Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

SPE-174039-MS

Fast Production Optimization with Decline Curve Analysis Under Facility


Constraints: A Field Case Study
P. Kritsadativud, and B. Jafarpour, University of Southern California; P. Ekkawong, PTT Exploration and
Production Public Company Limited

Copyright 2015, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Western Regional Meeting held in Garden Grove, California, USA, 27–30 April 2015.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Production optimization can play a major role in increasing recovery and decreasing operation cost. In
many oilfields, the geology, production operations, and their related constraints are very complex. These
complexities can complicate the formulation and solution of the pertinent optimization problems and
increase the computational cost of finding a solution. Although full reservoir simulation provides detailed
analysis and prediction of reservoir performance, the significant uncertainty and complexity of reservoir
models can make the simulation results and their interpretations questionable. Moreover, in some cases,
a reservoir model may not even be available to perform full simulation for performance optimization. The
cost and complexity of developing full-scale simulation models, together with the considerable compu-
tational overhead associated with production optimization (especially under geologic uncertainty), call for
development of fast proxy models for production optimization. To this end, various reduced-order and
surrogate models have been designed to approximate the production behavior of a reservoir at a fraction
of the computation required for full simulation.
We present an efficient production optimization scheme by integrating constrained optimization with
fast decline curve analysis for predicting well production performance. The proposed production optimi-
zation approach is formulated as a constrained optimization problem by defining a desired objective
function and a set of existing field/facility constraints. An efficient gradient-based optimization algorithm
is then adopted to solve the resulting optimization problem for a single timestep. The optimization is then
coupled with the decline curve analysis to predict future production rates. The optimization process is
performed recursively in time for a specified duration. The predictions with the decline curve analysis are
reasonable so long as the operating conditions remain unchanged. Using field data, we demonstrate that
the proposed formulation can provide fast solutions to large-scale production optimization problems. The
results in this paper suggest that the developed technique can be applied to improve production
performance and operation efficiency with a minimal computational cost when compared to production
optimization with full-scale reservoir simulation. It also offers the flexibility to adjust the problem
formulation under various field conditions and is particularly useful when a full-scale reservoir model
does not exist simulate the reservoir response for production optimization.
2 SPE-174039-MS

Introduction
The complexity of geologic models and facilties and operations constraints often complicate the formu-
lation and solution of production optimization problems. Reservoir model development is an intensive and
costly process that is also subject to several sources of uncertainty, an inherent property in modeling
subsurface systems. Besides the complexities and costs involved in developing full reservoir simulation
mdoels, using such models to investigate and identify optimal production strategies is a computationally
demanding taks, especially when robustness against geologic, economic, and field uncertainties are to be
incorporated. In additional to uncertainties involved in constructing geologic and simulation models, in
many cases, such as fields operated by small producers, a full reservoir simulation model may not exist.
These diffcultties have led to research and development efforts to build efficient approximate proxy (or
surrogate) models that can be easily applied to production optimization.
Here, we present an efficient scheme that integrates constrained optimization with decline curve
analysis for efficient production performance forecasting and optimization. The approach is also appli-
cable to oilfields with gas lift as it can identify the optimal gas lift distribution to maximize a desired
performance metric subject to production facility constraints. The use of well-established decline curve
analysis allows for faster computation, which is desirable for real-time decision making. While the
proposed method does not allow for detailed modeling of the underlying physics in the reservoir, it can
be used as a fast tool for planning field operations or for finding better initial solutions for computationally
demanding large-scale optimization problems with full reservoir simulation.
Various production optimization aspects have been discussed in the literature. In this paper, we focus
on studies related to allocating the limited amount of gas lift to specific wells in order to maximize oil
production. The gas lift optimization studies can be categorized into four types: two based on the number
of wells, and two based on the number of timesteps. The first category is single well analysis with a single
timestep. Martinez et al. (1994) applied genetic algorithm to optimize the distribution of gas lift amongs
several wells. Ray et al. (2006) used multiobjective evolutionary algorithms to solve gaslift optimization
problems by describing the relationship between gaslift injection rate and produced oil as a piece-wise
linear function. For single well analysis with multiple-timestep forecasting, the existing studies combine
optimization methods with reservoir simulation for production forecasting. Fang and Lo (1996) imple-
mented the simplex production optimization algorithm using a blackoil simulator to maximize oil
production. The reservoir performance, wellbore hydraulics, surface facility constraints and gas lift
injection were integrated into a generalized well management scheme. Their approach was validated
against two full field models, in which it could provide higher oil production.
The third category involves optimization of multiple connected wells through a surface pipeline
network in a single timestep. This occurs when the flow interactions among wells are strong. In Dutta-Roy
et al. (1997), the flow interaction among wells could affect the gaslift performance relationship when the
backpressure in the shared flowline was relatively large. To allocate gaslift injection optimally, a
full-network solution was proposed and Sequential Quadratic Programing (SQP) was used to solve the
problem. This application is implemented in the optimization module of the integrated production
modeling software (IPM) (see, e.g., Ozdogan (2008)) to capture the interaction among wells in the same
pipeline. The fourth category is obtained when the full-network optimization process in the third category
is linked to reservoir simulation to forecast and optimize production for multiple timesteps. Several
techniques including the use of proxy functions are proposed (Lu and Flemming, 2012). Dehdari and
Oliver (2012) proposed a new method to eliminate nonnegative constraints and achieve faster updates in
solving the iterations of the SQP algorithm, when compared with the conventional gradient calculation.
Most of these techniques use detailed reservoir simulation models to honor the governing fluid flow
physics that come with their corresponding computational overhead.
SPE-174039-MS 3

Although reservoir simulation can provide more accurate results that follow well-established physical
laws when simulation models accurately represent the physical reservoir, the existing uncertainties in
describing the complexity of geologic formations and their heterogeneous properties, and multiphase fluid
flow within them, make the forecast outcomes questionable. Moreover, full simulation models may not
even be available in some cases. Therefore, the use of IPM with simple material balance equations instead
of full simulation can alleviate the complexity of reservoir simulation and yet produce reasonable
predictions for planning production operations. In fields with a large number of reservoirs, even the use
of IPM may require a significant amount of computation. Alternatively, performance predictions with
decline curve analysis can offer an efficienct approach for short-term forecasting when operating
conditions remain unchanged.
Integration of optimization algorithms with performance prediction methods was proposed by Arnon-
din (1995). An Excel spreadsheet was developed to use production decline data for predicting the
maximum platform oil production while honoring gas-compression and water handling constriants. A
major limitation in that case was the number of equations in solving the optimization problem. Ekkawong
et.al (2015) implemented a production optimization workflow using linear programing and decline curve
analysis. They applied their method to a gas field with small compartmentalized reservoirs with several
constraints. The latter formulation was developed for linear problems and did not include non-linear
constraints, such as gas lift allocation. The proposed method in this paper does not have these limitations
and is applicable to various field conditions with both linear and non-linear constraints. As shown later
in thos paper, application of the proposed approach can result in significant production improvements over
manual adjustment based on sound engineering judgement.
Methodology

We present the proposed methodology in three parts. The first part outlines the problem formulation
as a nonlinear constrained optimization problem. The second part explains the gaslift optimization
procedure in which the gaslift performance relationships are defined. These two parts are then combined
to solve well optimum rates for a single timestep. The third part extends the optimization process by
incorporating production decline analysis to predict and optimize future well performances.
Description of Optimization Problem
The decision variables of the optimization problem include the optimum oil, gas, and gross production
rates as well as a gaslift injection rate from each well. The goal is to optimize a desired measure of
performance while reconciling the conflicting operational demands that are posed as constraints. In typical
oilfields, the wells can be either controlled manually or automatically using computer software. In either
case, the information about maximum production potential for each well, subject to facilities constraints,
is needed and can be determined by well testing (flowing conditions at minimum allowable pressure) or
by calculating it from wellhead performance relationship (WPR) models. This maximum potential can be
set as an upper limit. The lower limit can be set to zero, corresponding to well shutin. Other equality and
inequality constraints are derived from facility limits and/or availability of resources, for example gas and
water for injection.
The optimization problem for an oilfield with gaslift can be formulated to maximize the total oil
production subject to the existing constraints. In that case, the mathematical formulation of the problem
can be expressed as
(1a)

(1b)
4 SPE-174039-MS

(1c)

(1d)

(1e)

(1f)

(1g)

where the specified list of constraints consists of water treatment facility constraints (1b), liquid
handling facility constraints (1c), gas hadling facility constraints (1d), availability of the injection gas (1e),
gaslift performance relationship (1f), and gaslift valve mechanical limitation (1g). Alternatively, one could
minimize water production while maintaining a minimum oil production target. In that case, the
formulation can be modified to
(2a)

(2b)

where Equation (2a) denotes the summation of water production from every well in the field and
Equation (2b) represents the inequalities constraints to main a minimum oil production level. Note that all
the other constraints from (1b) through (1g) are still included. In a similar formulation, one can minimize
gas production while maintaining minimum oil production target. In that case, the formulation is changed
to:
(3)

in which the constraints remain the same as in water minimization objective function and the objective
function is sepecified to minimize gas production. Finally, one could specify a combination of these
objective functions either in the form of a multiobjective optimization or by combining the objective
function using their assigned weights. For example, to maximize oil production while minimizing water
and gas production one could maximize an objective function of the form ␩o * Qo – ␩w * Qw – ␩ g * Qg
subject to a similar set of constraints, where, ␩o, ␩w, and ␩g represent positive weights given to oil, water,
and gas production, respectively. More details about multiobjective optimization can be found in
Hajizadeh et al. (2011) and Christie et al. (2013).

Optimization of gas lift


In gas lift operations, the oil rate depends on the gas injection rate. The functional relations can be either
calculated using the Nodal Analysis or multirate-gas injection tests. Several equations are proposed to
represent this relationship in the literature. Nishikiori et al. (1989) proposed an extension of the equal
slope allocation method to find the optimum gas lift injection rate using nonlinear optimization. Alarcon
et al. (2002) proposed a mathematical model for gas lift performance by improving the model in
(Nishikiori et al., 1989) in which the well performance relationship is described as
(4)

In this paper, we use the gaslift performance relationship in Equation (4) as one of the constraints (1f)
in our optimization problem. Since production rates are the decision variables in our optimization, gas and
water production rates can be calculated in terms of oil production rate and from the definition of gas oil
SPE-174039-MS 5

ratio (GOR) and watercut (WCUT). A standard iterative gradient-based algorithm can be employed t solve
the resulting optimization problem. The solution of these optimization problems gives the optimum rate
allocatrions assuming a steady state production scenario. As such, the results can be used by field
operators to control the wells for a short perior of time when the production trends remain unchanged. To
model the transient behavior in the field, we combine this optimization with decline curve analysis.
Integration of production optimization and decline curve analysis
By combing decline curve analysis with the optimization approach outlined in the previous section the
transient behavior of the reservoir can be included to predict the future production from the wells. The
type of production decline (exponential, hyperbolic, and harmonic) and its coefficient can be determined
from the production history. If the decline curve analysis assumptions are valid, the method can be applied
to predict future production and the potential decline trend. This will then be used to adjust optimal rate
allocations accordingly. The maximum potential oil production rate in the future can now be predicted
(from decline curves) and be used as an upper bound constraint for future time steps. Gas and water
production rates are calculated using the GOR and WCUT values and the predicted oil production rate.
The optimization workflow is illustrated in Figure 1.

Figure 1—Optimization workflow with forecasts given by decline curve analysis

Field Case Study


We present an example based on data from part of a mature oilfield. This field consists of many
production sites but only one location (site -C), as shown in Figure 2, is selected for this study. There are
10 active production wells with various types of artificial lift such as beam pump (BP), gas lift (GL), and
electrical submersible pump (ESP). The production test data are shown in Table 1. Some of the wells are
not producing at full choke (128/64”). The well potential (the rate at fully opened choke) is estimated
using a wellhead performance relationship model. The constaints for this well site are as follow:
6 SPE-174039-MS

– Maximum water handling constraints 600 STB/D


– Maximum liquid handling constraints 1500 STB/D
– Maximum gas handling constraints 3000 MSCF/D
– Injection gas availability 3000 MSCF/D
– Maximum gas injection at each well (valve mechanical limitation) 500 MSCF/D

Figure 2—Schematic of wells and production facilities model from production in site-C with different types of artificial lift (Beam Pump
- C01, C02; Gas Lift - C03, C04, C05, C06, C07, C08, C09; Electrical Submersible Pump - C10)

Table 1—Production test data from production site C


Well Potential @
Production Test Data test date

Flow Test Choke FTHP Gross Net GOR Gaslift Wcut Gross Net
Well Mechanis Date Size /64 psia STBD STBD SCF/STB MSCF % STBD STBD

C-01 BP 1-Jan-14 128 114 183 7 65972 0 96.4 183 7


C-02 BP 1-Jan-14 128 124 176 94 2882 0 46.6 176 94
C-03 GL 2-Jan-14 64 126 527 147 492 500 72.2 550 147
C-04 GL 2-Jan-14 128 211 153 129 13628 200 15.3 153 129
C-05 GL 3-Jan-14 128 94 104 88 3756 500 15.9 104 88
C-06 GL 3-Jan-14 128 120 49 33 2128 300 32.3 49 33
C-07 GL 4-Jan-14 128 143 141 130 4383 400 7.7 141 130
C-08 GL 4-Jan-14 128 136 50 48 3712 300 3.3 50 48
C-09 GL 5-Jan-14 128 172 432 284 4909 300 34.3 432 284
C-10 ESP 5-Jan-14 128 140 95 36 7053 0 62.5 95 36
SPE-174039-MS 7

The gaslift performance relationship is obtained from multi-injection rate tests. The model in Equation
(4) is then used to fit the data. The well performance plots for each well are shown in Figure 3.

Figure 3—Gaslift performance relations based on well data and Equation (4).

To estimate the optimum well rates, two approaches can be used. First, modeling the well and network
system using IPM. In this approach, the parameters of the model are first adjusted to match the production
responses. The wells can then be controlled and optimized if sufficient data is available to calibrate the
behavior of the wells. This method requires a tedious history matching process, which can be slow
depending on the size of the unknown variables. Moreover, in general, the solution can be non-unique
since an underdetermined history matching inverse problem has to be solved, with the potential pitfall that
the obtained solutions do not provide accurate future predictions. In the second approach, which is
adopted in this paper, the maximum oil production potential for every well is first estimated for a common
date, and the proposed optimization is then applied to determine oil rate for that timestep. This aprpaoch
results in much faster calculations and is more intuitive and simpler to implement. The initial rate from
the prediction is also close to the actual rate since it is based on the latest well performance results.
However, the method is based on several assumptions, including
X well models are needed to estimate maximum oil production potential
X communication among wells in flowline has negligible back pressure effects
X frequent production tests are available and the rates follow decline curve assumptions for short term
predictions
X communications between the wells in the reservoir are represented in the production response
X wells are producing under the same operating condition (decline curve analysis assumptions are
valid)
In Site-C, the well model to estimate the maximum oil production potential is available. The pipeline
network model is also constructed and the communication among the wells in the flowlines has a
negligible back pressure effect. This is mainly because multiple wells are gathered at a downstream
separator under a specified pressure control, or the production through the wellhead is flowing at the
critical flowrate since the pressure drop inside the tubing is more dominant than in the surface pipeline
network. Moreover, it is observed for short-term periods (less than 3 month) that the operating condition
does not change significantly. Therefore, the above assumptions hold and the proposed approach can be
applied.
8 SPE-174039-MS

Results
The Sequential Quadratic Programming (SQP) algorithm is used to solve the optimization problem
formulated in the previous section. The tolerance of the objective function is selected to be 1e-12. The
results for oil production maximization scheme are shown in Figure 4 and Figure 5. Figure 4 shows the
results for individual well rates while Figure 5 depicts the total rate from all the wells. The constraints are
all satisfied and the objective function has converged to a solution. An exponential decline model based
on the production history was selected and used. The GOR and WCUT are increased with increasing
cumulative oil production. The 90-day production forecasts are shown in Figure 6. The results show that
the maximum oil production (green line) is achieved during the gaslift, and gas and water are produced
at the rate below their specified limits. Oil production is limited during the first 20 days due to the
maximum gas and water production constraints (cyan line). After this time period oil is produced at its
maximum potential and gradually declines following the exponential trend.

Figure 4 —The optimum production/injection rates after maximizing total oil production for each individual well

Figure 5—Total field liquid rate and gas rate after maximizing total oil production
SPE-174039-MS 9

Figure 6 —Optimum rate prediction for 90 days after maximizing total oil production

A second case is run to minimize gas production while maintaining oil production at the minimum
value of 500 STB/D. The results for the first timestep of the optimization are shown in Figure 7 and Figure
8, while the forecasts are shown in Figure 9. These optimization results show that gas production is kept
at its lowest possible amount during the early timesteps until there is no well with low GOR value. Then
gas production will then begin to gradually increase to its maximum potential. Although oil production
has the potential to exceed 500 STB/D, it needs to be produced at 500 STB/D to minimize the associated
gas production. All other constraints such as maximum gross, maximum water, maximum gas, and
maximum gaslift are satisfied.

Figure 7—The optimum production/injection rates after minimizing gas production for each individual well
10 SPE-174039-MS

Figure 8 —Total field liquid and gas rates after minimizing gas production

Figure 9 —Optimum rate prediction for 90 days after minimizing total gas production

It is noteworthy, that the proposed approach allows for new production stream (for example, new well
start-up, additional perforation, or the reroute of the new pipeline to an existing network) to be included
in the future timestep. An example of such events is shown in Figure 10 where the objective function is
to maximize oil production. In this case, Well C-09 is currently shutin for well intervention job and is
scheduled to be back on the stream on day 14. As a result, the maximum oil potential is increased on day
14 due to production from Well C-09. At that time, the optimum oil production does not yet reach its
maximum potential because of the maximum gas production constraint (red line).
SPE-174039-MS 11

Figure 10 —Optimum rate prediction for 90 days after maximizing total oil production with a new well to join the stream on day 14.

The computational cost of the optimization and prediction framework is far less than what is needed
in reservoir simulation and integrated production modeling since the calculation is only based on
extrapolation of current well performances, especially considering the efforts and time that are allocated
to constructing geological models, validating PVT fluid properties, history matching, and reservoir
simulation. The runtime for 90 days of optimization problem for 10 wells with Intel® core™ i7-2640M
2.8GHz CPU is only 48.5 seconds. The plot of the objective function versus the number of iterations for
maximizing oil production is shown in Figure 11. In this case, the number of iterations needed before
converging to the solution is 52. The most time-consuming step of the workflow is the SQP optimization
algorithm as it rapidly increases with increasing number of inequality and non-linear constraints.

Figure 11—Objective function value vs number of iterations of the first timestep optimization corresponding to the oil production
maximization.

Conclusions
We presented a new method for fast production optimization and forecasting by combining decline curve
analysis for production forecasting and constrained optimization. The formulation begins by defining the
objective function and the constraints in a mathetical form and solving them, using standard optimization
techniques, for a single timestep to identify the optimal rate allocations. To include the production
transient effect, decline curve analysis is then included in the formulation to assist the optimization by
12 SPE-174039-MS

forecasting future outcomes. The proposed approach does not require reservoir model construction and
calibration and, hence, is extremely fast and simple to implement. Some of its advantages include
predicting the production profile based on the latest well performance, and flexibility to modify and
customize the formulation for a specific set of field conditions and requirements. However, certain
assumptions (the main ones listed in the body of the paper) are needed before this approach can be applied.
In particular, the use of decline curve analysis implies that its underlying assumptions must hold. Possible
future extension of this approach include incorporating other types of artificial lift, including constraints
to honor well connectivity in the pipeline network, and additional constriants for certain field development
schemes such as waterflooding or enhanced oil recovery.

Nomenclature
qp,i ⫽ Flow rate of phase p (liquid, water, oil, produced gas, injected gas) in well i
qp,max ⫽ Maximum rate of phase p from a single well
Qp,max ⫽ Maximum rate of phase p from total wells
Qp,target ⫽ Target rate of phase p from total wells
Qp ⫽ Rate of phase p from total wells
Wcuti ⫽ Water cut of well i
GORi ⫽ Gas oil ratio of well i
n ⫽ Total number of wells
␩ ⫽ Normalized weight factor (for weighted sum approach multiobjective optimization)

References
1 Martinez E., Moreno W., Moreno J., Maggiolo R. 1994. Application of Genetic Algorithm on the
Distribution of Gas-Lift Injection Paper SPE 26993 presented at the III Latin American/
Caribbean Petroleum Engineering Conference held in Buenos Aires, Argentina, 27-29 April 1994
2 Ray J.E, and Sarker R. 2006. Multiobjective Evolutionary Approach to the Solution of Gas Lift
Optimization Problems presented in IEEE Congress on Evolutionary Computation Vancouver,
BC, Canada, 16-21 July 2006
3 Fang W. Y., and Lo K.K. 1996. A Generalized Well-Management Scheme for Reservoir
Simulation, SPE J., SPE Reservoir Engineering, May 1996, 116 –120
4 Dutta-Roy K., Kattapuram J. 1997. A New Approach to Gas-Lift Allocation Optimization Paper
SPE38333 presented at 1997 SPE Western Regional Meeting held in Long Beach California,
25-27 June 1997
5 Ozdogan, U., Keating, J. F., Knobles, M., Chawathe, A., & Seren, D. 2008. Recent Advances and
Practical Applications of Integrated Production Modeling at Jack Asset in Deepwater Gulf of
Mexico Paper SPE 113904 presented at 2008 SPE Europec/ EAGE Annual Conference and
Exhibition held in Rome, Italy, 9-12 June 2008
6 Lu Q., Fleming G.C. 2012.Gas Lift Optimization Using Proxy Functions in Reservoir Simulation
Paper SPE 1140935 presented at the SPE Reservoir Simulation Symposium, Texas, 21–23
February 2011.
7 Dehdari V., Oliver D. 2012.Sequential Quadratic Programming for Solving Constrained Produc-
tion Optimization -Case Study from Brugge Field Paper SPE 141589 presented at the SPE
Reservoir Simulation Symposium, Texas, 21-23 February 2011.
8 Arnondin M.C. 1995.Integration of Production Analyst and Microsoft Excel’s Solver for Pro-
duction Forecast and Optimization Paper SPE27566 presented at the SPE European Petroleum
Computer Conference, Aberdeen, 15-17 March 1995
SPE-174039-MS 13

9 Ekkawong P., Kritsadativud P., Lerlertpakdee P., Amornprabharwat A. 2015.Innovative and


Automated Workflow for Fast Production Optimization and Forecast in Gulf of Thailand Gas
Fields Using Linear Programing Opitmization. Paper SPE 173450 presented at SPE Digital
Energy Conference and Exhibition, The Woodlands, Texas, USA 3-5 March 2015
10 Hajizadeh Y., Christie M., Demyanov V. 2011.Towards Multiobjective History Matching: Faster
Convergence and Uncertainty Quantification Paper SPE 141111 presented at SPE Reservoir
Simulation Symposium, Texas, USA 21-23 February 2011
11 Christie M., Eydinov D., Demyanov V., Talbot J., Arnold D. 2013.Use of Multi-Objective
Algorithms in History Matching of a Real Field Paper SPE 163580 presented at SPE Reservoir
Simulation Symposium, Texas, USA, 18-20 February 2013
12 Nishikiori, N., Redner R. A., Doty D. R., Schmidt Z. 1989. An Improved Method for Gas Lift
Allocation Optimization Paper SPE 19711 presented at the 64th Annual Technical Conference
and Exhibition of SPE held in San Antonio, TX, 8-11 October 1989
13 Alarcon G., Torrez. C.F., Gomez L.E. 2002. Global Optimization of Gas Allocation to a Group
of Wells in Artificial Lift Using Nonlinear Constrained Programing. Journal of Energy Resources
Technology, December 2002, 262–268
14 Cummings C.A., Gentry R.W. 1986. A Method for Predicting Solution Gas-Drive Production
Decline Paper SPE 15021 presented at the Permian Basin Oil & Gas Recovery Conference of the
Society of Petroleum Engineers, Midland, Texas, USA 13-14 March 1986.
15 Frederick J.L., Kelkar M. 2005. Decline-Curve Analysis for Solution Gas Drive Reservoirs Paper
SPE 94859 presented at the 2005 SPE Annual Technical Conference and Exhibition held in
Dallas, Texas, USA 9-12 October 2005.

You might also like