You are on page 1of 236

THE UNIVERSITY OF TULSA THE GRADUATE SCHOOL

MATCHING OF PRODUCTION PERFORMANCE USING DYNAMIC NODAL ANALYSIS

by Arsene B. Bitsindou

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Discipline of Petroleum Engineering The Graduate School The University of Tulsa 2002

THE UNIVERSITY OF TULSA THE GRADUATE SCHOOL

MATCHING OF PRODUCTION PERFORMANCE USING DYNAMIC NODAL ANALYSIS

by ARSENE B. BITSINDOU

A DISSERTATION APPROVED FOR THE DISCIPLINE OF PETROLEUM ENGINEERING

By Dissertation Committee , Chairperson

ii

ABSTRACT

Bitsindou B. Arsene (Doctor of Philosophy in Petroleum Engineering) Matching of Production Performance Using Dynamic Nodal Analysis. Directed by Dr. Mohan Kelkar (266 words) This work explores procedures to optimize oil reservoir production for single and multi-well production. The work is divided in two parts: The first part develops a procedure to match the production from a single well using a concept of dynamic nodal analysis. By combining the desirable features of nodal

analysis, material balance technique and the decline curve analysis, the method is able to match the historical performance of the well, and is able to predict the future performance of the oil well under the existing as well as altered conditions. The second part develops a procedure to match multi-well production from an oil reservoir producing under natural depletion or water drive mechanism. It extends the classical tank model to a multi-tank model scheme by dividing the original reservoir into small radial regions around the wells (well regions) and a larger region (reservoir region) that feeds the small regions. By discretizing the flow equations for three-phase flow and determining the transfer coefficients from reservoir region to well region, the method is

iii

able to match the historical performance of the wells, as well as predict the future performance of the wells under the existing and altered conditions. The procedure developed in this dissertation assumes a simplified reservoir model. However, by incorporating reservoir as well as production components in the analysis, it provides a mechanism by which reservoir performance can be optimized by either adjusting a production component (e.g., well head pressure, tubing, choke, etc) or a reservoir component (e.g., skin factor removal through stimulation). Further, the history matching process requires a lot less information compared to conventional reservoir simulation.

iv

ACKNOWLEDGEMENT

I would like to take this opportunity to thank Dr. Mohan Kelkar for his invaluable guidance and support during the course of this study. I also express my gratitude to Dr Leslie G. Thompson, Dr Mauricio Prado and Dr Kevin ONeil for their valuable time in reviewing this work and providing suggestions. I would like to express my appreciation to all the other faculty members who contributed to my education as a TU graduate student. Also, I would like to thank my graduate student colleagues who made my life easier at TU. This dissertation is dedicated to my parents, Donat and Julienne, whose support and encouragement will always be appreciated.

TABLE OF CONTENTS

TITLE PAGE.............................................................................................................................i APPROVAL PAGE .................................................................................................................ii ABSTRACT ............................................................................................................................iii ACKNOWLEDGEMENTS .................................................................................................... v TABLE OF CONTENTS .......................................................................................................vi LIST OF TABLES .................................................................................................................xii LIST OF FIGURES ..............................................................................................................xiv

CHAPTER I 1.1 1.2

INTRODUCTION............................................................................... 1 Single-Well Reservoir Scheme........................................................................1 Multi-Well Reservoir Scheme......................................................................... 7

CHAPTER II

SINGLE WELL PERFORMANCE MATCH: APPROACH......................................................................................10

2.1 2.2 2.3 2.4 2.5

Oil in Place Estimation ..................................................................................11 Relative Permeability Curves Estimation .....................................................15 History Match of Observed Production Data................................................18 Future Performance Prediction......................................................................20 Procedure .......................................................................................................21 vi

2.6 2.7

Validation of the Model.................................................................................21 Sensitivity Analysis .......................................................................................27 2.7.1 Sensitivity of History Match Results With Respect to Objective Function Selection......................................................27 2.7.1.1 2.7.1.2 2.7.1.3 Least Square Difference Method ...................................27 Absolute Difference Method..........................................28 Robust Method ...............................................................29 2.7.1.3.1 2.7.1.3.2 2.7.1.3.3 2.7.1.3.4 2.7.1.3.5 2.7.1.3.6 2.7.2 Component F1 ..................................................... 30 Component F2 .................................................... 31 Component F3 .................................................... 31 Component F4 ..................................................... 32 Component F5 ..................................................... 32 Component F6 .................................................... 33

Sensitivity of History Match Results With Respect to Errors ...........................................................................................38

2.7.3 2.7.4

Statistical Analysis: Confidence Interval Computation ...................42 Sensitivity of History Match Results With Respect to Type of Decline Curve................................................................42

2.8

Summary ........................................................................................................46

vii

CHAPTER III RESULTS 3.1

SINGLE WELL PERFORMANCE MATCH: 47 Synthetic Data From Dynamic Nodal Analysis Forward Model .................48 3.1.1 3.1.2 3.1.3 Oil in Place ........................................................................................48 Relative Permeability Calculations...................................................48 History Match....................................................................................51

3.2

Synthetic Data: Eclipse Homogeneous Model..............................................55 3.2.1 3.2.2 3.2.3 Oil in Place ........................................................................................55 Relative Permeability Calculations...................................................55 History Match....................................................................................56

3.3

Synthetic Data: ECLIPSE heterogeneous model..........................................60 3.3.1 3.3.2 3.3.3 Oil in Place ........................................................................................61 Relative Permeability Calculations...................................................61 History Match....................................................................................61

3.4

Summary ........................................................................................................65

CHAPTER IV

MULTI-WELL PERFORMANCE MATCH: APPROACH......................................................................................66

4.1 4.2 4.3

Oil in Place Estimation ..................................................................................67 Relative Permeability Curves Estimation .....................................................67 History Match of Observed Production Data................................................68 viii

4.4 4.5 4.6 4.7

Mathematical Modeling.................................................................................68 History Match ................................................................................................72 Future Performance Prediction......................................................................73 Summary ........................................................................................................73

CHAPTER V 5.1

MULTI-WELL PERFORMANCE MATCH: RESULTS...............74 Synthetic Data: Horizontal Reservoir ...........................................................74 5.1.1 5.1.2 5.1.3 Oil in Place Calculation.....................................................................76 Calculation of Relative Permeability Curves ...................................76 History Match....................................................................................78

5.2

Synthetic Data: Dipping Reservoir................................................................87 5.2.1 5.2.2 5.2.3 Oil in Place Calculation.....................................................................87 Calculation of Relative Permeability Curves ...................................88 History Match....................................................................................90

5.3

Field Example ..............................................................................................100 5.3.1 5.3.2 Brief History of the Field ................................................................100 Data Available.................................................................................101 5.3.2.1 5.3.2.2 5.3.2.3 5.3.2.4 Well Map ......................................................................101 Depth Information ........................................................101 PVT Data ......................................................................103 Pressure Data ................................................................107 ix

5.3.2.5 5.3.3

Production Data ............................................................109

History Match Process ....................................................................109 5.3.3.1 5.3.3.2 Oil in Place Calculation................................................109 Calculation of Relative Permeability Curves.....................................................................109 5.3.3.3 History Match ...............................................................112 5.3.3.3.1 5.3.3.3.2 Field Production History Match .................... 114 Well Production History Match .................... 119 5.3.3.3.2.1 Well Unit W-

0314304 7 .................................................... 119 5.3.3.3.2.2 MOORE

UNIT 82-892.............................................. 122 5.3.3.3.2.3 Well MOORE

UNIT W-59617 3 ...................................... 125 5.3.3.3.2.4 Well USA W-

86-341 26-1 ................................................. 128

CHAPTER VI

CONCLUSIONS AND RECOMMENDATIONS ........................131

REFERENCES .....................................................................................................................134

APPENDIX A

EXTENDED MUSKAT METHOD...............................................140

APPENDIX B

MULTI-TANK SCHEME .............................................................148

APPENDIX C

DYNAMIC NODAL ANALYSIS .................................................184

APPENDIX D

FIELD CASE: HISTORY MATCH RESULTS ............................201

xi

LIST OF TABLES

2.1 2.2

Synthetic data: input parameters..........................................................................22 Sensitivity of history match results with respect to objective function selection...........................................................................................34

2.3

Sensitivity of history match results with respect to errors in the rate data ..........................................................................................................38

3.1 3.2 3.3

Relative permeability parameters estimation ......................................................49 History match results ...........................................................................................51 Homogeneous reservoir: relative permeability parameters estimation .......................................................................................................55

3.4 3.5

Homogeneous reservoir: history match results ...................................................56 Heterogeneous reservoir: relative permeability parameters estimation .......................................................................................................61

3.6

History Match for Synthetic data obtained from heterogeneous reservoir..........................................................................................................62

5.1 5.2 5.3 5.4

Synthetic data: input parameters..........................................................................75 Relative permeability regression parameters ......................................................76 Radius of well-regions estimation .......................................................................79 Regression results for transfer coefficient from reservoir-region to well-region .................................................................................................79 xii

5.5

Regression results for transfer coefficient from well-region to well. ................................................................................................................79

5.6 5.7 5.8

Relative permeability regression parameters ......................................................88 Radius of well-regions regression limits .............................................................91 Regression results for transfer coefficient from reservoir to wellregion..............................................................................................................91

5.9

Regression results for transfer coefficient from well-region to well .................................................................................................................92

5.10 5.11 5.12 5.13 5.14 5.15

Well Depths Information ...................................................................................102 PVT data.............................................................................................................104 PVT correlations ................................................................................................107 Relative permeability regression parameters ....................................................110 Regression results for radius of well-regions ....................................................112 Regression results for transfer coefficient from well-region to well ...............................................................................................................113

5.16

Regression results for Transfer coefficient from reservoir-region to well-region ...............................................................................................113

xiii

LIST OF FIGURES

1.1 2.1

Components in the Production System..................................................................2 Reservoir pressure: eclipse forward simulation versus dynamic nodal analysis forward simulation.................................................................24

2.2

Rate: eclipse forward simulation versus dynamic nodal analysis forward simulation.........................................................................................25

2.3

GOR: eclipse forward simulation versus dynamic nodal analysis forward simulation.........................................................................................26

2.4

Sensitivity of reservoir pressure with respect to objective function selection...........................................................................................35

2.5 2.6 2.7

Sensitivity of rate with respect to objective function selection ..........................36 Sensitivity of GOR with respect to objective function selection........................37 Sensitivity of rate with respect to objective function selection and errors in the rate data...............................................................................39

2.8

Sensitivity of GOR with respect to objective function selection and errors in the rate data.............................................................................40

2.9

Sensitivity of reservoir pressure with respect to objective function selection and errors in the rate data.................................................41

2.10 2.11

Sensitivity of reservoir pressure with respect to type of decline ........................43 Sensitivity of rate with respect to type of decline ...............................................44 xiv

2.12 3.1

Sensitivity of GOR with respect to type of decline.............................................45 Match of krg/kro ratio obtained from observed data and the krg/kro ratio calculated from correlations.....................................................50

3.2

Synthetic data: sensitivity of reservoir pressure history match with respect to different guess values of regression

parameters ......................................................................................................52 3.3 Synthetic data: sensitivity of rate history match with respect to different guess values of regression parameters...........................................53 3.4 Synthetic data: sensitivity of GOR history match with respect to different guess values of regression parameters...........................................54 3.5 Synthetic data (homogeneous reservoir): reservoir pressure history match..................................................................................................57 3.6 3.7 3.8 Synthetic data (homogeneous reservoir): rate history match..............................58 Synthetic data (homogeneous reservoir): GOR history match ...........................59 Synthetic data (heterogeneous reservoir): permeability

distribution .....................................................................................................60 3.9 Synthetic data (heterogeneous reservoir): reservoir pressure history match..................................................................................................62 3.10 3.11 Synthetic data (heterogeneous reservoir): rate history match.............................63 Synthetic data (eclipse heterogeneous model): GOR history match ..............................................................................................................64 xv

4.1 5.1

Multi-well scheme representation .......................................................................69 Match between estimated gas/oil relative permeability ratio and gas/oil relative permeability ratio obtained from observed data (synthetic data, dipping reservoir) .........................................................77

5.2

Synthetic data (horizontal reservoir): reservoir pressure history match ..............................................................................................................80

5.3 5.4 5.5 5.6 5.7 5.8 5.9 5.10

Well 1 Synthetic data (horizontal reservoir): rate history match........................81 Well 1 Synthetic data (horizontal reservoir): GOR history match .....................82 Well 2 Synthetic data (horizontal reservoir): rate history match........................83 Well 2 Synthetic data (horizontal reservoir): GOR history match .....................84 Well 3 Synthetic data (horizontal reservoir): rate history match........................85 Well 3 Synthetic data (horizontal reservoir): GOR history match .....................86 Dipping reservoir: position of the wells ..............................................................87 Match between estimated gas/oil relative permeability ratio and gas/oil relative permeability ratio obtained from observed data (synthetic data, dipping reservoir) .........................................................89

5.11 5.12 5.13 5.14 5.15

Synthetic data (dipping reservoir): pressure history match ................................93 Well 1 Synthetic data (dipping reservoir): rate history match............................94 Well 1 Synthetic data (dipping reservoir): GOR history match ........................95 Well 2 Synthetic data (dipping reservoir): rate history match............................96 Well 2 Synthetic data (dipping reservoir): GOR history match .........................97 xvi

5.16 5.17 5.18 5.19 5.20

Well 3 Synthetic data (dipping reservoir): rate history match............................98 Well 3 Synthetic data (dipping reservoir): GOR history match .........................99 Well map ............................................................................................................100 Pressure profile as a function of time ................................................................108 Match between estimated relative permeability ratio and observed relative permeability ratio ............................................................111

5.21 5.22 5.23 5.24

Cumulative oil production .................................................................................115 Cumulative production gas/oil ratio ..................................................................116 Cumulative gas production ................................................................................117 Match between predicted reservoir pressure and observed reservoir pressure.........................................................................................118

5.25

Match between predicted oil rate and observed oil rate for well unit W-0314304 7 ........................................................................................120

5.26

Match between predicted GOR and observed GOR for well unit W-0314304...................................................................................................121

5.27

Match between predicted oil rate and observed oil rate for well MOORE UNIT 82-89 2...............................................................................123

5.28

Match between predicted GOR and observed GOR for well MOORE UNIT 82-892................................................................................124

5.29

Match between predicted oil rate and observed oil rate for well MOORE UNIT W-59617 3.........................................................................126 xvii

5.30

Match between predicted GOR and observed GOR for well MOORE UNIT W-59617 3........................................................................127

5.31

Match between predicted oil rate and observed oil rate for well USA W-86-341 26-1....................................................................................129

5.32

Match between predicted GOR and observed GOR for well USA W-86-341 .....................................................................................................130

B.1 C.1 C.2 D.1

Multi-well scheme representation .....................................................................149 Typical inflow and outflow curves ....................................................................199 Example of an unstable production condition (Liquid loading) .......................200 Well Taylor Unit W-034304 1: match between predicted rate and observed rate .........................................................................................202

D.2

Well Taylor Unit W-034304 1: match between predicted GOR and observed GOR.......................................................................................203

D.3

Well Peregrine W32266A 1: match between predicted rate and observed rate ................................................................................................204

D.4

Well Peregrine W32266A 1: match between predicted GOR and observed GOR..............................................................................................205

D.5

Well Moore Unit FEE 1: match between predicted rate and observed rate ...............................................................................................206

D6

Well Moore Unit FEE 1: match between predicted GOR and observed GOR..............................................................................................207 xviii

D.7

Well Taylor Unit FEE 4: match between predicted rate and observed rate ................................................................................................208

D.8

Well Taylor Unit FEE 4: match between predicted GOR and observed GOR..............................................................................................209

D.9

Well USA W-42621A 27-1: match between predicted rate and observed rate ................................................................................................210

D.10

Well USA W-42621A 27-1: match between predicted GOR and observed GOR..............................................................................................211

D.11

Well Woods 36-1: match between predicted rate and observed rate................................................................................................................212

D.12

Well Woods 36-1: match between predicted GOR and observed GOR .............................................................................................................213

D.13

Well Federal W-86337 15-1: match between predicted rate and observed rate ................................................................................................214

D.14

Well Federal W-86337 15-1: match between predicted GOR and observed GOR..............................................................................................215

D.15

Well Federal W-47686 22-1: match between predicted rate and observed rate ................................................................................................216

D.16

Well Federal W-47686 22-1: match between predicted GOR and observed GOR..............................................................................................217

xix

xx

CHAPTER I INTRODUCTION

In order to produce economically a hydrocarbon reservoir, a reliable procedure is needed to reproduce the historical observed production and predict the future performance. This work presents two new procedures that can reproduce the production history and predict the future performance. The first procedure, that uses dynamic nodal analysis technique, deals with single-well reservoir scheme, whereas the second procedure, that uses multi-tank approach is directed to multi-well reservoir production.

1.1 Single-Well Reservoir Scheme The production optimization of an oil well requires an appropriate selection of the individual components in the production system. Currently, nodal analysis is used to accomplish this task. Nodal analysis1,2,3 involves calculating the pressure drop in individual components within the production system so that pressure value at a given node in the production system (e.g., bottom hole pressure) can be calculated from both ends (separator and reservoir) [See Figure 1]. The rate for which pressure calculated at the node from both ends is the same; this is the rate at which the well will produce. By adjusting individual components, the sensitivity of individual components on the overall production can be investigated, and hence an optimum selection of components can be obtained at a given time. The major drawback of the 1

2 conventional nodal analysis is that it only provides you a snapshot picture of the well production. It does not provide you with information about how the production will change as a function of time. For example, if you change a tubing, the nodal analysis may provide which is the best tubing size at present time; however, it may not be able to tell you which tubing size is the best over the life of the well based on the future production. Even generating future inflow performance curves (which characterize how the reservoir will behave in the future at discrete times) may not help since we will not be able to estimate how the rate has changed over the time intervals.

Figure 1.1: Components in the Production System (After Brown1).

3 To include the effect of time on the production performance, the most commonly used technique is the decline curve analysis. Decline curve analysis2 involves matching the prior production data using one of the decline types (exponential, hyperbolic or harmonic), and using the estimated decline parameters, predicting the future performance under existing conditions. Decline curve analysis is a very powerful tool, and has been used extensively to predict the future performance of an oil well. The major drawback of the decline curve analysis is that it only accounts for the reservoir performance by ignoring the effects of tubing size, choke, surface pipeline or other components in the production system. In addition, although it is true that decline curve analysis can predict the future performance under existing conditions, it may not predict how the well will behave in the future if the production conditions are altered. These alterations include changing skin factor, changing choke size, changing tubing size, or changing the surface compressor. The conventional material balance technique12,13,14,15 which uses diagnostic plots have also proven to be useful in understanding the behavior of the oil/gas wells. These plots, for example, include p/z (reservoir pressure over compressibility factor) versus gas production to predict how much gas the well will eventually produce or, for oil wells, a plot of F/(Eo+mEg+Efw) versus Np will provide the initial oil in place. where F=Np(Bo+(Rp-Rs)Bg)+WpBw Eo=(Bo-Boi)+(Rsi-Rs)Bg Eg=Boi(Bg/Bgi-1) Efw=(1+m)Boi(CwSwc+Cf)p/(1-Swc). The terms in the above equations conform to the conventional meaning: (I.1) (I.2) (I.3) (I.4)

4 Np=cumulative oil produced [stb] Wp= cumulative water produced [stb] Bo=oil formation volume factor at current average reservoir pressure [bbl/STB] Boi= oil formation volume factor at initial reservoir pressure [bbl/STB] Bw=water formation volume factor at current average reservoir pressure [bbl/STB] Rs=solution gas/oil ratio at current average reservoir pressure [scf/stb] Rsi=solution gas/oil ratio at initial reservoir pressure [scf/stb] Bg=gas formation volume factor at current average reservoir pressure [rb]/[scf] Bgi=gas formation volume factor at initial reservoir pressure [rb]/[scf] Swc=average connate saturation [-] m =ratio of free gas-in-place to original oil-in-place [bbl/bbl] Cw=water compressibility [psia-1] Cf=formation compressibility [psia-1] These techniques can also account for, through a trial and error procedure, the presence of water influx12,49. The drawback of the material balance technique is that it does not account for time. It can predict the production as a function of reservoir pressure, but not as a function of time. Further, it also only accounts for reservoir component, and not for any other components in the production system. The effect of alterations on the oil well performance cannot be predicted using the material balance technique. The inclusion of time in terms of predicting the future performance is critical from economics point of view. This cannot be accomplished using this technique. To overcome the drawbacks presented in the above methods, we need a technique which can:

5 Predict the future performance as a function of time in the presence of various production components including the reservoir Match the prior production data in the presence of various production components so that appropriate parameters can be assigned for future production prediction. This is similar to decline curve analysis except that we need to include the production components in the system. Quantify the uncertainty with respect to various parameters (e.g., reservoir permeability, skin factor, tubing roughness, drainage area, the type of pressure drop correlation) by generating alternate possibilities of parameters which can match the production data. Predict the future performance under existing conditions as well as altered conditions to compare the production scenarios in the future. Quantify the uncertainty in predicting the future performance which can be combined with the price of oil/gas to conduct a risk analysis. Optimize the producing well configuration so that the net profit over the life of the well is maximized. Some of the specific examples where the proposed technique can be applied are: Fracturing or Stimulating an oil well: A service company will always compare the production with and without the stimulation to sell a particular stimulation procedure. However, stimulation, typically does not increase the reserves. It only accelerates the production. Therefore, after stimulation, the oil well will decline faster than at the current conditions. For proper economic evaluation, it is critical that we examine the incremental oil production - difference between production with stimulation minus

6 production without stimulation (which is positive at the beginning but will become negative at later times) as a function of time. Currently we do not have a procedure which includes the production components in the system in determining the incremental production as a function of time. Changing of Production Components: The prediction of the oil well performance in the future is critical under existing as well as modified conditions. For example, we would like to know how an alteration of production components will affect the wells performance before the actual work-over is performed. These alterations include

changing the choke size, changing the tubing size or reducing the well head pressure.3 Based on the production scenarios under existing as well as altered conditions, a proper method can be selected for continued oil production. History Matching of Prior Production Data: To instill confidence in the predictive ability of any program, the user should be able to match the prior production from the same oil well. Decline curve analysis essentially matches the prior production data by using a specific model and then predicts the future performance under existing conditions. However, the decline curve analysis only predicts one future performance based on the prior data. In reality, we know that significant uncertainties exist with respect to the input parameters used for predicting the past performance. For example, the same prior production data can be matched by either altering the permeability or skin factor, or by changing the tubing correlation or the roughness factor. Changing the drainage area or thickness or the porosity or saturation can all alter the possible reserves the well is capable of producing. However, the productivity of well can be only significantly affected by the thickness of the well. If we want to quantify the

7 uncertainties in predicting the future performance, we need to develop alternate scenarios - all matching the prior performance. Subsequently, these scenarios can be used to predict the future performance of an oil well under existing as well as modified conditions. This type of information is extremely useful in economic risk and

uncertainty analysis. None of the existing techniques - conventional nodal analysis, decline curve analysis and the material balance analysis - will be able to solve the above stated problems adequately. To use one of these techniques, we either have to forego the time element, the effect of production components, or the effect of altered production conditions on the overall production. What we need is a technique which can combine the advantages of simple material balance technique with the decline curve analysis as well as the nodal analysis so that the overall procedure can be automated to address many of the problems discussed above. In our study, we propose a method to incorporate material balance, decline curve and nodal analysis techniques. By combining these three procedures, we demonstrate that well production can be optimized. 1.2 Multi-Well Reservoir Scheme Different methods are usually used to evaluate reservoir production from a multiwell reservoir. The most common methods are: material balance (tank model) methods and numerical reservoir simulation methods16,17,18,19,20,21. The numerical simulation methods account correctly the physics of the multiphase flow in reservoir and honor the material balance equation. Thus, they can predict the GOR,

8 the water cut and the pressure and saturation changes. Their drawback is that they are computationally intensive and require input data that are not always available. The tank model method14 is easy to use. It honors the material balance equation. Production history and PVT data need to be available in order to use the method. Its main drawback is that time does not explicitly appear in the equation. Also it can only capture the overall reservoir pressure and saturation changes. It cannot capture localized changes (at well locations, for example). In order to overcome the drawbacks presented in the above methods, we need a technique which is easy to use, is not computationally intensive, and need only limited number of input data. We need a method that can: Match the prior production data so that appropriate model parameters can be assigned for future production prediction. Predict the future performance as a function of time. Quantify the uncertainty with respect to various parameters (e.g., reservoir permeability, skin factor,) by generating alternate possibilities of parameters which can match the production data. Predict the future performance under existing conditions as well as altered conditions to compare the production scenarios in the future. Quantify the uncertainty in predicting the future performance, which can be combined with the price of oil to conduct a risk analysis. This work presents a modified tank model that captures produced GOR and WOR at well locations by using a multi-tank model.

9 This work is divided into several chapters. After this introduction chapter, Chapter II describes the algorithm for single-well performance match using the dynamic nodal analysis technique and details the mathematical models as well as the regression analysis used in this technique. Chapter III presents the results of the application of the dynamic nodal analysis technique for a single-well scheme. Chapter IV describes the algorithm for multi-well performance match using the multi-tank scheme and details the mathematical model as well as the regression analysis used in this technique. Chapter V presents the results of the application of the multi-tank technique. Finally, in Chapter VI, conclusions and recommendations are provided.

CHAPTER II SINGLE WELL PERFORMANCE MATCH: APPROACH

This chapter discusses the procedure used for single well match. We will discuss the procedure in this chapter followed by examples in the next chapter. This chapter is divided into eight sections. In the first section, we discuss the procedure for estimating oil in place using material balance technique. We have incorporated three principal mechanisms in calculating oil in place these include gas and oil expansion, formation expansion and water influx. We use the procedure provided by Sills and use non-linear regression to determine the oil in place. In the second section, we discuss the estimation of relative permeability curves using production data. We have extended Muskats procedure to calculate the relative permeability values for gas, oil and water using available production data. We assume relative permeability model based on Standings curves, and estimate the relevant parameters using non-linear regression. In the next section, we develop the procedure for history matching the production data. We couple the nodal analysis with material balance analysis, and match the observed oil rates and pressure data by adjusting reservoir parameters. This is also done using non-linear regression analysis. In section four, we discuss the procedure we use to predict the future performance of a single well once the production data are satisfactorily matched. In section five, we present all the necessary components used in history matching procedure. Any additional assumptions required to run the program are also presented in this section. 10

11 As part of the validation of our program, in section six, we compare the results of our simplified forward model with the results of the forward model from ECLIPSE simulator. Using the same input, we compare the single well production results from ECLIPSE simulator with single cell simulator used in dissertation. The reasonable match between the two shows that the approach used in this dissertation is satisfactory. In the next section, we conduct further validation of the proposed method by conducting various sensitivity studies. We test various alternative forms of objective functions and demonstrate that the objective function we have selected does an excellent job of matching the results. We also compare the results by randomly adding errors in a synthetic model and show the objective form selected is robust enough to match the observed data. In the same section, we also present confidence intervals after the objective function is reduced to desired levels. The range of confidence interval indicates a satisfactory match with an acceptable uncertainty range. We also compare the results with different decline curve methods to evaluate the sensitivity of assuming different decline curve methods on the final results. We show that so long as the time steps are small, assuming different decline curve analysis method does not have significant impact on the final result. In the last section, we summarize the results from this chapter.

2.1 Oil in Place Estimation The oil in place is estimated using Sills method12 which expresses the material balance equation in the form:

F = NE o ,aq + US ( p, t ) Where:

(II-1)

12

F = cumulative reservoir voidage N = original oil-in-place Eo,aq = combined oil and aquifer expansion U = aquifer influx constant S = aquifer influx function. The analytical expressions of F, N, Eo,aq, U and S are given below. The procedure used is the following: 1. Initialize the process by reading the cumulative oil, gas and water production of the reservoir as well as the corresponding reservoir pressure values 2. Compute the cumulative reservoir voidage F and the combined oil and aquifer expansion Eo,aq as:

F = N p (Bo Rs B g ) + (G p Gi )B g + (W p Wi )Bw
m 1 E o ,aq = 2ce S + 1 S S 1 S wo wg og E o = Bo Boi + (Rsi Rs )B g
( pi p )(c f + S wo c w ) E fwo = Boi 1 S wo

(II-2)

m(E g + E fwg ) + Boi + E o + E fwo B gi

(II-3)

(II-4)

(II-5)

E g = B g B gi +

(1 S

B g i S og E o
wg

S og )Boi

(II-6)

13

( pi p )(c f + S wg c w ) E fwg = B gi 1 S wg S og
U = 1.119 fH (C w + C f ) Ro2

(II-7)

(II-8) (II-9)

f = (encroachmen tan gle) 0 / 360 0

S(p,t)=aquifer influx function. In this work, this function is computed using a polynomial approach for the Van Everdingen and Hurst water encroachment model. This approach is discussed in a paper by Klins, Bouchard and Cable49. The terms in the above equations conform to the conventional meaning: Cw=compressibility of aquifer Cf = compressibility of formation = porosity of aquifer H= aquifer thickness Bo= oil formation volume factor [bbl/stb] Bg = gas formation volume factor [bbl/mscf] Bw = water formation volume factor [bbl/stb] Rs = solution gas oil ratio [mscf/stb] ce = effective aquifer compressibility [1/psia] S = aquifer function [psi] m = reservoir volume ratio of original free gas-in-place to original oil-in-place, GBgi/NBoi

14

Np = cumulative oil production [stb] Gp = cumulative gas production Gi = cumulative gas injection Wp = cumulative water produced Wi = water injection [stb] Sog = initial gas cap oil saturation [fraction] Swg = initial gas cap water saturation [fraction] Swo = initial oil zone water saturation [fraction] Eo = oil and solution gas expansion term [bbl/stb] Efwg = gas zone formation and water expansion term Efwo = oil zone formation and water expansion term [bbl/stb] Eg = free gas expansion term [bbl/mscf] Eo = oil and solution gas expansion [bbl/stb] 3. Using the non-linear regression procedure (Levenberg-Marquardt

algorithm)23,24,25,26,27 compute N if reservoir is producing under natural depletion drive or N and RD if reservoir is producing under water influx drive mechanism. RD represents the ratio of aquifer radius to reservoir radius.

15 2.2 Relative Permeability Curves Estimation

The Muskat method14 has been extended to account for the water influx. By combining the material balance equations for oil, gas and water through the reservoir and using the definition of the GOR and WOR, the gas/oil and water/oil relative permeability ratios as well as the oil, gas and water saturations are determined as functions of reservoir pressure. The procedure used to estimate the relative permeability curves is the following: 1. Initialize the process by knowing the original oil in place. This is obtained through material balance calculation. 2. Assume that the GOR as well as the cumulative oil, gas and water production of the well is known as a function of reservoir pressure. 3. Assume the oil, gas and water saturations are known at a given pressure P. Assume a pressure decrement, P, such that the reservoir pressure decreases from P to P-P. 4. Calculate the change in oil, gas and water saturations using the following equation (see derivation in appendix A). The right hand side of the following two equations are evaluated using saturation values obtained at previous time step.
S o A( P) + S o B( P) S o = P k rg k ro (1 S o S w )C ( P) + S o D( P) k rg o k rw o + + 1 k ro g k ro w k rw B We S w E ( P) w k ro V p dP

(II-10)

16

k rg k B dWe S o A( P) + S o B( P) (1 S o S w )C ( P) + S o D( P) rw S w E ( P) w k ro k ro V p dP S w = P k rg o k rw o + + 1 k ro g k ro w k rw S o D( P) + S w E ( P) + F ( P) k ro

(II-11) and Sg=1-So-Sw Where A( P) = B g dRs Bo dP (II-13) (II-12)

B( P) =

o 1 dBo g Bo dP
1 dB g B g dP

(II-14)

C ( P) =

(II-15)

D( P) =

o 1 dBo w Bo dP
1 dBw 0.0 = negligible Bw dP

(II-16)

E ( P) =

(II-17)

F ( P) =

Bw We V p dP

(II-18)

5. Calculate the gas/oil relative permeability ratio and the water/oil relative permeability ratio using the following equations:

17

k rg k ro and

= (GOR Rs )

g Bg o Bo

(II-19)

B k rw = WOR w w k ro o Bo

(II-20)

6. Repeat step 3 through 5 so that the relative permeability ratios as well as the corresponding oil gas and water saturations are computed for all the observed reservoir pressure values. 7. Determine the relative permeability curves coefficients by curve fitting the observed relative permeability ratios using a relative permeability model. In our approach the three-phase relative permeability is calculated using the Standing model: krg = krg (Sg) krw= krw (Sw) kro= kro (Sg, Sw) Where krg, krw and kro are modeled as follows: (II-21) (II-22) (II-23)

So o k ro = k ro 1 S iw

2+ 2+ S o + S m S iw S w S iw 1 S 1 S iw iw 2+ S o + S w S iw 1 1 S iw

(II-24)

k rg

S + Sm 1 o g = k rg S S iw m
2 + 3

(II-25)

k rw

S S iw = w 1 S iw

(II-26)

18

The saturation terms in the above equations conform to the conventional meaning: Sg = gas saturation Sw = water saturation Sg = gas saturation. Siw = irreducible water saturation The parameters Sm, kroo, krgo, and have the following conventional meaning: Sm=1-Sgc, where Sgc is the gas critical saturation kroo = end point of the oil relative permeability curve krgo = end point of the gas relative permeability curve = size distribution index These four relative permeability curves parameters (Sm, kroo, krgo, and ) are determined using a non-linear regression procedure (Levenberg-Marquardt procedure)25 which minimizes the differences between the relative permeability ratios derived from observed production data and the relative permeability ratios calculated using the above relative permeability correlations.

2.3 History Match of Observed Production Data

The overall approach to accomplish the stated goals in the Introduction chapter is to combine the reservoir material balance technique with conventional nodal analysis technique to predict the performance of an oil well as a function of time. The elements of

19

decline curve analysis are used in the time evaluation. The steps involved in the procedure are as follows. 1. Initialize the process by calculating the initial oil present as well as inputing the dimensions of existing production components. 2. Calculate the initial rate at which well can produce using conventional nodal analysis. 3. Assume a small decrement in the reservoir pressure. Using the extended Muskat method, compute the GOR and WOR at which the well will produce by estimating the saturation of oil, gas and water in the reservoir at the new reservoir pressure. 4. Calculate the new oil rate at which the well will produce using nodal analysis. 5. Using material balance, calculate the amount of oil present in the reservoir at the new pressure. calculated. 6. Knowing the initial and the final rates as well as the type of decline (based on decline curve analysis), calculate the incremental time over which the oil is produced. 7. Repeat the procedure for the entire time interval over which production data are available. 8. If the actual rate and the predicted rate at different time intervals do not match, use a constrained non-linear regression procedure (Levenberg-Marquardt procedure plus Imaging Extension method)25,26,27 to vary input parameters to obtain a match. These parameters will include drainage volume, well productivity index, tubing correlations as well as tubing roughness, etc. Change the parameters and go to step 2. Repeat Steps 28 till a reasonable match is obtained. Using the difference, an incremental amount of oil produced can be

20

9. The above steps can be repeated with emphasis on different input parameters so that various alternate combinations of input parameters can satisfy the production data.

2.4 Future Performance Prediction

1. Once a satisfactory match between the predicted and the observed rates is obtained, predict the future performance of the well under the existing conditions. The

procedure will be the same as described in Steps 1-8 in the History Matching section. Repeat the steps till an abandonment pressure is reached. 2. Consider different scenarios for variations in production procedures. These include, for example, changing the number of perforations, stimulating the well, fracturing the well, etc. 3. Predict the future performance under the new operating conditions using the same procedure as explained in Step 1. 4. Repeat Step 3 for alternate combinations of input parameters to quantify uncertainties in the prediction of future performance. 5. Compare the production under the new scenario with the base case to calculate the incremental oil/gas production as a function of time. 6. Repeat Step 5 for different input configurations. 7. Use the information generated in Steps 5 and 6 to study the economic feasibility of making the changes in the production configuration.

21 2.5 Procedure

In order to fulfill the above tasks, a FORTRAN program is developed that integrates the material balance computations and production system analysis. For the production system analysis, we need to include reservoir3,4,6,7,8, tubing1, gravel pack1, perforations9, subsurface safety valve (SSSV)34,35, choke36, pipeline1 and the separator. The pressure drop predictions in individual components are well established29,30,31,32,33. equations are available in the literature. Most of these

We already have available the FORTRAN

programs to calculate the pressure drop predictions in most of these components. We also have established a procedure as to how to calculate the rate under given production components which can be easily used for the proposed method as well.

We will assume that the operator has already conducted a decline curve analysis using many of the example, Fetkovich et al. recommend exponential decline for high pressure oil/gas wells (> 5,000 psia), hyperbolic decline with a b value of between 0.4 and 0.5 for typical oil/gas wells, and a value greater than 0.5 and less than 1.0 for multiple layered oil/gas reservoirs. All the procedures for pressure drop computations in the system components as well as the non-linear regression procedure used in this work are detailed in appendix C.

2.6 Validation of the Model

To validate the model presented in this work, the results obtained from a forward simulation using the dynamic nodal analysis procedure have been compared to the results obtained from an ECLPSE forward model28, using the same input data set in both simulations. The following Table 2.1 presents the synthetic data used in both ECLIPSE and dynamic nodal analysis simulations.

22 Table 2.1

Synthetic data: input parameters. Type of decline Pressure decrement [psia] Optimization tolerance Reservoir Initial pressure [psia] Initial temperature [F] Pay [ft] Skin Drainage radius [ft] Permeability [md] Porosity [fraction] Water saturation [fraction] Fluid properties Specific gravity of produced gas Oil density [API] Specific gravity of produced water Completion Hole diameter [in] Casing diameter [in] Perforated interval [ft] Perforation diameter [in] Perforation tunnel length [in] Perforation density [SPF] Mode of perforation Tubing inside diameter [in] Tubing roughness [ft] Tubing length [ft] Hole inclination angle [degree] Pressure drop correlation: Production Gas/Oil ratio at bubble point, [scf/stb] Well head pressure, [psia] Well head temperature, [F] Reference separator pressure, [psia] Reference separator temperature, [deg F] = 12.25 = 9.625 = 40 = 0.720 = 12.33 = 12 = underbalanced = 1.945 = 0.00015 = 8688.0 = 90 Beggs and Brill = = = = = 400 100 100 14.7 60.0 = = = = = = = = = = = 2200 140 40.0 0.0 500 32.33 0.19 0.15 0.65 35 1.0 = exponential = 50 = 0.000001

23

The results obtained for the forward ECLIPSE simulation and for the dynamic nodal analysis simulation are presented in figure 1.1, figure 2.2 and figure 2.3.

24

Predicted reservoir pressure [forward dynamic nodal analysis simulation] Predicted reservoir pressure [forward ECLIPSE simulation]
2500

2000

Pressure [psia]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [days]

Figure 2.1. Reservoir pressure: eclipse forward simulation versus dynamic nodal analysis forward simulation.

25

Predicted rate [forward dynamic nodal analysis simulation ] Predicted rate [forward ECLIPSE simulation]
1200

1000

800

Rate [stb/d]

600

400

200

0 0 200 400 600 800 1000 1200

Time [days]

Figure 2.2. Rate: eclipse forward simulation versus dynamic nodal analysis forward simulation.

26

Predicted GOR [forward dynamic nodal analysis simulation] Predicted GOR [forwad ECLIPSE simulation]
4500

4000

3500

3000

GOR [scf/stb]

2500

2000

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [days]

Figure 2.3. GOR: eclipse forward simulation versus dynamic nodal analysis forward simulation.

27

As can be seen, the results obtained from the dynamic nodal analysis model and ECLIPSE agree each other quite well. The small discrepancies observed in the two models are a result of single cell model in our program versus multi-cell model in ECLIPSE.

2.7 Sensitivity Analysis 2.7.1 Sensitivity of History Match Results With Respect to Objective Function Selection

In order to get a match between the observed data and the model predicted values for reservoir pressures and rates, we need to minimize the differences between the observed data and the predicted values of reservoir pressure and rate. This is accomplished by minimizing an objective function. Three different methods have been used to define the objective function: the least square difference method, the absolute difference method, and the robust method.

2.7.1.1 Least Square Difference Method

The objective is defined as the euclidean norm of a vector FVEC defined as:

28
Qobs ,1 Qmod,1 Q obs , 2 Qmod, 2 . . Qobs ,n Qmod,n FVEC = Pr ,obs1 Pr ,mod 1 P Pr ,mod 2 r ,obs 2 . . Pr ,obsn Pr ,mod n

(II-26)

The objective function Fobj is then defined as: Fobj = (Qobs ,i Qmod,i ) + (Pr ,obs ,i Pr ,mod,i )
n n i =1 i =1 2 2

(II-27)

where n = total number of observed data points Qobs,i = observed oil rate corresponding to the ith data point Qmod,i = model predicted oil rate corresponding to the ith data point Pr,obs,i = observed reservoir pressure corresponding to the ith data point Pr,mod,i = model predicted reservoir pressure corresponding to the ith data point

2.7.1.2 Absolute Difference Method

The objective function is defined as the absolute norm of a vector FVEC that is defined as:

29
Qobs ,1 Qmod,1 Q obs , 2 Qmod, 2 . . Qobs ,n Qmod,n FVEC = Pr ,obs1 Pr ,mod 1 P Pr ,mod 2 r ,obs 2 . . Pr ,obsn Pr ,mod n

(II-28)

The objective function Fobj is then defined as:

Fobj = Qobs ,i Qmod,i + Pr ,obs ,i Pr ,mod,i


i =1 i =1

(II-29)

where n = total number of observed data points Qobs,i = observed oil rate corresponding to the ith data point Qmod,i = model predicted oil rate corresponding to the ith data point Pr,obs,i = observed reservoir pressure corresponding to the ith data point Pr,mod,i = model predicted reservoir pressure corresponding to the ith data point

2.7.1.3 Robust Method

The objective function is defined as the euclidean norm of a vector FVEC defined as:

30
F1 F 2 F FVEC = 3 F4 . F5 F6

(II-30)

The components Fj of vector FVEC are chosen such that the computation is more resistant to errors in the observed data and is less sensitive to outliers. The definition of the functions Fj is presented below:

2.7.1.3.1 Component F1

This component express the fact that ideally the correlation coefficient between the observed and model predicted rate is equal to 1. Mathematically, we define,

F1 = (Qobs , Qmod ) 1

(II-31)

where (Qobs, Qmod) represents the correlation coefficient between the observed rate data and the model predicted rate values:

(Qobs , Qmod ) =

COV (Qobs , Qmod ) Qobs * Qmod

(II-32)

The advantage of using the correlation coefficient is that it is resistant to noise in the data. It is not sensitive to outliers. It should be noted that high correlation coefficient does not necessarily mean that the values are similar.

31

The basic assumption here is that we are modeling the measured data correctly that the errors in the measured data are normally distributed with mean zero, and the errors are not correlated.

2.7.1.3.2 Component F2

This component represents the fact that ideally the plot of Qmod versus Qobs is a straight line of slope one ( with intercept equal to zero) :

F2 = SLOPE1
where SLOPE is defined as:
SLOPE = COV (Qobs , Qmod )

(II-33)

2 obs

(II-34)

COV(Qobs, Qmod) is the covariance between the observed and model predicted rates.

2.7.1.3.3 Component F3

This component is chosen to represent the fact that ideally the intercept of the straight line Qmod versus Qobs is equal to zero.
Intercept = Q mod SLOPE * Q obs = 0

(II-35)

So, F3 is defined as:


F3 = Qmod 1 Qobs

(II-36)

32 2.7.1.3.4 Component F4

This component express the fact that ideally the correlation coefficient between the observed and model predicted reservoir pressure is equal to 1. Mathematically, we define,

F1 = ( Pobs , Pmod ) 1

(II-37)where (Pobs, Pmod

( Pobs , Pmod ) =

COV ( Pobs , Pmod ) Pobs * Pmod

(II-38)

The basic assumption here is that we are modeling the measured data correctly that the errors in the measured data are normally distributed with mean zero, and the errors are not correlated.

2.7.1.3.5 Component F5

This component represents the fact that ideally the plot of Pmod versus Pobs is a straight line of slope one ( with intercept equal to zero) :
F2 = SLOPE 1

(II-39)

where SLOPE is defined as: SLOPE = COV ( Pobs , Pmod ) (II-40)

2 obs

COV(Pobs, Pmod) is the covariance between the observed and model predicted pressures.

33 2.7.1.3.6 Component F6

This component is chosen to represent the fact that ideally the intercept of the straight line Pmod versus Pobs is equal to zero. Intercept = P mod SLOPE * P obs = 0 So, F6 is defined as:
F6 = Pmod 1 Pobs

(II-41)

(II-42)

The objective function is then defined as


Fobj = Fi 2
i =1 6

(II-43)

The dynamic nodal analysis history match procedure was run using the three different objective function definitions presented above. The synthetic data presented in table 2.1 are used as observed data. The results of the simulations are given in the following table 2.2 and in figure 2.4, figure 2.5 and figure 2.6 Table 2.2 Sensitivity of history match results with respect to objective function selection.
Robust method Half length True Initial Confidence value guess Minimum Maximum Final value Final value Final value interval 22 -2 9 22 -2 9 42 2 16 32.33 0 12.33 32.33 0 12.33 32.33 0 12.33 0.009 0.0043 0.0036 Square Absolute differences differences

Parameters Permeability [md] 32.33 Skin [-] 0 Perforation length [inch] 12.33

34

Pressure history match using robust method Observed pressure Pressure History match using square differences Pressure history match using absolute differences
2500

2000

P ressu re [p sia]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [day]

Figure 2.4. Sensitivity of reservoir pressure with respect to objective function selection.

35

Rate History match using robust method History match using square differences History match using absolute differences Observed rate
1200

1000

800

Rate [stb/d]

600

400

200

0 0 200 400 600 800 1000 1200

Time [day]

Figure 2.5. Sensitivity of rate with respect to objective function selection.

36

GOR history match using robust method GOR history match using square differences GOR history match using absolute differences Observed GOR
4500 4000 3500 3000

G O R [scf/stb]

2500 2000 1500 1000 500 0 0 200 400 600 800 1000 1200

Time [days]

Figure 2.6. Sensitivity of GOR with respect to objective function selection.

37

As can be seen, the results obtained from all three objective function methods agree with each other quite well. The permeability, the skin as well as the length of perforation predicted from the three objective function methods are the same. The rate, the pressure as well as the GOR curve agree each other very well.

2.7.2 Sensitivity of History Match Results With Respect to Errors

In order to further assess the sensitivity of the history match with respect to the selection of the objective function, some errors are introduced randomly in the observed rate data in order to verify the effect of bad data or outliers on the dynamic nodal analysis computer program. The errors added are between 30% and +30% of the original observed rate data. The results of the history match are presented in table 2.3 and in figure 2.7, figure 2.8 and figure 2.9. Table 2.3 Sensitivity of history match results with respect to errors in the rate data.
Initial Final Parameters True value guess Minimum Maximum value Permeability [md] 32.33 22 22 42 36.08 Skin [-] 0 -2 -2 2 0.937 Perforation length [inch] 12.33 9 9 16 9.006 Half length Confidence interval 0.806 0.4421 0.011

38

Predicted rate using robust method [after adding errors between 30% and +30% to the original data] Predicted rate using square differences [after adding errors between -30% and +30% to the original data] Predicted rate using absolute differences [after adding errors between -30% and +30% to the original data] Observed rate [after adding errors between -30% and +30%] Predicted rate [no error added to the original data]
1200 1000

R ate [stb/day]

800 600 400 200 0 0 200 400 600 800 1000 1200

Time [day]
Figure 2.7. Sensitivity of rate with respect to objective function selection and errors in the rate data.

39

Predicted GOR using the robust method [after adding errors between -30% to +30% to the original data Predicted GOR using square differences [after adding errors between -30% and +30% to the original data] Predicted GOR using absolute differences [after adding -30% to +30% to the original data] Observed GOR Predicted GOR [no error added to the original data]
4500 4000 3500

GOR [scf/stb]

3000 2500 2000 1500 1000 500 0 0 200 400 600 800 1000 1200

Time [day]

Figure 2.8. Sensitivity of GOR with respect to objective function selection and errors in the rate data.

40

Predicted pressure using robust method [after adding errors between -30% and +30% to the original data Predicted pressure using square differences [after adding errors between -30% and +30% to the original data] Predicted pressure using absolute differences [after adding errors between -30% to +30% to the original data] Observed pressure
2500

2000

Pressure [psia]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [day]

Figure 2.9. Sensitivity of reservoir pressure with respect to objective function selection and errors in the rate data.

41

As it can be seen the predicted rate and predicted pressure obtained after adding errors between +30% and 30% to the original observed rate agree very well each other for all three objective function methods described in the above section. The match between the predicted GOR and the observed GOR is also fairly good, although not perfect. This is due to the fact that the errors added to the original observed rate values have not been incorporated in the observed GOR.

2.7.3 Statistical Analysis: Confidence Interval Computation

Confidence interval has been computed based on the results of the history match obtained from the robust method26,27. By definition, a 95% confidence interval is a range of values within which there is a 95% probability of finding the actual parameter value. The computation of the confidence interval is performed using the subroutine CONFIDL26,27. The results of the computation are shown in table 2.2 and in table 2.3. In both tables, it can be seen that the confidence interval are smaller than or equal to the interval defined by the minimum and maximum values of the regression parameters.

2.7.4 Sensitivity of History Match Results With Respect to Type of Decline Curve

Exponential decline, hyperbolic decline (b=0.5) and harmonic decline have been used to compute the time increments during the history match computations for the same observed data set. The results are shown in figure 2.10, figure 2.11 and figure 2.12.

42

Predicted reservoir pressure [exponential decline] Predicted reservoir pressure [ harmonic decline] Predicted reservoir pressure [hyperbolic decline] Observed reservoir pressure
2500

2000

Pressure [psia]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [day]

Figure 2.10. Sensitivity of reservoir pressure with respect to type of decline.

43

Predicted rate [exponential decline] Predicted rate [harmonic decline] Predicted rate [hyperbolic decline] Observed rate
1200

1000

800

Rate [stb/day]

600

400

200

0 0 200 400 600 800 1000 1200

Time [day]

Figure 2.11. Sensitivity of rate with respect to type of decline.

44

Predicted GOR [exponential decline] Predicted GOR [harmonic decline] Predicted GOR [hyperbolic decline] Observed GOR
4500 4000 3500 3000

GOR [scf/stb]

2500 2000 1500 1000 500 0 0 200 400 600 800 1000 1200

Time [day]

Figure 2.12. Sensitivity of GOR with respect to type of decline.

45

As it can be seen, the production history match as well as pressure history match and the GOR history match are the same for all the three types of decline. So, any of these three types of decline can be used for the dynamic nodal analysis procedure to compute the time decrements because the time steps involved are very small.

2.8 Summary

In this chapter, the dynamic nodal analysis procedure for oil reservoir has been presented. The mathematical models used in the history match and forecast algorithms as well as the regression analysis method used and the conventional nodal analysis technique are presented. Further details can be found in Appendix C.

CHAPTER III SINGLE WELL PERFORMANCE MATCH: RESULTS

In this chapter the dynamic nodal analysis described in Chapter II is applied to production systems. Those production systems include synthetic data obtained from the forward model as well as from a reservoir simulator ECLIPSE. This chapter is divided into four sections. In the first section, we apply the dynamic nodal analysis to synthetic data that have been obtained using the results from running the dynamic nodal analysis program itself in forward mode. In the second section, the dynamic nodal analysis is applied to synthetic data that have been generated from an ECLIPSE simulation of a hypothetical well producing in a homogeneous reservoir. In the third section, the dynamic nodal analysis is applied to synthetic data that have been obtained from an ECLIPSE simulation of a hypothetical well producing in a heterogeneous reservoir. Through application under various conditions, we demonstrate that our methods works well in predicting the historical performance. In the last section, we summarize the results from this chapter.

47

48 3.1 Synthetic Data From Dynamic Nodal Analysis Forward Model These synthetic data have been generated using the results from a forward simulation of a hypothetical well. They represent an oil well which was open to production for 2.8 years. The characteristics of the reservoir as well as the description of the completion are summarized in Table 2.1 shown in chapter II.

3.1.1 Oil in Place The oil in place is computed as explained in Chapter II. The results of the computation gives an oil in place of 754727 STBO. The true value is 754653 STB. Therefore, the estimated originally oil in place is in good agreement with the true value.

3.1.2 Relative Permeability Calculations The regression procedure was run to estimate the coefficients that appear in the relative permeability curves correlations, as discussed in chapter II. The results of the history match using different initial regression parameter values are shown in the following table 3.1 and in figure 3.1.

49 Table 3.1 Relative permeability parameters estimation.

Parameters Sm k rg0 k ro
0

True value 2 0.95 0.65 0.85

Case I Initial Final guess value 2 2 0.95 0.95 0.65 0.85 0.65 0.85

Case II Initial Final guess value 4 1.66 0.9 0.95 0.5 0.75 0.65 0.99

Case III Initial Final guess value 5 3.16 0.9 0.95 0.9 0.95 0.65 0.63

50

krg/kro ratio derived from observed data predicted krg/kro ratio (case I) Predicted krg/kro ratio (case II) Predicted krg/kro ratio (case III)
0.12

0.1

0.08

Krg/Kro

0.06

0.04

0.02

0 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85

Oil saturation

Figure 3.1: Match of krg/kro ratio obtained from observed data and the krg/kro ratio calculated from correlations.

51 3.1.3 History Match The computer program was run with the regression parameters selected to be permeability, skin and length of perforation. The results of the history match with different initial regression parameter values are shown in table 3.2. Table 3.2 History match results.

Case I Initial Parameters True value guess Permeability [md] 32.33 22 Skin [-] Perforation length [inch] 0 12.33 -2 9 Final value Initial guess 10 4 14

Case II Final value Initial guess 40 10 5

Case III Final value

32.33 0.0099

39 .97 4.52 0.55 1.62 10.26 1.52

39.81.56 2.46 0.03 3.55

0 0.0043 12 .33 0.0036

5.04 2.43

As can be seen, the results obtained are fairly good. The match for permeability, skin and length of perforation is very good between case I and case II. However, the match between case III and the true values is not very good. This points out to non-unique solutions and local minima problems evident in gradient methods. The match of rate, pressure and GOR values are shown in figure 3.2, figure 3.3, figure 3.4. The predicted rate and pressure profiles match very well the observed rate and pressure data. The match of the predicted GOR and the observed GOR is reasonably good as well.

52

Observed reservoir pressure [from forward model] Predicted reservoir pressure (case I) Predicted reservoir pressure (case II) Predicted reservoir pressure (case III)
2500

2000

Pressure [psaia]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [days]

Figure 3.2. Synthetic data: sensitivity of reservoir pressure history match with respect to different guess values of regression parameters.

53

Observed rate [from forward model] Predicted rate (case I) Predicted rate (case II) Predicted rate (case III)
1400

1200

1000

Rate [stb/d]

800

600

400

200

0 0 200 400 600 800 1000 1200

Time [day]

Figure 3.3: Synthetic data: sensitivity of rate history match with respect to different guess values of regression parameters.

54

Observed GOR [from forward model] predicted GOR (case I) Predicted GOR (case II) Predicted GOR (case III)
4500 4000 3500 3000

GOR [scf/stb]

2500 2000 1500 1000 500 0 0 200 400 600 800 1000 1200

Time [days]

Figure 3.4: Synthetic data: sensitivity of GOR history match with respect to different guess values of regression parameters.

55 3.2 Synthetic Data: Eclipse Homogeneous Model These synthetic data have been generated using the results from an ECLIPSE simulation of a hypothetical well. They represent an oil well which was open to production for 3 years.

3.2.1 Oil in Place The oil in place is computed as explained in Chapter II using material balance equation. The results of the computation gives an oil in place of 753857 STBO. The true value is 754455 STBO. As can be seen these two values are in very good agreement.

3.2.2 Relative Permeability Calculations The coefficients , Sm, Krg0, Kro0 that appear in the relative permeability curve correlations, have been estimated using the regression procedure described in chapter II. The results are shown in the following table 3.3 Table 3.3 Homogeneous reservoir: relative permeability parameters estimation.
Limits Initial guess Minimum Maximum Estimates 4 0.5 6 2.39 0.9 0.85 1 0.976 0.5 0.75 0.4 0.7 1 1 0.461 0.718

Regression parameters Sm Krg Kro


0 0

TRUE 2 0.95 0.65 0.85

56 3.2.3 History Match The computer program was run with the regression parameters selected to be permeability, skin and length of perforation. The results of the history match are shown in table 3.4 and in figure 3.5, figure 3.6, figure 3.7. Table 3.4 Homogeneous reservoir: history match results.
Limits Regression parameters TRUE Initial guess Minimum Maximum Estimates 31.08 1.14 Permeability [md] 32.33 10 1 50 Skin [-] 0 4 -5 5 1.33 0.42 Length of perforation [inch] 12.33 14 9 16 9.99 0.36

57

Observed reservoir pressure (eclipse )


2500

Predicted reservoir pressure

2000

Reservoir pressure [psia]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [days]

Figure 3.5: Synthetic data (homogeneous reservoir): reservoir pressure history match.

58

Observed rate (eclipse)


1400

Predicted rate

1200

1000

Rate [stb/day]

800

600

400

200

0 0 200 400 600 800 1000 1200

Time [days]

Figure 3.6: Synthetic data (homogeneous reservoir): rate history match.

59

Observed GOR (eclipse)


4500 4000 3500 3000

Predicted GOR

GOR [scf/stb]

2500 2000 1500 1000 500 0 0 200 400 600 800 1000 1200

Time [days]

Figure 3.7: Synthetic data (homogeneous reservoir): GOR history match.

60 3.3 Synthetic Data: ECLIPSE Heterogeneous Model These synthetic data have been generated using the results from an eclipse simulation of a hypothetical well at the center of a heterogeneous reservoir. They represent an oil well which was open to production for 1.1 years. The permeability distribution is shown in the following figure 3.8. The well is located in grid block (6,6,1).

Heterogeneous Reservoir: Permeability Distribution


32 28 24 20
0 500.0 1000 1500 2000 2500 3000 3500 4000 4500 5000

Y dimension

16 12 8 4

12

16

20

24

28

32

X dimension

Figure 3.8: Synthetic data (heterogeneous reservoir): permeability distribution.

61 3.3.1 Oil in Place The oil in place is computed as explained in Chapter II. The results of the computation gives an oil in place of 754452 STBO. The true value is 754396 STB. As can be seen these two values are in very good agreement.

3.3.2 Relative Permeability Calculations The coefficients , Sm, Krg0, Kro0 that appear in the relative permeability curve correlations, have been estimated using the regression procedure described in chapter II. The results are shown in the following table 3.5 Table 3.5 Heterogeneous reservoir: relative permeability parameters estimation.
Limits Minimum Maximum 1 3 0.9 1 0.5 0.8 0.7 0.95

Regression parameters Sm Krg0 Kro


0

TRUE 2 0.95 0.65 0.85

Initial guess 2.5 0.9 0.6 0.9

Estimates 2.11 0.98 0.548 0.95

3.3.3 History Match The computer program was run with the regression parameters selected to be permeability, skin and length of perforation. The results of the history match are shown in table 3.6 and in figure 3.9, figure 3.10, figure 3.11.

62 Table 3.6 History Match for Synthetic data obtained from heterogeneous reservoir.
Limits Minimum Maximum 26 4930 -2 2 9 16

Regression parameters TRUE Permeability [md] 1 to 5000 Skin [-] 0 Length of perforation [inch] 12.33

Initial guess 32.33 0 14

Estimates 54.1 -1.9 14.41

Observed reservoir pressure [Eclipse] History match Forecast Truth [Eclipse]


2500

2000

Reservoir pressure [psia]

1500

1000

500

0 0 200 400 600 800

Time [days]

Figure 3.9: Synthetic data (heterogeneous reservoir): reservoir pressure history match.

63

Observed rate [Eclipse] History match Forecast Truth [Eclipse]


3500 3000 2500 2000 1500 1000 500 0 0 200 400 600 800

Rate [stb/day]

Time [days]

Figure 3.10: Synthetic data (heterogeneous reservoir): rate history match.

64

Observed GOR [Eclipse] History match Forecast Truth [Eclipse]


4500 4000 3500

GOR [scf/stb]

3000 2500 2000 1500 1000 500 0 0 200 400 600 800

Time [days]

Figure 3.11 Synthetic data (eclipse heterogeneous model): GOR history match.

65 3.4 Summary The dynamic nodal analysis procedure has been applied to a single oil well model to a homogeneous reservoir as well as to a heterogeneous reservoir. The results obtained using this procedure agree very well with the results obtained from ECLIPSE simulator. This confirms the applicability of the procedure.

CHAPTER IV MULTI-WELL PERFORMANCE MATCH: APPROACH

This chapter discusses the procedure used for matching performance of reservoir produced by using multiple wells. We will discuss the procedure in this chapter followed by examples in the next chapter. This chapter is divided into seven sections. In the first section, we discuss the procedure for estimating oil in place using material balance technique. This procedure is the same as the one used in chapter II. However, in the multiwell scheme, the field cumulative production data are used instead of individual well production. In the second section, we discuss the procedure used for the estimation of relative permeability curves using production data. Again, the procedure is the same as the one used in chapter II. However, for the multi-well scheme, the field cumulative

production data are used instead of individual well production. In the next section, we present the model used for history matching the production data and discuss the assumptions made in this process. In section four, we present the mathematical model used in the multi-tank scheme. In this model, the system is divided into three elements: the wells, the well-regions and the reservoir-region. During production, fluids (oil, gas or water) flow from the reservoir-region to the well-regions and from well-regions to the wells. The details of the mathematical derivations are presented in appendix B. In the next section, we develop the procedure for history matching the production data. We match the observed oil rates data by adjusting reservoir 66

67 parameters (transfer coefficients, radius of well-regions and radius of reservoir region). This is done using non-linear regression analysis. In section six, we discuss the procedure we use to predict the future performance of the wells once the observed production data are satisfactorily matched. In the last section, we summarize the results from this chapter.

4.1 Oil in Place Estimation The procedure used in chapter two for the estimation of oil in place for the single well scheme is also used for multi-well scheme. In the multi-well scheme, the method is applied using the cumulative field data (field cumulative production of oil, field cumulative production of gas, field cumulative production of water and reservoir pressure), instead of individual well production data.

4.2 Relative Permeability Curves Estimation The procedure used in chapter two for the estimation of relative permeability curves for the single-well scheme, is also used for the multi-well scheme. In the multi-well scheme, the method is applied using the cumulative field data: (field GOR, field WOR, reservoir pressure), instead of individual well production data.

4.3 History Match of Observed Production Data In the multi-well scheme, the reservoir is subdivided into circular regions around the wells (well regions); the rest of the reservoir constitutes the reservoir region (see Figure 4.1 below).

68 The following assumption are made: 1. The pressure is uniform in each well region and in the reservoir-region and is equal respectively to the average pressure of the well region or to the average pressure of the reservoir region. 2. The depth of each well region and of the reservoir region is evaluated at the center of the well-region or at the center of the reservoir-region respectively. 3. The transmissibilities between the reservoir-region and the well regions are evaluated using upstream weighting criteria. 4. No cross flow between the well regions is allowed.

4.4 Mathematical Modeling Figure 4.1 presents a sketch of the mathematical modeling used in the multi-well scheme. The figure shows two parts separated by a big arrow. The upper part represents the system that is made of the actual reservoir and the actual wells. The lower part shows the model used to derive the mathematical description of the system. In this model, the system is divided into three elements: the wells, the well-regions and the reservoir-region. During production, fluids (oil, gas or water) flow from the reservoir-region to the well-regions and from well-regions to the wells.

69 reservoir well

Well-region Reservoir-region

Well

Figure 4.1: Multi-well scheme representation.

70 By applying the mass balance equation and Darcys law to each well-region and to the reservoir-region, we get the following system of two flow equations (see details in appendix B)

AS i n * Po , R n +1 + AZ i n * Po ,i n +1 = AQi n NW n +1 n n +1 ASRR n * Po , R + AZRi * Po ,WRi = AQR n i =1

(IV-1)

The coefficients ASin, AZin, AQni, ASRRn, AZRin, AQRn are computed using saturation and pressure values at previous time step (time Tn ). The expression of the coefficients ASin, AZin, AQni, ASRRn, AZRin, AQRn are given respectively in equation B106A, equation B-106B and equation B-106C. The above system of equations can be written in the matrix form as follows:

AS1n n AS 2 L K L L AS n NW ASRR n

AZ 1n 0 0 L L 0 0 AZR1n

0
n AZ 2 0

L L L L n AZR2

L L L 0 L L L L L L L L L L L L L L L L L L L L

+ Pon, R1 AQ1n n +1 n Po ,1 AQ2 L L L L L L = (IV-2) 0 L L 0 L L L +1 n n 0 AZ NW Pon, NW 1 AQ NW n +1 n L AZR NW Po , NW AQR n

L L L L 0

0 0

The LU decomposition method is used to solve for P(n+1). Once the pressure values P(n+1) are determined using saturation values at previous time step Tn , the saturation values at present time Tn+1 are computed as follows:

71
n S w n +1 YY1 * t + B BW w WRi =

(S )

n +1 w WRi

WRi WRi

(IV-

3)

(S )
n +1 g

WRi

n 1 Rso ZZ 3 * t VV 1 + S g B Bo g WRi = n +1 1 Rso B g Bo WRi WRi

(IV-4)

(S )
n +1 w

n Rsw S w n +1 Bw ZZ 2 * t + B w R = R

(IV-5)

(S )
n +1 g

n 1 Rso ZZ 4 * t VV 2 + S g B Bo g R = n +1 1 Rso B g Bo R R

(IV-6)

where : (Sw n+1 )WRi represent water saturation in well-region i at time Tn+1 (Sg n+1 )WRi represent gas saturation in well-region i at time Tn+1

72

(Sw n+1 )R represent water saturation in reservoir-region at time Tn+1 (Sg n+1 )R represent gas saturation in reservoir-region at time Tn+1

4.5 History Match

1. Initialize the process by imputing the dimensions of each region (well regions and reservoir region) as well as defining the ranges of variation of the oil, gas and water transfer coefficients from the reservoir region to each well region and from the well regions to the wells. The range of variation of the radii of each well region is also defined. 2. Select a time step19,20 and solve for the pressure for each region using saturation values obtained at the previous time step. 3. Compute oil, gas and water saturations using pressure values computed in the same time step (present time). 4. Repeat the procedure for the entire time interval over which production data are available. 5. Knowing the oil, gas and water rates, calculate the bottom hole pressure knowing the well head pressure. 6. Calculate the oil, gas and water rates at each well for the given bottom hole flowing pressure as well as the average reservoir pressure. 7. If the actual rate and actual reservoir pressure and the predicted rate and the predicted reservoir pressure values at different time intervals do not match, use constrained nonlinear regression procedure (Levenberg-Marquardt procedure plus Imaging Extension

73

method)25,26,27 to vary the regression parameters in order to obtain a match. These parameters will include the well region radius and the transfer coefficients values. Change the parameters and go to step 1. Repeat Steps 1-6 till a reasonable match is obtained.

4.6 Future Performance Prediction

1. Once a good match between the predicted and the observed rates and reservoir pressure is obtained, predict the future performance of the well under the existing conditions. The procedure will be the same as described in Steps 1-5 in the History Matching section. Repeat the steps till an abandonment pressure is reached. 2. Predict the future performance under the new operating conditions using the same procedure as explained in Step 1. 3. Repeat Step 2 for alternate combinations of input parameters (Step 1 - History Match) to quantify uncertainties in the prediction of future performance. 4. Compare the production under the new scenario with the base case to calculate the incremental oil production as a function of time. 5. Use the information generated in Steps 2 and 3 to study the economic feasibility of making the changes in the production configuration. The derivation of the mathematical model is detailed in appendix B.

4.7 Summary

In this chapter, the multi-tank model for multi-well production analysis has been presented. The mathematical model used has been described and is detailed in appendix B.

CHAPTER V MULTI-WELL PERFORMANCE MATCH: RESULTS

In this chapter the multi-tank procedure described in Chapter IV is applied to production systems. Those production systems include synthetic data as well as actual field data. This chapter is divided into three sections. In the first section, we apply the multitank model to synthetic data that have been generated using the results from an ECLIPSE forward simulation of a hypothetic horizontal reservoir producing from three wells. In the second section, the multi-tank model is applied to synthetic data that have been generated from an ECLIPSE simulation of a hypothetical dipping reservoir that is produced from three wells. In the third section, the multi-tank model is applied to an actual field that produces from twelve wells. In all three cases, we obtain a reasonable match with the observed production data - validating our procedure.

5.1 Synthetic Data: Horizontal Reservoir These synthetic data have been generated using the results from an ECLIPSE forward simulation of hypothetic wells. They represent three oil wells which were open to

74

75 production for 3 years. The characteristics of the reservoir are summarized Table 5.1. These synthetic data are used as observed data. Table 5.1 Synthetic data: input parameters. Reservoir Initial pressure at datum [psia] Datum depth (vertical) [ft] Reservoir depth (top) [ft] Initial temperature [F] Pay [ft] Skin Well 1 Well 2 Well 3 Drainage radius of the reservoir [ft] Permeability [md] Porosity [fraction] Water saturation [fraction] Fluid properties Specific gravity of produced gas Oil density [API] Specific gravity of produced water Reference separator pressure, [psia] Reference separator temperature, [deg F]

= = = = = = = = = = = =

2200 5000 5000 140 40.0 0 5.0 -3.0 866.07 32.33 0.19 0.15

= = = = =

0.65 35 1.0 14.7 60.0

76 5.1.1 Oil in Place Calculation Using the observed pressure data and the observed cumulative oil and gas production data, the original oil in place has been determined using the material balance equation. The computation gives a value of the original oil in place (OOIP) of 2266246 STBO. The true value is 2263415 STBO.

5.1.2 Calculation of Relative Permeability Curves Using the observed pressure data and the observed cumulative production and GOR data, the relative permeability ratio curves have been estimated and shown in Figure 5.1. The values of the coefficients , kroo , krgo, and Sm (Sm =1-Sgc ) obtained are shown in the following Table 5.2. As can be seen on figure 5.1, the predicted Krg/Kro curve reproduces very well the Krg/Kro curve obtained from observed data.

Table 5.2 Relative permeability regression parameters.


Limits Minimum Maximum 1 6 0.9 1 0.4 0.5 0.8 1

Regression parameters Sm Krg Kro


o o

TRUE 2 0.95 0.65 0.85

initial guess 2.5 0.98 0.4 0.8

Estimates 2 0.95 0.65 0.85

77

Krg/Kro ratio from observed data Predicted Krg/Kro ratio


0.035

0.03

0.025

Krg/kro ratio [-]

0.02

0.015

0.01

0.005

0 0.6 0.65 0.7 0.75 0.8 0.85 0.9

Oil saturation [-]

Figure 5.1: Match between estimated gas/oil relative permeability ratio and gas/oil relative permeability ratio obtained from observed data (synthetic data, dipping reservoir).

78 5.1.3 History Match The computer program was run with the regression parameters selected to be the radii of well regions, the transfer coefficients from reservoir-region to well-regions and the transfer coefficients from well-regions to corresponding wells. The results of the history match are shown in table 5.3, table 5.4, table 5.5, and in figure 5.2, figure 5.3, figure 5.4, figure 5.5, figure 5.6, figure 5.7 and figure 5.8. Table 5.3 represents the results of the estimation of the radius of each well-region. As can be seen, the estimated value (final value) for the radius of each well-regions is very close to the initial value. Therefore, in this case, the history match is not very sensitive to the well-region radius. Table 5.4 shows the results of the estimation of the transfer coefficients from reservoir- region to well-regions. Table 5.5 represents the results of the estimation of the transfer coefficient from each well-region to the corresponding well. Figure 5.2 shows the history match results for the reservoir pressure. As can be seen, the observed reservoir pressure values and the predicted values match very well. Figure 5.3 and figure 5.4 represent the history match results for the rate and the GOR for well 1. As can be seen, the observed values and predicted values match very well. Figure 5.5 and figure 5.6 represent the history match results for the rate and the GOR for well 2. Again, the observed values and predicted values match very well.

79 Figure 5.7 and figure 5.8 represent the history match results for the rate and the GOR for well 3. The match between the observed values and the predicted values is very good.

Table 5.3 Radius of well-regions estimation.


Initial value Minimum value Maximum value Final value 100 99.9 313 100.94 100 99.9 313 105.14 100 99.9 313 103.92

Well region 1 Well region 2 Well region 3

Table 5.4 Regression results for transfer coefficient from reservoir-region to well-region.
Initial value 5.520 5.520 5.520 Minimum Maximum value value Final value 0.552 551.958 5.264 0.552 551.958 5.594 0.552 551.958 5.065

Well WELL 1 WELL 2 WELL 3

Table 5.5 Regression results for transfer coefficient from well-region to well.
Minimum value 0.019 0.009 0.051 Maximum value 191.605 93.632 514.803

Well WELL 1 WELL 2 WELL 3

Initial value 1.916 0.936 5.148

Final value 1.796 0.868 5.103

80

Observed reservoir pressure Predicted reservoir pressure


2500

2000

Pressure [psia]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [days]

Figure 5.2: Synthetic data (horizontal reservoir): reservoir pressure history match.

81

Observed rate
1200

Predicted rate

1000

800

Rate [day]

600

400

200

0 0 200 400 600 800 1000 1200

Time [day]

Figure 5.3: Well 1 Synthetic data (horizontal reservoir): rate history match.

82

Observed GOR
1800 1600 1400 1200

Predicted GOR

GOR [scf/stb]

1000 800 600 400 200 0 0 200 400 600 800 1000 1200

Time [days]

Figure 5.4: Well 1 Synthetic data (horizontal reservoir): GOR history match.

83

Observed rate
700

Predicted rate

600

500

Rate [stb/day]

400

300

200

100

0 0 200 400 600 800 1000 1200

Time [days]

Figure 5.5: Well 2 Synthetic data (horizontal reservoir): rate history match.

84

Observed GOR
2000 1800 1600 1400

Predicted GOR

GOR [scf/stb]

1200 1000 800 600 400 200 0 0 200 400 600 800 1000 1200

Time [days]

Figure 5.6: Well 2 Synthetic data (horizontal reservoir): GOR history match.

85

Observed rate
3000

Predicted rate

2500

2000

Rate [scf/day]

1500

1000

500

0 0 200 400 600 800 1000 1200

Time [days]

Figure 5.7: Well 3 Synthetic data (horizontal reservoir): rate history match.

86

Observed GOR
1800 1600 1400 1200

Predicted GOR

GOR [scf/day]

1000 800 600 400 200 0 0 200 400 600 800 1000 1200

Time [days]

Figure 5.8: Well 3 Synthetic data (horizontal reservoir): GOR history match.

87 5.2 Synthetic Data: Dipping Reservoir These synthetic data have been generated using the results from an ECLIPSE forward simulation of an hypothetical dipping reservoir. They represent an oil reservoir which was open to production for 3 years through three wells. . The characteristics of the reservoir are the same as those of the horizontal reservoir presented in the previous section. The datum depth is 5000 ft and the initial reservoir pressure at datum is 2200 psia. The well positions are summarized in figure 5.9. These synthetic data are used as observed data.

Well 2

5120 ft Well 1 5270 ft Well 3 reservoir 5390 ft

Figure 5.9: Dipping reservoir: position of the wells.

5.2.1 Oil in Place Calculation Using the observed pressure data and the observed cumulative oil and gas production data, the original oil in place has been determined using the material balance equation. The

88 computation gives a value of the original oil in place (OOIP) of 2144405 STBO. The true value is 2118896 STBO.

5.2.2 Calculation of Relative Permeability Curves Using the observed pressure data and the observed cumulative production and GOR data the relative permeability ratio curves have been estimated and shown in Figure 5.10. The values of the coefficients , kroo , krgo, and Sm (Sm =1-Sgc ) obtained are shown in the following Table 5.6. As can be seen on figure 5.10, the predicted Krg/Kro curve reproduces very well the Krg/Kro curve obtained from observed data.

Table 5.6 Relative permeability regression parameters.


Limits Minimum Maximum 1 6 0.8 1 0.4 0.7 1 1

Regression parameters Sm Krg Kro


o o

TRUE 2 0.95 0.65 0.85

initial guess 4 0.9 0.5 0.75

Estimates 5.6 1 0.54 0.86

89

Krg/Kro ratio from observed data Predicted Krg/Kro ratio


0.018 0.016 0.014 0.012

Krg/Kro ratio

0.01 0.008 0.006 0.004 0.002 0 0.6 0.65 0.7 0.75 0.8 0.85 0.9

Oil saturation
Figure 5.10: Match between estimated gas/oil relative permeability ratio and gas/oil relative permeability ratio obtained from observed data (synthetic data, dipping reservoir).

90 5.2.3 History Match The computer program was run with the regression parameters selected to be radii of well-regions, transfer coefficients from reservoir-region to well-regions, and transfer coefficients from well-regions to corresponding wells. The results of the history match are shown in table 5.7, table 5.8, table 5.9 and in figure 5.11, figure 5.12, figure 5.13, figure 5.14, figure 5.15, figure 5.16 and in figure 5.17. Table 5.7 represents the results of the estimation of the radius of each well-region. Table 5.8 shows the results of the estimation of the transfer coefficients from reservoir- region to well-regions. Table 5.9 represents the results of the estimation of the transfer coefficient from each well-region to the corresponding well. Figure 5.11 shows the history match results for the reservoir pressure. As can be seen, the observed reservoir pressure values and the predicted values match very well. Figure 5.12 and figure 5.13 represent the history match results for the rate and the GOR for well 1. As can be seen, the observed rate values and predicted rate values match very well. However the predicted GOR values are slightly lower than the observed GOR values. This is due to the fact that the program tends to exhibit lower GOR values for deeper wells. Figure 5.14 and figure 5.15 represent the history match results for the rate and the GOR for well 2. The observed values and predicted values match quite well.

91 Figure 5.16 and figure 5.17 represent the history match results for the rate and the GOR for well 3. The match between the observed rate values and the predicted rate values is very good. However the predicted GOR values are lower than the observed GOR values. This is due to the fact that the program tends to exhibit lower GOR values for deeper wells.

Table 5.7 Radius of well-regions regression limits.


Initial value 100 100 100 Minimum value 99.9 99.9 99.9 Maximum value 300 300 300 Final value 100.14 267.56 106

Well region 1 Well region 2 Well region 3

Table 5.8 Regression results for transfer coefficient from reservoir to well-region.
Initial value 16.344 16.344 16.344 Minimum Maximum value value Final value 0.163 1634.439 7.361 0.163 1634.439 2.139 0.163 1634.439 4.834

Well WELL 1 WELL 2 WELL 3

92 Table 5.9 Regression results for transfer coefficient from well-region to well.
Minimum value 0.016 0.008 0.032 Maximum value 155.788 84.175 318.229

Well WELL 1 WELL 2 WELL 3

Initial value 1.558 0.842 3.182

Final value 1.390 0.824 3.227

93

Observed pressure
2500

Predicted pressure

2000

Pressure [psia]

1500

1000

500

0 0 100 200 300 400 500 600 700

Time [days]

Figure 5.11: Synthetic data (dipping reservoir): pressure history match.

94

Observed rate
1400

Predicted rate

1200

1000

Rate [stb/day]

800

600

400

200

0 0 100 200 300 400 500 600 700

Time [days]

Figure 5.12: Well 1 Synthetic data (dipping reservoir): rate history match.

95

Observed GOR
1400

Predicted GOR

1200

1000

GOR [scf/stb]

800

600

400

200

0 0 100 200 300 400 500 600 700

Time [days]

Figure 5.13: Well 1 Synthetic data (dipping reservoir): GOR history match.

96

Observed rate
800

Predicted rate

700

600

Rate [stb/day]

500

400

300

200

100

0 0 100 200 300 400 500 600 700

Time [days]

Figure 5.14: Well 2 Synthetic data (dipping reservoir): rate history match.

97

Observed GOR
1400

Predicted GOR

1200

1000

GOR [scf/stb]

800

600

400

200

0 0 100 200 300 400 500 600 700

Time [days]

Figure 5.15: Well 2 Synthetic data (dipping reservoir): GOR history match.

98

Observed rate
3000

Predicted rate

2500

2000

Rate [stb/day]

1500

1000

500

0 0 100 200 300 400 500 600 700

Time [days]

Figure 5.16: Well 3 Synthetic data (dipping reservoir): rate history match.

99

Observed GOR
1200

Predicted GOR

1000

800

GOR [scf/stb]

600

400

200

0 0 100 200 300 400 500 600 700

Time [days]

Figure 5.17: Well 3 Synthetic data (dipping reservoir): GOR history match.

100 5.3 Field Example 5.3.1 Brief History of the Field The Second Bench reservoir in the First Frontier Formation is located in Wyoming, in the Moore Area (Wyoming). The well map is shown in figure 5.18.

Figure 5.18: Well map.

The Taylor Unit #4 (Section 27, Township 41 North, Range 76 West) was a discovery well. The field was discovered in 1982. The second well drilled in the reservoir was Moore Unit #1 (Section 2, Township 40 N, Range 76 W). The well was drill stem

101 tested in May 1983, and was subsequently pressure surveyed in May 1995. In addition, two wells, Moore Unit #3, and USA Federal # 15-1, were also tested in May 1995. Overall, 12 wells have been producing in the formation under a natural depletion drive mechanism.

5.3.2 Data Available Limited data are available from the field. It is briefly described below:

5.3.2.1 Well Map The well map presented above (Figure 5.18) shows the well locations in the field.

5.3.2.2 Depth Information In our program, relative locations of the wells are not needed since we assume no communication between different well regions. However, the depths of individual wells are important in determining the relative gas and oil migration. Therefore, depths for each well, as recorded in the well map, are shown in Table 5.10 below. It is logical to assume that the wells which are in structurally high position (shallow wells) will produce at higher GOR and the wells which are in structurally low position (deep wells) will produce at lower GOR.

102 Table 5.10 Well Depths Information.

Well UNIT W-03143047 TAYLOR UNIT W034304 1 PEREGRINE W32266A 1 MOORE UNIT FEE 1 MOORE UNIT 82-89 2 TAYLOR UNIT FEE 4 MOORE UNIT W-59617 3 USA W-42621A 27-1 WOODS 36-1 FEDERAL W-86337 15-1 FEDERAL W-47686 22-1 USA W-86341 26-1

Depth [ft] (SS) 7075 7075 7277 7246 7148 7023 7277 7197 7218 7124 7157 7214

103 5.3.2.3 PVT Data The PVT properties for the field were not available. However, the API gravity of oil is 43.4 and the gas gravity is 0.7314. The reservoir temperature is 250 [F] and the GOR above bubble point pressure is 1454.1 [scf/stb]. Using these parameters, the PVT data for the solution gas/oil ratio (Rs), the oil formation volume factor (Bo), the gas formation volume factor (Bg), as well as the oil viscosity (o), the gas viscosity (g) and the water viscosity (w) have been estimated using literature correlations. In Table 5.11 below, we provide the values of these parameters as a function of pressure.

104 Table 5.11 PVT data.


oil gas water Rs viscosity viscosity viscosity [scf/stb] [cp] [cp] [cp] 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1454.10 1426.12 1387.73 1349.95 1312.80 1276.29 1240.44 1203.55 1166.90 1131.86 0.21317 0.21194 0.21072 0.20951 0.20831 0.20711 0.20593 0.20475 0.20358 0.20241 0.20126 0.20012 0.19898 0.19786 0.19674 0.19563 0.19454 0.19345 0.19238 0.19131 0.19026 0.18922 0.18819 0.18717 0.18616 0.18517 0.18419 0.18396 0.18538 0.18738 0.18943 0.19151 0.19363 0.19578 0.19808 0.20046 0.20282 0.03027 0.03009 0.02992 0.02974 0.02956 0.02938 0.02920 0.02902 0.02884 0.02866 0.02847 0.02829 0.02810 0.02792 0.02773 0.02754 0.02736 0.02717 0.02698 0.02679 0.02659 0.02640 0.02621 0.02601 0.02582 0.02563 0.02542 0.02538 0.02169 0.02157 0.02144 0.02132 0.02120 0.02107 0.02095 0.02083 0.02071 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323

Pressure Bo Bg [psia] [bbl/stb] [cf/scf] 6231 6168 6104 6041 5977 5914 5850 5787 5724 5660 5597 5533 5470 5406 5343 5280 5216 5153 5089 5026 4962 4899 4836 4772 4709 4645 4582 4567 4518 4455 4391 4328 4265 4201 4138 4074 4011 1.8385 1.8399 1.8413 1.8427 1.8442 1.8457 1.8472 1.8487 1.8503 1.8519 1.8536 1.8553 1.8571 1.8588 1.8607 1.8626 1.8645 1.8665 1.8685 1.8706 1.8727 1.8749 1.8771 1.8795 1.8818 1.8843 1.8868 1.8874 1.8706 1.8475 1.8249 1.8028 1.7811 1.7599 1.7382 1.7167 1.6962 0.00347 0.00349 0.00351 0.00353 0.00355 0.00357 0.00359 0.00361 0.00363 0.00366 0.00368 0.00370 0.00373 0.00375 0.00378 0.00381 0.00384 0.00386 0.00389 0.00392 0.00396 0.00399 0.00402 0.00406 0.00409 0.00413 0.00417 0.00418 0.00421 0.00425 0.00429 0.00434 0.00439 0.00444 0.00449 0.00454 0.00459

105 Table 5.11 (contd) PVT data.


oil gas water Rs viscosity viscosity viscosity [scf/stb] [cp] [cp] [cp] 1098.32 0.20516 0.02058 0.23323 1066.20 0.20749 0.02046 0.23323 1035.39 0.20981 0.02034 0.23323 1005.83 0.21212 0.02022 0.23323 977.45 0.21441 0.02010 0.23323 949.02 0.21679 0.01997 0.23323 921.96 0.21914 0.01985 0.23323 896.27 0.22144 0.01973 0.23323 871.85 0.22371 0.01961 0.23323 848.64 0.22594 0.01949 0.23323 826.54 0.22812 0.01937 0.23323 805.49 0.23028 0.01925 0.23323 785.59 0.23237 0.01914 0.23323 767.59 0.23432 0.01902 0.23323 750.11 0.23627 0.01890 0.23323 733.13 0.23821 0.01878 0.23323 716.63 0.24015 0.01867 0.23323 700.58 0.24209 0.01855 0.23323 684.97 0.24403 0.01844 0.23323 669.78 0.24596 0.01832 0.23323 655.00 0.24789 0.01821 0.23323 641.13 0.24975 0.01810 0.23323 632.22 0.25096 0.01799 0.23323 622.20 0.25236 0.01788 0.23323 611.14 0.25393 0.01777 0.23323 599.09 0.25567 0.01766 0.23323 586.14 0.25759 0.01756 0.23323 572.35 0.25969 0.01745 0.23323 557.79 0.26197 0.01735 0.23323 542.53 0.26443 0.01725 0.23323 526.66 0.26708 0.01715 0.23323 510.81 0.26981 0.01705 0.23323 494.26 0.27277 0.01695 0.23323 477.03 0.27596 0.01685 0.23323 459.21 0.27940 0.01675 0.23323 440.90 0.28309 0.01664 0.23323 421.25 0.28724 0.01654 0.23323 400.85 0.29177 0.01643 0.23323 381.48 0.29630 0.01632 0.23323 363.06 0.30084 0.01621 0.23323

Pressure Bo Bg [psia] [bbl/stb] [cf/scf] 3947 1.6767 0.00465 3884 1.6581 0.00471 3821 1.6404 0.00477 3757 1.6234 0.00484 3694 1.6071 0.00491 3630 1.5909 0.00498 3567 1.5755 0.00505 3503 1.5610 0.00513 3440 1.5472 0.00521 3377 1.5342 0.00530 3313 1.5218 0.00539 3250 1.5101 0.00548 3186 1.4990 0.00558 3123 1.4890 0.00568 3059 1.4793 0.00579 2996 1.4699 0.00591 2932 1.4609 0.00603 2869 1.4521 0.00615 2806 1.4435 0.00629 2742 1.4352 0.00643 2679 1.4272 0.00658 2615 1.4196 0.00674 2552 1.4148 0.00691 2488 1.4094 0.00708 2425 1.4034 0.00727 2362 1.3969 0.00747 2298 1.3900 0.00769 2235 1.3826 0.00791 2171 1.3748 0.00815 2108 1.3667 0.00841 2044 1.3582 0.00869 1981 1.3498 0.00898 1918 1.3411 0.00930 1854 1.3320 0.00964 1791 1.3227 0.01000 1727 1.3131 0.01040 1664 1.3029 0.01082 1600 1.2924 0.01129 1537 1.2824 0.01179 1474 1.2730 0.01234

106 Table 5.11 (contd) PVT data.


oil gas water Rs viscosity viscosity viscosity [scf/stb] [cp] [cp] [cp] 345.86 329.98 314.21 298.56 283.06 269.19 254.40 238.79 222.37 205.10 187.91 170.90 154.12 137.25 120.59 103.74 86.32 68.93 52.51 36.97 23.50 12.70 2.32 0.30530 0.30963 0.31414 0.31884 0.32375 0.32837 0.33356 0.33935 0.34584 0.35315 0.36100 0.36940 0.37841 0.38830 0.39904 0.41106 0.42499 0.44077 0.45784 0.47649 0.49518 0.51235 0.53118 0.01610 0.01599 0.01588 0.01577 0.01566 0.01555 0.01545 0.01534 0.01523 0.01513 0.01503 0.01492 0.01483 0.01473 0.01464 0.01455 0.01446 0.01437 0.01429 0.01422 0.01416 0.01411 0.01406 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323 0.23323

Pressure Bo Bg [psia] [bbl/stb] [cf/scf] 1410 1347 1283 1220 1156 1093 1029 966 903 839 776 712 649 585 522 459 395 332 268 205 141 78 15 1.2642 1.2561 1.2481 1.2403 1.2325 1.2256 1.2182 1.2105 1.2024 1.1939 1.1856 1.1773 1.1692 1.1612 1.1532 1.1453 1.1371 1.1290 1.1214 1.1143 1.1082 1.1033 1.0986 0.01293 0.01359 0.01432 0.01512 0.01601 0.01701 0.01813 0.01940 0.02086 0.02253 0.02448 0.02678 0.02953 0.03288 0.03705 0.04236 0.04939 0.05910 0.07341 0.09658 0.14055 0.25609 1.38269

107 The PVT values have been estimated using the correlations in Table 5.12.

Table 5.12 PVT correlations.

PVT parameter Solution gas/oil ratio Oil formation volume factor Z factor Dead oil viscosity Live oil viscosity below bubble point pressure Live oil viscosity above bubble point pressure Gas viscosity Water viscosity

Correlation Lasater
38,43 45 45

Standing Standing Glaso

40

Beggs and Robinson Vasquez and Beggs Lee et al. Van Wingen
44

42

39,46

5.3.2.4 Pressure Data The observed data are shown in the Figure 5.19. Some of pressure values are estimated using available production data to increase the number of points.

108

Observed reservoir pressure [psia]


6500 6000

Reservoir pressure [psia]

5500 5000 4500 4000 3500 3000 2500 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.19: Pressure profile as a function of time.

109 5.3.2.5 Production Data In addition to the above data, monthly production data are available from each of the wells in public domain. Limited flowing tubing pressure data are available, however, sufficient information was not available to calculate the bottom hole pressure data. In the presence of limited well flowing pressure information, lot of bottom hole pressure data were assumed.

5.3.3 History Match Process As discussed in the previous chapter, the history matching process was conducted in three steps. In the first step, oil in place is calculated based on material balance. In the second step, the relative permeability values for gas and oil are estimated by matching field-wide GOR with the simulated values. In the third step, we matched the individual production rates using the optimization process.

5.3.3.1 Oil in Place Calculation Using the observed pressure data and the observed cumulative oil and gas production data, the original oil in place has been determined using the material balance equation. The result gives a value of original oil in place (OOIP) of 17,766,390 [STBO].

5.3.3.2 Calculation of Relative Permeability Curves Using the observed pressure data and the observed cumulative production and GOR data the relative permeability ratio curves have been estimated and shown in Figure

110 5.20. The values of the coefficients , kro , krg , and sm (sm =1-sgc ) obtained are shown in
o o

the following Table 5.13.

Table 5.13 Relative permeability regression parameters.


Relative permeability curves coefficients Kro0 Krg0 Initial value 0.34 1.0 4.0 0.95 Final value 0.348 0.938 1.694 1.0

sm =1-sgc

To improve the match between the ratio of gas relative permeability and oil relative permeability values derived from observed data and calculated values, the gas relative permeability has been fitted by the following equation:
2 krg = 1.7843s g 0.0306 s g + 0.0124

The curves obtained are shown in Figure 5.20 below:

111

Krg/Kro ratio from observed data Predicted krg/kro ratio


0.4 0.35 0.3 0.25

Krg/kro [-]

0.2 0.15 0.1 0.05 0 0.5 0.55 0.6 0.65 0.7 0.75 0.8

Oil saturation [-]

Figure 5.20: Match between estimated relative permeability ratio and observed relative permeability ratio.

112

5.3.3.3 History Match The computer program was run to compute the history match of the production data. The regression parameters used are: the radius of well-regions, the transfer coefficients from well-regions the the wells and the transfer coefficients from the reservoir-region the the well-regions. The results obtained for these parameters are shown in the following tables 5.14, table 5.15 and table 5.16.

Table 5.14 Regression results for radius of well-regions.


Initial value 374 119 185 147 217 122 135 246 159 276 300 377 Minimum value 100 100 100 100 100 100 100 100 100 100 100 100 Maximum value Final value 400 400 400 400 400 400 400 400 400 400 400 400 179.786 385.58 235.665 207.722 290.661 235.448 341.028 316.797 180.349 228.387 263.713 212.212

Well UNIT W-0314304 7 TAYLOR UNIT W034304 1 PEREGRINE W32266A 1 MOORE UNIT FEE 1 MOORE UNIT 82-89 2 TAYLOR UNIT FEE 4 MOORE UNIT W-59617 3 USA W-42621A 27-1 WOODS 36-1 FEDERAL W-86337 15-1 FEDERAL W-47686 22-1 USA W-86341 26-1

113

Table 5.15 Regression results for transfer coefficient from well-region to well.
Minimum value 0.035 0.043 0.039 0.041 0.038 0.043 0.042 0.037 0.041 0.037 0.036 0.035 Maximum value 348.769 429.413 394.305 411.923 383.512 427.543 418.266 374.573 407.074 368.252 361.491 349.148

Well UNIT W-0314304 7 TAYLOR UNIT W034304 1 PEREGRINE W32266A 1 MOORE UNIT FEE 1 MOORE UNIT 82-89 2 TAYLOR UNIT FEE 4 MOORE UNIT W-59617 3 USA W-42621A 27-1 WOODS 36-1 FEDERAL W-86337 15-1 FEDERAL W-47686 22-1 USA W-86341 26-1

Initial value 3.488 4.294 3.943 4.119 3.835 4.275 4.183 3.746 4.071 3.683 3.615 3.491

Final value 26.864 55.472 17.718 50.642 42.882 77.278 1.877 14.368 43.368 161.041 7.069 10.599

Table 5.16 Regression results for Transfer coefficient from reservoir-region to well-region.
Initial value 9.250 6.174 7.082 6.579 7.490 6.223 6.402 7.826 6.765 8.196 8.432 9.304 Minimum value 0.093 0.062 0.071 0.066 0.075 0.062 0.064 0.078 0.068 0.082 0.084 0.093 Maximum value 924.965 617.441 708.195 657.928 748.975 622.340 640.179 782.562 676.483 819.644 843.186 930.415

Well UNIT W-0314304 7 TAYLOR UNIT W034304 1 PEREGRINE W32266A 1 MOORE UNIT FEE 1 MOORE UNIT 82-89 2 TAYLOR UNIT FEE 4 MOORE UNIT W-59617 3 USA W-42621A 27-1 WOODS 36-1 FEDERAL W-86337 15-1 FEDERAL W-47686 22-1 USA W-86341 26-1

Final value 97.227 24.898 78.978 25.509 47.269 57.693 50.404 46.847 0.287 17.407 14.752 77.262

114 5.3.3.3.1 Field Production History Match The match between the observed oil cumulative production and the predicted oil cumulative production is shown in Figure 5.21. As can be seen, the match is very good. The match between the observed cumulative gas/oil ratio and the predicted cumulative gas/oil ratio is shown in Figure 5.22. The match between the observed gas cumulative production and the predicted gas cumulative production is shown in Figure 5.23. The discrepancy between the two curves is mainly due to two reasons. First, the program cannot reproduce the GOR for wells that are structurally high but present a low observed GOR values. Second, the program tends to exhibit GOR values that are not very high for the deep wells. The match between the observed reservoir pressure and the predicted reservoir pressure is shown in Figure 5.24. The discrepancy between the two curves is mainly due to the fact that the program predict slightly less gas production as discussed in the section above.

115

Observed cumulative oil production data used in history match Predicted cumulative oil production data obtained from history match
2000000 1800000 1600000

Cumulative produced oil [stb]

1400000 1200000 1000000 800000 600000 400000 200000 0 0 500 1000 1500 2000

Time [days]

Figure 5.21: Cumulative oil production.

116

Observed cumulative production gas/oil ratio Predicted cumulative production gas/oil ratio
3500

3000 2500

RP [scf/stb]

2000

1500

1000 500

0 0 500 1000 1500 2000

Time [days]

Figure 5.22: Cumulative production gas/oil ratio.

117

Observed cumulative gas production data used in history match Predicted cumulative gasproduction obtained from history match
7000000000 6000000000

Cumulative produced gas [scf]

5000000000 4000000000 3000000000 2000000000 1000000000 0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.23: Cumulative gas production.

118

Observed reservoir pressure Predicted reservoir pressure


7000

6000

5000

Pressure [psia]

4000

3000

2000

1000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.24: Match between predicted reservoir pressure and observed reservoir pressure.

119

5.3.3.3.2 Well Production History Match The results of the history match have been generated for each of the twelve wells producing from the field. In this section the results of four wells (out of a total number of 12 wells) are presented in figure 5.25, figure 5.26, figure 5.27, figure 5.28, figure 5.29, figure 5.30, figure 5.31 and figure 5.32. Those wells are: Unit W-0314304 7, Moore Unit 82-89 2, Moore Unit W-59617 3, and USA W-86341 26-1. The history match results for the other wells are shown in appendix D.

5.3.3.3.2.1 Well Unit W-0314304 7 The match of the oil rate for well Unit W-0314304 7 is quite good. However we did not get a very good match for the GOR. This is due to the fact that the program cannot reproduce the GOR for wells that are structurally shallow but present a low observed GOR values. This is the case of well Unit W-0314304 7 that is structurally high (shallow well). The results of the history match are shown in figure 5.25 and figure 5.26.

120

Observed rate
140

Predicted rate

Well: UNIT W-03143047


120

100

Rate [stb/d]

80

60

40

20

0 0 500 1000 1500

Time [days]

Figure 5.25: Match between predicted oil rate and observed oil rate for well unit W0314304 7.

121

Observed GOR
80000

Predicted GOR

WELL: UNIT W-03143047


70000

60000

GOR [scf/stb]

50000

40000

30000

20000

10000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.26: Match between predicted GOR and observed GOR for well unit W-0314304 7.

122

5.3.3.3.2.2 MOORE UNIT 82-892 The match of the oil rate for well MOORE UNIT 82-892 is quite good. However we did not get a very good match for the GOR. This is due to the fact that the program tends to exhibit GOR values that are not very high for the deep wells. This is the case of well MOORE UNIT 82-892 which is a deep well. The results of the history match for this well are shown in figure 5.27 and figure 5.28.

123

Observed rate
1000

Predicted rate

WELL: MOORE UNIT 82-89 2


900 800 700

Rate [stb/day]

600 500 400 300 200 100 0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.27: Match between predicted oil rate and observed oil rate for well MOORE UNIT 82-89 2.

124

Observed GOR
9000

Predicted GOR

WELL: MOORE UNIT 82-89 2


8000 7000 6000

GOR [scf/stb]

5000 4000 3000 2000 1000 0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.28: Match between predicted GOR and observed GOR for well MOORE UNIT 82-892.

125

5.3.3.3.2.3 Well MOORE UNIT W-59617 3 The match of the oil rate for well MOORE UNIT W-59617 3 is quite good. The predicted GOR is lower than the observed GOR. This is due to the fact that this well is one of the deepest wells for this reservoir. The results of the history match are shown in figure 5.29 and figure 5.30.

126

Observed rate
1200

Predicted rate

WELLMOORE UNIT W-59617 3


1000

800

Rate [stb/day]

600

400

200

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.29: Match between predicted oil rate and observed oil rate for well MOORE UNIT W-59617 3.

127

Observed rate
14000

Predicted rate

WELL: MOORE UNIT W-59617 3


12000

10000

GOR [scf/stb]

8000

6000

4000

2000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.30: Match between predicted GOR and observed GOR for well MOORE UNIT W-59617 3.

128

5.3.3.3.2.4 Well USA W-86-341 26-1 The match of the oil rate for well USA W-86-341 26-1 is quite good. The match of the GOR is also reasonably acceptable. This well is structurally producing from the middle of the reservoir. The results of the history match are shown in figure 5.31 and figure 5.32.

129

Observed rate
1000 900

Observed rate

WELL: USA W-86341 26-1


800 700

Rate [stb/day]

600 500 400 300 200 100 0

200

400

600

800

1000

1200

1400

1600

Time [days]

Figure 5.31: Match between predicted oil rate and observed oil rate for well USA W-86341 26-1.

130

Observed GOR
3500

Predicted GOR

3000

WELL:USA W-86341 26-1

2500

GOR [scf/stb]

2000

1500

1000

500

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure 5.32: Match between predicted GOR and observed GOR for well USA W-86-341 26-1.

CHAPTER VI CONCLUSIONS AND RECOMMENDATIONS

6.1 Conclusions In this dissertation two procedures to match oil reservoir performance have been discussed. The first one addresses the single well production case while the second one deals with multi-well production scheme. The first procedure presents an algorithm to match the production from single well using a concept of dynamic nodal analysis. By combining the desirable features of nodal analysis, material balance technique and the decline curve analysis, the method is able to match the historical performance of the well data, and predict the future performance of the oil well under the existing as well as altered conditions. The method is validated by comparing the results of the model with the results generated from ECLIPSE for homogeneous reservoir as well heterogeneous reservoirs. The second procedure develops a model to match multi-well production from an oil reservoir that produces under natural or water drive mechanism. It extends the classical tank model to a multi-tank model scheme by dividing the original reservoir into small radial regions around the wells (well regions) and a larger region (reservoir region) that feeds the small regions. The three-phase (oil, gas and water) flow equation that describes this model has been derived and discretized. By determining the transfer coefficients from reservoir

region to well region as well as the size of each well-region, the method is able to match the 131

132 historical performance of the wells, as well as predict the future performance of the wells under the existing and altered conditions. The method has been validated by comparing the results obtained from the model with the results generated from ECLIPSE as well as an exhaustive field study. These procedures can provide valuable means to help the engineer in decision making. Opening an oil well to production always involves considerable expenses whereas a simple model can be run many times at a lower cost to try many different possible scenarios in order to make technical and economical decisions. It should be noted that the prediction of the future performance based on history match of well performance is not unique. There are many sets of system parameters that can match the past performance of the well. There is always some uncertainty associated with the model used to arrive at a satisfactory historical performance match. Based upon the history match results, the engineer can obtain a range of future performances, and hence can make a decision in light of uncertainties.

6.2 Recommendations The presented work can be improved by implementing the following features in the computer program.

133 Horizontal oil well inflow performance can be added to the computer program for the single-well scheme, in order to expand its use to oil wells that have this geometry. The multi-well scheme can be extended to use exclusively in forward mode by including shape factors to compute directly the transfer coefficients. In this way, this scheme can be used for new reservoirs that have not been opened to production yet. The multi-well scheme can be extended to gas reservoirs. The multi-well scheme can be extended to multiplayer reservoirs.

REFERENCES

1. Brown, K.E.: The technology of artificial lift methods. Pennwell Publishing Company, Tulsa (1984), volume 4 2. Bitsindou, A.: Gas Well Production Optimization Using Nodal Analysis, MS Thesis, University of Tulsa, Tulsa (1998) 3. Raghavan R., Well Test Analysis: Wells Producing by Solution Gas Drive , paper SPE 5588 presented at the 1975 SPE Annual Technical Conference and exhibition, Dallas, Sept. 28-Oct. 1 4. Chopra Anil K., and Carter Robert D., Proof of two phase Steady-State Theory for Flow Through Porous Media, paper SPE 14472. SPE Formation Evaluation, December 1986. 5. Perez, G. and Kelkar, B.G.: A Simplified Method to Predict Over-all Production Performance, Journal of Canadian Petroleum Technology, January-February, 1990, Volume 29, No. 1 6. Jacquard P. Theorie de linterpretation des measures de pression. Revue de lInstitut Francais du Petrole, 19(3): 297-334, 1964.

134

135 7. Pudjo Sukarno: Inflow Performance Relationship Curves In Two-Phase and ThreePhase Flow Conditions, PHD Dissertation, University of Tulsa, Tulsa (1986) 8. Jones, L.G., Blount, E.M. and Glaze, C.E.: Use of Short Term Multiple Rate Flow Test to Predict Performance of Wells Having Turbulence, paper SPE 6133 presented at the 1976 Annual Technical Conference and Exhibition, New Orleans, Oct. 3-6. 9. McLeod, Harry O. Jr. The effect of Perforating Conditions on Well Performance, J. Pet. Tech. (January 1983) 10. Fetkovich, M.J., Fetkovich, E.J. and Fetkovich, M.D.: Useful Concepts for DeclineCurve Forecasting, Reserve Estimation, and Analysis, paper SPE 28628 presented at the 1994 SPE Annual Technical Conference and exhibition, New Orleans, Sept. 25-28 11. Fetkovich, M.J.: Decline Curve Analysis Using Type Curves, JPT, June 1980, 10651077 12. Sills, S. R.: Improved Material Balance Regression Analysis for Water Drive Oil and Gas Reservoirs, paper SPE 28630 13. Arthur, J.E., China, H.S. and Temeng. K.O.: Material Balance Modelling and Performance Prediction of a Composite Gas Reservoir, paper SPE 26194 presented at the 1993 SPE Gas Technology Symposium, Calgary, Alberta , Canada, June 28-30. 14. Dake, L. P. The Practice of Reservoir Engineering. Elsevier, Amsterdam, The Netherlands, 1994. 15. Havlena, D. Odeh, A.S. The material Balance as an Equation of a straight line, Transactions of the AIME (1963)

136 16. Aziz K. and Settari A. Petroleum reservoir Simulation. Elsevier Applied Science Publishers, New York, 1979. 17. Thomas, G.W.: Principles of Hydrocarbon Reservoir Simulation, International Human Resources Development Corporation, Boston. 18. Chierici, G.L.: Principi di Ingegneria dei Giacimenti Petroliferi, AGIP (1990), Volume 2. 19. Ole K. Jensen, An Automatic Timestep Selection Scheme for Reservoir Simulation, paper SPE 9373 presented at the 1980 SPE Annual Technical Conference and exhibition, Dallas, Sept. 21-24 20. Sammon P.H., and Rubin B., Practical Control of Timestep Selection in Thermal Simulation, paper SPE 12268. SPE. Reservoir Engineering, March 1986. 21. D. W. Peaceman. Interpretation of well-block pressures in numerical reservoir simulation with non-square grid blocks and anisotropic permeability. Soc. Pet. Eng. J., pages 531-543, 1983. 22. Standing M.B. Notes On Relative Permeability Relationships. (1975) 23. Oliver D. Inverse Problem Theory (class notes), University of Tulsa,1999. 24. Tarantola A. Inverse Problem Theory: Methods for Data Fitting and Model Parameter Estimation. Elsevier, Amsterdam, The Netherlands, 1987. 25. Garbow, B.S., Hillstrom, K.E. and More J.J.: LMDIF1, Argonne National Laboratory. Minpack Project, March 1980.

137 26. Carvalho, R. Thompson, L.G., Redner, R. and Reynolds, A.C.: Simple Procedure for Imposing Constraints for Nonlinear Least Square Optimization, paper SPE 29582 27. Carvalho, R.: Nonlinear Regression: Application to Well Test Analysis, PHD Dissertation, University of Tulsa, Tulsa (1993) 28. Schlumberger Oilfield Services: Eclipse E100, Reference Manual, 2001 29. Beggs, H.D. and Brill, J.P.: A Study of Two-Phase Flow in Inclined Pipes, J. Pet. Tech. (May 1973), 607-617. 30. Hagedorn, A.R. and Brown, K.E.: Experimental Study of Pressure Gradients Occurring During Continuous Two-Phase Flow in Small-Diameter Vertical Conduits, J. Pet. Tech. (April 1965), 475-484. 31. Gray, H.E. Vertical Flow Correlation in Gas Wells, In User Manual for API 14B, Subsurface Controlled Safety Valve Sizing Computer Program, App. B. (June 1974) 32. Dukler, A.E.: Gas-Liquid Flow in Pipelines, I. Research Results, AGA-API Project NX-28 (May 1969). 33. Brill, J.P. and Beggs, H.D.: Two-Phase Flow in Pipes, Tulsa (Feb. 1984). 34. Users Manual for API 14B Subsurface Controlled Subsurface Safety Valves Sizing Computer Program, API Manual 14BM (June 1974) Sec. B4. 35. Brill, J.P. Beggs, H.D. and Sylvester, N.D.: Orifice Coefficients for Two-Phase Flow Through Velocity Controlled Subsurface Safety Valves, Final Report on API OSAPR Project No. 1 (April 1976).

138 36. Sachdeva, R., Schmidt, Z., Brill, J.P. and Blais, R.N.: Two-Phase Flow Through Chokes, paper SPE 15657 presented at the 1986 Annual Technical Conference and Exhibition, New Orleans, Oct, 5-8. 37. Press, W., Flannery, B., Teukolsky, S., Vetterling, W.: Numerical Recipes. Cambridge University Press. 1986. 38. Lasater, J.A.: Bubble Point Pressure Correlation, Trans., AIME (1958) 213, 379-381 39. Vasquez, M. and Beggs, H.D.: Correlations for Fluid Physical Property Prediction, JPT (June 1980) 968-970. 40. Glas, .: Generalized Pressure-Volume-Temperature Correlations, JPT (May, 1980) 785-795 41. Baker, O. and Swerdloff, W.: Finding Surface Tension of Hydrocarbon Liquids, Oil and Gas J. (Jan. 2 1956) 125. 42. Beggs H.D. and Robinson, J.R.: Estimating the viscosity of Crude Oil Systems, JPT (Sept. 1975) 1140-1141. 43. Yarborough, L. and Hall, K.R.: How to Solve Equation of State for Z-Factors, Oil and Gas J. (Feb. 18 1974) 86-88. 44. Lee, A.L., Gonzalez, M.H. and Eakin, B.E.:The Viscosity of Natural Gases, JPT (Aug. 1966) 997-1000. 45. Standing, M.B.: Volumetric and Phase Behavior of Oil Field Hydrocarbon Systems, 8th Printing, Society of Petroleum Engineers, 1977, P.121

139 46. Vasquez, A.: Correlations for Fluid Physical Property Prediction, M.S. Thesis, The University of Tulsa. 47. Frick, T.C. and Taylor, R.W.: Petroleum Production Handbook,McGraham-Hill Book Company, Inc. (1962), Volume 2 48. Fanchi, J. R., Harpole, K., Keplinger and Associates Inc. Tulsa Oklahoma, Bujnowski, S., The BDM Corporation Bartlesville Oklahoma: BOAST: a Three-dimensional, Three-phase Black Oil Applied Simulation Tool (Version 1.1), United States Department of Energy. (1982), Volume 1 49. Klins, M.A., Bouchard, A.J., Cable, C.L.: A Polynomial Approach to the Van Everdingen-Hurst Dimensionless Variables for Water Encroachment, paper SPE 15433

APPENDIX A

EXTENDED MUSKAT METHOD

This method is discussed in Dakes book14 as it applies to gas/oil systems. We have extended this method under three phase flow conditions. A.1 Oil material balance
Nr = N N p = Vp So Bo

(A-1)

where N = OOIP [stb] Nr = total oil remaining in reservoir [stb] Np = total oil produced [stb] So=oil saturation [-] Vp=pore volume [stb] Bo=oil formation volume factor [bbl/stb] differentiating, we get:

dN r V p dS o V p S o dBo = 2 dP Bo dP Bo dP
A.2 Gas material balance
Gr = N r Rs = Vp So Bo Rs + (1 S o S w ) Vp Bg

(A-2)

(A-3)

where Gr = total gas remaining in reservoir

140

141 Rs= soilution gas/oil ratio [scf/stb] Bg= gas formation volume factor [bbl/scf] So=oil saturation [-] Sw= water saturation [-] differentiating, we get R dS S dRs Rs S o dBo 1 dS o 1 dS w (1 S o S w ) dBg dGr = Vp s o + o 2 2 Bo dP Bo dP dP dP Bo dP Bg dP Bg dP Bg dGr GOR = dP dN r dP Rs dS o S o dRs Rs S o dBo 1 dS o 1 dS w (1 S o S w ) dB g + 2 2 Bo dP Bo dP dP Bo dP B g dP B g dP Bg = V p dS o V p S o dBo 2 Bo dP Bo dP (A-4)

(A-5)

Alternatively, GOR is defined as:

GOR =

qg qo

k rg o Bo + Rs k ro g B g

(A-6)

Equating (A-5) and (A-6), we have

k rg k ro

Rs dS o S o dRs Rs S o dBo 1 + 2 Bo dP Bo dP Bo dP B g o Bo + Rs = 1 dS o g Bg Bo dP

dS o 1 dS w (1 S o S w ) dB g 2 dP B g dP dP Bg S o dBo 2 Bo dP (A-7)

1 dS o S o k rg o Bo + Rs B dP B 2 k B o o ro g g Rs dS o S o dRs Rs S o dBo + 2 Bo dP Bo dP Bo dP

1 dS o 1 dS w (1 S o S w ) dB g 2 B g dP B g dP dP Bg

dBo = dP

(A-8)

142

k rg o Bo 1 dS o k rg o Bo S dBo + Rs + Rs o2 = k B B dP k B B dP ro g g ro g g o o Rs dS o S o dRs Rs S o dBo 1 dS o 1 dS w (1 S o S w ) dB g + 2 2 Bo dP Bo dP dP Bo dP B g dP B g dP Bg

(A-10)

k rg o Bo 1 Rs 1 1 + Rs + + B k ro g B g o Bo B g B g k rg o Bo S dBo S o dRs Rs S o + Rs o2 + 2 k B B dP B dP Bo o ro g g o dS o dP After simplification, we get


dS o dP

dS w = dP dBo (1 S o S w ) dB g 2 dP dP Bg (A-11)

k rg o Bo 1 1 1 dS w k rg o Bo S o dBo S o dRs (1 S o S w ) dB g + + = + 2 k B B dP B g B g dP k ro g B g Bo 2 dP Bo dP Bg ro g g o (A-12) k rg o 1 1 dS w k rg o S o dBo S o dRs (1 S o S w ) dB g + + = + 2 k B dP B g B g dP k ro g B g Bo dP Bo dP Bg ro g g (A-13)

dS o dP

dS o dP

k rg o dS k rg o S o dBo S o B g dRs (1 S o S w ) dB g + 1 + w = + k dP k ro g Bo dP Bo dP Bg dP ro g (A-14)

A.3 Water material balance


Wr = We W p = Vp Sw Bw

(B-15)

where We = Water influx [sbl] Wr = total waterl remaining in reservoir [sbl] Bw= water formation volume factor [bbl/sbl]

143 Differentiating equation B-15, we get: dWr V p dS w V p S w dBw = 2 dP Bw dP Bw dP (A-16)

W p = We Wr
differentiating equation B-17, we get

(B-17)

dWe dWr dP dP dP N p = N Nr =
differentiating equation B-19, we get
dN p dN dN r dP dP dP dN = r dP =

dW p

(A-18) (B-19)

(B-20)

By definition of WOR, we have


dW p WOR = dP dN p dP dWe dWr dP dP = dN r dP

(A-21)

dWe V p dS w V p S w dBw 2 dP Bw dP Bw dP WOR = V p dS o V p S o dBo B dP B 2 dP o o

(A-22)

144 1 dWe 1 dS w V p dP Bw dP WOR = 1 dS o S o B dP B 2 o o Where


Vp = NBoi 1 S wi

S w dBw 2 Bw dP dBo dP

(A-23)

(A-24)

Alternatively, WOR is defined as:


WOR = qw qo

k B = rw o o k ro w Bw

(A-25)

Equating (A-23) and (A-25), we have 1 dWe 1 dS w o Bo V p dP Bw dP = w Bw 1 dS o S o B dP B 2 o o


S w dBw 2 Bw dP dBo dP

k rw k ro

(A-26)

k rw o Bo 1 dS o S o dBo 1 dWe 1 dS w S w dBw = k ro w Bw Bo dP Bo 2 dP V p dP Bw dP Bw 2 dP k rw o Bo 1 dS o k rw o Bo S o dBo 1 dWe 1 dS w S w dBw + = 2 k ro w Bw Bo dP k ro w Bw Bo dP V p dP Bw dP Bw 2 dP k rw o dS o k o S o dBo S w dBw 1 dWe 1 dS w + = rw + + k ro w Bw dP Bw dP k ro w Bw Bo dP Bw 2 dP V p dP

(A-27)

(A-28)

(A-29)

Multiplying both sides by Bw we have:


k rw o dS o dS w k S dBo S w dBw Bw dWe + = rw o o + + k ro w dP dP k ro w Bo dP Bw dP V p dP

(A-30)

145 We have a system of two linear equations with two unknowns dSo/dP and dSw/dP:
k rg o dS k rg o S o dBo S o B g dRs (1 S o S w ) dB g dS + 1 o + w = + dP dP k ro g Bo dP Bo dP Bg dP k ro g k rw o dS o + dS w = k rw o S o dBo + S w dBw + Bw dWe k dP dP k ro w Bo dP Bw dP V p dP ro w

(A-31)

Subtracting the two equations, we have:

dS o dP k rg k ro

k rg o k + 1 + rw k k ro ro g o S o dBo S o B g + g Bo dP Bo

o dS o = w dP
dRs (1 S o S w ) dB g k rw o S o dBo S w dBw Bw dWe + dP Bg dP k ro w Bo dP Bw dP V p dP

(A-32)
dS o dP k rg o k rw o + + 1 k ro g k ro w

k rg o k rw o S o dBo S o B g dRs (1 S o S w ) dB g S w dBw Bw dWe = + + Bo dP Bg dP Bw dP V p dP k ro g k ro w Bo dP (A-33)


k rg o k rw o S o dBo S o B g dRs (1 S o S w ) dB g S w dBw Bw dWe + + Bo dP Bg dP Bw dP V p dP dS o k ro g k ro w Bo dP = dP k rg o k rw o + + 1 k ro g k ro w

(A-34) In terms of finite differences, we have:


S o A( P) + S o B( P) S o = P k rg k ro (1 S o S w )C ( P) + S o D( P) k rg o k rw o + + 1 k ro g k ro w k rw B dWe S w E ( P) w k ro V p dP

(A-35)

146 Where A( P) = B( P) = C ( P) =
D( P) = E ( P) =

B g dRs Bo dP

(A-36) (A-37)
(A-38) (A-39) (A-40) (A-41)

o 1 dBo g Bo dP
1 dB g B g dP

o 1 dBo w Bo dP
dBw 0.0 = negligible dP dWe dP

1 Bw B F ( P) = w Vp

From equation A-31 , we have:

dS w k rw o dS o k rw o S o dBo S w dBw Bw dWe = + + dP k ro w dP k ro w Bo dP Bw dP V p dP


Using equation A-34 in equation A-42, we get:

(A-42)

k rg o k rw o S o dBo S o B g dRs (1 S o S w ) dB g S w dBw Bw dWe + + Bo dP Bg dP Bw dP V p dP dS w k rw o k ro g k ro w Bo dP = dP k ro w k rg o k rw o + +1 k k ro w ro g k S dBo S w dBw Bw dWe rw o o + + k ro w Bo dP Bw dP V p dP


(A-43) In terms of finite differences, we have:

147
k rg k B dWe S o A( P) + S o B( P) (1 S o S w )C ( P) + S o D( P) rw S w E ( P) w k ro k ro V p dP S w = P k rg o k rw o + + 1 k ro g k ro w k rw S o D( P) + S w E ( P ) + F ( P) k ro (A-44) Where A( P) = B( P) = C ( P) =
D( P) = E ( P) =

B g dRs Bo dP

(A-45) (A-46)

o 1 dBo g Bo dP
1 dB g B g dP

(A-47) (A-48) (A-49) (A-50)

o 1 dBo w Bo dP
dBw 0.0 = negligible dP dWe dP

1 Bw B F ( P) = w Vp

APPENDIX B

MULTI-TANK SCHEME

B.1 Mathematical model Figure B.1 presents a sketch of the mathematical modeling used in the multi-well scheme. The figure shows two parts separated by a big arrow. The upper part represents the system that is made of the actual reservoir and the actual wells. The lower part shows the model used to derive the mathematical description of the system. In this model, the system is divided into three elements: the wells, the well-regions and the reservoir-region. During production, fluids (oil, gas or water) flow from the reservoir-region to the wellregions and from well-regions to the wells.

148

149

reservoir well

Well-region Reservoir-region

Well

Figure B.1: Multi-well scheme representation.

150 B.2 Assumptions

The following assumption are made: 1. The pressure is uniform in each well region and in the reservoir-region and is equal respectively to the average pressure of the well region or to the average pressure of the reservoir region. 2. The depth of each well region and of the reservoir region is evaluated at the the center of the well-region or the center of the reservoir-region respectively. 3. The transmissibilities between the reservoir-region and the well regions are evaluated using upstream weighting criteria. 4. No cross flow between the well regions is allowed.

151 B.3 Well-region side: flow from reservoir region to well region

B.3.1 Oil The oil material balance for the well-region i gives: (Qo, in) R WRi (Qo, out )WRi Wi = (OilAccum)WRi (B-1)

k r ,o ( o,R o,WRi ) (Qo, in) R WRi = J R WRi B o o R WRi = To , R WRi ( o , R o ,WRi ) Where k r ,o To , R WRi = J R WRi B o o R WRi k r ,o = J WRiWi B o o

(B-2)

(B-3)

(Qo, out )WRiWi

( o,WRi o,Wi ) WRiWi = To , WRiWi ( o ,WRi o ,Wi )

(B-4)

Where k r ,o To , WRiWi = J WRiWi B o o WRiWi By definition (we use g=gc) Z o = Po o 144 The accumulation term is given by:
S o t B o WRi t

(B-5)

(B-6)

(OilAccum)WRi = Vb ,WRi

(B-7)

Vb,Wri = bulk volume of well-region i .


2 Vb ,WRi = RWRi H

(B-8)

Equation (B-1) can then be written as


To , R WRi ( o , R o ,WRi ) To , WRi Wi ( o ,WRi o ,Wi ) = Vb ,WRi S o t B o t WRi

(B-9)

152
Z Z Z Z To , R WRi Po , R o Po ,WRi o To ,WRi Wi Po ,WRi o Po ,Wi o 144 R 144 WRi 144 WRi 144 Wi S t o B o WRi t

= Vb ,WRi

(B-10)
Z Z To , R WRi (Po , R Po ,WRi ) To , R WRi o o 144 R 144 WRi Z Z To ,WRiWi (Po ,WRi Po ,Wi ) + To ,WRi Wi o o 144 WRi 144 Wi S o t B o t WRi

(B-11)

= Vb ,WRi

Z Z Z Z GOWT1 = To , R WRi o o + To ,WRi Wi o o 144 R 144 WRi 144 WRi 144 Wi (B-12) S o t B o WRi t

To , R WRi (Po , R Po ,WRi ) To ,WRi Wi (Po ,WRi Po ,Wi ) + GOWT 1 = Vb ,WRi

(B-13)

To , R WRi Po , R (To , R WRi + To ,WRi Wi )Po ,WRi + To ,WRi Wi Po ,Wi + GOWT1 S o t B o WRi = Vb ,WRi t Let us set GOWT = To ,WRiWi Po ,Wi + GOWT1 (B-14)

(B-15) (B-16)
S o t B o WRi t

ZO = (To , R WRi + To ,WRiWi )

XO = To , R WRi

XO * Po , R + ZO * Po ,WRi + GOWT = Vb ,WRi

(B-17)

153
B.3.2 Gas

(Qg , in) R WRi (Qg , out )WRiWi = (GasAccum)WRi


kr,g (Qg , in) R WRi = J R WRi B g g ( g ,R g ,WRi ) R WRi ( o,R o,WRi ) + R WRi

Rso k r ,o + J R WRi B o o

Rsw k r , w ( w,R w,WRi ) J R WRi B w w R WRi = Tg , R WRi ( g , R g ,WRi ) + To , R WRi (Rso )R WRi ( o , R o ,WRi ) + Tw, R WRi (Rsw )R WRi ( w, R w,WRi )
R WRi

(B-18)

Where
k rg Tg , R WRi = J R WRi B g g

(B-19)

k To , R WRi = J R WRi ro B o o R WRi k Tw, R WRi = J R WRi rw B w w R WRi

(B-20)

(B-21)

kr,g (Qg , out )WRi Wi = J WRi Wi B g g

( g ,WRi g ,Wi ) WRi Wi ( o,WRi o,Wi ) WRi Wi

Rso k r ,o + J WRi Wi B o o

Rsw k r , w ( w,WRi w,Wi ) + J WRi Wi B w w WRi Wi = Tg , WRi Wi ( g ,WRi g ,Wi ) + To , WRi Wi (Rso )WRi Wi ( o ,WRi o ,Wi ) + Tw, WRiWi (Rsw )WRi Wi ( w,WRi w,Wi )

(B-22)

154
k rg Tg ,WRi Wi = J WRi Wi B g g WRi Wi

(B-23)

k To ,WRiWi = J WRi Wi ro B o o WRi Wi k Tw,WRi Wi = J WRi Wi rw B w w WRi Wi

(B-24)

(B-25)

(GasAccum)WRi = Vb ,WRi

S g Rso S o Rsw S w t + + B Bo Bw g WRi t

(B-26)

The mass balance equation for gas becomes:


Tw, R WRi (Rsw )R WRi ( w, R w,WRi ) Tg , R WRi ( g , R g ,WRi ) + To , R WRi (Rso )R WRi ( o , R o ,WRi ) + Tg , WRiWi ( g ,WRi g ,Wi ) To , WRi Wi (Rso )WRi Wi ( o ,WRi o ,Wi ) S g Rso S o Rsw S w t + + B Bo Bw g WRi t

Tw, WRi Wi (Rsw )WRi Wi ( w,WRi w,Wi )

(B-27)

= Vb ,WRi

The potential is (we use g=gc) Z g = Pg g 144

(B-28)

o = Po o

Z 144 Z 144

(B-29)

w = Pw w

(B-30)

155 Z Z Tg , R WRi Pg , R g Pg ,WRi g 144 R 144 WRi Z Z + To , R WRi (Rso )R WRi Po , R o Po ,WRi o 144 R 144 WRi Z Z + Tw, R WRi (Rsw )R WRi Pw, R w Pw,WRi w 144 WRi 144 R Z Z Tg ,WRiWi Pg ,WRi g Pg ,Wi g 144 Wi 144 WRi Z Z To ,WRiWi (Rso )WRiWi Po ,WRi o Po ,Wi o 144 Wi 144 WRi Z Z Tw,WRiWi (Rsw )WRiWi Pw,WRi w Pw,Wi w 144 WRi 144 Wi S g Rso S o Rsw S w t + + B Bo Bw WRi g = Vb ,WRi t Z Z Tg , R WRi (Pg , R Pg ,WRi ) Tg , R WRi g g 144 R 144 WRi Z Z + To , R WRi (Rso )R WRi (Po , R Po ,WRi ) To , R WRi (Rso )R WRi o o 144 R 144 WRi Z Z + Tw, R WRi (Rsw )R WRi (Pw, R Pw,WRi ) Tw, R WRi (Rsw )R WRi w w 144 R 144 WRi Z Z Tg ,WRiWi (Pg ,WRi Pg ,Wi ) + Tg ,WRiWi g g 144 WRi 144 Wi Z Z To ,WRiWi (Rso )WRiWi (Po ,WRi Po ,Wi ) + To ,WRiWi (Rso )WRiWi o o 144 WRi 144 Wi Z Z Tw,WRiWi (Rsw )WRiWi (Pw,WRi Pw,Wi ) + Tw,WRiWi (Rsw )WRiWi w w 144 WRi 144 Wi S g Rso S o Rsw S w t + + B Bo Bw g WRi t (B-32)

(B-31)

= Vb ,WRi

156 Let us GGWT1 be


Z Z Z Z GGWT1 = Tg , R WRi g o g To , R WRi (Rso )R WRi o 144 R 144 WRi 144 R 144 WRi Z Z Tw, R WRi (Rsw )R WRi w w 144 R 144 WRi Z Z Z Z + Tg ,WRi Wi g g + To ,WRi Wi (Rso )WRi Wi o o 144 WRi 144 Wi 144 WRi 144 Wi Z Z + Tw,WRi Wi (Rsw )WRi Wi w w 144 WRi 144 Wi (B-33) So we have Tg , R WRi (Pg , R Pg ,WRi ) + To , R WRi (Rso )R WRi (Po , R Po ,WRi ) + Tw, R WRi (Rsw )R WRi (Pw, R Pw,WRi ) + GGWT1 S g Rso S o Rsw S w t + + B Bo Bw g WRi t (B-34) using capillary pressure, we have Pcow = Po Pw Pcgo = Pg Po so Pw = Po Pcow Po = Pg Pcgo
Pg , R Pg ,WRi = (Po , R + Pcgo , R ) (Po ,WRi + Pcgo ,WRi ) = (Po , R Po ,WRi ) + (Pcgo , R Pcgo ,WRi ) Pw, R Pw,WRi = (Po , R Pcow, R ) (Po ,WRi Pcow,WRi )

Tg ,WRi Wi (Pg ,WRi Pg ,Wi ) To ,WRiWi (Rso )WRi Wi (Po ,WRi Po ,Wi ) Tw,WRi Wi (Rsw )WRiWi (Pw,WRi Pw,Wi )

= Vb ,WRi

(B-35) (B-36) (B-37) (B-38)

(B-39)

= (Po , R Po ,WRi ) (Pcow, R Pcow,WRi )

(B-40)

Pg ,WRi Pg ,Wi = (Po ,WRi + Pcgo ,WRi ) (Po ,Wi + Pcgo ,wi ) = (Po ,WRi Po ,Wi ) + (Pcgo ,WRi Pcgo ,wi ) Pw,WRi Pw,Wi = (Po ,WRi Pcow,WRi ) (Po ,Wi Pcow,Wi ) = (Po ,WRi Po ,Wi ) (Pcow,WRi Pcow,Wi )

157 (B-41)

(B-42)

Substituting, we have

Tg , RWRi ((Po, R Po,WRi ) + (Pcgo, R Pcgo,WRi )) + To, RWRi (Rso )RWRi (Po, R Po,WRi ) Tg ,WRiWi ((Po,WRi Po,Wi ) + (Pcgo,WRi Pcgo,Wi )) To,WRiWi (Rso )WRiWi (Po,WRi Po,Wi ) Tw,WRiWi (Rsw )WRiWi ((Po,WRi Po,Wi ) (Pcow,WRi Pcow,Wi )) + GGWT1 S g Rso S o Rsw S w t + + B Bo Bw g WRi t
(B-43) Let us define GGWT2 as
GGWT 2 = Tg , RWRi (Pcgo , R Pcgo ,WRi ) Tw, RWRi (Rsw )R WRi (Pcow, R Pcow,WRi ) Tg ,WRiWi (Pcgo,WRi Pcgo,Wi ) + Tw,WRi Wi (Rsw )WRi Wi (Pcow,WRi Pcow,Wi )

+ Tw, RWRi (Rsw )RWRi ((Po, R Po,WRi ) (Pcow, R Pcow,WRi ))

= Vb,WRi

(B-44)

So we have
Tg , R WRi (Po , R Po ,WRi ) + To , R WRi (Rso )R WRi (Po , R Po ,WRi ) + Tw, R WRi (Rsw )R WRi (Po , R Po ,WRi ) Tg ,WRi Wi (Po ,WRi Po ,Wi ) To ,WRiWi (Rso )WRi Wi (Po ,WRi Po ,Wi )

Tw,WRi Wi (Rsw )WRi Wi (Po ,WRi Po ,Wi ) + GGWT 3

(B-45)

= Vb ,WRi

S g Rso S o Rsw S w t + + B Bo Bw g WRi t

158 where GGWT3=GGWT1+GGWT2 So (B-46)

(T [T

g , R WRi

+ To , R WRi (Rso )R WRi + Tw, R WRi (Rsw )R WRi )(Po , R Po ,WRi )

g ,WRi Wi

+ To ,WRi Wi (Rso )WRi Wi + Tw,WRiWi (Rsw )WRi Wi (Po ,WRi Po ,Wi ) (B-47)

+ GGWT 3 S g Rso S o Rsw S w t + + B Bo Bw g WRi t

= Vb ,WRi

Let us set XG = Tg , R WRi + To , R WRi (Rso )R WRi + Tw, R WRi (Rsw )R WRi

YG = Tg ,WRiWi + To ,WRiWi (Rso )WRiWi + Tw,WRiWi (Rsw )WRiWi

(B-48) (B-49)

XG (Po , R Po ,WRi ) + YG(Po ,WRi Po ,Wi ) + GGWT 3 = Vb ,WRi

S g Rso S o Rsw S w t + + B Bo Bw g WRi t (B-50)

XG * Po , R + (YG XG ) * Po ,WRi YG * Po ,Wi + GGWT 3 = Vb ,WRi

S g Rso S o Rsw S w t + + B Bo Bw g WRi t (B-51)

Let us set GGWT = GGWT 3 YG * Po ,Wi


ZG = YG XG

(B-52) (B-53)
S g Rso S o Rsw S w t + + B Bo Bw g WRi t

so, we have

XG * Po , R + ZG * Po ,WRi + GGWT = Vb ,WRi

(B-54)

B.3.3 Water

Material balance

159 (Qw, in) R WRi (Qw, out )WRi Wi = (WaterAccum)WRi k r ,w ( w,R w,WRi ) (Qw, in) R WRi = J R WRi B w w R WRi = Tw, R WRi ( w, R w,WRi ) Where k r ,w Tw, R WRi = J R WRi B w w R WRi k r ,w ( w,WRi w,Wi ) (Qw, out )WRi Wi = J WRi Wi B w w WRi Wi = Tw, WRi Wi ( w,WRi w,Wi ) Where k r ,w Tw, WRi Wi = J WRi Wi B w w WRi Wi By definition
w = Pw

(B-55)

(B-56)

(B-57)

(B-58)

(B-59)

wZ
144

(B-60)

Accumulation term:
S w t B w WRi t

(WaterAccum)WRi = Vb ,WRi

(B-61)

Vb,Wri = accumulation term


2 Vb ,WRi = RWRi H

(B-62)

Equation (B- ) can then be written as


S w t B w WRi t

Tw, R WRi ( w, R w,WRi ) Tw, WRi Wi ( w,WRi w,Wi ) = Vb ,WRi

(B-63)

160
Z Z Z Z Tw, R WRi Pw, R w Pw,WRi w Tw,WRi Wi Pw,WRi w Pw,Wi w 144 R 144 WRi 144 WRi 144 Wi S w t B w WRi t

= Vb ,WRi

(B-64)
Z Z Tw, R WRi (Pw, R Pw,WRi ) Tw, R WRi w w 144 R 144 WRi Z Z Tw,WRi Wi (Pw,WRi Pw,Wi ) + Tw,WRi Wi w w 144 WRi 144 Wi S w t B w WRi t

(B-65)

= Vb ,WRi

Z Z Z Z GWWT1 = Tw, R WRi w w w + Tw,WRi Wi w 144 WRi 144 Wi 144 R 144 WRi (B-66) S w t B w WRi t

Tw, R WRi (Pw, R Pw,WRi ) Tw,WRiWi (Pw,WRi Pw,Wi ) + GWWT1 = Vb ,WRi

(B-67)

Capillary pressure Pcow = Po Pw so Pw = Po Pcow


Pw, R Pw,WRi = (Po , R Pcow, R ) (Po ,WRi Pcow,WRi )

(B-68) (B-69)

= (Po , R Po ,WRi ) (Pcow, R Pcow,WRi )

(B-70)

Pw,WRi Pw,Wi = (Po ,WRi Pcow,WRi ) (Po ,Wi Pcow,Wi ) = (Po ,WRi Po ,Wi ) (Pcow,WRi Pcow,Wi )

(B-71)

161 Substituting, we have: Tw, R WRi [(Po , R Po ,WRi ) (Pcow, R Pcow,WRi )] Tw,WRiWi [(Po ,WRi Po ,Wi ) (Pcow,WRi Pcow,Wi )] S w t B w WRi t (B-72)

+ GGWT1 = Vb ,WRi

Let us define GWWT2 as GGWT 2 = Tw, R WRi (Pcow, R Pcow,WRi ) + Tw,WRi Wi (Pcow,WRi Pcow,Wi ) + GWWT1 (B-73)

So we have
Rsw S w t B w WRi t

Tw, R WRi (Po , R Po ,WRi ) Tw,WRi Wi (Po ,WRi Po ,Wi ) + GGWT 2 = Vb ,WRi

(B-74)

So Tw, R WRi Po , R (Tw, R WRi + Tw,WRiWi )Po ,WRi + Tw,WRi Wi Po ,Wi + GOWT 2 S w t B w WRi t (B-75)

= Vb ,WRi

Let us set GWWT = Tw,WRi Wi Po ,Wi + GWWT 2 ZW = (Tw, R WRi + Tw,WRi Wi )


S w t B w WRi t

(B-76) (B-77)

XW = Tw, R WRi

XW * Po , R + ZW * Po ,WRi + GWWT = Vb ,WRi

(B-78)

162
B.3.4 Summary (well-region side)

Oil:
S o t B o WRi t

XO * Po , R + ZO * Po ,WRi + GOWT = Vb ,WRi

(B-79)

Water:
S w t B w WRi t

XW * Po ,R + ZW * Po ,WRi + GWWT = Vb ,WRi

(B-80)

Gas:
S g Rso S o Rsw S w t + + B Bo Bw g WRi t

XG * Po , R + ZG * Po ,WRi + GGWT = Vb ,WRi

(B-81)

The combination of the three equations gives:

(B B R ) (XO * P + ZO * P + GOWT ) (B B R ) (XW * P + ZW * P + GWWT ) + (B ) ( XG * P + ZG * P + GGWT )


o g so WRi o,R o ,WRi W g sw WRi o,R o ,WRi g WRi o,R o ,WRi

(B-82)

= Vb ,WRi C t

t Po ,WRi t

This is the pressure equation corresponding to the well region i. Let us set
AO = (Bo B g Rso )WRi AG = (B g )WRi AW = (BW B g Rsw )WRi

(B-83) (B-84)

So , we get

AO * ( XO * Po , R + ZO * Po ,WRi + GOWT )

163

AG * ( XG * Po , R + ZG * Po ,WRi + GGWT ) = Vb ,WRi C t t Po ,WRi t

AW * (XW * Po , R + ZW * Po ,WRi + GWWT ) (B-85)

( AO * XO + AW * XW + AG * XG ) * Po,R + ( AO * ZO + AW * ZW + AG * ZG ) * Po ,WRi
+ AO * GOWT + AW * GWWT + AG * GGWT P = Vb ,WRi Ct t o ,WRi t (B-86)

Let us set AS = AO * XO + AW * XW + AG * XG AR = AO * ZO + AW * ZW + AG * ZG AT = AO * GOWT + AW * GWWT + AG * GGWT Then, the pressure equation becomes: AS * Po , R + AR * Po ,WRi + AT = Vb ,WRi C t t Po ,WRi t

(B-87) (B-88) (B-89)

(B-100)

B.3.5 Finite difference formulation: the IMPES method

Using the IMPES scheme (implicit pressure explicit saturation) AS n * Po , R


n +1

+ AR n * Po ,WRi

n +1

+ AT n =

Vb ,WRi C t ,WRi t

(P

n +1

o ,WRi

Po ,WRi

(B-101)

AS n * Po , R

n +1

Vb ,WRi C t ,WRi Vb ,WRi C t ,WRi n n +1 n Po ,WRi + AR n Po ,WRi = AT t t

(B-102)

Let us set AZ n = AR n Vb ,WRi C t ,WRi t Vb ,WRi C t ,WRi Po ,WRi


n

(B-103) (B-104)

t The pressure equation at well region i becomes:

AQ n = AT n

164 AS n * Po , R or AS i * Po , R
n n +1 n +1

+ AZ n * Po ,WRi

n +1

= AQ n

(B-105)

+ AZ i * Po ,i
n

n +1

= AQi

(B-106)

Where Po,R is the average pressure in the reservoir region And Po,I = Po,Wri is the average pressure in the well region i The expression of the coefficients ASin , AZin , and AQin are respectively the same as in equation B-87, B-103 and B-104. They are evaluated at well-region I and at time step tn .
AS in = ( AS )i
n

AZ = AZ
n i

AQ

n i

( ) = (AQ )
n i n

(B-106A) (B-106B) (B-106C)

165
B.4 Reservoir-region side

B.4.1 Oil

The oil material balance gives: (Qo, in) R (Qo, out ) R WR = (OilAccum) R (Qo, in) R = 0.0
(Qo, out ) R WR = (Qo ,in )R WRi
NW i =1

(B-107) (B-108)

k r ,o = J R WRi B i =1 o o
NW NW i =1

( o,R o,WRi ) R WRi

(B-109)

= To , R WRi ( o , R o ,WRi ) Where k r ,o To , R WRi = J R WRi B o o By definition Z o = Po o 144 Accumulation term:


S o t B o R t

R WRi

(B-110)

(B-111)

(OilAccum) R = Vb , R

(B-112)

Vb,R = Bulk volume of reservoir region Vb , R = R 2 H Vb ,WRi


i =1 NW

2 2 = R 2 RWRi RWi i =1
NW

(B-113)

Equation (B-1) can then be written as

0.0 [To , R WRi ( o , R o ,WRi )] = Vb , R


NW i =1

S o t B o R t

(B-114)

166
Z Z To , R WRi Po , R o Po ,WRi o = Vb , R 144 R 144 WRi S t o B o R t

(B-115) Z Z [To , R WRi (Po , R Po ,WRi )] To , R WRi o o 144 R 144 WRi S o t B o R = Vb , R t


NW Z Z GOWTR = + To , R WRi o o i =1 144 R 144 WRi

(B-116)

(B-117)

XORi = To , R WRi So, we have:


S o t B o t R

[XOR (P
NW i =1 i

o, R

Po ,WRi )] + GOWTR = Vb , R

(B-118)

B.4.2 Gas

(Qg , in) R (Qg , out ) R WR = (GasAccum) R (Qg , in) R = 0.0

(B-119) (B-120)

167
(Qg , out ) R WR = (Qg , in )R WRi
i =1 NW

kr,g ( g ,R g ,WRi ) J R WRi B g g R WR NW Rso k r ,o ( o,R o,WRi ) + = + J R WRi B i =1 o o R WR R k J R WRi sw r , w ( w,R w,WRi ) B w w R WRi
S g Rso S o Rsw S w t + + B Bo Bw g R t

(B-121)

(GasAccum) R = Vb , R

(B-122)

Let us set :
k rg Tg , R WRi = J R WRi B g g R WRi

(B-123)

k To , R WRi = J R WRi ro B o o R WRi k Tw, R WRi = J R WRi rw B w w R WRi

(B-124)

(B-125)

The mass balance equation for gas in Reservoir Region becomes:

( g ,R g ,WRi ) + To, RWRi (Rso )RWRi ( o,R o,WRi ) Tg , R WRi + Tw, R WRi (Rsw )R WRi ( w, R w,WRi )
= Vb , R S g Rso S o Rsw S w t + + B Bo Bw g R t (B-126)

The potential is (Assuming g=gc)

168

g = Pg g o = Po o

Z 144 Z 144 Z 144

(B-127)

(B-128)

w = Pw w

(B-129)

Z Z Pg ,WRi g Tg , R WRi Pg , R g 144 WRi 144 R NW Z Z + T (R ) P P o , R WRi so R WRi o , R o 144 R o ,WRi o 144 WRi i =1 Z Z + Tw, R WRi (Rsw )R WRi Pw, R w Pw,WRi w 144 R 144 WRi S g Rso S o Rsw S w t + + B Bo Bw R g t

(B-130)

= Vb , R

Z Z g Tg , R WRi (Pg , R Pg ,WRi ) Tg , R WRi g 144 R 144 WRi NW Z Z + To , R WRi (Rso )R WRi (Po , R Po ,WRi ) To , R WRi (Rso )R WRi o o i =1 144 R 144 WRi Z Z + Tw, R WRi (Rsw )R WRi (Pw, R Pw,WRi ) Tw, R WRi (Rsw )R WRi w 144 w 144 R WRi S g Rso S o Rsw S w t + + B Bo Bw R g t (B-130) Let us GGWTR1i be

= Vb , R

169
Z Z GGWTR1i = Tg , R WRi g g 144 R 144 WRi Z Z To , R WRi (Rso )R WRi o o 144 R 144 WRi Z Z Tw, R WRi (Rsw )R WRi w w 144 R 144 WRi

(B-131)

So we have
Tg , R WRi (Pg , R Pg ,WRi ) + To , R WRi (Rso )R WRi (Po , R Po ,WRi ) + Tw, R WRi (Rsw )R WRi (Pw, R Pw,WRi ) + GGWTR1i S g Rso S o Rsw S w t + + B Bo Bw g WRi t

(B-132)

= Vb ,WRi

using capillary pressure, we have Pcow = Po Pw Pcgo = Pg Po so Pw = Po Pcow Po = Pg Pcgo


Pg , R Pg ,WRi = (Po , R + Pcgo , R ) (Po ,WRi + Pcgo ,WRi ) = (Po , R Po ,WRi ) + (Pcgo , R Pcgo ,WRi ) = (Po , R Po ,WRi ) (Pcow, R Pcow,WRi )

(B-133) (B-134) (B-135) (B-136)

(B-137)

Pw, R Pw,WRi = (Po , R Pcow, R ) (Po ,WRi Pcow,WRi )

(B-138)

Substituting, we have:

Tg , R WRi ((Po , R Po ,WRi ) + (Pcgo, R Pcgo,WRi )) + To , R WRi (Rso )R WRi (Po , R Po ,WRi ) + Tw, R WRi (Rsw )R WRi ((Po , R Po ,WRi ) (Pcow, R Pcow,WRi )) + GGWTR1 i S g Rso S o Rsw S w t + + B Bo Bw g R t

170

(B-139)

= Vb , R

Let us define GGWTR2 as GGWTR 2 i = Tg , R WRi (Pcgo, R Pcgo,WRi ) Tw, R WRi (Rsw )R WRi (Pcow, R Pcow,WRi ) (B-140)

So we have
Tg , R WRi (Po , R Po ,WRi ) + To , R WRi (Rso )R WRi (Po , R Po ,WRi ) + Tw, R WRi (Rsw )R WRi (Po , R Po ,WRi ) + GGWTR3 i S g Rso S o Rsw S w t + + B Bo Bw g R t

(B-141)

= Vb , R

where GGWTR3,i=GGWTR1i+GGWTR2i So
Tg , R WRi + To , R WRi (Rso )R WRi + Tw, R WRi (Rsw )R WRi Po , R Tg ,WRi Wi + To ,WRiWi (Rso )WRi Wi + Tw,WRi Wi (Rsw )WRiWi Po ,WRi i =1 + GGWTR3i
NW

(B-142)

= Vb , R

S g Rso S o Rsw S w t + + B Bo Bw g R t

(B-143)

Tg , R WRi + To , R WRi (Rso )R WRi + Tw, R WRi (Rsw )R WRi Po , R + Tg ,WRiWi + To,WRiWi (Rso )WRiWi + Tw,WRiWi (Rsw )WRiWi Po,WRi i =1 GGWTR3i
NW

[ [

171

= Vb , R

S g Rso S o Rsw S w t + + B Bo Bw g R t

(B-144)

Let us set XGRi = Tg , R WRi + To , R WRi (Rso )R WRi + Tw, R WRi (Rsw )R WRi

(B-145)

GGWTR = { GGWTR3i }
i =1

NW

(B-146) S g Rso S o Rsw S w t + + B Bo Bw g R t

[XGR (P
NW i =1 i

o, R

Po ,WRi )] + GGWTR = Vb , R

(B-147)

B.4.3 Water

Material balance (Qw, in) R (Qw, out ) R WR = (WaterAccum) R


(Qw, in) R =

(B-148) (B-149)

( tWe )R
t

NW k ( w, R w,WRi ) (Qw, out ) R WR = J R WRi r , w B i =1 w w R WRi

= Tw, R WRi ( w, R w,WRi ) Where k r ,w Tw, R WRi = J R WRi B w w R WRi By definition


w = Pw

(B-150)

(B-151)

wZ
144

(B-152)

172 Accumulation term:


S w t B w R t

(WaterAccum) R = Vb , R

(B-153)

Vb,R = bulk volume of reservoir region Vb , R = R 2 H Vb ,WRi


i =1 NW

2 = R 2 RWRi H i =1
NW

(B-154)

Equation (B- ) can then be written as

( tWe )R
t

[Tw, R WRi ( w, R w,WRi )] = Vb , R


NW i =1

S w t B w R t

(B-155)

(tWe )R
t

Z Z Tw, R WRi Pw, R w Pw,WRi w = Vb , R 144 WRi 144 R i =1


NW

S w t B w R t

(B-156)

( tWe )R

= Vb , R

NW Z Z Tw, R WRi (Pw, R Pw,WRi ) Tw, R WRi w w t i =1 144 R 144 WRi S w t B w R

(B-157)

Let us set
Z Z GWWTR1i = Tw, R WRi w w 144 R 144 WRi

(B-158)

173

( tWe )R
t

[Tw, R WRi (Pw, R Pw,WRi ) + GWWTR1i ] = Vb , R


NW i =1

S w t B w R t

(B-159)

Capillary pressure
Pw, R Pw,WRi = (Po , R Pcow, R ) (Po ,WRi Pcow,WRi )

= (Po , R Po ,WRi ) (Pcow, R Pcow,WRi )

(B-160)

Substituting, we have:

( tWe )R
t

T { w, R WRi [(Po , R Po ,WRi ) (Pcow, R Pcow,WRi )] + GWWTR1i } = Vb , R

S w t B w R t (B-161)

Let us define GWWT2i as

GWWTR 2 i = Tw, R WRi (Pcow, R Pcow,WRi ) + GWWTR1i

(B-162)

So we have

( tWe )R
t
So

[Tw, R WRi (Po , R Po ,WRi ) + GWWTR 2 i ] = Vb , R


NW i =1

S w t B w R t

(B-163)

( tWe )R
t

[Tw, R WRi (Po , R Po ,WRi )] GWWTR 2 i = Vb , R


NW NW i =1 i =1

S w t B w R t

(B-164)

Let us set

GWWTR =

( tWe )R
t

GWWTR 2 i
i =1

NW

(B-165)

174

XWRi = Tw, R WRi S w t B w R t

(B-166)

[XWR (P
NW i =1 i

o,R

Po ,WRi )] + GWWTR = Vb , R

(B-167)

B.4.4 Summary (reservoir-region side)

Oil

[XOR (P
NW i =1 i

o, R

Po ,WRi )] + GOWTR = Vb , R

S o t B o R t

(B-168)

Water

[XWR (P
NW i =1 i

o,R

Po ,WRi )] + GWWTR = Vb , R

S w t B w R t

(B-169)

Gas

[XGR (P
NW i =1 i

o, R

Po ,WRi )] + GGWTR = Vb , R

S g Rso S o Rsw S w t + + B Bo Bw g R t

(B-170)

Combining the three balance equations for oil, gas and water for the Reservoir side, we have:

(B

NW B g Rso )R [XORi (Po , R Po ,WRi )] + GOWTR i =1

NW + (Bw B g Rsw )R [XWRi (Po , R Po ,WRi )] GWWTR i =1

(B-171)

t (Po , R ) NW + (B g )R [XGRi (Po , R Po ,WRi )] + GGWTR = Vb , R C t t i =1

175 This is the pressure equation corresponding to the reservoir region side. Let us set
AWR = (BW B g Rsw )R AGR = (B g )R AOR = (Bo B g Rso )R

(B-172) (B-173) (B-174)

So , we get
NW AOR [XORi (Po , R Po ,WRi )] + GOWTR i =1 NW + AWR [XWRi (Po , R Po ,WRi )] GWWTR i =1

(B-175)

t (Po , R ) NW + AGR [XGRi (Po , R Po ,WRi )] + GGWTR = Vb , R C t t i =1

NW ( AOR * XORi + AWR * XWRi + AGR * XGRi )(Po , R Po ,WRi ) i =1 + ( AOR * GOWTR + AWR * GWWTR + AGR * GGWTR ) t (Po , R ) = Vb , R C t t

(B-176)

Let us set: ASRi = AOR * XORi + AWR * XWRi + AGR * XGRi


ATR = AOR * GOWTR + AWR * GWWTR + AGR * GGWTR

(B-177) (B-178) (B-179)

[ASR (P
NW i =1 i NW

o, R

Po ,WRi )] + ATR = Vb , R C t , R

t (Po , R ) t t (Po , R ) t

(ASR * P ) (ASR * P
NW i =1 i o,R i =1 i

o ,WRi

) + ATR = V

b, R

C t , R

(B-180)

176
B.4.5 Finite difference formulation: the IMPES method (reservoir-region)
NW Vb , R C t , R NW n +1 n n +1 n ASRi Po , R ( ASRi ) (Po ,WRi ) + ( ATR ) = t i =1 i =1 n

(P

n +1

o, R

Po , R

)
(B-181)

n n n NW NW Vb , R C t , R Vb , R C t , R n +1 n n +1 n n Po , R ASRi Po , R ( ASRi ) (Po ,WRi ) = ( ATR ) t t i =1 i =1 (B-182)

Let us set
Vb , R C t , R NW ASRR = ASRi t i =1
n n n

(B-183)

AZRi = ( ASRi )
n n

(B-184)
Vb , R C t , R t
n n

AQR = ( ATR )
n

Po , R

(B-185)

ASRR n * Po , R

n +1

+ AZRi * Po ,WRi
n i =1

NW

n +1

= AQR n

(B-186)

B.5 Summary

We have (NW+1) equations in (NW+1) unknowns.


n +1

AS i * Po , R

+ AZ i * Po ,i
n NW i =1

n +1

= AQi

(B-187) = AQR n (B-188)

ASRR n * Po, R

n +1

+ AZRi * Po ,WRi
n

n +1

177 AS i n * Po , R n +1 + AZ i n * Po ,i n +1 = AQi n NW n +1 n n +1 ASRR n * Po , R + AZRi * Po ,WRi = AQR n i =1


+ Pon, R1 AQ1n n +1 n Po ,1 AQ2 L L L L L L L L L = 0 L L L L L 0 0 L L L L L L +1 n n L L L 0 AZ NW Pon, NW 1 AQ NW n +1 n L L L L AZR NW Po , NW AQR n

(B-189)

AS1n n AS 2 L K L L AS n NW ASRR n

AZ 1n 0 0 L L 0 0 AZR1n

0 n AZ 2 0 L L L L n AZR2

L L L L 0 L L L L L L L

0 0 L

(B-190)

M ( n ) P ( n +1) = Q ( n ) The LU decomposition method is used to solve for P(n+1)

(B-191)

For NW=7 (7 wells), we have for example: AS1n n AS 2 AS 3n n AS 4 AS n 5 n AS 6 AS n 7 ASRR n


+ 0 Pon, R1 AQ1n n 0 Pon,1+1 AQ2 + 0 Pon, 2 1 AQ3n n +1 n 0 Po ,3 AQ4 = + 0 Pon, 4 1 AQ5n + 0 Pon,5 1 AQ6n + AZ 1n Pon, 6 1 AQ7n n +1 AZR7n Po , 7 AQR n

AZ 1n 0 0 0 0 0 0 AZR1n

0 AZ 1n 0 0 0 0 0 n AZR2

0 0 AZ 1n 0 0 0 0 AZR3n

0 0 0 AZ 1n 0 0 0 n AZR4

0 0 0 0 AZ 1n 0 0 AZR5n

0 0 0 0 0 AZ 1n 0 AZR6n

(B-192)

178
AS1n n AS 2 AS 3n n AS 4 = AS n 5 AS 6n AS n 7 ASRR n
+ Pon, R1 n +1 Po ,1 + Pon, 2 1 n +1 P = on,3 1 P + on, 4 1 + Po ,5 P n +1 o,6 + Pon, 7 1

AZ 1n 0 0 0 0 0 0 AZR
n 1

0 AZ 0 0 0 0 0 AZR
n 2 n 1

0 0 AZ 1n 0 0 0 0 AZR
n 3

0 0 0 AZ 0 0 0 AZR
n 4 n 1

0 0 0 0 AZ 1n 0 0 AZR
n 5

0 0 0 0 0 AZ 1n 0 AZR6n

M ( n)

0 0 0 0 0 0 AZ 1n AZR7n

(B-193)

P ( n +1)

(B-194)

Q (n)

AQ1n n AQ2 AQ3n n AQ4 = AQ n 5 AQ6n AQ n 7 AQR n

(B-195)

179
B.6 Computation of saturations B.6.1 Computation of saturations in well-regions B.6.1.1 Water saturation
S w t B w WRi t

XW * Po , R

n +1

+ ZW * Po ,WRi

n +1

+ GWWT = Vb ,WRi

(B-196)

In the equation above, the coefficients XW, ZW and are evaluated using the saturation values at time tn and the pressure values at time tn+1. Let us set

(XW * P YY1 =

n +1

o, R

+ ZW * Po ,WRi Vb ,WRi

n +1

+ GWWT

(B-197)

S w t B w WRi YY1 = t S w S w YY1 * t = B B w WRi w WRi


n S w n +1 YY1 * t + B BW w WRi = n +1 n

(B-198)

(B-199) WRi WRi

(S )

n +1 w WRi

(B-200)

B.6.1.2 Gas saturation

180
S g Rso S o Rsw S w t + + B Bo Bw g WRi t

XG * Po , R

n +1

+ ZG * Po ,WRi

n +1

+ GGWT = Vb ,WRi

(B-201)

In the equation above, the coefficients XG, ZG and GGWT are evaluated using the saturation values at time tn and the pressure values at time tn+1. Let us set ZZ 3 = XG * Po , R
n +1

+ ZG * Po ,WRi Vb ,WRi

n +1

+ GGWT

(B-202)

S g Rso S o Rsw S w t + + B Bo Bw g WRi ZZ 3 = t

(B-203)

Where S0=1-Sg-Sw
S g Rso (1 S g S w ) Rsw S w + + ZZ 3 * t = t B Bo Bw g WRi
S g Rso Rso S g Rso S w Rsw S w ZZ 3 * t = t + + B Bo Bo Bo Bw g WRi 1 Rso Rso S R + = t S g + S w w so B B Bo w Bo WRi g Bo

(B-204)

(B-205)

1 Rso + t Rso + S w Rsw Rso ZZ 3 * t = t S g B B Bo WRi w Bo g Bo WRi Rso R R VV 1 = t + S w sw so B Bo WRi w Bo Rso Rso R R R R + S w sw so = + S w sw so B B Bo WRi Bo Bo WRi w w Bo
n +1 n

(B-206)

(B-207)

181 1 Rso ZZ 3 * t VV 1 = t S g B Bo g WRi 1 Rso S g 1 Rso = S g B B g Bo WRi g Bo WRi


n 1 Rso ZZ 3 * t VV 1 + S g B Bo g WRi = n +1 1 Rso B g Bo WRi WRi n +1 n

(B-208)

(S )
n +1 g

WRi

(B-209)

B.6.2 Computation of saturations in reservoir-regions

B.6.2.1 Water saturation

[XWR (P
NW i =1 i

n +1

o,R

Po ,WRi

n +1

)]+ GWWTR = V

b,R

S w t B w R t

(B-210)

In the equation above, the coefficients XWRi, and GWWTR are evaluated using the saturation values at time tn and the pressure values at time tn+1. Let us set

ZZ 2 =

[XWR (P
NW i =1 i

n +1

o,R

Po ,WRi Vb , R

n +1

)]+ GWWTR
(B-211)

S w t B w R ZZ 2 = t

(B-212)

182
S w ZZ 2 * t = t B w R S w S w = B B w R w R
n S w n +1 Bw ZZ 2 * t + B w R = R n +1 n

(B-213)

(S )
n +1 w

(B-214)

B.6.2.2 Gas saturation

[XGR (P
NW i =1 i NW

o, R

Po ,WRi )] + GGWTR = Vb , R

S g Rso S o Rsw S w t + + B Bo Bw g R t

(B-215)

ZZ 4 =

[XGR (P
i =1 i

o,R

Po ,WRi )] + GGWTR Vb , R (B-216)

S g Rso S o Rsw S w t + + B Bo Bw g R ZZ 4 = t S g Rso S o Rsw S w + + ZZ 4 * t = t B Bo Bw g R

(B-217)

(B-218)

Where S0=1-Sg-Sw

S g Rso (1 S g S w ) Rsw S w ZZ 4 * t = t + + B Bo Bw g R S g Rso Rso S g Rso S w Rsw S w = t + + B Bo Bo Bo Bw R g 1 Rso Rso R R + ZZ 4 * t = t S g + S w sw so B B Bo Bo w R g Bo 1 Rso + t Rso + S w Rsw Rso = t S g B B B Bo R w o g Bo R Rso R R VV 2 = t + S w sw so B B Bo R w o Rso Rso R R R R + S w sw so = + S w sw so B B B B Bo R Bo R w w o o
n 1 Rso ZZ 4 * t VV 2 + S g B g Bo R = n +1 1 Rso B g Bo R R n +1 n

183

(B-219)

(B-220)

(B-221)

(S )
n +1 g

(B-222)

APPENDIX C DYNAMIC NODAL ANALYSIS

C.1 History Match procedure The procedure used to compute the history match is summarized in the following steps: 1. Assume that the production history is known. Thus, for each observed production time Tobs1, Tobs2,, Tobs j,, Tobs n, the corresponding observed rate Qobs1, Qobs2, , Qobsj, , Qobsn is known. 2. Assume that at time Tj the following data are known: reservoir pressure Pj fluid properties as a function of pressure and temperature The type of decline (harmonic, hyperbolic or exponential) as well as the rate of decline. The pressure drop correlations as a function of rate for each Q. 3. The oil in place Nj at this time Tj is computed from material balance equation: 4. Compute GOR and WOR at this time Tj when the reservoir pressure is Pj. using equations A-6 and A-25. 184

185 5. Calculate the rate Qj at which the well will produce under the existing conditions. This is done by using the nodal analysis technique. As stated earlier, in this study the node is chosen at the bottom hole. The nodal analysis technique is presented in section 3 of this appendix. 6. Assume a small decrement in reservoir pressure Pj. The new reservoir pressure is then Pj+1 = Pj- Pj . At this reservoir pressure, calculate the new oil in place Nj+1 using the material balance equation The total amount of oil produced when the reservoir pressure decreases from Pj to Pj+1 is:

N = N j N j +1

(C-1)

7. Calculate the rate Qj+1 at which the well will produce under the present reservoir pressure Pj+1. This is done by nodal analysis at bottom hole. 8. Knowing the total amount of oil produced () and the oil flow rate Qj and Qj+1 at reservoir pressures Pj and Pj+1, we can calculate the elapsed time T required to reach that production.

For exponential decline10,11:

D=

Q j Q j +1 N

Q j Q j +1 N j N j +1

(C-2)

T =

Q 1 ln j D Q j +1

(C-3)

For harmonic decline10,11:

186

D=

Qj N

ln

Qj Q j +1

(C-4)

T =

1 Q j Q j +1 [ D Q j +1

(C-5)

For hyperbolic decline10,11: Q 1b j +1 D= * 1 Q (1 b) * N j


Qj
b Q j +1 1 T = * 1 + Q b*D j

(C-6)

(C-7)

The total calculated time when the reservoir pressure is Pj+1 can be calculated as:

T j +1 = T j + T

(C-8)

9. Assume a new reservoir pressure Pj+1 : Pj+1 = Pj-P where P is the pressure decrement. Repeat the process from step 4 to step 8 until the total calculated time Tj+k is greater or equal to the observed production time. 10. At this point, we have the model predicted times T1, T2, , Tj, Tj+k, and the corresponding rates and reservoir pressures: Q1, Q2, , Qj, , Qj+k,

187

P1, P2, , Pj, , Pj+k,

For each observed time Tobs j, we calculate the corresponding model predicted rate Qj and corresponding pressure Pj by interpolating the model predicted rates and pressures respectvely. At this point, we check how the calculated flow rate Qj compares with the historical observed production rate Qobs j at the same time Tobs j. This check represents the history match of the observed data. If significant differences exist between the calculated and the observed production rates, then some selected reservoir parameters have to be adjusted in order to match the historical performance. In order to match the historical observed performance, a non-linear regression calculation is performed to minimize the difference between calculated and observed production rates. This regression analysis is discussed in section 3 of this chapter. Once a satisfactory match between the predicted and the observed performance is obtained, we can proceed with forecast of future performance calculations. In the presence of water influx, the procedure is the same as in the case of natural depletion drive. However the description (ratio of aquifer radius to reservoir radius (RD), aquifer permeability, porosity, thickness and angle of encroachment) of the aquifer has to be supplied. The water influx, We is computed as:

We = US ( p, t )

(C-9)

188

where
U = 1.119 fH (C w + C f ) Ro2

(C-10) (C-11)

f = (encroachmen tan gle) 0 / 360 0

S(p.t)=aquifer influx function. In this work, this function is computed using a polynomial approach for the Van Everdingen and Hurst water encroachment model. This approach is discussed in a paper by Klins,Bouchard and Cable49. Cw=compressibility of aquifer water Cf = compressibility of formation

=porosity of acquifer
H= aquifer thickness

C.2 Future Performance Prediction

1. The future performance of the well under the existing conditions as well as under altered conditions can be calculated. The procedure is the same as described from step 2 to steps 8 in the History Matching section. Repeat the steps till an abandonment rate is reached. 2. Consider different scenarios for variations in production procedures. These include, for example, changing the number of perforations, stimulating the well, fracturing the well, installing the compressor at the surface.

189

3. Predict the future performance under the new operating conditions using the same procedure as explained in Step 1. 4. Repeat Step 3 for alternate combinations of input parameters to quantify uncertainties in the prediction of future performance. 5. Compare the performance under the new scenario with the base case to calculate the incremental gas production as a function of time. 6. Repeat step 5 for different input configuration. 7. Use information generated in step 5 and step 6 to study the economic feasibility of making the changes in the production configuration.

C.3 Regression Analysis

The basic objective of using the non-linear regression in this problem is to determine the optimum set, , of reservoir/completion parameters such that the observed data match as closely as possible to the calculated data from the model. In this study, the parameters on which the regression is performed consist of any set of 3 variables chosen among the following parameters: permeability, skin, radius of drainage, pay, perforated interval, radius of perforations, diameter of perforations, porosity, water saturation, and density of perforations. For example, one can choose such that

={permeability, skin, length of perforation penetration }. In this case the regression


calculations will be performed on the following variables: permeability, skin and length of perforation penetration.

190

In this study, the Levenberg-Marquardt algorithm LMDIF125, has been used. This algorithm has been selected because it does not require the derivatives of the functions to minimize. The purpose of LMDIF1 is to minimize the sum of the squares of m non-linear functions in n variables. This is done by the more general least square solver LMDIF. The user must provide the subroutines that compute the functions. The jacobian is then calculated by a forward-difference approximation. As stated earlier, in this work, the variables on which to regress are any set of 3 variables chosen by the user among the following parameters: permeability, skin, radius of drainage, pay, perforated interval, radius of perforations, diameter of perforations, porosity, water saturation, length of perforation penetration, and density of perforations. The m non-linear functions F1(), F2(), , Fm () can be considered as the components of a vector FVEC. The objective function is then computed as the euclidian norm of FVEC, that is: Objective function = Fj2 The functions Fj are chosen such that the computation is more resistant to errors in the observed data and is less sensitive to outliers. The definition of the functions Fj is presented below.

191 C.3.1 Function F1

This function compares observed data with the predicted data. Ideally the correlation coefficient between the observed and model predicted performance is equal to 1. Mathematically,

F1 ( ) = (Qobs , Qmod ) 1

(C-12)

(Qobs , Qmod ) =

COV (Qobs , Qmod ) Qobs * Qmod

(C-13)

FVEC (1) = F1 ( )

(C-14)

The advantage of using the correlation coefficient is that it is resistant to noise in the data. It is not sensitive to outliers. It should be noted that high correlation coefficient does not necessarily mean that the values are similar. The basic assumption here is that we are modeling the measured data correctly that the errors in the measured data are normally distributed with mean zero, and the errors are not correlated.

C.3.2 Function F2

This function is chosen to represent the fact that ideally the plot of Qmod versus Qobs is a straight line of slope one ( with intercept equal to zero).
FVEC (2) = F2 ( ) = SLOPE 1

(C-15)

SLOPE =

COV (Qobs , Qmod )

2 obs

(C-16)

192

where COV is the covariance between the observed and model predicted rates. So,
FVEC (2) = COV (Qobs , Qmod ) 1

2 obs

(C-17)

C.3.3 Function F3

This function is chosen to represent the fact that ideally the intercept of the straight line Qmod versus Qobs is equal to zero.
INTERCEPT = Q mod SLOPE * Q obs = 0

(C-18)

F3 =

Qmod 1 Qobs

(C-19)

because ideally the slope is equal to 1: SLOPE=1.

FVEC (3) = F3

(C-20)

C.3.4 Function F4

This function compares observed pressure data with the predicted pressure data. Ideally the correlation coefficient between the observed and model predicted performance is equal to 1. Mathematically,

F4 ( ) = ( Pobs , Pmod ) 1

(C-21)

( Pobs , Pmod ) =

COV ( Pobs , Pmod ) Pobs * Pmod

(C-22)

193
FVEC (4) = F4 ( )

(C-23)

The advantage of using the correlation coefficient is that it is resistant to noise in the data. It is not sensitive to outliers. It should be noted that high correlation coefficient does not necessarily mean that the values are similar. The basic assumption here is that we are modeling the measured data correctly that the errors in the measured data are normally distributed with mean zero, and the errors are not correlated.
C.3.5 Function F5

This function is chosen to represent the fact that ideally the plot of Pmod versus Pobs is a straight line of slope one ( with intercept equal to zero). FVEC (5) = F5 ( ) = SLOPE 1 SLOPE = COV ( Pobs , Pmod ) (C-24)

2 obs

(C-25)

where COV is the covariance between the observed and model predicted pressures. So, FVEC (5) = COV ( Pobs , Pmod ) 1

2 obs

(C-26)

C.3.6 Function F6

This function is chosen to represent the fact that ideally the intercept of the straight line Pmod versus Pobs is equal to zero.
INTERCEPT = P mod SLOPE * P obs = 0

(C-27)

194
F6 = Pmod 1 Pobs

(C-28)

because ideally the slope is equal to 1: SLOPE=1. FVEC (6) = F6 (C-29)

It should be emphasized that m, the number of functions Fj, should be greater or equal to the number of variables, n, on which the regression is performed. Also, the user specifies the tolerance FTOL which is used in the regression. The program terminates when the algorithm estimates either that the relative errors in the sum of squares, Fj2, is at most FTOL or that the relative error in the regression variables between successive iterations is at most FTOL. On termination, the regression algorithm output an integer variable INFO whose value means the following: INFO = 0: improper input parameters. INFO = 1: algorithm estimate that the relative error in the sum of squares is at most FTOL INFO = 2: algorithm estimates that the relative error between the calculated values of the regression parameters and the ideal solution is at most FTOL. INFO = 3: condition for info =1 and info = 2 both hold. INFO = 4: FVEC is orthogonal to the columns of the jacobian to machine precision. INFO = 5: number of calls to the function that compute FVEC has reached or exceed 200*(n+1) INFO = 6: FTOL is too small. No further reduction in the sum of squares is possible.

195

INFO = 7: FTOL is too small. No further improvement in the approximate solution is possible.

C.4.1 Parameter constraints

The Levenberg Marquardt algorithm33 that we use is unconstrained : i.e., variables can be chosen to minimize the objective function with value between infinite. Obviously, for our problem, we need to ensure that the values of the variables lie in the predefined interval of uncertainty and that these values are meaningful. For example we may want the regressed permeability value to be between Kmax and Kmin. In order to keep the values of the regression variables in certain predefined intervals, there are a couple of methods to use. It has been shown that the use of the penalty functions improves the convergence of the iterative procedure; however, it is also reported that the penalty function method may not prevent the values of the regression variables to be out of the predefined domain when the initial estimates of the regression variables are far from the solution. In this study, the imaging extension19 procedure is used.

C.4.2 Imaging extension method

The idea behind the method is to extend the objective function in such a way that the new objective function is defined everywhere (i.e., unconstrained) and that the solution of this new unconstrained problem is related to the solution of the original constrained problem.

196

The procedure for translating the unconstrained variable estimate LMDIF1 calculated by the regression algorithm LMDIF1 to the corresponding physically constrained value of the parameter c is the following: For LMDIF1 > max , compute :
N = int LMDIF 1 min max min

(C-30)

For LMDIF1 < max , compute :


min N = int LMDIF 1 min max 1

(C-31)

After calculating N, c can be obtained as: For N odd:

c = min + max + N ( max min ) LMDIF 1


For N even

(C-32)

c = LMDIF 1 N ( max min )

(C-33)

For more details about the imaging extension method, the reader is referred to the paper by R. Carvalho, L. G. Thompson, R. Redner and A. C. Reynolds cited in reference.

C.5 Nodal analysis technique

Nodal analysis1,2,5 provides a method to determine the rate at which a producing system will produce under certain production conditions. In order to evaluate that

197

producing rate, the production system is divided into two parts at a fixed node and the performance curves of each part are compared. These two performance curves are denoted as inflow (flow into the node) and outflow (flow out of the node) performance curves. For convenience, the node is chosen at the bottom hole. This choice does not affect the results of the performance computations. With the node at bottom hole, the inflow performance curve represents the pressure loss across the reservoir, the perforations and the gravel pack. It can be mathematically expressed in dimensionless form as:
Q PWF versus Qmax Pr I

(C-34)

Where
Pr P PWF WFS = 1 Pr Pr I PWFS PWF Pr P P = 1 r WFS Pr Pperf P r Pgp P (C-35) r

Qmax is the maximum flow rate at which the well can flow. The outfow performance curve describes the pressure loss in the tubing29.30.31,32,33, the bottom hole restriction (subsurface device)34,35,36, the safety valve, the well head choke and the surface pipeline. It can be mathematically expressed in dimensionless form as: PWF P r Where Q versus QMAX O

(C-36)

198
PWF P P = TBG + REST P r O P R Pr PSV + P r PCHOKE PPIPELINE + + P Pr r

(C-37)

A typical plot of the inflow curve as well as the two commonly observed outflow curves are shown in figure C.1 The overall performance of the producing system is obtained when the inflow and outflow curves intercept. This implies that the flow rate and the bottom hole flowing pressure are obtained by solving the equation: PWF P r P = WF P I r O

(C-38)

This equation is solved numerically using the secant method. As it can be seen on figure C.1 and figure C.2, this equation can have two different roots or one single root. If the equation has two different roots, one of the roots corresponds to an unstable production condition while the other root corresponds to a stable production condition. This situation is typical of system producing in two-phase flow with high gas velocity. If the equation has a single root, one of the following situations can happen: The derivative of the outflow curve at the root is positive. In this case the system produces under a stable condition. This is typical of systems close to single-phase flow. The derivative of the outflow curve at the root is negative. In this case the system produces under an unstable condition (liquid loading).

199

Figure C.1.

Typical inflow and outflow curves16.

200

P wf P r

Inflow curve

Outflow curve C

0 Unstable rate
Q Qmax

Figure C.2

Example of an unstable production condition (Liquid loading)

APPENDIX D

FIELD CASE: HISTORY MATCH RESULTS

In this appendix, the results of eight wells (out of a total number of 12 wells) are presented. Those wells are: Taylor Unit W-034304 1 (figure D1 and figure D2) Peregrine W32266A 1 (figure D3 and figure D4) Moore Unit FEE 1 (figure D5 and figure D6) Taylor Unit FEE 4 (figure D7 and figure D8) USA W-42621A 27-1 (figure D9 and figure D10) Woods 36-1 (figure D11 and figure D12) Federal W-86337 15-1 (figure D13 and figure D14) Federal W-47686 22-1 (figure D15 and figure D16) The history match results for the other four wells are shown in chapter V.

201

202

Observed rate
80

Predicted rate

70

WELL: TAYLOR UNIT W034304 1


60

Rate [stb/day]

50

40

30

20

10

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D1: Well Taylor Unit W-034304 1: match between predicted rate and observed rate.

203

Observed GOR
60000

Predicted GOR

WELL: TAYLOR UNIT W034304 1


50000

40000

GOR [scf/stb]

30000

20000

10000

0 0 500 1000 1500

Time [days]

Figure D2: Well Taylor Unit W-034304 1: match between predicted GOR and observed GOR.

204

Observed rate
100 90 80 70

Predicted rate

WELL: PEREGRINE W32266A 1

Rate [stb/day]

60 50 40 30 20 10 0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D3: Well Peregrine W32266A 1: match between predicted rate and observed rate.

205

Observed rate
7000

Predicted rate

6000

WELL: PEREGRINE W32266A 1

5000

GOR [scf/stb]

4000

3000

2000

1000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D4: Well Peregrine W32266A 1: match between predicted GOR and observed GOR.

206

Observed rate
1000 900 800 700

Predicted rate

WELL: MOORE UNIT FEE 1

Rate [stb/day]

600 500 400 300 200 100 0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D5: Well Moore Unit FEE 1: match between predicted rate and observed rate.

207

Observed GOR
14000

Predicted GOR

12000

WELL: MOORE UNIT FEE 1


10000

GOR [scf/day]

8000

6000

4000

2000

200

400

600

800

1000

1200

1400

1600

Time [days]

Figure D6: Well Moore Unit FEE 1: match between predicted GOR and observed GOR.

208

Observed rate
400

Predicted rate

350

WELL: TAYLOR UNIT FEE 4

300

Rate [stb/day]

250

200

150

100

50

0 0 500 1000 1500

Time [days]

Figure D7: Well Taylor Unit FEE 4: match between predicted rate and observed rate.

209

Observed GOR
70000

Predicted GOR

WELL: TAYLOR UNIT FEE 4


60000

50000

GOR [scf/stb]

40000

30000

20000

10000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D8: Well Taylor Unit FEE 4: match between predicted GOR and observed GOR.

210

Observed rate
900 WELL: USA W-42621A 27-1 800 700 600 Rate [stb/day] 500 400 300 200 100 0 0 200 400 600 800 Time [days] 1000

Predicted rate

1200

1400

1600

Figure D9: Well USA W-42621A 27-1: match between predicted rate and observed rate.

211

Observed GOR
3500

Predicted GOR

WELL USA W-42621A 27-1


3000

2500

GOR [scf/stb]

2000

1500

1000

500

500 Time [days]

1000

1500

Figure D10: Well USA W-42621A 27-1: match between predicted GOR and observed GOR.

212

Observed rate
80 70

Predicted rate

WELL: WOODS 36-1


60 50

Rate [stb/day]

40 30 20

10 0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D11: Well Woods 36-1: match between predicted rate and observed rate.

213

Observed GOR
30000

Predicted GOR

WELL: WOODS 36-1


25000

20000

GOR [scf/stb]

15000

10000

5000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D12: Well Woods 36-1: match between predicted GOR and observed GOR.

214

Observed rate
160

Predicted rate

WELL: FEDERAL W-86337 15-1


140

120

100

Rate [days]

80

60

40

20

0 0 500 1000 1500

Time [days]

Figure D13: Well Federal W-86337 15-1: match between predicted rate and observed rate.

215

Observed GOR
25000

Predicted GOR

WELL: FEDERAL W-86337 15-1


20000

GOR [scf/stb]

15000

10000

5000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D14: Well Federal W-86337 15-1: match between predicted GOR and observed GOR.

216

Observed rate
600

Predicted rate

WELL: FEDERAL W-47686 22-1


500

400

Rate [stb/day]

300

200

100

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D15: Well Federal W-47686 22-1: match between predicted rate and observed rate.

217

Observed GOR
7000

Predicted GOR

WELL: FEDERAL W-47686 22-1


6000

5000

GOR [scf/stb]

4000

3000

2000

1000

0 0 200 400 600 800 1000 1200 1400 1600

Time [days]

Figure D16: Well Federal W-47686 22-1: match between predicted GOR and observed GOR.

You might also like