gas Oil Ratio Gas Oil Separation Plant Well Pressure Productivity or Injectivity Index Well Flow Rate Water Cut Wellhead Pressure

20 SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002

Haugen, E.D. et al. 1995 Simulation of Independent Reservoir Coupled by Global Production and Injection Constrains, SPE 29106 presented at 13th SPE Symposium on Reservoir Simulation, San Antonio, Texas, Feb. 12-15. Hepguler, G. et al. 1997. Integration of a Field Surface and Production Network with a Reservoir Simulator, SPE 38937, SPE Computer Applications, p.88-92, June. Khodri, D. et al. 1997. Modeling of Reservoir to Surface network Coupling: A Case Study of Gas Storage Management, SPE 38864 presented at the 1997 SPE Annual Technical Conference and Exhibition, San Antonio, Texas, Oct. 5-6. Schiozer, D.J., and Aziz, K. 1994. Use of Domain Decomposition for Simultaneous Simulation of Reservoir and Surface Facilities, SPE paper 27876 presented at Western Regional Meeting, Long Beach, Calif., March 23-25. Stoisits, R.F. et al. 1992. Application of Nonlinear Adaptive Modeling for Rigorous Representation of Production Facilities in reservoir Simulation, SPE 24898 paper presented at the 67th Annual Technical Conference and Exhibition of SPE, Washington, D.C., UAS, Oct. 4-7. Trick, M.D. 1998. A Different Approach to Coupling a reservoir Simulation with a Surface Facilities Model, SPE 40001 paper presented at 1998 SPE Gas Technology Symposium, Calgary, Alberta, Canada, March 15-18. Wallace, D.J., and Van Spronsen, E. 1983. A Reservoir Simulation Model with Platform Production/Injection Constraints For Development Planning and Reservoir Management at the Kuparuk Revier Field, SPE paper 20045 presented at Reservoir Simulation Symposium, San Francisco, Calif., Nov. 15-18.

SAUDI ARAMCO JOURNAL OF TECHNOLOGY FALL 2002 21

You might also like