Copyright 2004, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the SPE Asia Pacific Conference on Integrated Modelling for Asset Management held in Kuala Lumpur, Malaysia, 29-30 March 2004. This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836 U.S.A., fax 01-972-952-9435.

Abstract An integrated reservoir study represents the one reservoir modeling technique available to a reservoir management team that incorporates ALL of the information available regarding a petroleum reservoir. As such, the integrated study has the potential to provide the team with the highest resolution most accurate description of their field that is currently available. However, that high resolution comes at the expense of a highly complex, data intensive process. This paper was an attempt to simplify and codify the complex process of performing an integrated reservoir study.

The paper presents an overview of the integrated reservoir modeling process including how that process has changed from the early days of reservoir simulation to the present day. Two models are presented that illustrate the study process for a 1980 time frame multi reservoir study and a 2002 time frame geostatistically based compositional reservoir model..

Emphasis is placed on the changes in workflows from the

early models -- simple mapping of properties and manual digitization of maps to simulation model grids -- to the more complex models of today -- with property distributions based on object modeling and geostatistical analysis to distribute reservoir properties.

Through the examples, the paper illustrates the point that

reservoir studies have evolved from a time when the geologists and engineers often knew more about their reservoirs than the available modeling tools would allow them to implement in their models – dual porosity and flow across faults, for example – to the point where today we often have the modeling capability to model more complex phenomena than our knowledge of the reservoir may warrant – for example, object modeling with no data on the characteristics

of lithofacies in our reservoir, or dual porosity flow models with no data on fracture spacing and distribution.

The Integrated Study Conceptually Although the term “Integrated Study” only gained widespread use in the 1990’s 2,3,4, the conceptual tasks needed to build a model of a reservoir have been identified and performed since the advent of multi-well full field simulation studies. Those tasks -- structural interpretation, petrophysical analysis, stratigraphic analysis, fluid PVT analysis, and reservoir simulation – have formed the basis for building models since the 1970’s and continue today to be the backbone of the “Integrated Study” concept.

What has changed in the Integrated Study concept is:

• The switch from the analog to digital form for the data input and analysis results of the “geo-science” tasks of the studies

• The development of ever more mechanistically sophisticated analysis tools, for example

• Object modeling in facies identification work • Geostatistical methods in distributing

petrophysical properties inter-well, • Arbitrary connections in simulators to allow

modeling of faults and fractures • Many orders of magnitude faster computers and

greater data storage capacity

As a result of the sophisticated tools and greater computing capacity, the focal point of all of the model building process has moved from the reservoir simulation ”dynamic” model in the integrated study concept of the 1980’s, Figure 1, to the geologic “static” model today. This movement of the focal point is, we believe, a major reason for the greater interaction between the disciplines and tasks that is the emphasis of the 2003 integrated study concept, Figure 2. Integrated Study Tasks All of the disciplines identified in Figure 2 that contribute to an integrated study have specific tasks that MUST be performed if the Integrated Study is to be successful. They also have a timing interdependence so that the results of one

SPE 87032

"State-of-the-Art" Integrated Studies Methodologies - An Historical Review Vernon S. Breit SPE, Joe A. Dozzo SPE - International Reservoir Technologies, Inc.

2 SPE 87032

analysis – such as production well test analysis – will be available for other tasks – such as stratigraphic analysis. The specific tasks for each discipline in a typical Integrated Study are detailed below and their inter-relationships in time are shown in the accompanying Gantt chart, Fig. 3. Data Review Static Model

Geophysical analysis Calibrate seismic to wells, vsp’s, and synthetics Horizon picks – key third order surfaces Fault interpretation including well ties Structural analysis integrated with wells and dip meter Complete fault-juxtaposition studies

3D Velocity modeling Generate 3D velocity model – time/depth tied to wells

Sedimentology Core description on 1 or 2 wells Core/log comparisons Interpretation of depositional environments Map depositional environments for reservoir units Condition geo-statistical object models

Sequence Stratigraphy Construct a comprehensive grid of cross sections

Structural Stratigraphic

Integrate with seismic data Integrate with petrophysics and engineering contacts

Petrophysics Digitize all appropriate log traces Log editing, environmental corrections, normalization Permeability transforms by litho-facies Depth shifting of core data Core/log and log/log porosity transforms, if required Shale volume, effective porosity, water saturation

Well History Reviews by Well Review Completion data Workovers and re-completions Stimulations and blow-outs Verify bottom hole locations Locate fault cuts in logs Calculate TVD’s

Production History Analysis Decline curves by well Decline curves by reservoir area Evaluate potential coning/creating problems Pressure test analysis Build a database of pressure data Correct to datum Interpret data tests for pressure and reservoir properties

PVT Analysis Determine areal and vertical variations in properties Correlate fluid properties to flow units Develop EOS or black oil tables for use in modeling.

Relative Permeability & Capillary Pressure Analysis Establish rock types Correlate to sedimentological & petrophysical analysis Determine end points and hysteresis

Zonation and Contact Definition Correlate reservoir flow units to stratigraphy

Set contact levels by fault block and flow unit Integration of all disciplines

Material Balance Calculations by Zone Construct coarse models for material balance analysis Develop corner point grids for sloping faults Make final list for zonations in history match model

3D Structural/Stratigraphic Framework Depth convert faults & horizons using velocity model Apply residual ties to horizons& faults from well control Populate model with stratigraphic surfaces

Zone Mapping for Model Gridding Net-to-gross ratio Porosity Permeability Contacts Depositional environments (Facies)

Geo-statistical Modeling Determine methodology to quantify depositional model Assess existing data quality and distribution Calculate geo-statistical property distributions Condition distribution of log-derived parameters

With seismic attributes, Well test analysis, etc.

Sensitivities of OOIP to geo-statistical realizations and contact uncertainty

Original Oil-In-Place Calculations Determine OOIP by zone, and by fault block Production allocations by flow unit Recovery to-date by flow unit

Simulation Model Model Construction

Up-scaling of porosity and net-to-gross Up-scaling of permeability Grid-size sensitivity analysis

History Matching to the Well Level Match overall field pressure levels to total fluid withdrawals Match overall field w-cut and GOR by changes in

Relative permeability, PVT and Overall Kv /Kh ratios

Match individual well responses by localized horizontal permeability changes and vertical permeability changes

Prediction of Future Performance Well productivity tuning Well-bore and surface facilities model set-up Multiple prediction cases

Reserve and Costs Analysis for Each Case Studied

Economic Analysis of all Cases

Final Report in Printed and HTML Electronic Form

SPE 87032 3

Detailed Task Dexcriptions Phase I - Data Review The first phase of the Full Field Review should be a thorough multi-disciplinary review of both the quantity and quality of all seismic, geologic and engineering data available.

METHODOLOGY A multi-disciplinary team consisting of geoscientists, reservoir, and facilities engineers evaluates the data available in digital form. Further, the team should identify key data requiring conversion to digital form and refine the initial estimate of the time and cost required to convert the requisite data for further use.

The team will assess the status of the current 3-D seismic volume, determine if inversion of the existing volume would be cost effective and/or if further reprocessing is required. The team reviews the petrophysical data available digitally as well as the availability of special and conventional core data in digital form to calibrate a petrophysical model. Finally, assess the engineering data available in digital form for the field to build a static model and conduct a simulation study. These data include fluid properties, pressure measurements, well tubing configurations, production and perforation/workover histories. The facilities engineer reviews field operations to identify potential bottlenecks and identify any opportunities to increase either efficiency or capacities.

Phase II - Static Model The objective of Phase II is to develop a geologic model of the reservoir that is of high enough resolution to be the basis for both future full field modeling efforts and high resolution mechanistic modeling efforts..

METHODOLOGY 5 This section presents a discussion of the methods and workflow for the generation of a 3-D high-resolution structural framework. The structural frame will consist of a model that contains all the fault planes and geologic layers in depth. Then the calculation of properties to fill the structural framework is detailed.

Geophysics / structural frame An integrated geologic-geophysical workflow should be used to build the 3-D structural model.. This method will be progressive in that it will advance from a coarser to finer scale.

The first and coarsest scale of modeling is at the seismic horizon level. We start with these horizons because these horizons are controlled by seismic and the stratigraphic picks in wells, they provide the most accurate structural representation. Initial seismic and well log interpretations are developed, and differences are resolved during the iterative integration process.

Figure 4 shows the workflow to integrate and perform quality control of the geophysical and geologic interpretations. There are two main QC procedures that will be used to find differences in interpretations, a visual QC of every well tie on the seismic section, then after the seismic interpretation is depth converted, the elevation differences between stratigraphic and fault picks on logs and on seismic are calculated. Then any differences greater than a predefined cutoff are determined and remedied.

Synthetics, time depth tables and velocity model An extensive integrated workflow is used to generate the final 3-D velocity model., Figure 5, for the seismic data volume over the field. This process uses all well logs, synthetics, seismic, and time-depth tables in the field. The model is initiated using the time-depth tables. These time-depth tables will then be modified during the horizon interpretation as necessary, starting in the shallow horizons and working down. The resulting adjusted time-depth tables will be used to create the 3-D velocity model.

Fault interpretation If the faults are high angle, and no fault picks need to be honored in the wells, a vertical fault model may be built in the interests of time and budget. Alternatively, if the exact fault geometry is determined to be critical to modeling reservoir pressures, fluid flow or individual well performance, the study will use a complex workflow, Figure 6, that utilizes a 3-D mapping package to model faults. The final product is an integrated fault model that exactly honors the faults picked on logs.

Interpret selected horizons

Integrated geologic-geophysical horizons provided a coarse structural frame for the overall horizon model. Figure 7 shows the workflow to construct the integrated horizons. Shown in the figure are the visual and the digital QC processes that will be used to verify that the interpretations and velocity model are valid. The horizons will be interpreted in all usable well logs and in the 3-D seismic volume in the field. Stratigraphic units will be tied to previous historical studies and correlated in well logs. The correlation is based on sequence stratigraphic concepts. Key horizons will be picked in the seismic data. The integration of these horizons with stratigraphic picks on the well logs will be validated with synthetics. The integration of the geological and geophysical horizon interpretations will provide an ongoing quality check. To the extent possible, the geologist and geophysicist will work concurrently on the same horizons to maximize interaction and enhance quality of each other’s efforts. Additionally, isopach maps will be constructed periodically to ensure that trends and thickness variations are reasonable.

Petrophysics / Log Analysis Build and QC log database

Corrected log curves should be converted to LAS format. Where unedited logs exist, these curves will be checked and any errors noted. Gamma Ray curves will be normalized and the normalized curves used in developing the stratigraphic model.

4 SPE 87032

Add production logs to database and analysis Cased-hole logs may add important value to saturation calculations, fluid contact tracking and monitoring well / casing integrity. All pertinent cased hole logs will be added to the logging database.

Facies model, correlate core phi-k to facies

Facies identified in core will be compared to log data to determine log signatures of each facies. Further, core porosity versus permeability correlations will be investigated to see how / if they vary by facies.

Build & calibrate petrophysics model to core data, run

model in all wells A petrophysical model will be developed using all the wells with both wire-line logs and core data. Corrected wire-line logs will be analyzed for porosity, net sand thickness and permeability. The resulting values will be used during the property-mapping phase of the project.

Build HTML log displays

An HTML based tools share up-to-date data and interpretations to all team members. The tool we construct is typically called the Data-Viewer and contains log displays, maps, production plots, text and tables. The Data-Viewer provides a highly interactive user-friendly environment for viewing and comparing project data and is an invaluable aid to all phases of a project.

Geology/Stratigraphy Build stratigraphic model

Integrated geologic-geophysical horizons provide a coarse structural frame for the overall horizon model. The horizons are interpreted in all usable well logs and in the 3-D seismic volume in the field. Stratigraphic units should be tied to the historical studies and correlated in well logs. The correlation is based on sequence stratigraphic concepts. Conceptual depositional models will be developed using observations from core descriptions, log character, and available literature. Key horizons will be picked in the seismic data. The integration of these horizons with stratigraphic picks on the well logs will be validated with synthetics. The integration of the geological and geophysical horizon interpretations will provide an ongoing quality check..

Review correlations for consistency with engineering data

Just as significant effort will be devoted to ensure consistency between the geophysical and well log derived horizons, appropriate effort must devoted to integrate stratigraphic horizon picks and geologic layers with engineering data. RFT’s static pressures and pressure build-up data can all be used to help validate the horizon picks and geologic layer definition.

Build structural model

As the development of horizons and structural frame from the geo-science and stratigraphic team members converges, a final integrated structural model will be constructed. This final integrated, internally consistent and quality checked model will be the basis for the high resolution 3 dimensional

static model to be delivered at the end of Phase II.

Map properties (true vertical thickness, net to gross, porosity, permeability) Property maps will be generated for all geologic layers. The properties will be deterministically mapped honoring the conceptual depositional models developed for each layer. The properties will include: geologic layer true vertical thickness, depositional facies if appropriate, net to gross, porosity and permeability.

True vertical thickness maps will be generated for each geologic layer. The structural mapping of each geologic layer will be a complex iteration between using the seismically derived structure maps as a guide for the infill geologic layer structures and isopach maps. The resulting structural framework will tie all stratigraphic picks in wells

Depositional facies maps will be created for each of

the geologic layers. Basis for input for these maps will be the existing core descriptions within this study area. Facies identified in core will be compared to log data to determine log signatures of each facies and deterministically mapped.

Net/gross ratio values for each well in each geologic layer

will be calculated using the interpreted petrophysical curves and the stratigraphic picks for each geologic layer. . The product of the net/gross sandstone maps and interval true vertical thickness maps produce the net sandstone maps required for simulation. These maps will also provide rock volumes that are permeable to reservoir fluids, regardless of structural position.

Porosity and permeability will be determined from the

logs at all wells, correlated to facies or reservoir quality, and distributed stochastically between wells using geo-statistical software. Porosity and permeability will be distributed within the framework of the true vertical thickness, net-to-gross and facies maps.

Contact review and definition Fluid contact definition is essential for both a correct OOIP determination and an accurate history match during the reservoir simulation phase of the study. The depths to the free (100%) water zones are used directly in the model, along with J-function relationships, to determine the heights of the oil-water transition zone in each grid cell. Well logs and engineering data should be used to identify the fluid contacts. RFT pressures, and relative differences in observed contacts will be used to identify vertical seals that isolate groups of geologic layers within fault blocks.

Basic Production and Reservoir Engineering Build well histories

A production and completion database will be generated from well and production reports. Perforation and sleeve history data are entered into database for retrieval during electronic reporting and for use in preparing the history match recurrent

SPE 87032 5

data. Perforation and pressure information is also entered into the database and is plotted along with wire-line logs to be used to identify compartmentalization and behind pipe reserves.

Reconcile monthly production to well histories

An independent quality check is performed on the monthly production data and perforation / workover histories developed from the well files. This step ensures all well activity is captured for use in the simulation model.

Production log analysis All available production logs will be included in the full field review. During this phase of the study, the production logs will be used to help identify fluid contacts and their movement over time. Further, the production logs will be examined for evidence of perforation / casing integrity and identification of potential behind pipe opportunities.

Review and mapping of static pressure data

All static pressure data will be reviewed, quality checked and datum corrected for used in identifying hydraulic compartmentalization within the geologic model. These data will also be entered into the database for use in the history match.

RFT/MDT– all RFT/RIF/MDT

Pressure data will be reviewed, quality checked and datum corrected for used in identifying hydraulic compartmentalization within the geologic model. These data will also be entered into the Schedule database for use in the history match.

DST and PBU analysis

All static pressure data will be reviewed, quality checked and datum corrected for used in identifying hydraulic compartmentalization within the geologic model. If sufficient digital rate, time and pressure data are available, updated analyses will be conducted and the resultant KH’s will be used to help independently validate the up-scaled Integrated 3D Static Model simulation layer KH’s.

PVT data review and basic analysis

The PVT sample reports will be analyzed and prepared for use in all engineering calculations, including reservoir simulation.

SCAL data review and basic analysis Special core data - relative permeability and capillary pressure - are used to provide engineering data necessary for reservoir simulation is performed during this phase. These data, along with analysis of open and cased-hole wire-line logs, will also be used to determine displacement endpoints and efficiencies.

SCAL lithofacies correlations

Additionally, the special core analysis data will be evaluated as part of the facies model to ensure displacement efficiencies are modeled properly throughout the reservoir.

DCA and material balance studies During Phase II of the Study, conventional engineering calculations will be performed to establish initial estimates of both ultimate and remaining reserves based on updated In-Place volumes from the new Integrated 3D Static Model.

Integrated 3 Dimensional Static Model Create a 3-D modeling grid based on fault model and

stratigraphic horizons The 3-D modeling grid is the framework in which the reservoir properties will be distributed. First a 3-D fault model will be constructed by reconciling all intersections among fault surfaces and horizons. A 3-D grid is then designed with an appropriate layering scheme and resolution to capture the geologic variability.

Establish facies proportions from well data & facies maps The distribution of reservoir facies is performed with a geo-statistical technique known as object modeling. For each geologic layer, global facies proportions will be determined from the well observations. Facies maps will be generated to examine regional trends and global facies proportions may be adjusted if maps reveal a sampling bias. The facies maps will be used in the object modeling to control the lateral distribution of facies within a given geologic layer.

Create vertical trends from well data & relative depth

The vertical distribution of facies within a geologic layer will be examined from the well observations. Facies proportions are evaluated as a function of depth, either relative to the top and base of the geologic layer or relative to a reference horizon. These vertical trends are calculated for each facies for each zone and used to control the vertical distribution of facies.

Generate well statistics and establish object geometry In object modeling, the facies are represented as geometric objects, which represent geologic bodies. A conceptual model will be developed to describe the geometry (size, shape, orientation) of the facies bodies. These geometric parameters are generally specified as probability functions that describe the range of possible values. Insight for the conceptual model can be gained by examination of the well logs and cross-sections. Analogue data may also be used.

Perform object-based facies modeling The facies will be distributed within each zone using an object-modeling algorithm. The algorithm honors the well observations, facies trends, global facies proportions, and geometric parameters. The resulting 3-D facies model will describe the primary reservoir architecture within each geologic layer.

Perform porosity, permeability and saturation modeling The facies model provides a template for distributing petrophysical properties. The petrophysical properties will be mapped within each facies using a geo-statistical technique called sequential Gaussian simulation (SGS). The SGS algorithm honors well observations and specified variograms. The variograms, which describe the spatial variation of a

6 SPE 87032

given property, will be interpreted from the log data. Correlations will be developed for initial and residual fluid saturations as a function of rock quality and depth utilizing SCAL data, where available. The correlations will be applied to the property arrays in the static model to estimate the original fluid distribution and the saturation end-points.

Upscale fine grid static model for use in history match Once completed, the static model will be used in the initialization of the history match model. Properties in the static model will be up-scaled to the flow simulation grid using techniques that preserve the geologic heterogeneity.

Phase III - Dynamic Model Phase III is the development and history matching of a full field model and execution of various prediction cases to optimize a future operating strategy. Typically, the history match and prediction cases are the exclusive domain of production and reservoir engineers. The continued tight integration of the static and dynamic models is required to achieve the highest quality history match and maintain the integrity of all the underlying data analysis.

METHODOLOGY Fluid properties for reservoir modeling

The fluid samples from the field will be modeled and EOS package. The PVT sample analysis will be reviewed and the best samples will be used in the theoretical modeling. Laboratory measured properties will be used to tune the equation of state describing the reservoir fluids. The EOS software will then be used to generate the PVT properties for reservoir simulation. The oil, water and gas properties will be used in all calculations and for all phases of project work when fluid properties are required.

Relative permeabilities for reservoir modeling Existing SCAL data from the field is reviewed to determine the endpoints under water displacement and gas displacement drive mechanisms. Water-Oil and Gas-Oil relative permeability tables can be constructed for reservoir modeling using various Corey exponents as sensitivities. An end point rescaling option can be used in the simulator to handle sensitivities in relative permeabilities. Relative permeability endpoints may be correlated separately by facies if needed.

Build model and initialize When the geological structural model is complete, the reservoir geometry model can be constructed, Figure 8. The advantage of a full field model will be that accurate and proper constraints can be applied at the well, platform and field levels in the model.

Fault planes and their throws will be incorporated in the

simulation model via the 3 dimensional structural framework. Layer thickness will be determined after the completion of the facies modeling. With current hardware and software capabilities, black oil reservoir simulation models of up to 500,000 cells can be handled in a timely manner. Properties from the 3D static model will be up-scaled and input to the

simulation model. This will include porosities, net to gross ratios, permeabilities and water saturations.

Build simulation rate deck The production history of field should be updated and then output in a suitable format for a deck building software such as SCHEDULE. Workover histories will also be incorporated here.

History match model The dynamic reservoir simulation model will be used for history matching after initializing and all parameters in the model checked and verified. The model is pressure matched first and then the matching of the well water cuts and gas-oil ratios follows. An important factor in the history match can be the accounting of extraneous fluid flow between horizons behind pipe. Therefore, accurate and proper accounting of the workover history in each well will be important.

Construct and run prediction cases When the history match is complete, the model will be used for future prediction runs. An initial prediction case will be defined and liaison with and feedback from field personnel will be required to define the initial prediction cases and any alternative exploitation schemes to be analyzed.

Integrated Study Examples 1980 Model The study objective was to determine whether gas injected in one reservoir has migrated to a second reservoir in the Cretaceous reservoirs in the study area. A geologic model based on a “bird foot delta” depositional model, Fig. 9, was based on the self-potential and resistivity log response curves. In the resulting geologic model the bar fingers and the channel fill sands each have their preferred grain-size gradation with a logical relation to shaliness, saturation, porosity, and permeability. This model exhibited a strong change in rock properties laterally form the axis of the channels, Fig. 10. Several faults were indicated in the area and the faults appeared .to be sealing and separate the reservoirs into local traps. Well log and/or core data was available on 201 wells; that data was normalized and log porosities adjusted to match core values. Permeabilities for the modeling effort were derived from a core porosity/permeability relationship based on core data from 61 wells. All the geologic and petrophysical data analysis was incorporated into 18 hand drawn maps of structure, sand thickness, porosity and saturation.

The geologic model showed that two sand bodies were present in most but not all areas. The two sands were separated by shales in most locations with a very few exceptions. A simulation model of the two sand bodies was built using the geological model’s value of net thickness, structure, porosity and water saturations as a starting point. The geologic properties were transformed from the hand drawn maps to a simulation grid by hand digitization, Fig. 11.

SPE 87032 7

Vertical permeabilities were set to zero in all areas except for those where the two layers were in vertical contact – see gray areas on the grid map.

The resulting simulation model contained 2244 grid cells. Two models were built, one with flow between reservoirs and one with sealing faults between reservoirs. Following a lengthy history match, during which horizontal and vertical permeabilities were manually adjusted, the model was able to satisfactorily reproduce the observed pressure behavior of the reservoirs, Fig. 12, and the oil, water and gas production at the field level but not at the well level.

The time necessary to perform the structural interpretation, stratigraphy, petrophysics, and simulation described was 1.5 years. 2002 Model An updated integrated study of an Eocene age reservoir was performed in 2001 and 2002. The reservoir had a depositional environment that was interpreted to be a tide-dominated deltaic system not unlike the depositional environment in the 1980 model discussed earlier in this report. The study was undertaken to identify opportunities to increase oil production and locate additional reserves. Specifically, the study was to recommend improved recovery techniques to maximize the field's potential. The study followed the study methodology also discussed earlier in the report.

Petrophysical modeling had been performed on the first 80

wells in a previous study. A review of those results showed that modeled total porosity was essentially effective in clean sands, however, a V-shale correction needed to be applied in shalier sands. When calibrated to pressure-corrected core porosities, the Steiber Vshale method worked best as a correction.

Nine flow units were correlated in 95 wells in the field.

Eight flow units were mapped across the entire field. The correlations were based on maximum flooding surfaces and pressure barriers and included sands of similar petrophysical properties, but were independent of individual facies. They were not correlated based on solely depositional environments that may lump sands of differing petrophysical properties or flow behavior within a single well. Facies maps were made to ensure the proper distribution and orientation of sand bodies as well as assigning the proper petrophysical properties within those sands.

Facies were interpreted for each well from the log curves. Deterministic facies maps reflecting a tide-dominated environment and honoring regional depositional trends were made for each flow unit.

The structural framework consisted of nine structural meshes, one on top of each flow unit along with the bottom of the reservoir porosity. Faults were modeled as having consistent dip and slip vectors. In general, the seismic quality in un-deformed regions was good, but the entire volume is under-migrated, which made mapping exact fault locations and correlating seismic horizons near faults difficult. The

resulting structure is shown in the view of the simulation grid, Fig. 14. Once stratigraphic correlations were complete, the picks were posted on the 3D seismic volume to ensure that stratigraphic continuity and faults were honored.

A detailed three-dimensional full-field geologic static model was built to characterize the structural and petrophysical parameters controlling fluid flow in the reservoir. A hybrid stochastic/deterministic method for distributing reservoir facies and petrophysical properties was developed to facilitate the conditioning of the model to a large number of wells (91). Modeling regions were defined from traditional facies maps and an object-based facies model was developed for each region. The technique also allowed inter-fingering at facies boundaries. The model was further constrained by vertical proportion curves, which controlled the vertical distribution of facies within a zone, Fig 13. Petrophysical properties were then distributed within each facies using a sequential Gaussian technique. There were approximately 8.6 million cells in the model in 315 layers. The modeling was performed using RMS v5.1.1. Construction proceeded in three major phases: structural modeling (gridding), facies modeling (large-scale heterogeneity), and petrophysical modeling (fine-scale heterogeneity). The eight reservoir zones, representing the major flow units and flow barriers in the reservoir, were modeled independently to allow greater specificity of the modeling parameters to the depositional processes. The parameters that were varied between zones included facies maps, global facies volume fractions, vertical facies trends, sand body geometry, and direction of sediment transport.

The reservoir simulation grid followed the same reservoir

zonation (eight zones) as the geologic modeling grid. However, the areal resolution was reduced to 150m x 150m from 50m x 50m, i.e., approximately nine cells in the geologic grid for every one cell in the simulation grid. The vertical resolution was also reduced such that as few as four or as many as twenty-five layers were in the fine-scale model for each layer in the coarse-scale model. The resulting simulation model contained approximately 111,000 cells, Fig. 14.

The up-scaling from the 8.6 million cells in the static

model to the 111,000 cells in the simulation exercise targeted four effective properties: porosity, net-to-gross ratio, horizontal permeability and vertical permeability. For the calculation of porosity and net-to-gross, a net-reservoir parameter was defined such that it was equal to one if porosity ≤ 6% and otherwise equal to zero. The effective porosity was then calculated as the arithmetic average porosity for the net-reservoir cells. The net-to-gross ratio was then calculated as the (bulk-rock) volume-weighted average of the net-reservoir parameter. A diagonal-tensor type pressure-solver was used to determine the effective permeability. The method is aimed at finding the permeability of the homogeneous medium (coarse cell) that gives the same flux as the heterogeneous medium (fine-scale) under the same boundary conditions.

The resulting thirty-layer simulation model was

history matched to historical pressure and production

8 SPE 87032

data. The history match of observed liquid production, in the reservoir was excellent, Fig. 15.

The history matched simulation model was used to

evaluate a "Base Case" or continued current operations for 20 years. To evaluate a full field WAG in the reservoir, the history-matched model was converted to a compositional model. The compositional model was used to predict recovery for many different implementation schemes: including flank injection. Cases were evaluated using natural gas and nitrogen injection. The results of predictions for several gas injection and water injection schemes are shown in Fig. 16.

The time necessary to perform the structural interpretation,

stratigraphy, petrophysics, and simulation described was 1.4 years.

References 1. Galloway, W.E. 1975. Process framework for describing the

morphologic and stratigraphic evolution of deltaic depositional systems, In: Deltas, Models for Exploration, M.L. Broussard (ed.) Houston Geological Society.

2. Grant, I., Marshall, J.D., Dietvorst, P., Hordijk, J., “Improved

Reservoir Management by Integrated Study Cormorant Field, Block 1”, SPE Paper 20891, Europec 90, The Hague Netherlands, Oct. 1990.

3. Hampton, D.W., et. al., “An integraged Study of a Campeche Bay

Fracture Carbonate Reservoir”, SPE Paper 28672, SPE International Petroleum Conference of Mexico, Veracruz, Mexico, Oct. 1994.

4. Schildberg, Y., Poncet, J., Bandiziol, D., Deboanisne, R., Vittori, J.

“Integration of Geostatistics and well Test to Validate a Priori Geological Models for the Dyanmic Simulation: Case Study”, SPE Paper 38752, SPE Annual Technical Conference and Exhibition, San Antonio, Texas, USA, Oct, 1997.

5. Taboada, R., Condat, P., et.al., “El Tordillo Reservoir Static

Characterization Study: El Todillo Field, Argentina” SPE Paper 69660, SPE Latin American and Caribbean Petroleum Engineering Conference, Buenos Aires, Argentina, Mar. 2001.

SPE 87032 9

Seismic Interpretation

Structural Geology

Petrophysics

Stratigraphy

Reservoir Simulation

Model

FieldDevelopment

Plan

Drilling/Production

FacilitiesFluid Properties

Seismic Interpretation

Structural Geology

Petrophysics

Stratigraphy

Reservoir Simulation

Model

FieldDevelopment

Plan

Drilling/Production

FacilitiesFluid Properties

FIGURE 1: 1980 INTEGRATED STUDY SCHEMATIC

FIGURE 2: 2003 INTEGRATED STUDY SCHEMATIC

Seismic Interpretation Structural

Geology

Petrophysics

Stratigraphy

High Resolution Reservoir

Static Model

Field Development

Plan

Drilling /Prod Testing

Facilities

Full Field Reservoir Simulation

Model

Pattern Simulation

Models

Seismic Interpretation Structural

Geology

Petrophysics

Stratigraphy

High Resolution Reservoir

Static Model

Field Development

Plan

Drilling /Prod Testing

Facilities

Full Field Reservoir Simulation

Model

Pattern Simulation

Models

10 SPE 87032

FIGURE3: TYPICAL STUDY TIMELINE

SPE 87032 11

FIGURE 4: INTEGRATED GEOLOGIC/GEOPHYSICAL WORKFLOW FIGURE 5: 3-D VELOCITY MODEL WORKFLOW

Vis ual QC,Adjus t T-D Tables

Se is micData

Horizons

Faults

Dip MetersWell Logs

Faults

Horizons

Dip MetersWell Logs

Faults

Horizons

Build Veloc ityModel

Depth ConvertSe is mic Interp.

Model Se is micHorizons

and Faults

Dig ital QC ofSe is mic and Log

Interpretation

Pas sQCFail

Pas s , Tie Well and Se is mic in

Structural Model

Move to nextSe is mic Horizon

Top HorizonHorizon 2Horizon 3

.

.Bas e Horizon

Fail, Fix Interpretationsand/or T-D Tables

and repeat Dig ital QC

WellLogs

Time DepthTables

3-DSeismic

InterpretHorizon

Pick

InterpretHorizon

ObtainTime/Depth

Pairs

HorizonElevations

Time and DepthShift T-D tables,Populate wells

Tie Synthetics,IdentifyHorizon

CreateSynthetics

GridTime

DevelopVelocity-Time

Function

DeterminePsuedo well

locations

Generate andPopulate T-D tables,

For psuedo wellsInitialize 3-DVelocity Model

12 SPE 87032

FIGURE 6: INTEGRATED 3-D FAULT FRAMEWORK WORKFLOW FIGURE 7: INTEGRATED 3-D HORIZON FRAMEWORK WORKFLOW

QC differencebetween seismic

and well log faultsin Excel

SeismicFault

Interpret.

Depth Convertin TDQ

Export usingFIE

ModelFault Frame

OK?

Yes, AdjustSeismic Faultsto Honor Well

Log Faults

Increase data densityand calculate tip loop

polygons in Excel

No, Fix Interpretationsand/or T-D Tablesand redo process

BoreholeTrajectories

CalculateBorehole

Intersections

Well LogFaults

LoadIntersections

into OW

VelocityModel

QC differencebetween seismic

and well log interp.on map, Excel

SeismicHorizon

Interpert.

Depth Convertin TDQ

Export, adjustdatum, merge

surveys

OK?

Yes, adjustseismic horizonhonor well picks

Model andcorrect for survey

mis-ties

No, reviewand fix

interpretations

Calculate elevationof seismic horizon

at each pick

Loadseismic elev

into OW

Redo FaultFramework

Well LogInterpret.

Visual QC, adjust T-D Tables, rebuild

velocity model

QC differencebetween seismic

and well log interp.on map, Excel

SeismicHorizon

Interpert.

Depth Convertin TDQ

Export, adjustdatum, merge

surveys

OK?

Yes, adjustseismic horizonhonor well picks

Model andcorrect for survey

mis-ties

No, reviewand fix

interpretations

Calculate elevationof seismic horizon

at each pick

Loadseismic elev

into OW

Redo FaultFramework

Well LogInterpret.

Visual QC, adjust T-D Tables, rebuild

velocity model

g

SPE 87032 13

FIGURE 8: SIMULATION MODEL CONSTRUCTION WORKFLOW FIGURE 9: CONCEPTUAL 1980 FACIES MODEL

Production Rates

HIGH RESOLUTION

STATIC MODEL

Fault Traces Structural Grid

Structural Data

Fracture properties

Matrix properties Grid Properties

PVT DATA

Equation-of-State

Fluid-Rock PropertiesRelative permeability

capillary pressure

VFP Tables

Perforated Intervals Dynamic Model

SIMULATIO

N MO

DELSeparator Model

WELL TEST DATA

PROD/INJ DATA

SCALE-U

P

FLOG

RID

SCAL DATA

Production Rates

HIGH RESOLUTION

STATIC MODEL

Fault Traces Structural Grid

Structural Data

Fracture properties

Matrix properties Grid Properties

PVT DATA

Equation-of-State

Fluid-Rock PropertiesRelative permeability

capillary pressure

VFP Tables

Perforated Intervals Dynamic Model

SIMULATIO

N MO

DELSeparator Model

WELL TEST DATA

PROD/INJ DATA

SCALE-U

P

FLOG

RID

SCAL DATA

Wave-Dominated

Tide-Dominated

Sediment Input-Dominated

1

Wave-Dominated

Tide-Dominated

Sediment Input-Dominated

Wave-Dominated

Tide-Dominated

Sediment Input-Dominated

1

14 SPE 87032

FIGURE 10: EXAMPLE 1980 PROPERTY DISTRIBUTION FIGURE 11: EXAMPLE 1980 SIMULATION GRID

Area of CommunicationBetween Layers

Boundary of Hydro-Carbon Accumulation

Area of CommunicationBetween Layers

Boundary of Hydro-Carbon Accumulation

SPE 87032 15

FIGURE12: EXAMPLE 1980 MODEL PRESSURE MATCH FIGURE13: EXAMPLE 2002 PROPERTY DISTRIBUTION

0 5 10 15 20 25 30 35 40YEARS

PRES

SUR

E @

DA

TUM

, PS I

G

1700

1600

1500

1400

1300

1200

1100

1000

900

800

700

600

500

400

300

OBSERVED PRESSURE RANGESMODEL 1 CALCULATED AVERAGE PRESSUREMODEL 2 CALCULATED AVERAGE PRESSURE

0 5 10 15 20 25 30 35 40YEARS

PRES

SUR

E @

DA

TUM

, PS I

G

1700

1600

1500

1400

1300

1200

1100

1000

900

800

700

600

500

400

300

OBSERVED PRESSURE RANGESMODEL 1 CALCULATED AVERAGE PRESSUREMODEL 2 CALCULATED AVERAGE PRESSURE

Vertical Facies Proportions, Well Data

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

0.95

0.85

0.75

0.65

0.55

0.45

0.35

0.25

0.15

0.05

Rel

ativ

e D

epth

, ZD

EP

Facies Proportion [%]

CHANNEL (29.9%) MUD (23.7%)PROXIMAL (20.5%) MUD2 (P) (16.8%)DISTAL (3.2%) MUD3 (D) (5.9%)

Z02 - C-20

Lyr 30

Lyr 13

FACIES AT TOP OF STRATIGRAPHIC INTERVAL

Vertical Facies Proportions, Well Data

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

0.95

0.85

0.75

0.65

0.55

0.45

0.35

0.25

0.15

0.05

Rel

ativ

e D

epth

, ZD

EP

Facies Proportion [%]

CHANNEL (29.9%) MUD (23.7%)PROXIMAL (20.5%) MUD2 (P) (16.8%)DISTAL (3.2%) MUD3 (D) (5.9%)

Z02 - C-20

Lyr 30

Lyr 13

FACIES AT TOP OF STRATIGRAPHIC INTERVAL

16 SPE 87032

FIGURE 14: EXAMPLE 2002 SIMULATION GRID

SPE 87032 17

FIGURE 15: EXAMPLE 2002 HISTORY MATCH FIGURE 16: EXAMPLE 2002 PREDICTED INCREMENTAL RECOVERIES

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

STB/Day

Simulated Oil RateObserved Oil RateSimulated Water RateObserved Water Rate

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 5 10 15 20 25 30 35 40Year

Fraction

Simulated Water CutObserved Water Cut

0

10000

20000

30000

40000

50000

60000

70000

80000

90000

STB/Day

Simulated Oil RateObserved Oil RateSimulated Water RateObserved Water Rate

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 5 10 15 20 25 30 35 40Year

Fraction

Simulated Water CutObserved Water Cut

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 5 10 15 20 25 30 35 40Year

Fraction

Simulated Water CutObserved Water Cut

-20000

-10000

0

10000

20000

30000

40000

50000

60000

70000

0 5 10 15 20

INC

RE

MEN

ATA

L C

umul

ativ

e O

il R

ecov

ery,

MST

B

Flank WAG 1:3Flank WAG 1:3 NitrogenGas InjectionWater Injection - VREPWater Injection

-20000

-10000

0

10000

20000

30000

40000

50000

60000

70000

0 5 10 15 20

INC

RE

MEN

ATA

L C

umul

ativ

e O

il R

ecov

ery,

MST

B

Flank WAG 1:3Flank WAG 1:3 NitrogenGas InjectionWater Injection - VREPWater Injection