TuMod Integrated Turbidite Modelling

Similar documents
Submarine Debris flow Project Proposal to Force August 2018/v1.02

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

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

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

Deterministic, Process Based Modeling of the Deepwater Fill of the Peïra Cava Basin, SE France*

PETROLEUM GEOSCIENCES GEOLOGY OR GEOPHYSICS MAJOR

Quantitative Seismic Interpretation An Earth Modeling Perspective

Seismic Expressions of Submarine Channel - Levee Systems and Their Architectural Elements

Deep Water Systems and Sequence Stratigraphy. By: Matt Kyrias, Chris Majerczyk, Nick Whitcomb, Wesley Vermillion

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

Parameter Estimation and Sensitivity Analysis in Clastic Sedimentation Modeling

Search and Discovery Article #40536 (2010) Posted June 21, 2010

Bulletin of Earth Sciences of Thailand

Controls on clastic systems in the Angoche basin, Mozambique: tectonics, contourites and petroleum systems

DATA ANALYSIS AND INTERPRETATION

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

RESERVOIR CHARACTERISATION

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

Third EAGE CO2 Geological Storage Workshop, Edinburgh, March

Case Study of the Structural and Depositional-Evolution Interpretation from Seismic Data*

2002 GCSSEPM Foundation Ed Picou Fellowship Grant for Graduate Studies in the Earth Sciences Recipient. David Pyles

TECHNICAL STUDIES. rpsgroup.com/energy

Evolution of the Geological Model, Lobster Field (Ewing Bank 873)

Reservoir characterization

Prof Bryan T CRONIN Principal Geologist 2 Tullow Ghana Ltd

Geosciences Career Pathways (Including Alternative Energy)

Introduction to sequence stratigraphy and its application to reservoir geology

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

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

Sequence Stratigraphy: An Applied Workshop

Modeling of Intra-Channel Belt Depositional Architecture in Fluvial Reservoir Analogs from the Lourinha Formation, Portugal*

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

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

The Impact of Parasequence Stacking Patterns on Vertical Connectivity Between Wave-Dominated, Shallow Marine Parasequences, Book Cliffs, Eastern Utah

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

Available online at ScienceDirect. Energy Procedia 114 (2017 )

GeoCanada 2010 Working with the Earth

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

BERG-HUGHES CENTER FOR PETROLEUM AND SEDIMENTARY SYSTEMS. Department of Geology and Geophysics College of Geosciences

23855 Rock Physics Constraints on Seismic Inversion

A.K. Khanna*, A.K. Verma, R.Dasgupta, & B.R.Bharali, Oil India Limited, Duliajan.

Use of Cellular Automata Flow Model in the Hybrid Geostatistical Models: A Proposal

Meandering Miocene Deep Sea Channel Systems Offshore Congo, West Africa

Outcrops from Every Continent and 20 Countries in 140 Contributions. Tor H. Nilsen, Roger D. Shew, Gary S. Steffens, and Joseph R.J. Studlick.

INT 4.5. SEG/Houston 2005 Annual Meeting 821

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

Rock physics and AVO analysis for lithofacies and pore fluid prediction in a North Sea oil field

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

Quantitative Relation of the Point BarWidth and Meander Belt Width of Subsurface Reservoir

2011 SEG SEG San Antonio 2011 Annual Meeting 1134

TEAM MAK EAGE SC. Project Description: Research in Albertine Graben an important location for oil and gas in Uganda.

Application of Predictive Modeling to the Lower Cretaceous Sedimentary Sequences of the Central Scotian Basin

NEW GEOLOGIC GRIDS FOR ROBUST GEOSTATISTICAL MODELING OF HYDROCARBON RESERVOIRS

Dalhousie University- Petroleum Geoscience Field Methods- Trinidad Summary Report

Bikashkali Jana*, Sudhir Mathur, Sudipto Datta

Earth models for early exploration stages

Reservoir connectivity uncertainty from stochastic seismic inversion Rémi Moyen* and Philippe M. Doyen (CGGVeritas)

Relinquishment Report. Licence P2016 Block 205/4c

Bulletin of Earth Sciences of Thailand

MASTER THESIS. Faculty of Science and Technology. Study program/specialization: Spring semester, MSc Petroleum Geosciences Engineering

Image: G. Parker. Presenters: Henry Chan, Kayla Ireland, Mara Morgenstern, Jessica Palmer, Megan Scott

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

UNIT 3 GEOLOGY VOCABULARY FLASHCARDS THESE KEY VOCABULARY WORDS AND PHRASES APPEAR ON THE UNIT 3 CBA


Seismic Attributes and Their Applications in Seismic Geomorphology

AAPG European Region Annual Conference Paris-Malmaison, France November RESOURCES PERSPECTIVES of the SOUTHERN PERMIAN BASIN AREA

Geography 3202 Unit 4 S.C.O. 4.3 & 4.5. Primary Resource Activities Offshore Oil And Gas

Applications of Borehole Imaging to Hydrocarbon Exploration and Production

Time to Depth Conversion and Uncertainty Characterization for SAGD Base of Pay in the McMurray Formation, Alberta, Canada*

Sequence Stratigraphy. Historical Perspective

Stratigraphic Trap Identification Based on Restoration of Paleogeophology and Further Division of System Tract: A Case Study in Qingshui Subsag*

UK Field Training Courses

Exploration, Drilling & Production

We A Multi-Measurement Integration Case Study from West Loppa Area in the Barents Sea

Synthetic Seismic Modeling of Turbidite Outcrops

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

Sedimentary Basins. Gerhard Einsele. Springer-Verlag Berlin Heidelberg New York London Paris Tokyo Hong Kong Barcelona Budapest

PATTERN BASED GEOLOGICAL MODELING OF DEEP WATER CHANNEL DEPOSITS IN THE MOLASSE BASIN, UPPER AUSTRIA. Lisa Stright

The Marine Environment

CHAPTER III. METHODOLOGY

SPECIALIST GEOLOGY SERVICES

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

RC 3.3. Summary. Introduction

Seismic validation of reservoir simulation using a shared earth model

Stratigraphic Architecture and Key Stratigraphic Surfaces Formed by Punctuated Flow Events - An Experiment on Fluviodeltaic Responses*

Storage 6 - Modeling for CO 2 Storage. Professor John Kaldi Chief Scientist, CO2CRC Australian School of Petroleum, University of Adelaide, Australia

Southern Songkhla Basin, Gulf of Thailand

Ocean Basins, Bathymetry and Sea Levels

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

Deep-Water Reservoir Potential in Frontier Basins Offshore Namibia Using Broadband 3D Seismic

Quarterly Report April 1 - June 30, By: Shirley P. Dutton. Work Performed Under Contract No.: DE-FC22-95BC14936

24. Ocean Basins p

CLASTICS FIELD TRIP. Dynamic stratigraphy, facies, architecture and fracture analysis of coastal depositional systems

Surface Processes Focus on Mass Wasting (Chapter 10)

Modeling Lateral Accretion in the McMurray Formation at Grizzly Oil Sands Algar Lake SAGD Project

Exploration research at SINTEF Petroleum

Extended Abstract for presentation at EAGE Meeting Paris 13/ History of Norwegian Petroleum Exploration and its impact on Norwegian Geosciences

Stochastic Simulation of Inclined Heterolithic Stratification with Streamline-based Stochastic Models

Depositional Environments. Depositional Environments

Seismic stratigraphy, some examples from Indian Ocean, interpretation of reflection data in interactive mode

Transcription:

A project proposal from Norsk Regnesentral (NR) and the University of Bergen (UiB) Project description PART 1: The KMB project 1. Objectives Develop methodologies, algorithms, and software tools for modelling deep-water (turbidite) reservoirs. Sub-goals: Increase knowledge on the link between turbidite processes and depositional geometries and how to model them numerically. Combine stochastic modelling approaches with process based models that describe the appropriate sand-body geometries. Develop methods for conditioning stochastic and process models for turbidites to 3D seismic data. Make computer programs for modelling turbidite deposits. Develop broader concepts for combining process and stochastic modelling for other depositional systems. Ensure relevance and applicability of software through case studies. Complete a Dr. Scient thesis. 2. Frontiers of knowledge and technology There is a significant volume of literature on deep marine deposits including outcrop studies on their geometries and numerical and physical studies of the sedimentary processes responsible that deposit them. However, modelling systems at an oil field-scale remains a key challenge. Recent developments at NR and Roxar (Hauge et al. 2003) have begun to address the issue of reliable stochastic modelling of these systems but significantly work is required to capture the complex amalgamation between different turbidite events. NR have a strong tradition of geological facies modelling and has made theoretical and commercialized contributions to this field in more than a decade. NR and UiB have collaborated successfully on a number of recent projects. Some relevant recent publications include: R. Hauge, A.R. Syversveen and A. MacDonald: SPE 84053 Modelling Facies Bodies and Petrophysical Trends in Turbidite Reservoirs SPE ATCE 2003. Proceedings. Denver October 7, 2003. Skare, Øivind; Bølviken, Erik and Holden, Lars: «Improved sampling-importance resampling and reduced bias importance sampling». Scandinavian Journal of Statistics Vol. 2003, Vol 30, No. 4, December 2003. Holden, Lars: «Convergence of Markov Chains in the relative supremum norm». J. of Applied Probability Vol. 37, Nr. 4, 2000. Holden, Lars; Hauge, Ragnar; Skare, Øivind and Skorstad, Arne: «Modeling of fluvial reservoirs with object models». Mathematical Geology Vol. 30, Nr. 5, 1998. Skorstad, Arne; Hauge, Ragnar and Holden, Lars: «Well conditioning in a fluvial reservoir model». Mathematical Geology Vol. 31, Nr. 7, 1999. Howell, John (University of Bergen); MacDonald, Alister (Roxar); Fjellvoll, Bjørn; Skorstad, Arne; Fordham, Alex (University of Liverpool, UK); Flint, Stephen (University of Liverpool, UK) and The Page 1 of 10

Saigup Consortium (EU): «Testing reservoir sensitivity in heterogeneities shallow marine reservoirs». AAPG Annual Meeting 2003. Salt Lake City, USA May 11-14, 2003. Skare, Øivind; Skorstad, Arne; Hauge, Ragnar and Holden, Lars: «Conditioning a fluvial model on seismic data». Fifth International Geostatistics Congress Proceedings. Wollongong September 22-27, 1996. Hauge, Ragnar; Skorstad, Arne and Holden, Lars: «Conditioning a fluvial reservoir on stacked seismic amplitudes». IAMG '98, Isola d'ischia Proceedings of the 4th annual conference of the International Association for Mathematical Geology 1998. Abrahamsen, Petter and Benth, Fred Espen: «Kriging with Inequality Constraints». Mathematical Geology Vol. 33, No. 6, 2001. Heimsund, S Hansen EWM, Bass JH and Nemec W (2003) Numerical CFD simulations of turbidity currents and comparisons with laboratory data, Slope Workshop extended abstract, 13 Referenced publications not written by personnel directly involved in the project: Johnson S.D, Flint S, Hinds D & De Ville Wickens H 2001. Anatomy, geometry and sequence stratigraphy of basin floor to slope turbidite systems, Tanqua Karoo, South Africa. Sedimentology, Vol. 48, p. 987 MacDonald, A.C. & J.O. Aasen 1994, A Prototype Procedure for Stochastic Modeling of Facies Tract Distribution in Shoreface Reservoirs, in Stochastic Modeling and Geostatistics, J. M. Yarus & R. L. Chambers (Eds.), AAPG Computer Applications in Geology, No 3, pp. 91-108 3. Research tasks The three most challenging research tasks are: Integrating a heuristic process model for facies geometry with a stochastic model to allow flexible conditioning to observations. Utilize high quality seismic data to guide the geometry of facies bodies such as channels and lobes. To scale process modelling algorithms developed for detailed analysis of very small time steps (one flow lasting minutes or hours) to thousands of flows over geological time (thousands to millions of years). Besides this, there will be numerous challenges ranging from making software useful for a broader audience to solving mathematical problems related to convergence of statistical algorithms. One PhD is planned. 4. Research approach, methods The best method for accurately representing the distribution of petrophysical properties within a reservoir is by representing the geometries of the facies and then populating facies bodies with petrophysical properties. Therefore, any future tools for modelling turbidite reservoirs will need to be based upon understanding the facies and their spatial distribution. Such facies based approaches are traditionally geometric (conditioned to outcrop and other data). We propose a method that is based upon process-based algorithms that will also condition to geometric considerations. Models will also need to include a significant stochastic component to capture the heterogeneity and uncertainty. Turbidite exploration is typically based on high quality seismic data and limited well data. Any future modelling strategy will need to be able to heavily sample and utilise 3D seismic datasets; the next generation of stochastic models will sample and condition to bodies extracted from seismic data in addition to wells. Key Issues to be addressed: There are a number of aspects unique to turbidites. We have identified the following as central to developing new modelling tools: Page 2 of 10

1. The scale of depositional systems whole turbidite systems can be 100s of km long. Typically, oilfields sit within a relatively small part of a single depositional system; however, it is essential to understand, at least in part, the rest of the depositional system to understand the reservoir geometries. 2. A wide variety of depositional sub-environments exist, including stacked slope channels within canyons and mini-basins; channels, over-bank and lobe systems that pass down dip into more sheet-like systems with or without minor channels. The reservoir characteristics of each of these elements are distinctly different. 3. Seismic data is a key to the delineation of turbidite systems. It is also central to understanding the distribution of depositional facies (reservoir) elements within the system. 4. Turbidite depositional systems are highly sensitive to initial and evolving sea-floor topography within the basin. Consequently, they lend themselves to a more process-based approach than has traditionally been used for the stochastic modelling of shoreface (MacDonald and Aasen 1994) and fluvial (Holden et al 1998) systems which favour a more geometric approach. Within this study, we propose to develop tools to model deep water systems. Our approach will build upon a combination of a stochastic approach and a process based approach. The proposed approach will include the following steps: 1. The definition of a Container mapped from seismic-stratigraphic surfaces. The base of the container is mapped as the base of the slope-canyon and the initial topography of the ocean floor below the fan system. The top is defined as the seismically defined top of the fill and spill package above the canyon or the top of the fan. In order to enable a geologically realistic filling of the container, it may be significantly larger than the proposed final model. 2. Zonation of the model. The reservoir should be subdivided into distinct genetic packages based upon the recognition of stratigraphically significant surfaces. The nature of these surfaces may vary along the profile. In the fan type systems they will be based upon the hierarchy of starvation surfaces defined by Johnson et al (2001) whilst in the slope systems they will represent phases of canyon fill. 3. The key challenge is filling of the container with geological realistic geometric entities. This is a major aspect of the proposal and will be expanded upon further in the following two sections. The section entitled modelling the components of a deep marine system deals with the different methods that will be applied to the slope, upper fan and outer fan system while the stochastic modelling section addresses the detail of the new process based approach. 4. Gridding. Grid design is a key issue in modelling turbidites. We propose novel techniques of data storage that will use less memory and allow more detailed grids to be produced and populated. Models built of outcrops during the study will be used to test this grid design against more conventional gridding methods. Modelling the components of a deep marine system A key aspect of the approach described above is the geologically realistic population of the grid. A number of different approaches based upon mimicking the geological process that fill the basin has been identified. These are summarized below within the three key depositional settings: slope, upper fan, and lower fan. Slope Canyon/Mini-basin setting In the canyon, fills of the slopes three main facies are defined. The base of the canyon is typically filled with debris flow deposits and bypass facies, the main canyon fill is sand channels and the top of the canyon is occupied by heterolithic channels and overbank deposits. The heterolithic channel fills and overbanks may expand beyond the top of the canyon and form a fill and spill cap. Outside the canyon and its fill and spill cap, the deposits are comprised of non-reservoir, background facies. These include muddy slopeturbidites, hemi-pelagic shale and mud rich debris flows and slides. Initially a non-reservoir background will be set, the facies within the container are then populated, using the techniques described below conditioned from available well and seismic data. Vertical trends within the Page 3 of 10

canyon and the relative proportions of three main facies will be a variable that can be adjusted between models. Three methods will be evaluated for the population of the canyon fill with channel bodies: Using existing tools: A stochastic approach of placing channel objects within a shale background, similar to that used for fluvial reservoirs. This is a well-proven methodology for modelling channel systems. Population of a sand rich background with shales and barriers to represent the residual material after channel erosion. For this purpose, the utility of the traditional shapes typically used (discs, plates, lozenges) will be evaluated against a more geologically realistic sheet with holes. The lateral extent and degree of vertical bed amalgamation will be conditioned using recently published statistical trends for bed amalgamation. A new process based approach to populating channel systems, which involves producing a topographic surface within the grid, and then placing a flow line along the axis of lowest topography. User-defined parameters for the flow determine a degree of initial erosion followed by deposition within the channel and beyond. The properties of the channel fill and overbank spill record the waning flow, which is also defined by the user. The top of the depositional packages defines a new topographic surface onto which the next event is superimposed. The fill of the container is thus built up (see discussion below). This is a new approach to reservoir modelling of these systems and once the algorithms are written, a significant part of the work will involve the testing and refining of the concepts against outcrop and subsurface examples. Upper fan: Channel levee type systems: In populating grids for the upper fan, it is important to capture the relationship between the channel, overbank, and lobe systems. The key issues are the relationship between the channel and the surrounding material i.e. erosional or aggradational; the controls on channel position and stacking (evolving basin floor topography) and the proportion of channel fill to overbank deposits. As with the canyon/slope systems there are a number of approaches that can be applied including the population of stochastic objects within a seismically defined container and a process-based approach. We propose to test the following approaches: Geometrically defined facies belt and channel scheme, using the existing algorithms for lobate deltaic and fluvial systems. The key development will be the integral linking of channels to sheets with a channel automatically placed along the lobe or finger axis. The scale of the channel will also be linked to the size of the lobe. A stochastic technique based upon the distribution of amalgamated and non-amalgamated sheets of sand using the published distributions. This will capture channel areas as vertically amalgamated and overbank and lobe areas as less amalgamated. The process modelling approach described previously will be extended into this upper fan area. Preferential flow lines defined from topographic surfaces will be used to define a flow path. The initial strength of the flow will determine whether is it erosional or purely depositional and a flow strength decay algorithm will determine the point at which it becomes purely depositional. The size of the flow versus the size of the erosionally defined channel will determine the point at which it spills onto the surrounding overbank areas and becomes less confined. The top surface of the flow becomes an input parameter for the next depositional episode. The initial input will be seismically mapped reflectors at the base of the container. Internal seismic reflectors will be honoured to account for structurally controlled growth or thinning within the modelled interval. As previously, the modelling will be conditioned to both seismic and well observations of objects. Outer fan: Sheet systems: The new stochastic process model will also be used to model sediment sheets in the outer fan system. Modelling the outer fan will be predominantly based upon modelling lobes of sediment linked to minor, isolated but often very continuous channel bodies. The key challenges here are to capture the shape of the lobes and the genetic link between lobe and channel and to capture the Page 4 of 10

pinchout geometries. Grid design becomes increasingly important towards the lobe margins and the sensitivity towards different gridding strategies will be studied. The process/stochastic model The final models result from a two-fold approach that incorporates very detailed fluid flow process modelling and large scale conditioned stochastic modelling. Recent studies (Heimsund et al. 2002) have successfully simulated detailed aspects of turbidity flows from a small flume tank using computer models. Flows have been simulated using compositional fluid dynamics which is widely used for tackling engineering issues but rarely applied to sedimentology. The FLOW-3D software was adapted to simulate flows using finite-difference approximations to populate a 3-D grid. Renormalized Group Theory is used to simulate turbulence within two miscible fluids of different densities. Models account for gravity, pressure, and shear stress and produce both erosion and deposition within the sediment. Comparison with experimental data Velocity (m/s) 1.0 0.8 Experimental Simulated 0.6 0.4 0.2 0.0 0 10 20 30 40 T ime (s) 0 Width (m) 1 Figure showing screen shots from a simulated turbidite in Flow-3D. The velocity profile for the flow compares favourably with experimental lab data whilst the density and particle distribution produce Cross-sectional geometries realistic deposits. 0 Distance (m) ChannelChannel-levee section 4 Longitudinal section Figure showing the effects of a simulated flow encountering topography and the resultant deposits. To date the majority of simulated flows have been very small (meters) and over very short time periods (seconds to minutes). A key challenge is the upscaling of experimental results to both basin size and time scale. We do not propose to accurately recreate the conditions of each flow since the complexity of such Page 5 of 10

calculation would prohibit efficient conditioning to observed well and seismic data. Thus, a more heuristic approach will be chosen. Traditional geometric modelling involves placing stochastic shapes within a model to represent reservoir bodies. Dimensions and architecture of shapes are typically defined from empirical measurement of outcrops. Here we propose to arrive at the reservoir geometries by partially simulating the flow processes using the Flow-3D algorithms. A volume of sediment will be released down a slope defined from backstripped seismic surfaces. Depending upon its strength and flow characteristics, this turbidity flow will either erode or deposit sediment along the flow path, evolving and changing along its course. The speed of the flow will be a product of the gradient and the flow density. Deposition within all three key environments slope canyon fills, upper fan systems, and lower fan systems will be modelled as a single, genetically linked entity. The complete deep marine system will include in the order of 100 individual turbidite events. An idealized version of the modelling process can be visualized in the following steps: Page 6 of 10

The above process represents two turbidite events. In the upper part of the depositional system the deposits are confined to the deeply eroded channels, in the mid-fan there are channels and overbanks and in the lower fan the grid is dominantly populated with sheets of sand. In addition to depositing the sediment this depositional event produces a new seafloor topography that will form the basis for the next event. To form a complete deep marine system this sequence of events is repeated several hundred times. As the next event is conditioned from the inherited topography of the previous some repulsion of fans will be experienced. The deposits of the next turbidite event are represented in Fig 6. Note in the upper fan how the channels may erode through the clay sheets and form amalgamated sheets. This will be controlled by the relative rates of sand and shale deposition and will also be conditioned by predicted and observed bed amalgamation statistics. Deposits of the next turbidite events. Note the degree of erosion in canyon and the occurrence of stacked channel, whilst the upper fan is characterized by a combination of channels and lobes that are compensationally stacked. The outer fan is comprised of sheets. The final result is a succession of several hundred turbidite event beds, each described by a top and a base surface. In addition, the centreline for each turbidite (channel) will be stored. Note that we can use a very fine-meshed resolution for the grids since each turbidite has a limited lateral extension and smart storage mechanisms can be utilized. The basic input for the modelling is the initial surface that will be conditioned from seismic and represents a modelling zone boundary. The system will also condition to documented channel and sheet deposits in wells and ideally seismic data. A process of forward and reverse iteration will be developed to enable the process model to honour the existing data. The final gridding for the reservoir will depend on which part of the deep marine system constitutes the reservoir. We have already noted that a turbidite system is a lot larger than the reservoir so gridding should be adapted to the geometry found in the area of interest. Note that resolution should not be a problem since we can use very fine-meshed top and base surfaces for each turbidite. Conditioning models to seismic data Stochastic modelling tools used for reproducing shoreface and fluvial reservoirs will condition to well data and seismic attributes. The aim for all of the methods described above is to go one stage further and condition to 3D bodies extracted from seismic data. Such bodies capture, at least in part, the depositional elements within the survey (see figure below). Within a number of existing modelling systems, it is possible to extract these bodies and use them to populate the modelling grid. However, due to inherent errors in time slicing and subtle variations in attribute values, such bodies are often unduly discontinuous. Whilst it is possible to manually extrapolate between them this is time consuming and prone to user bias. Page 7 of 10

We propose to develop algorithms that will use the seismic body data to condition the stochastic models. Within these models, observed bodies will be correctly positioned and the areas between them, vertically and laterally, will be populated stochastically, honouring the data and the trends that come from them. This is a major new development and a key challenge, the results of which will have implications for modelling any systems with high quality 3D seismic data. Secondly, many turbidites systems are affected by syntectonic basin evolution. As stated previously basin topography results from both the effects of tectonics on the sea floor and the inherent evolution of the depositional system. To capture the tectonic component it is proposed to take seismic reflectors from the top, the base and from within the depositional container and map thickness trends. These trends will then be used to condition the topography for the process modelling and to control the application of bed amalgamation variograms. 5. Project organisation and management NR will be responsible for developing the new software tools. NR has a proven track record for finding methods and algorithms for modelling geology and developing experimental and commercial software for such purposes. UiB has significant experience in outcrop and subsurface studies of turbidite systems and will supervise the PhD student. NR and UiB have worked closely within the EU sponsored SAIGUP project and industry sponsored research. NR and UiB will be jointly responsible for progress and results in the project. Semi-annual meetings with the sponsoring oil companies guarantee that research and development is progressing in the correct direction. The supporting oil companies have significant experience with turbidite systems and the requirements of modelling them. There will be interaction and communication between the meetings when necessary. 6. International co-operation ConocoPhillips are a major international oil company and staff from Houston and the UK will have input to the project. Prof. John Howell at UiB is a British Citizen and has well-established turbidite links in the UK, Page 8 of 10

Spain and the US. Additionally results will be exposed to the international research community through workshops, conferences and journal publications. 7. Progress plan - milestones Month: UiB Reading and project definition Definition of field work Initial process modelling for channels Initial evaluation of field work Data processing and model building Model building Seismic work/joint with NR Testing initial algorithms with NR Further modelling Further process modelling Modelling second seasons data Final model testing Project write up and algorithm testing NR Software design Literature study Implement initial process model Initial well conditioning Initial seismic conditioning Refining initial algorithms Publishing and documentation Steering meetings First meeting Second meeting Third meeting Fourth meeting Fifth meeting Final meeting 1 2 3 4 5 6 7 8 9 1 1 1 1 2 3 4 5 6 7 8 9 1 1 1 1 2 3 4 5 6 7 8 9 1 1 1 0 1 2 0 1 2 0 1 2 8. Costs incurred by each research performing partner NR Personnel: 3 100 000 NR Administration, travel, conference attendance: 275 000 UiB PhD (salary, supervision): 1 650 000 UiB Field studies, travel, conference attendance: 525 000 Total 5 550 000 9. Financial contribution by partner Financing in NOK. 2004 2005 2006 2007 Total ConocoPhillips 500 000 500 000 500 000 1 500 000 Hydro 250 000 250 000 250 000 750 000 NFR 300 000 1 100 000 1 100 000 800 000 3 300 000 Total 1 050 000 1 850 000 1 850 000 800 000 5 550 000 Page 9 of 10

PART 2: Exploitation of results 10. Relevance for knowledge-building areas The project is in the field of petroleum and is relevant to both exploration and production. It will contribute to maintaining Norway s position as the number one in reservoir modelling. 11. Importance to Norwegian industry Better reservoir description in combination with modern well technology has boosted the recovery on The Norwegian Continental Shelf significantly. Deep marine deposits (turbidites) are significant petroleum reservoirs in Norway and internationally. There is currently no appropriate method for making numerical models capturing the heterogeneity and uncertainty in this kind of reservoir. Even marginal improvements in tools for describing this kind of depositional environment have a potential for releasing huge values. There is a profitable Norwegian software industry (e.g. Roxar and Schlumberger (former Technoguide)) that is able to commercialise software for reservoir description. Their combined annual revenue is in the order of 500 million NOK. 12. Relevance for Innovation programmes The project is relevant for the OG- (oil and gas) program. The main impact is the possibility for better petroleum reservoir handling and thereby enhanced oil recovery. The OG 21 initiative pinpoints five areas important to a sustained income to the Norwegian society from the petroleum industry. This project can contribute significantly to enhanced recovery and deep-water exploration since turbidite plays worldwide often is located outside the present continental shelf. 13. Environmental impact No direct environmental impact. 14. Information and dissemination of results Biannual meeting with the sponsors to communicate results and obtain feedback from professionals within the companies. Results will be published on relevant international workshops and conferences such as those organized by IAMG, SPE, AAPG, and EAGE. Results will published in international journals ranging from the more mathematically focused such as Mathematical Geology, to more Geology focused publications such as AAPG Bulletin, and more petroleum industry related such as Petroleum Geoscience. The ambition is to have three journal publications related to the PhD thesis and five journal publications related to the work at NR. Computer software will be available from the project. It is a goal to make the software available to academia and to make a commercial-quality product available to the oil industry. Page 10 of 10