89,99 €
Mehr erfahren.
- Herausgeber: John Wiley & Sons
- Kategorie: Wissenschaft und neue Technologien
- Sprache: Englisch
The essential resource to an integrated approach to reservoir modelling by highlighting both the input of data and the modelling results
Reservoir Modelling offers a comprehensive guide to the procedures and workflow for building a 3-D model. Designed to be practical, the principles outlined can be applied to any modelling project regardless of the software used. The author — a noted practitioner in the field — captures the heterogeneity due to structure, stratigraphy and sedimentology that has an impact on flow in the reservoir.
This essential guide follows a general workflow from data QC and project management, structural modelling, facies and property modelling to upscaling and the requirements for dynamic modelling. The author discusses structural elements of a model and reviews both seismic interpretation and depth conversion, which are known to contribute most to volumetric uncertainty and shows how large-scale stratigraphic relationships are integrated into the reservoir framework. The text puts the focus on geostatistical modelling of facies and heterogeneities that constrain the distribution of reservoir properties including porosity, permeability and water saturation. In addition, the author discusses the role of uncertainty analysis in the static model and its impact on volumetric estimation. The text also addresses some typical approaches to modelling specific reservoirs through a mix of case studies and illustrative examples and:
- Offers a practical guide to the use of data to build a successful reservoir model
- Draws on the latest advances in 3-D modelling software
- Reviews facies modelling, the different methods and the need for understanding the geological interpretation of cores and logs
- Presents information on upscaling both the structure and the properties of a fine-scale geological model for dynamic simulation
- Stresses the importance of an interdisciplinary team-based approach
Written for geophysicists, reservoir geologists and petroleum engineers, Reservoir Modelling offers the essential information needed to understand a reservoir for modelling and contains the multidisciplinary nature of a reservoir modelling project.
Sie lesen das E-Book in den Legimi-Apps auf:
Seitenzahl: 396
Veröffentlichungsjahr: 2018
Ähnliche
Table of Contents
Cover
Title Page
Copyright
Dedication
Preface
Chapter 1: Introduction
1.1 Reservoir Modelling Challenges
1.2 Exploration to Production Uncertainty
1.3 Content and Structure
1.4 What is a Reservoir Model?
1.5 The Modelling Workflow
1.6 An Integrated Team Structure for Modelling
1.7 Geostatistics
1.8 Data Sources and Scales
1.9 Structural and Stratigraphic Modelling
1.10 Facies Modelling
1.11 Property Modelling
1.12 Model Analysis and Uncertainty
1.13 Upscaling
1.14 Summary
Chapter 2: Data Collection and Management
2.1 Seismic Data
2.2 Well Data
2.3 Dynamic Data
2.4 Important Specialist Data
2.5 Conceptual Models
2.6 Summary
Chapter 3: Structural Model
3.1 Seismic Interpretation
3.2 Fault Modelling
3.3 Horizon Modelling
3.4 Quality Control
3.5 Structural Uncertainty
3.6 Summary
Chapter 4: Stratigraphic Model
4.1 How Many Zones?
4.2 Multi-Zone Grid or Single-Zone Grids?
4.3 Well-to-Well Correlation
4.4 Geocellular Model
4.5 Geological Grid Design
4.6 Layering
4.7 Grid Building Workflow
4.8 Quality Control
4.9 Uncertainty
4.10 Summary
Chapter 5: Facies Model
5.1 Facies Modelling Basics
5.2 Facies Modelling Methods
5.3 Facies Modelling Workflows
5.4 Flow Zones
5.5 Uncertainty
5.6 Summary
Chapter 6: Property Model
6.1 Rock and Fluid Properties
6.2 Property Modelling
6.3 Property Modelling Methods
6.4 Rock Typing
6.5 Carbonate Reservoir Evaluation
6.6 Uncertainty
6.7 Summary
Chapter 7: Volumetrics and Uncertainty
7.1 Work Flow Specification
7.2 Volumetric Model Work Flow
7.3 Resource and Reserves Estimation
7.4 Uncertainty Modelling
7.5 Summary
Chapter 8: Simulation and Upscaling
8.1 Simulation Grid Design
8.2 Upscaling Property Models
8.3 Work Flow Specification
8.4 Summary
Chapter 9: Case Studies and Examples
9.1 Aeolian Environments (Figure 9.1)
9.2 Alluvial Environments (Figure 9.3)
9.3 Deltaic Environments (Figure 9.4)
9.4 Shallow Marine Environment (Figure 9.6)
9.5 Deepwater Environments (Figure 9.8)
9.6 Carbonate Reservoirs (Figure 9.10)
9.7 Fractured Reservoirs (Figure 9.12)
9.8 Uncertainty Modelling
9.9 Summary
Afterword
References
Appendix A: Introduction to Reservoir Geostatistics
A.1 Basic Descriptive Statistics
A.2 Conditional Distributions
A.3 Spatial Continuity
A.4 Transforms
A.5 Lag Definition
A.6 Variogram Interpretation
A.7 Kriging
A.8 Simulation
A.9 Object Modelling
A.10 Summary
Index
End User License Agreement
Pages
xiii
xiv
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
205
206
206
207
207
208
208
209
209
210
210
211
211
212
212
213
213
214
214
215
215
216
216
217
217
218
218
219
219
220
220
221
221
222
222
223
223
224
224
225
225
226
226
227
227
228
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
307
308
309
310
311
312
Guide
Cover
Table of Contents
Preface
Begin Reading
List of Illustrations
Chapter 1: Introduction
Figure 1.1 Reservoir modelling workflow elements presented as a traditional linear process showing the links and stages of the steps as outlined in the following chapters.
Figure 1.2 Idealized evolution of resources with time over the oilfield life cycle showing the reduction in uncertainty after each stage.
Figure 1.3 Actual example of resource change during appraisal and development of an oil or gas field; appraisal often continues after project sanction as too many wells may erode project economics.
Figure 1.4 Histogram of the relative change in proven + probable estimates in ultimate recovery of oil over the period 1989–1996 for UK North Sea fields with reserves greater than 10 MMSTBOE.
Figure 1.5 The many components of a 3D reservoir model from seismic interpretation to well planning.
Figure 1.6 An example of a parallel workflow adopted by many companies for an efficient, integrated approach to reservoir modelling. The separate disciplines work together to design the model, analyse the available data and understand the uncertainties of the field under study.
Figure 1.7 The statistical analysis of 100 beach pebbles displayed as a histogram and used to describe simple statistical nomenclature; includes a normal distribution curves and tabulated results.
Figure 1.8 The scales of investigation of different types of data found in a reservoir study compared with the overall typical field size.
Figure 1.9 A matrix of horizontal and vertical heterogeneity classified by depositional environment.
Chapter 2: Data Collection and Management
Figure 2.1 Depth measurement and well path trajectory terminology.
Chapter 3: Structural Model
Figure 3.1 Examples of simple structural and stratigraphic trapping mechanisms that can be reproduced in 3D geocellular models.
Figure 3.2 Representation of the seismic response at lithological interfaces producing a reflection coefficient that may be used to generate a synthetic seismic seismogram used in depth conversion.
Figure 3.3 Example of sequence stratigraphy used in the interpretation of large seismic features typical of a lowstand to highstand cycle.
Figure 3.4 Examples of different velocity functions used for depth conversion showing the impact of additional data on the type of function applied; calibrated checkshot and VSP data are required for the detailed analysis of the overburden.
Figure 3.5 Common descriptive terminology for normal and reverse faults: the extrapolation distance is the interval the software interpolates in building a robust fault.
Figure 3.6 Modelling terminology used for the creation of horizon lines and surfaces.
Figure 3.7 Examples of horizon lines, fault surfaces and fault pillars used to edit simple and complex faulted structures.
Figure 3.8 The impact of only using development wells in depth conversion; the structure collapses because the depth conversion is based on a limited overburden distribution because offset wells are not included. In this case, the hydrocarbons in place were reduced by 30%.
Chapter 4: Stratigraphic Model
Figure 4.1 Classification and impact of different types of horizons used in modelling.
Figure 4.2 Comparison between lithostratigraphic correlation and sequence stratigraphy of the Brent Group, fluvio-deltaic system, North Sea.
Figure 4.3 Horizon, zone and sub-grid nomenclature used in geocellular modelling.
Figure 4.4 Depth reference terminology for use in well correlation.
Figure 4.5 Grid orientation and axes nomenclature used in geocellular grid construction.
Figure 4.6 Types of heterogeneity at different scales from the microscopic to basin.
Figure 4.7 Examples of different types of genetic units that can be modelled using facies object methods.
Figure 4.8 Different types of layers used in geocellular modelling.
Figure 4.9 Different layering and surface attachment methods used to represent medium-scale geological features.
Figure 4.10 Grid cell quality control examples of imperfect cells that reduce the efficiency of dynamic simulation.
Chapter 5: Facies Model
Figure 5.1 Walther's law in action; facies found adjacent laterally will be seen in the same order vertically. This example is of a progradational shoreface marine sequence.
Figure 5.2 Conceptual model of a desert setting including sand dunes, dry channels, alluvial fans and playa lake environments.
Figure 5.3 (a) Simplified lithology determination from wireline logs; natural gamma, bulk density, photoelectric and neutron porosity logs.
Figure 5.4 Upscaling or blocking raw discrete (facies) and continuous (porosity) well data.
Figure 5.5 Shifting and scaling blocked well data to correct the mismatch between raw data and model sub-grids.
Figure 5.6
Experimental variograms:
The nomenclature, orientation and application of different variogram types.
Figure 5.7 Example of an indicator model of an alluvial floodplain comprising shales (green), overbank deposits (brown) and channel sands (yellow). Note that the channel sands are not continuous across the model.
Figure 5.8 Examples of truncated Gaussian simulation using trends model progradation and retrogradation of shallow marine environments.
Figure 5.9 Examples of different object shapes that can be modelled to represent facies.
Figure 5.10 Sand body width-to-thickness measurements classified by depositional environment: data compiled from numerous outcrop sources.
Figure 5.11 An indicator simulation model of a floodplain environment with three facies: shale, overbank and extensive coal deposits.
Figure 5.12 The same floodplain model used as a background in which low sinuosity channels are modelled as objects.
Figure 5.13 Flow zones and rock types, an alternative way to describe the reservoir architecture before property modelling.
Figure 5.14 The representative elementary volume (REV) concept and scales of investigation and measurement in heterogeneous and homogenous media.
Chapter 6: Property Model
Figure 6.1 Grain size and sorting of sediments in different depositional environments.
Figure 6.2 (a) Porosity: the relationship between volume of pore space and total volume of rock is a function of grain size, sorting and packing at the time of deposition. Post-depositional processes such as compaction and diagenesis can alter the original relationship. (b) Water saturation: the proportion of the total reservoir pore volume filled with water: the remaining pore volume is filled with oil or gas, not necessarily hydrocarbon gas. (c) Permeability: the ability of a reservoir to conduct fluids through an interconnected pore network.
Figure 6.3 (a): Description of capillary pressure based on a simple experiment of water rising in a tube; (b) description of wettability as the interaction between a surface and an adsorbed fluid.
Figure 6.3 (a): Description of capillary pressure based on a simple experiment of water rising in a tube; (b) description of wettability as the interaction between a surface and an adsorbed fluid.
Figure 6.4 Identify the point of inflection for a suite of wireline logs to determine bed boundaries and eliminate the ‘shoulder effects’; part of the blocking of continuous raw log data.
Figure 6.5 Porosity distribution: mapped, interpolated and stochastically distributed showing the increasing degree of heterogeneity in the property.
Figure 6.6 Schematic of a variogram showing the nugget, sill and range.
Figure 6.7 Examples of porosity distribution by kriging and simulation showing the greater variability in distribution away from well control in the latter realizations.
Figure 6.8 Facies-constrained porosity distribution: (a) the interpolated porosity model honours the well data but results in a smooth distribution between the wells; (b) and (c) a simple threefold facies scheme of channel, overbank and floodplain allows the porosity seen in the well to be distributed more meaningfully, capturing the rapid changes laterally in the model.
Figure 6.9 Facies-constrained porosity model showing channel, overbank and floodplain distribution.
Figure 6.10 Total versus effective porosity systems; log analysis gives total porosity including clay-bound immoveable water. Core analysis may also give total porosity depending on how the plugs have been cleaned and dried. For volumetric estimates, we should use effective properties, so we should model overburden corrected effective porosity.
Figure 6.11 An example of a typically skewed permeability distribution with approximately 50% of the observations being <20 mD.
Figure 6.12 Physics of the reservoir; representation of fluid distribution with an oil reservoir based on the relationship between water saturation, capillary pressure and the free water level datum.
Figure 6.13 Example of a water saturation model using a saturation height relationship; the free water level is identified where
S
w
= 1.
Figure 6.14 The relationship between capillary pressure, height and permeability demonstrating the impact of rock quality on water saturation.
Figure 6.15 Net-to-gross (NTG) terminology, whatever approach you take be consistent.
Figure 6.16 Total property modelling (TPM) avoids the need to apply NTG until upscaling properties for simulation.
Figure 6.17 Carbonate rock-type classification based on Lucia (1999). The example shown is from a non-vuggy dolomitic limestone and significant displacement boundaries are established at 20 and 100 µm pore throat sizes, defining the separate permeability fields.
Chapter 7: Volumetrics and Uncertainty
Figure 7.1 Scales of measurement: from core data through log, seismic and well test to demonstrate the several orders of magnitude difference between the various sources.
Figure 7.2 Standard oilfield terminology for resources and reserves.
Figure 7.3 Deterministic and stochastic reserves terminology.
Figure 7.4 Petroleum Reserves Management System (PRMS, 2011) terminology for resources and reserves.
Figure 7.5 Uncertainty in oilfield reserves classified by depositional environment.
Figure 7.6 A deterministic method for determining the low–mid–high range of hydrocarbon volumes. This might be the starting point for any volumetric exercise involving 3D modelling.
Chapter 8: Simulation and Upscaling
Figure 8.1 Upscaling of reservoir properties is dependent on sampling method, scale and region.
Figure 8.2 Grid resolution aligned to the major structural features creates a more robust and realistic faulted grid.
Figure 8.3 When a grid is aligned to the primary flow direction, a better grid for dynamic simulation is created.
Figure 8.4 The SmartModel concept promotes building the geocellular and simulation grids with the same orientation and complementary dimensions so that upscaling and down-gridding methodologies can be improved.
Figure 8.5 Numerical dispersion is created when the simulation grid requires rotation such that the flow paths between wells become distorted.
Figure 8.6 The results of upscaling porosity from the fine scale to the coarse scale retain all the primary property distribution and the same overall pore volume.
Figure 8.7 Permeability upscaling is a more challenging task and may require different averaging methods or dynamic pressure solver techniques.
Figure 8.8 Examples of two common averaging methods: (a) arithmetic–harmonic; (b) harmonic–arithmetic.
Figure 8.9 Different boundary condition that can be applied to pressure solver upscaling of permeability.
Figure 8.10 Examples of different upscaling regions: local, regional and global.
Figure 8.11 A comparison of resampling and direct sampling methods for cell centre-based and corner-based methods.
Chapter 9: Case Studies and Examples
Figure 9.1 (a) A schematic aeolian dune form showing the different elements of deposition each with potential different reservoir properties that might require rock typing and separate property distributions.
Figure 9.2 (a) SIS facies model of the Hyde Field based on the description in Sweet
et al
. (1996). (See Figure 9.1b for facies description and colours.)
Figure 9.3 (a) An annotated map of the River Indus picking out the major channel forming features of a high-energy, seasonal fluvial system. (b) A low-sinuosity, low-NTG fluvial system (sand volume ∼ 25%) with attached levee deposits. (c) Individual channel bodies are picked out as isolated bodies; only the pink-coloured bodies are well connected.
Figure 9.4 (a) The Brent Field conceptual model, based on the South Cormorant Field, UK North Sea.
Figure 9.5 These three images show how a complex deposition system like the Brent Group can be constructed in stages; (a) the use of TGSim to model facies belts representing the Rannoch–Etive sequence. The basal Broom package is in blue; (b) the use of channel objects modelled in a background of floodplain deposits created by indicator simulation; (c) the Tarbert interval is built using an SIS algorithm with a strongly N–S major direction.
Figure 9.6 (a) Shallow marine deposition takes place below wave base and results in a gradual progradation from upper shoreface to marine transition with a predictable change in grain size, sorting and clay content; (b) this is reflected in the upward coarsening profile and in the case of the Fulmar Formation, UK North Sea, a classical bow-shaped log profile; (c) these predictable characteristic are reflected in the poro–perm cross-plot, (d) where the higher energy shoreface deposit forms a separate cluster of data.
Figure 9.7 Examples of shallow marine models; (a) progradation sequence of fine sands, very fine sands and argillaceous sands modelled with TGSim to create belts; (b) additional high-energy clean shoal sands models as objects with strong linear orientation; (c) porosity model using SGS to distribute facies-specific ranges in porosity between 0.01 and 0.30 p.u.
Figure 9.8 (a) Schematic deepwater depositional model showing the variety of different potential environments from the shoreline/slope to the abyssal plain; (b) classical example of the Bouma depositional model for turbidite facies. The vertical profile controls reservoir quality, but it is seldom that all elements of the cycle are preserved.
Figure 9.9 (a) A two-zone model representing deepwater deposition. In the upper zone, the yellow sands are distributed using elliptical objects between 1000 and 5000 m long and 100 and 1000 m wide up to a sand volume of 50%. In the lower zone, a similar volume of sand is distributed using an indicator simulation method in which the major variogram direction is 5–10 times greater than the minor orientation; (b) the lower image is of a sand-rich, channelized turbidite system where the attached levees are up to 10 times the width of the channels. The background shales potentially act as vertical barriers to flow.
Figure 9.10 (a) A carbonate ramp conceptual model using sequence stratigraphy to build a series of prograding packages with episodic transgressions leading to non-carbonate sedimentation; (b) a model comprising three sequences of reservoir limestones (blue/green) and non-reservoir evaporites (purple/grey). The reservoir zones have been constructed using SIS and without trends to capture the different types of distribution, randomly mosaic (lower) and aggradational build-ups (upper). The three colours in the reservoir zones represent good-, moderate- and poor-quality rock types, each could subsequently be modelled with appropriate ranges of porosity.
Figure 9.11 (a) An example of the Lucia rock-type classification for non-vuggy dolomites based on core porosity and permeability data; (b) rock types distributed according to a well-defined facies scheme of low-energy peri-tidal to open offshore environments.
Figure 9.12 (a) Fracture density as seen in core and image data from three wells in a granite basement field. These are used to characterize fracture types for subsequent modelling; (b) the required geometry of each fracture type and how they may be modelled as vertical objects within a small prototype model.
Figure 9.13 A simple approach to structural modelling based on depth conversion uncertainty resulting in 3 deterministic models of top reservoir.
Figure 9.14 Results of a model driven workflow approach (Leahy and Skorstad, 2013); (a) shows the areas of seismic uncertainty when picking surfaces due to poor resolution; (b) shows the multi-realization surfaces having run the stochastic routines for both surface and fault uncertainty; (c) is a histogram of the range in volumetric results based on the workflow.
Figure 9.15 Uncertainty in channel geometry and orientation makes a significant difference in volumetric estimation and reservoir connectivity.
Figure 9.16 Examples of different distributions used in facies modelling to introduce a stochastic element into object geometry and orientation.
Figure 9.17 Conceptual model, uncertainty workflow and permeability distributions for three shallow marine depositional scenarios.
Figure 9.18 Time-of-flight results displayed as a function of the drained volume for each of the realizations as well as frequency and cumulative probability plots.
List of Tables
Chapter 1: Introduction
Table 1.1 List of key information required before starting a reservoir modelling project
Table 1.2 Data sources for modelling
Table 1.3 Typical list of products from a reservoir modelling study that may be requested by a peer reviewer
Chapter 2: Data Collection and Management
Table 2.1 Listing of curves and specific log names that should be defined in the modelling database
Table 2.2 List of major logs used for a CPI and available for the modelling database
Table 2.3 A list of the standard mnemonics used for core analysis data
Chapter 3: Structural Model
Table 3.1 Example of fault classification and naming procedure
Chapter 4: Stratigraphic Model
Table 4.1 The requirement for building a 3D reservoir model increases with permeability heterogeneity (1–3 orders of magnitude), fluid type and production mechanism
Chapter 6: Property Model
Table 6.1 Permeability ranges for different qualitative descriptions of permeability
Table 6.2 Conversion of laboratory capillary pressure data to reservoir conditions.
Table 6.3 Bulk volume water at irreducible water saturation as a function of grain size and type of carbonate porosity
Chapter 7: Volumetrics and Uncertainty
Table 7.1 Reserves definitions as recommended in PRMS
Table 7.3 Prospective resources definition as recommended in PRMS
Chapter 9: Case Studies and Examples
Table 9.1 Distribution of facies by reservoir zones from core and log data; these form targets for modelling
Table 9.2 Input variables for uncertainty modelling of a shallow marine deposition system
Reservoir Modelling: A Practical Guide
Steve Cannon
Principal Consultant Steve Cannon Geoscience UK
This edition first published 2018
© 2018 John Wiley & Sons Ltd
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.
The right of Steve Cannon to be identified as the author of this work has been asserted in accordance with law.
Registered Office(s)
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
Editorial Office
The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.
Wiley also publishes its books in a variety of electronic formats and by print-on-demand. Some content that appears in standard print versions of this book may not be available in other formats.
Limit of Liability/Disclaimer of Warranty
While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.
Library of Congress Cataloging-in-Publication Data
Names: Cannon, Steve, 1955- author.
Title: Reservoir modelling : a practical guide / Steve Cannon, principal consultant (Steve Cannon; Geoscience).
Description: First edition. | Hoboken, NJ : Wiley, 2018. | Includes bibliographical references. |
Identifiers: LCCN 2017056087 (print) | LCCN 2017056889 (ebook) | ISBN 9781119313434 (pdf) | ISBN 9781119313441 (epub) | ISBN 9781119313465 (cloth)
Subjects: LCSH: Reservoirs-Mathematical models. | Hydraulic structures-Mathematical models.
Classification: LCC TC167 (ebook) | LCC TC167 .C36 2018 (print) | DDC 627/.86015118-dc23
LC record available at https://lccn.loc.gov/2017056087
Cover Design: Wiley
Cover Image: (Reproduced) Courtesy of Emerson-Roxar
To all the Cannons, Nichols, Whitleys, Reeves and Watsons who\break have supported my geological studies, especially on the beach at Porthmadog and many other outcrops around the world!
Preface
This book has matured over 40 years of practical oilfield experience in mud logging and well site operations, from core analysis to sedimentology and reservoir modelling to field development: I have been fortunate to have had the opportunity to be employed in a variety of different roles for a wide range of companies and organizations. All of this has culminated in the opportunity to teach a successful course on integrated reservoir modelling, which forms the foundation of this book.
By profession, I am a geologist, by inclination a petrophysicist and I am a reservoir modeller by design. In reality, I promote the building of fit-for-purpose reservoir models to address specific uncertainties related to hydrocarbon distribution and geological heterogeneity that impacts fluid flow in the reservoir. A simple mantra for reservoir modelling, as in life, is ‘keep it simple’: we never have enough knowledge or data to rebuild the subsurface only to try and make a meaningful representation of the reservoir.
My background in reservoir evaluation gives me the experience to promote 3D modelling as a solution to most field development and production challenges as long as the question being asked is properly defined. Reservoir simulation projects are clearly designed to address specific issues, so should geological models, be it volumetric estimation, well planning or production optimization. This book is focused on the development of structurally complex, clastic, offshore fields rather than large onshore producing fields. This is largely because of the difference in well numbers and spacing; geostatistical software modelling products were developed specifically for these challenges. That the same tools have been expanded for use in giant onshore fields with a large well count has made 3D geo-modelling the tool of choice for reservoir characterization and dynamic simulation.
The person building a reservoir model can be part of a multidisciplinary team, the ideal situation in my view: or a geologist who knows how to use the software and is part of a linear workflow that starts with the geophysicist and ends with a reservoir engineer; in this case, each discipline often uses a different software product and there is minimal discussion at each stage of the process. Increasingly, the seismic interpreter can build the structural model as the first step and the geologist builds and populates the grid. Whichever situation you find yourself in, it is essential to take the rest of the stakeholders with you at each stage of the model.
The book does not promote one type of method over another or specify one commercial product above another; I am grateful to a number of organisations that have provided me with the tools of my trade, especially Schlumberger and Emerson-Roxar. My background as a consultant with Roxar Software Solutions from 2000 to 2008 defines my preference for object modelling of geological facies, rather than pixel-based methods, but in reality, the software tools available to the modeller allow a wealth of options. I would like to thank Aonghus O'Carrol, Dave Hardy, Neil Price, Doug Ross, Tina Szucs and all the people who have told me to ‘RTBM’ and play with the software. My thanks also to Steve Pickering and Loz Darmon from Schlumberger-NExT who encouraged me to develop the course and supported me during the delivery of the material to over 200 students worldwide and to Rimas Gaizutis who may recognize some of these ideas from working together in the past.
Finally, I am not an academic and this is not an academic treatise but a practical handbook. Many people will disagree with my philosophy when it comes to reservoir modelling, but when you are limited by: time, data or resources, pragmatism and compromise are the order of the day. A wise man once wrote, ‘all models are wrong, though some can be useful’ (Box, 1979).
Steve Cannon
2018
Chapter 1Introduction
The purpose of this practical guide is to summarize the procedures and workflow towards building a 3D model: the principles are applicable to any modelling project regardless of the software; in other words, this is an attempt at a practical approach to a complex and varied workflow (Figure 1.1). What we are not trying to do in this book is to build detailed geological models of depositional environments but to capture the heterogeneity due to structure, stratigraphy and sedimentology that has an impact on flow in the reservoir.
Figure 1.1 Reservoir modelling workflow elements presented as a traditional linear process showing the links and stages of the steps as outlined in the following chapters.
The key to building a reservoir model is not the software; it is the thought process that the reservoir modeller has to go through to represent the hydrocarbon reservoir they are working on. This starts with a conceptual model of the geology and a diagram of the ‘plumbing’ model to represent how fluids might flow in the reservoir. Modern integrated modelling software starts with seismic input in terms of both interpreted horizons and faults and seismic attribute data that characterizes reservoir from non-reservoir and ends by linking to dynamic simulation; the so-called seismic-to-simulation solution. I have always been concerned that geophysicists and reservoir engineers might forget the geology that actually creates their oil or gas accumulation.
Wikipedia defines reservoir modelling as ‘the construction of a computer model of a petroleum reservoir, for the purposes of reserves estimation, field development planning, predicting future production, well placement and evaluating alternative reservoir management.’ The model comprises an array of discrete cells arranged as a 3D grid populated with various attributes such as porosity, permeability and water saturation. Geological models are static representations of the reservoir or field, whereas dynamic models use finite difference methods to simulate the flow of fluids during production. You could of course construct a reservoir model using paper and coloured pencils, but analysis of that model is challenging!
Geo-modelling is ‘the applied science of creating computerized representations of the Earth’s crust based on geophysical and geological observations.' Another definition is ‘the spatial representation of reservoir properties in an inter-well volume that captures key heterogeneities affecting fluid flow and performance.’ However you define it, geo-modelling requires a balance between hard data, conceptual models and statistical representation. Whether you are working on a clastic or carbonate reservoir, the workflow is the same, although the challenges are different: in carbonate reservoirs, characterizing the petrophysical properties properly is paramount because diagenesis will usually destroy any primary deposition controls on reservoir quality. We will look at carbonate reservoir characterization separately.
A few key statements should be made at the outset:
•
Every field is unique and therefore has different challenges
•
Every challenge will have a unique solution
•
Every solution is only valid for the given situation and therefore …
•
KEEP IT SIMPLE …… at least to begin with.
1.1 Reservoir Modelling Challenges
Building a model of an oil and gas reservoir is complex and challenging as much because of the variety of data types involved as the many different steps required. The process is made easier if you can establish why you are building the model; what is the objective of the model? Today, we generally build 3D geocellular models for volumetric estimation, dynamic simulation, well planning and production optimization or to understand the uncertainty inherent in any hydrocarbon reservoir. Above all, a successful 3D model aids in the communication of concepts and the interpretation of data used to characterize a potential or producing oil or gas field.
We model reservoirs in 3D because nature is three dimensional and because the reservoir is heterogeneous and we have restricted opportunities for sampling. Additionally, to understand flow in the reservoir, we need to consider connectivity in three dimensions, rather than simple well-to-well correlation. Having built a 3D representation of the reservoir, it can be used to store, edit, retrieve and display all the information used to build the model; in effect, a model is a means to integrate data from all the subsurface disciplines, so the data are not just stored in the minds of geologists.
Reservoir modelling is also a challenge because we are dealing with a mix of geological and spatial properties and also the complex fluids present in the reservoir. The data available to build a representative model are generally either sparse, well data or poorly resolved, seismic data. The resulting model is dependent on the structural complexity, the depositional model, the available data and the objectives of the project. Building a usable reservoir model is always a compromise: we are trying to represent the reservoir not replicate it.
The advances in computer processing power and graphics over the past 20 years has meant that geoscientists can build representative models of a reservoir to capture the variability present at all the appropriate scales from the microscopic to the field scale. However, as reservoirs are complex, we need to be highly subjective about the scale at which we model and the level of detail we incorporate: a gas reservoir may well be a tank of sand but faults may compartmentalize that tank into a number of separate accumulations.
1.2 Exploration to Production Uncertainty
Even before the first exploration well is drilled on a prospect, a geologist will have estimated the likely volume of oil or gas contained in the structure; by comparing one field with another, a reservoir engineer may even have estimated a recovery factor. The volume estimated will have an upside and a downside to provide a range of values. At this stage, the volume range may have been calculated using deterministic or probabilistic methods, or a mixture of both, and will generally be quite a large spread. At each stage in the life cycle of the field, the median value and the range should ideally change in a predictable way as uncertainty is reduced through appraisal drilling and data acquisition (Figure 1.2). When there is sufficient confidence in the estimates, a development decision is made and a project can begin to spend real money! In reality, the evidence from many different field developments is that the ranges in volume are always being revised to account for new data, new ideas or new technology. This often has the effect of delaying the timing of decision-making, especially in smaller fields, where the risk of getting it wrong can have a bigger impact on value (Figure 1.3).
Figure 1.2 Idealized evolution of resources with time over the oilfield life cycle showing the reduction in uncertainty after each stage.
Figure 1.3 Actual example of resource change during appraisal and development of an oil or gas field; appraisal often continues after project sanction as too many wells may erode project economics.
In 1997, the UK government commissioned a survey to review the track record of field development in the UK sector of the North Sea with the aim of determining how reserves estimation changed from the time of project sanction to field maturity. Fields that were sanctioned between 1986 and 1996 and containing greater than 10 MMBOE were reviewed. This was done to establish the confidence the operator had in the estimated ultimate recovery they reported, the methods adopted to make the estimation and the major uncertainties in their estimation.
Until 1996, 65% of respondents said that they used deterministic methods to provide a single value with explicit upside and downside ranges; 53% reported using Monte Carlo/parametric methods, and 30% said that they adopted probabilistic estimation using multiple geological or simulation models: several companies use a mix of all the methods, which is why the percentages add up to more than 100% (Thomas, 1998). In the same survey, 30% of the respondents said that gross geological structure accounted for much of the uncertainty in the estimation of ultimate recoverable reserves; the remainder believed that the reservoir description accounted for the uncertainty. For fields under appraisal, the level of uncertainty was greater than those in production and that, in general, the estimates tended to be pessimistic rather than optimistic.
While analysing the results of the survey, it became apparent that reserves estimates have varied by plus or minus 50% in more than 40% of the fields after project sanction; this was particularly true for fields where the estimates were based on deterministic models (Figure 1.4). The economic impact on field development cannot be ignored; 60–80% more wells were required to achieve the reported reserves estimates, together with very expensive retrofitting of equipment on offshore platforms. Many of the fields surveyed were compartmentalized, fluvio-deltaic reservoirs of the Brent Province that required significant additional investment over their lifetimes. With the increased use of 3D geocellular models over the past twenty years, one would anticipate an improvement in the estimation of in-place hydrocarbon volumes and the ultimate recovery of reserves. We only really know the ultimate recovery from a field when it has been abandoned, and even then someone might try to redevelop a field when the economic situation is improved.
Figure 1.4 Histogram of the relative change in proven + probable estimates in ultimate recovery of oil over the period 1989–1996 for UK North Sea fields with reserves greater than 10 MMSTBOE.
Source: Thomas (1998). Copyright 1998, EAGE publications.
It is important to realize that all fields are unique and that it necessary to understand the geology, the reservoir fluids, the data available and the development scenario proposed to maximize the economic return of the discovered resources. Resources only become reserves when an approved field development plan is in place and the money required to meet the development costs has been committed. It also helps if all the stakeholders, operator, partners and government bodies are in agreement with the objectives of the proposed development, not always a gimme!
1.3 Content and Structure
The chapters in this guide follow a general workflow from data QC and project management, structural modelling, facies and property modelling to upscaling and the requirements for dynamic modelling. Throughout, the recognition and handling of uncertainty at all stages is stressed, with some attempt made to deal with the issues.
•
Chapter 2
reviews the basic data required to build a reservoir model and how to set up, QC and manage the project database.
•
Chapter 3
discusses structural elements of a model including a review of the seismic interpretation and depth conversion, which are known to contribute most to volumetric uncertainty.
•
Chapter 4
looks at the internal reservoir architecture and how the large-scale stratigraphic relationships are integrated into the reservoir framework.
•
Chapter 5
takes a look at facies modelling; the different methods and the need for understanding the geological interpretation of cores and logs.
•
Chapter 6
is focussed on modelling the main reservoir properties; porosity, permeability and water saturation. Carbonate reservoir description is also covered in this section.
•
Chapter 7
discusses the role of uncertainty analysis in the static model and its impact on volumetric estimation.
•
Chapter 8
looks at upscaling both the structure and the properties of a fine-scale geological model for dynamic simulation.
•
Chapter 9
addresses some typical approaches to modelling specific reservoirs through a mix of case studies and examples.
This book is written from the perspective of a specialist reservoir modeller embedded in a subsurface asset team. As such, the modeller is responsible for taking the interpretations of each of the other specialists and melding them into a successful model to be used for volumetric analysis, dynamic simulation, well planning or production optimization. The key to this is being prepared to make compromises at almost every stage: the seismic interpreter should identify all the potential faults, and the sedimentologist should record all the possible facies, but each member of the team should be aware that not every detail can or should be modelled.
Some of the essential information needed to understand a reservoir for modelling is presented in Table 1.1: this also displays the multidisciplinary nature of a reservoir modelling project. All the subsurface disciplines are involved in defining what the project might achieve in terms of understanding a new field development and what uncertainties may still remain, whereas the petroleum engineers require volumes and production profiles to design the appropriate facilities to handle the hydrocarbon throughput.
Table 1.1 List of key information required before starting a reservoir modelling project
Drive mechanism
Fluid expansion, solution gas, aquifer and so on
Reservoir fluid
Dry gas, condensate, light oil, heavy oil
Reservoir framework
Normal faults, rollover anticline, thrust/slide
Reservoir architecture
Single tank, multiple stacked, compartmentalized
Trapping mechanism
Structural, stratigraphic, diagenetic
Depositional environment
Clastic alluvial, deltaic, marine; carbonate ramp/reef
Reservoir conditions
HPHT, LPHT, normally pressured, aquifer support
Data types and coverage
2D, 3D, 4D seismic; wells, logs, cores; pressure data
Development scenario
Offshore-fixed platform, FPSO, subsea tieback; onshore well spacing, storage tanks, pipeline
One key element that is often neglected is the impact of imperfect or missing data; the known unknowns and the unknown surprises that manifest themselves as a field undergoes development; usually, the results of inadequate appraisal. It might be incomplete seismic coverage or absence of core data with which wireline log properties can be calibrated; these types of unknowns can compromise any meaningful estimate of in-place volumetrics. Hopefully, as a project moves through appraisal towards sanction, the uncertainties are reduced through focussed data acquisition – but not always: beware of the asset manager who believes that additional data acquisition erodes the value of a project; this usually means that the project is marginal at best.
1.4 What is a Reservoir Model?
A reservoir model can be a series of two-dimensional maps and well correlations, an inverted seismic volume defining the distribution of lithology and fluids in a section or a three-dimensional geocellular grid that combines all of the well and seismic data (Figure 1.5). However the reservoir is represented, the ultimate objective is to describe the type and scales of heterogeneity that affect fluid distribution and flow in the subsurface. The value of any model is dependent on the data available to build it and the ability to correctly interpret that data: not an easy task! A useful reservoir model is a balance between hard data, conceptual models and the statistical representation of the subsurface.
Figure 1.5 The many components of a 3D reservoir model from seismic interpretation to well planning.
Source: Reproduced with permission of Emerson-Roxar.
Six reasons for 3D modelling:
•
Incomplete information about dimensions, architecture and variability at all scales
•
Complex spatial disposition of reservoir building blocks or facies
•
Difficult to capture variability in rock properties and the structure of variability with spatial position and direction
•
Unknown relationship between rock property value and volume of rock for averaging (scale)
•
Relative abundance of static (point values for
k
,
ø
,
S
w
and seismic data) over dynamic reservoir data
•
Convenience and speed.
What do we actual mean by the word ‘model’? A geological model attempts to represent the relationship between different depositional elements of a stratigraphic sequence at a variety of different scales. This may be a conceptual model that seeks to capture the broad picture of our understanding in terms of the depositional environment or some hierarchical relationship between model parameters such as porosity and permeability. Mathematical models may be deterministic or stochastic depending on whether the input data can be associated with some degree of uncertainty: each deterministic model has only one realization, whereas a stochastic model may have many realizations, each honouring the statistical input and range of values. Models can be map-based (two dimensional) or grid-based (three dimensional). In the former, properties are varied only in the x and y directions using a number of different mapping algorithms to distribute the outcomes. Grid-based models also vary in the z direction (depth) as defined by the geometry of the geological structure.
A reservoir model should be built to answer a specific aspect of the subsurface that impacts on hydrocarbon distribution or fluid flow. When designing a reservoir model, the ultimate purpose of the model must be defined: should it address the structural uncertainty of a prospect, the distribution and connectivity of reservoir units or perhaps the definition of a number of infill well locations? Each of these challenges will require a different approach and functionality; however, the key will be flexibility of both the ideas and the solutions that are generated by the modelling team.
To this end, the reservoir model should address the following:
•
Reservoir envelope: top and base structure
•
Reservoir compartments: fault geometry
•
Internal framework: correlation scheme
•
Reservoir architecture: facies model
•
Petrophysical property distribution
•
Volumetric assessment
•
Retention of relevant fine-scale detail through upscaling.
Each step in the workflow or each phase of a study should have an agreed deliverable that defines the potential functionality of the model at that stage. The following uses are typical for a 3D model:
•
Geoscience database and validation of input data
•
Visualization of a mapped structure and associated 2D property maps
•
Visualization of seismic attribute data and analysis for reservoir property modelling and architecture
•
Deterministic 3D property model and volumetric assessment
•
Integration of conceptual model with core constrained 3D facies model
•
