PECIKO GEOLOGICAL MODELING: POSSIBLE AND RELEVANT SCALES FOR MODELING A COMPLEX GIANT GAS FIELD IN A MUDSTONE DOMINATED DELTAIC ENVIRONMENT

Similar documents
Opportunities in Oil and Gas Fields Questions TABLE OF CONTENTS

Study on the Couple of 3D Geological Model and Reservoir Numerical Simulation Results

From 2D Seismic to Hydrodynamic Modelling

An Overview of the Tapia Canyon Field Static Geocellular Model and Simulation Study

Reservoir characterization

3D geological model for a gas-saturated reservoir based on simultaneous deterministic partial stack inversion.

Simultaneous Inversion of Clastic Zubair Reservoir: Case Study from Sabiriyah Field, North Kuwait

The SPE Foundation through member donations and a contribution from Offshore Europe

Training Venue and Dates Ref # Reservoir Geophysics October, 2019 $ 6,500 London

RESERVOIR SEISMIC CHARACTERISATION OF THIN SANDS IN WEST SYBERIA

Quantitative Seismic Interpretation An Earth Modeling Perspective

CHAPTER III. METHODOLOGY

Future of Tunu Field Development: A Breakthrough of Gas Sand Identification Using Automated Seismic Assessment*

GeoCanada 2010 Working with the Earth

Search and Discovery Article #20222 (2013)** Posted November 25, 2013

Facies Analysis of the Lower Cretaceous Wilrich Member (Lower Falher) of the Spirit River Formation.

Process oriented modelling of heterolithic tidal reservoirs Example from Heidrun well

Porosity. Downloaded 09/22/16 to Redistribution subject to SEG license or copyright; see Terms of Use at

Statistical Rock Physics

Best Practice Reservoir Characterization for the Alberta Oil Sands

OUTCROP! CHARACTERISATION! OF! TRANSGRESSIVE! SANDSTONE! RESERVOIRS:! QUANTITATIVE!COMPARISON!OF!OUTCROP!ANALOGUES!

Relinquishment Report

Pre Stack Imaging To Delineate A New Hydrocarbon Play A Case History

Sarah Jane Riordan. Australian School of Petroleum University of Adelaide March 2009

PROCEEDINGS, INDONESIAN PETROLEUM ASSOCIATION Forty-First Annual Convention & Exhibition, May 2017

Modeling Lateral Accretion in McMurray Formation Fluvial- Estuarine Channel Systems: Grizzly Oil Sands May River SAGD Project, Athabasca

Application of multiple-point geostatistics on modelling groundwater flow and transport in the Brussels Sands

NEW GEOLOGIC GRIDS FOR ROBUST GEOSTATISTICAL MODELING OF HYDROCARBON RESERVOIRS

Reservoir Rock Properties COPYRIGHT. Sources and Seals Porosity and Permeability. This section will cover the following learning objectives:

Evan K. Franseen, Dustin Stolz, Robert H. Goldstein, KICC, Department of Geology, University of Kansas

North Dakota Geological Survey

Facies Modeling in Presence of High Resolution Surface-based Reservoir Models

MUHAMMAD S TAMANNAI, DOUGLAS WINSTONE, IAN DEIGHTON & PETER CONN, TGS Nopec Geological Products and Services, London, United Kingdom

Available online at ScienceDirect. Energy Procedia 114 (2017 )

Reservoir Management Background OOIP, OGIP Determination and Production Forecast Tool Kit Recovery Factor ( R.F.) Tool Kit

3D geologic modelling of channellized reservoirs: applications in seismic attribute facies classification

Multiple-Point Geostatistics: from Theory to Practice Sebastien Strebelle 1

F. Bacciotti K. D Amore J. Seguin

Downloaded 10/02/18 to Redistribution subject to SEG license or copyright; see Terms of Use at

The 3-D Seismic Geomorphology of Deep-Water Slope Channel Systems A Case Study from the Deep Water Nile Delta

Stephanie B. Gaswirth and Kristen R. Mara

Main Challenges and Uncertainties for Oil Production from Turbidite Reservoirs in Deep Water Campos Basin, Brazil*

Horizontal well Development strategy

SAND DISTRIBUTION AND RESERVOIR CHARACTERISTICS NORTH JAMJUREE FIELD, PATTANI BASIN, GULF OF THAILAND

Accommodation. Tectonics (local to regional) Subsidence Uplift

Imaging complex structure with crosswell seismic in Jianghan oil field

23855 Rock Physics Constraints on Seismic Inversion

3D Seismic Reservoir Characterization and Delineation in Carbonate Reservoir*

Analysis of influence factors of oil and gas reservoir description accuracy

Risk Factors in Reservoir Simulation

Geologic influence on variations in oil and gas production from the Cardium Formation, Ferrier Oilfield, west-central Alberta, Canada

Building an Integrated Static Reservoir Model 5-day Course

P026 Outcrop-based reservoir modeling of a naturally fractured siliciclastic CO 2 sequestration site, Svalbard, Arctic Norway

Serica Energy (UK) Limited. P.1840 Relinquishment Report. Blocks 210/19a & 210/20a. UK Northern North Sea

Reliability of Seismic Data for Hydrocarbon Reservoir Characterization

QUANTITATIVE INTERPRETATION

The Kingfisher Field, Uganda - A Bird in the Hand! S R Curd, R Downie, P C Logan, P Holley Heritage Oil plc *

Bulletin of Earth Sciences of Thailand

PETROLEUM GEOSCIENCES GEOLOGY OR GEOPHYSICS MAJOR

Hydrocarbon Potential of the Marginal Fields in Niger Delta Oza Field, a case study*

A021 Petrophysical Seismic Inversion for Porosity and 4D Calibration on the Troll Field

F003 Geomodel Update Using 4-D Petrophysical Seismic Inversion on the Troll West Field

Petrophysical Rock Typing: Enhanced Permeability Prediction and Reservoir Descriptions*

Net-to-gross from Seismic P and S Impedances: Estimation and Uncertainty Analysis using Bayesian Statistics

5 ORIGINAL HYDROCARBONS IN PLACE

APPENDIX C GEOLOGICAL CHANCE OF SUCCESS RYDER SCOTT COMPANY PETROLEUM CONSULTANTS

Bulletin of Earth Sciences of Thailand. Evaluation of the Petroleum Systems in the Lanta-Similan Area, Northern Pattani Basin, Gulf of Thailand

Bikashkali Jana*, Sudhir Mathur, Sudipto Datta

Modeling Lateral Accretion in McMurray Formation Fluvial-Estuarine Channel Systems: Grizzly Oil Sands May River SAGD Project, Athabasca*

Excellence. Respect Openness. Trust. History matching and identifying infill targets using an ensemble based method

CO 2 storage capacity and injectivity analysis through the integrated reservoir modelling

Downloaded 09/15/16 to Redistribution subject to SEG license or copyright; see Terms of Use at

Search and Discovery Article #20285 (2014)** Posted December 15, 2014

Exploration Significance of Unconformity Structure on Subtle Pools. 1 Vertical structure characteristics of unconformity

Appraising a late-middle-aged Brent Group field

Facies Classification Based on Seismic waveform -A case study from Mumbai High North

High-resolution Sequence Stratigraphy of the Glauconitic Sandstone, Upper Mannville C Pool, Cessford Field: a Record of Evolving Accommodation

Multiple horizons mapping: A better approach for maximizing the value of seismic data

Pros and Cons against Reasonable Development of Unconventional Energy Resources

The Alba Field: Improved Reservoir Characterisation using 4D Seismic Data. Elaine Campbell Oliver Hermann Steve Dobbs Andrew Warnock John Hampson

Search and Discovery Article #20097 (2011) Posted January 31, 2011

THE USE OF SEISMIC ATTRIBUTES AND SPECTRAL DECOMPOSITION TO SUPPORT THE DRILLING PLAN OF THE URACOA-BOMBAL FIELDS

Recent developments in object modelling opens new era for characterization of fluvial reservoirs

Geosciences Career Pathways (Including Alternative Energy)

Glauconitic Oil Reservoirs in Southern Alberta Creating the Correct Geological Model to Guide Development Drilling

High Resolution Field-based Studies of Hydrodynamics Examples from the North Sea

Optimisation of Well Trajectory and Hydraulic Fracture Design in a Poor Formation Quality Gas-Condensate Reservoir

Analysis of the Pattern Correlation between Time Lapse Seismic Amplitudes and Saturation

Downloaded 09/16/16 to Redistribution subject to SEG license or copyright; see Terms of Use at

Keywords. PMR, Reservoir Characterization, EEI, LR

EMEKA M. ILOGHALU, NNAMDI AZIKIWE UNIVERSITY, AWKA, NIGERIA.

Facies Analysis Of The Reservoir Rocks In The. Sylhet Trough, Bangladesh. Abstract

BLACK PLATINUM ENERGY LTD

Relinquishment Report for Licence Number P1471 Block 16/8f March 2009

Parameter Estimation and Sensitivity Analysis in Clastic Sedimentation Modeling

Prof Bryan T CRONIN Principal Geologist 2 Tullow Ghana Ltd

Constraining Uncertainty in Static Reservoir Modeling: A Case Study from Namorado Field, Brazil*

MUDLOGGING, CORING, AND CASED HOLE LOGGING BASICS COPYRIGHT. Coring Operations Basics. By the end of this lesson, you will be able to:

A E. SEG/San Antonio 2007 Annual Meeting. exp. a V. a V. Summary

Downloaded 10/25/16 to Redistribution subject to SEG license or copyright; see Terms of Use at

Transcription:

IATMI 2005-29 PROSIDING, Simposium Nasional Ikatan Ahli Teknik Perminyakan Indonesia (IATMI) 2005 Institut Teknologi Bandung (ITB), Bandung, 16-18 November 2005. PECIKO GEOLOGICAL MODELING: POSSIBLE AND RELEVANT SCALES FOR MODELING A COMPLEX GIANT GAS FIELD IN A MUDSTONE DOMINATED DELTAIC ENVIRONMENT Philippe Samson ; TOTAL E&P Indonésie Tantri Dewi-Rochette ; TOTAL E&P Indonésie Maurice Lescoeur; TOTAL E&P Indonésie Philippe Cordelier; TOTAL E&P Indonésie ABSTRACT Geological modeling is not an objective in itself. It may have different purposes such as estimating IGIP, validating well locations, optimizing well trajectory, reservoir studies and constraining geological models for production forecast. Geological model methodology depends on (i) the main objectives of the model, (ii) the available data, and (iii) the geological understanding of the field. The geological model evolves with the life of the field. Peciko is a challenging gas reservoir regarding geological modeling due to several factors. These include (i) complex geology, associated with a mud dominated deltaic environment of deposition for the reservoir section, (ii) a large gross gas column of 2000 m contained within several tens of reservoirs. (iii) poor seismic resolution, and (iv) an active drilling campaign with approximately 20 wells drilled per year. During early field appraisal, geological modeling is aimed at estimating the initial gas in place (IGIP). For this purpose, delineation wells allow the definition of a geological layering scheme and the construction of multi-2d models comprising 39 layers. Subsequently, during field development and production, the main objectives are to validate future well locations and to geologically constrain flow models for production optimization. Such a dynamic objective requires the main heterogeneity that controls fluid flow to be modeled. To achieve this objective, models have a fine elementary scale, corresponding with deltaic cycles. A total of 96 deltaic cycles corresponding to mouth bars deposit sequences are recognized within the field and are modeled with multi-2d techniques. 3D object modeling is a further step if the proposed models do not sufficiently constrain flow simulations. Such an approach raises additional challenges including the stochastic distribution of modeled objects and requires a full transverse uncertainty workflow. INTRODUCTION Peciko is a giant gas field, located south east of the actual Mahakam Delta. Peciko covers an area of 250 km2 and has a maximum gross reservoir column of over 2000 m and is structured as a gentle anticline plunging NNE. Peciko is formed by an interbedded sandstone and mudstone succession deposited in a mud dominated deltaic environment. This paper describes (i) the complexity of the geological model related to the depositional environment, and (ii) the evolution of the geological model and the controlling factors which drive this evolution. These factors include the degree of geological understanding which improves through time and controls the amount of detail of the geological model. Furthermore modeling methodology is specific to the objectives of the model. For instance, a model to estimate gas in place may not provide relevant flow barriers for reservoir flow simulation. PECIKO GEOLOGICAL CONCEPT Depositional Environment and Facies The Peciko field is formed by a sandstone and mudstone succession, interpreted as being deposited in a delta front environment (Figure 1). Elementary sand bodies are mainly interpreted as mouth bars, and seldom as 1

channels. Petrophysical facies representing differences in sand quality are based on wetclay and porosity well logs (Figure 2). These sandy facies called A, B and C sands are cut off through a continuous trend towards decreasing petrophysical quality. The corresponding facies in core show the increase in clay content through laminae (Figure 3). An outcrop example of a typical deltaic cycle, where 3 mouth bars have developed, is shown in Figure 4. The first and the third mouth bars show a classical coarsening up sequence, whereas the second mouth bar has a sharp base. This is thought to represent a rapid deposition of sediments, possibly related to flooding, rather than a channel (based on its position in the global deltaic cycle environment trend, interpreted as pro-delta to delta front). Figure 2. Petrophysical facies definition. A, B and C sands are defined through cut offs on wetclay and effective porosity logs. Figure 1. Sedimentological vs. reservoir petrophysical facies. A, B and C petrophysical facies can be matched with sedimentological facies identified on cores. However bioturbated vs. laminated mud-sand organization is not discriminated through petrophysical facies while it can drastically change the reservoir behavior of a C sand. Bioturbation of mud rich facies can destroy the good horizontal permeability of the initially laminated sediment organization. Figure 3. Sedimentological vs. reservoir petrophysical facies. A, B and C petrophysical facies can be matched with sedimentological facies identified on cores. However bioturbated vs. laminated mud-sand organization is not discriminated through petrophysical facies while it can drastically change the reservoir behavior of a C sand. Bioturbation of mud rich facies can destroy the good horizontal permeability of the initially laminated sediment organization. 2

Figure 4. Stack of mouth bars in a Deltaic Cycle. This outcrop picture illustrates the succession of mouth bars in a deltaic cycle. Different types of mouth bars may develop as illustrated by mouth bar 2 compared to 1 and 3. Sandstone body 2 has a sharp base but is still interpreted as a mouth bar and not a channel as it is part of a global trend from prodelta to delta front which does not reach the delta plain. This mouth bar is thought to represent a rapid deposition of sediments, possibly related to flooding. Orders of Depositional Sequences The first major depositional sequences identified in early wells are the stratigraphic units that correspond to 3rd order cycles. Peciko reservoirs are contained within 6 stratigraphic units. Within these stratigraphic units a layering is proposed based on correlation between 22 delineation wells resulting in a well spacing ranging from 3 to 5 km. A total of 39 layers are identified. Each of these layers is not strictly a stratigraphic sequence, as the layering is also driven by reservoir considerations including reservoir pressure and fluid distribution. However, in general this layering may correspond with 4th order cycles. With a development well spacing of 1400 m, it is possible to correlate at an even finer scale. A total of 96 deltaic cycles are now correlated throughout the field. A deltaic cycle is the elementary regressive-transgressive sequence within which several mouth bars may be stacked (see Figure 5). Deltaic cycles organize themselves into larger scale regressivetransgressive trends that correspond to the identified layers. Figure 5. Deltaic Cycles and their stacking within a layer (equivalent to a parasequence set). A deltaic cycle is a regressive-transgressive sequence. Up to 4 deltaic cycles may organize themselves into a layer corresponding to a 4th order regressivetransgressive sequence, which organize themselves into 3rd order stratigraphic units. Depositional Sequences, Sand-body Organization and Fluid Distribution The stratigraphic interpretation that guides Peciko geologic modeling is summarized in Figure 6. Within a deltaic cycle sand bodies are organized into a series of mouth bars downstream of distributary channels. Multiple deltaic cycles are themselves organized into layers, as shown in Figure 7. Deltaic cycles are generally consistent in term of their fluid content. Comparisons of fluid content and initial pressure measurement show that deltaic cycle limits usually act as flow barriers during gas accumulation and/or during field production. Two different scales of gas accumulations are found: 1. The Geological Pressure Unit (GPU) corresponds to one gas accumulation on a geological time scale. Information regarding these gas accumulations is provided by fluid content and initial pressure (i.e. measurement taken prior to production start of the considered reservoir). The equilibrium within such a gas pool has been reached through geological time. As such, only poor connectivity is required to reach this equilibrium. 2. The Reservoir Pressure Unit (RPU) corresponds to one gas accumulation at production time scale. Information regarding these gas accumulations is provided by 3

pressure measurement taken after production start. Lower permeability zones sufficient to connect sands in a GPU can act as barriers at production time scale, and therefore result in separate RPUs. As such a GPU is generally larger scale comprising several RPUs. Figure 6. Sketch of Peciko Sedimentological organization. Mouth bars organize themselves into stack of mouth bars that are developing during a deltaic cycle downstream of an active distributary channel. Figure 7. Comparison of layer scale and deltaic cycle scale. Correlations at deltaic cycle scale better explain fluid distribution. Working at this scale integrating pressure data helps improve both stratigraphic and fluid status correlations. Data available to interpret GPUs are limited to well formation tests (WFT) in wells drilled prior to production start. In contrast, data to interpret RPUs are more frequent and incorporate both WFT and production logging test (PLT) data from wells drilled after production start. However data for GPUs are absolute values while data for RPUs are relative, depending on the date of measurement, geology and production history of neighbor wells. The initial pressure analysis is done by pressure vs. depth plots, such as water head plots. The interpretation of initial pressure plots and addition of depleted measurements is shown in Figure 8. Figure 8. Initial pressure plot with GPU interpretation (left) and the same plot with depleted WFT data (right). In such graphs, initial pressure measurements organize themselves in gas and water linear trends suggesting geological pressure units. Data from production wells in reservoirs that might be depleted can be incorporated and used qualitatively to confirm or invalidate the compatibility of a pressure measurement with a given GPU or RPU. GEOLOGICAL MODEL DURING FIELD DELINEATION AND EARLY FIELD DEVELOPMENT To prepare the development of the field a succession of delineation wells are drilled and a first detailed geological model is built. To prepare the development of the field a succession of delineation wells are drilled and a first detailed geological model is built. Available Well Data The first version of this model is based on 22 delineation wells. The well spacing is typically 3 to 5 km. Available data from the wells include (i) cores from several wells, (ii) wireline logs, and (iii) WFT pressure measurement and fluid analysis. This measurement is made in the well at selected depths chosen by geologist and reservoir engineer in order to capture the pressure heterogeneity and the fluid content. Geological correlation of 22 delineation wells allows a model to be built containing 39 layers using one cell per layer. Pressure data indicate that a thick continuous mudstone interval between layers acts as a flow barrier. 4

Objectives The objective of the first version of the model is to estimate the initial gas in place and its spatial distribution to help define a development plan. After field development start, production wells begin to provide additional information and allow the geological model to be improved. The model is updated on a regular basis to better constrain the geological model and the reservoir flow simulation. Model Description Structural Framework The structural framework for the model is defined by 39 stratigraphic markers correlated between wells. These stratigraphic markers are interpolated between wells using 3-D seismic data. Unfortunately sand and fluid distribution in Peciko is below the acquired seismic data resolution. As a result of this only 3 seismic horizons are used as a basis for the model. Sand Distribution For each layer, one gross sand thickness map is made based on well log data. The cumulative thickness of the 3 identified petrophysical facies with reservoir potential (A, B and C sandy facies) is then used to calculate a NetSand value for each layer at each well location. Gross and Net Sand values are then integrated with sedimentological concepts, such as (i) sediment supply direction, (ii) proximal to distal environments, and (iii) mouth bars and stack of mouth bars dimensions to manually contour one NetSand Map per layer (Figure 9). For major, sandy layers, it is possible to correlate individual mouth bars and to map their shapes. In such cases, the NetSand map of a layer is obtained by stacking the individual mouth bar maps. Figure 9. Examples of NetSand maps at layer scale (left) and deltaic cycle scale (right). The deltaic cycle scale map confirms, for this layer, the notion of lateral distribution of disconnected stacks of mouth bars expressed in sketch of figure 6. It contrasts with the apparent homogeneity of the layer scale netsand map (as anticipated by the impact of model resolution expressed in figure 11). Fluid Distribution One of the Peciko s challenging characteristics is its complex fluid distribution. It is linked to the mud-dominated deltaic depositional environment of the reservoir succession. It is also controlled by high sand continuity in the WNW part of the field and a poor sand connectivity in the SE associated with massive mudstone units. This results in two phenomena: 1. A non horizontal, non unique gas water contact per layer: There is commonly an uncertainty associated with the gas water contact. Due to the environment of deposition giving rise to mudstones with relatively thin sand bodies, one can seldom identify a gas-water contact in the well. A gas-water contact is usually defined by lowest known gas at the base of a mouth bar and highest known water at the top of a deeper mouth bar. This intervening succession commonly consists of mudstone or poor quality thin sandstone bars that can only be interpreted as possibly containing gas. Even though this uncertainty exists, it is not possible to define a horizontal gas-water contact or to correlate gas-water contacts between wells to deduce a single realistic gas-water surface. 2. Perched water: In some instances, water located between two gas zones is interpreted as perched water or as water contained within isolated, by-passed sandstone bodies during gas migration. This may result in a complex vertical fluid distribution within a layer as shown in Figure 7. Since it is not realistic to interpret and model a gas-water contact at layer scale, an alternative 5

fluid modeling technique known as Filling Ratio mapping has been used. The filling ratio (FR) is computed at the wells using the NetSand (NS) and the NetPay (NP) as follows: FR = NP / NS. To model a FR map one must interpret the extension of the gas bearing pools. Each pool is regarded as a continuous gas body and is therefore a GPU possessing a common pressure gradient. A combination of four parameters defines the extent of each GPU: 1. The fluid properties 2. The gas water contact in the well and the associated top layer structure map 3. The initial fluid pressure 4. Distance from well control. For reserves certification, proven reserves (1P) are within 1.5 km of well control, whereas probable reserves (2P) are within 2.5 km of well control. Once the outline of each GPU is defined, it is used as a 0 limit for the Filling Ratio. Filling Ratio maps are then made for each layer by interpolation of the FR values computed at the wells, within this outline. An example of GPUs and their relation to fluid distribution and structure is given in Figure 10. Figure 10. Geological Pressure Units interpreted at layer scale, illustrated with two layers. The complex relation with structure is the expression of counter pressure trapping related to water expelled from shale SSE of the field and globally poor reservoir continuity in such mudstone dominated deltaic environment Petrophysical Parameters An average porosity (Phi) value is initially computed for the net sand per layer using well data. These porosity values are then interpolated laterally within the outline of the net sand map and by using the sand trend as guide. The net sand limit is equated to a minimum porosity value for sand of 5 p.u. Gas saturation (Sg) maps are then computed by a simple interpolation of the well values. Formation volume factor for gas (Bg) is computed through Bg=f(depth) laws applied to the structural map depth values. IGIP Maps Initial Gas in Place (IGIP) thickness maps are computed as the product of the previously described parameter maps: IGIP = NS * FR * Phi * Sg / Bg Successive Model Updates Drilling activity during field development results in a continuous flow of new well data. From a total of 22 wells prior to production start in 2000, the number has risen to over 80 wells at the end of 2004. The additional well data allow improvements to be made to the previous model. These improvements include the following: 1. The reduced well spacing (now reaching approximately 1400 m) allows improved lateral sampling and more detail to be added to the model. It also removes the need to impose limits (certification rule) of lateral extent to the reservoirs for 1P and 2P reserve estimates. 2. The concept of C connected sand is introduced to discriminate sand accumulations that contribute to reserves from those that may not. C sands 6

amalgamate sedimentological facies with different flow behavior. This facies is mudstone dominated and corresponds to sandstone and mudstone laminae deposits preserving good horizontal permeability. Bioturbation can re-arrange and mix sandstone and mudstone resulting in the destruction of the horizontal permeability without significantly increasing the vertical permeability. Laminated and bioturbated C facies are different sedimentological facies corresponding to the same petrophysical facies as they cannot be discriminated through petrophysical logs. Furthermore perforated C sands do not always produce hydrocarbons. This led to the interpretation that all C sands may not have reservoir potential. Some are producing through their direct perforations, and some produce through connections to A and B sands. C sands that do not produce are referred as non-connected C sands as opposed to connected C sands. Limiting factors of early models include the following: 1. Consistency of the fluid distribution at the layer scale. Some water bearing sandstone bodies lying vertically between gas-filled sandstone bodies are not actual trapped water. The vertical fluid distribution therefore requires a finer vertical subdivision than the previous 39 stratigraphic layers of early models. Therefore these early models fail to capture the vertical reservoir flow heterogeneity. 2. Layers are too thick, (20 to 100 m) resulting in continuous NetSand maps that fail to capture lateral sand discontinuities and resulting lateral flow heterogeneity (see geological concept in Figure 6 and the way it is translated at layer scale in Figure 11). 3. Additional well data provide added control for seismic interpretation. This resulted in an improved structural framework using 9 seismic horizons instead of the initial 3 horizons. Model Usage The first objective of the model was to estimate the IGIP volume and distribution. Model improvements over time have allowed information from the first years of production to be included. The concept of connected C sand also marks an evolution from a true gas-in-place analysis to the modeling of producible gas-inplace. The development plan includes the successive building of an optimized number of platforms. The overall project is divided into phases that must be validated. These early models were used to estimate gas-in-place and reserves in development areas, and to validate the location, size and economics of future platforms. Limiting Factors of Early Models The lateral and vertical resolution of early models is too poor to describe the heterogeneity that controls fluid flow in the reservoir. As a result, early models provide poor input for reservoir flow simulation. Figure 11. Model version of the sketch from figure 6 at layer scale, deltaic cycle scale and sand body scale. The vertical resolution of the geological model highly impacts the apparent heterogeneity of the field. At deltaic cycle scale the netsand maps can capture major flow barriers (limits of stack of mouth bars) while at layer scale the netsand maps hide these heterogeneities. A thinner resolution, the individual sand body scale, would capture limits of sand bodies that may also act as flow heterogeneities. However limits of sand bodies are expected to reduce but not to prevent flows. 3. The reservoir model derived from the geological model is too simple as no water is included. As Peciko aquifer is not strong, 7

this approach has been satisfactory so far, but it may not be so in the future. Although the geological models described above do not prevent the use of an aquifer in the reservoir simulation, a more detailed model is needed to appropriately describe the aquifer. In term of gas charge over geological time, the main controlling unit is the GPU. However, the thick model layers may contain several GPUs and therefore all internal flow barriers are not appropriately modeled. The model is also not capable of capturing heterogeneity at the RPU scale. GEOLOGICAL MODEL DURING PRODUCTION AND LATE FIELD DEVELOPMENT Available Data Continued drilling has allowed the average well spacing to be reduced to approximately 1.4 km over most of the field, at the end of 2004. This well spacing provides tremendous improvement for log correlation allowing the correlation of deltaic cycles. The previous 39 layers are now subdivided into 96 deltaic cycles. Fluid correlation within this layering scheme allows improvement in the fluid description and understanding at each well, while also providing a guide to the stratigraphic correlations themselves. Pressure data are also used for these correlations. Objectives At this stage of field development IGIP is not the primary issue, since IGIP is well defined by production data. The main remaining focus is now reserves and their optimization. As the crest of the reservoir is covered by production wells, field development is now focused on how to produce the flanks and to optimize production with infill wells in key areas. For example, it may be beneficial to add wells in some areas, reducing the well spacing to around 1000 m, in order to accelerate production, and to optimize the recovery factor. Although, water production is not currently a major issue, this may not be the case in the future, as depletion increases and additional flank wells are drilled. In order to predict water breakthrough, the model needs to explicitly model water and aquifers. Model Scale The new detailed layering provides deltaic cycles that are consistent in term of fluid distribution and pressure trends. This drastically reduces the amount of interpreted perched water (see Figure 7). At this scale each GPU is explicitly modeled and provides guides for RPU interpretation. The new layering scheme introduces additional barriers to horizontal flow that separate each deltaic cycle limit. As a result of the layers being thinner, the net sand maps tend to be less continuous; as these maps stack fewer individual mouth bars preserving the definition of vertical barriers (see Figure 7). A typical net sand map of a deltaic cycle which contains 4 sandstone bodies that are disconnected from each other and therefore define 4 GPUs, is illustrated in Figure 9 (right side). This understanding of the sandstone continuity is to be opposed to the global map at layer scale of the previous model (see left side of Figure 9). CONCLUSIONS The modeling scale highly impacts the image of the actual geology. Possible scales of work are numerous. The ability to work at one scale is related to the amount and the quality of data available. The scale of work also depends on the objective of the model. Several scales have been investigated on Peciko. A layer scale allowed estimating gas in place at the beginning of the field development. Object modeling techniques have also been investigated since that time, but building such a detailed model requires sufficient understanding to well constrain the stochastic techniques which are applied. The important issues when building a model are (i) the objective of the model, (ii) the heterogeneity regarding this objective, and (iii) the appropriate modeling scale for this heterogeneity. Initial models were built to evaluate IGIP. At that time layer scale was enough and geological understanding did not allow reaching a finer scale. Today models are focused on reserves and methods to optimize field production. According to today s understanding of the field, the expected dominant factor controlling reservoir heterogeneity is the stacked nature of 8

mouth bars. For stacked mouth bars that develop during a deltaic cycle, the appropriate scale for such models is the deltaic cycle scale. However, transmissibility reductions are expected to occur within a stack of mouth bar at the interface between individual mouth bars. This reduction is not expected to drastically alter fluid mobility. Should this hypothesis be too optimistic, a model at a finer scale would be required. WAY FORWARD In the case that further refinement is needed (see Figure 11), object based modeling could be a way forward. Such a model requires geostatistical techniques to be constrained by geological concepts at the deltaic cycle scale. Modeling throughout the life of the field generally leads to increased detail and resolution in the geological models. This process is a learning curve related to available data and improvements in geological understanding over time. Short cuts in this learning curve might prove dangerous for the reliability of the geological model. For example, object modeling has been considered since early in the Peciko field life, but should be attempted only when the deltaic cycles are well understood. It is today considered as a possibility for the next generation of geological models. Once the deltaic cycle scale is well constrained, object based modeling can be used to reach a high level of detail consisting of the mouth bars within each deltaic cycle. ACKNOWLEDGEMENTS The authors thank TOTAL, TOTAL E&P INDONESIE, Ditjen MIGAS, BPMIGAS and INPEX for their permission to publish this paper. Special thanks go to all Peciko asset members whose daily work made Peciko models possible. The authors also thank the petrophysics team, the reservoir transverse team and the exploration team from TOTAL E&P INDONESIE in Balikpapan for their valuable collaboration. 9