Sunrise Troubadour Gas Condensate Fields - OnePetro

SPE 64483 Identifying, Evaluating and Modelling Key Dynamic Parameters: Sunrise Troubadour Gas Condensate Fields P.M. Stephenson, SPE; R.B. Ainsworth; D.A.Johnson; J.M.P. Koninx; R.J. Seggie, Woodside Energy Ltd.

Copyright 2000, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the SPE Asia Pacific Oil and Gas Conference and Exhibition held in Brisbane, Australia, 16–18 October 2000. 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 The Sunrise and Troubadour Fields form a massive offshore gas-condensate resource located in the Timor Sea, 450 kilometres northwest of Darwin. The commercialisation, marketing and contract negotiations for such a world scale gas development requires confidence in reservoir performance and product composition, plus a clear appreciation of their uncertainties. This paper discusses how the key dynamic uncertainties have been identified, evaluated and modelled to determine their impact on development decisions. Early cross section models identified connected gasinitially-in-place and reservoir quality as key uncertainties, impacting field performance. Simultaneously, preliminary marketing efforts and facility design recognised fluid properties as a key parameter. This directed data acquisition through appraisal wells, which were fully cored and production tested. Iso-kinetic sampling and careful well test design has resulted in representative compositional data and identified compositional variations across the field. High rate dependent skins measured in appraisal wells have been evaluated and requantified for development wells. Detailed single well models have evaluated the effects of condensate banking. As marketing efforts progressed, full field simulation became warranted to capture remaining uncertainty on reservoir quality and distribution, and fault transmissibility. A suite of full field dynamic models built around the key dynamic uncertainties has provided confidence that uncertainties are manageable and that the development and customer commitments are robust. The case for further data acquisition, aimed at optimising the final development, can

readily be made using the suite of dynamic models which integrate the input from all sub-surface disciplines. Focused appraisal has reduced and quantified the key uncertainties. The leading edge evaluation methods have removed the need for several additional appraisal wells that might otherwise have been required. Effective communication and teamwork between subsurface, facility engineering, commercial and marketing has proved crucial to ensure that the appropriate uncertainties were addressed in a timely manner as each discipline matured their part of the project. Introduction The Sunrise-Troubadour gas-condensate fields are located 450 kilometres northwest of Darwin on the edge of the Australian continental shelf and 50 kilometres from the adjacent Timor Trench. Together they hold in-place volumes ranging from (p90 to p10) 11.6 to 24 Tcf gas and from 510 million to 1 billion bbls of condensate. The p50 recoverable volumes are 9.2 Tcf dry gas and 321 million bbls condensate. The gas condensate accumulations are located in some 35 metres of net pay, located in 4 reservoir units within the 80 metre thick Upper Plover, Jurassic reservoir (Fig. 1). These are separated from the water bearing Lower Plover by the 80 metre thick D.Caddaense shale. The main reservoir units are modelled as shoreface (10 to 100 mD) and incised valley fill (100-800 mD) deposits. Smaller proportions of hydrocarbons are also contained in lower shoreface and mouthbar deposits (1-10 mD). A fuller account of the geology is covered in [1]. The accumulation is a true giant, with a combined areal closure of over 1000 square kilometres and 180 metres of vertical closure. The larger Sunrise Field is 75 kilometres long and 50 kilometres wide at top reservoir (Fig. 2). It is fault bounded to the south but within dip closure in other directions. Troubadour is within dip closure all around. The two fields were discovered in 1974, however, a dispute between Australia and Indonesia over sovereignty halted further exploration until the Timor Gap Treaty (establishing the Zone of Co-operation) was signed in 1991. Appraisal continued with a well in 1995, 2-D seismic in 1997 and 4 wells in 1998/99. 3-D seismic is currently being acquired and processed. Appraisal spend for a field of this size is significant. Hence identification of key uncertainties

2

P.M. STEPHENSON; R.B. AINSWORTH; D.A. JOHNSON; J.M.P. KONINX; R.M. SEGGIE.

and subsequent planning of appraisal has been managed in line with market and engineering development [2]. This paper focuses on identifying, appraising and managing the dynamic uncertainties of this field. Identifying Key Parameters Early work on marketing and facilities design identified fluid properties as a key uncertainty. Simultaneously, early crosssection simulations identified reservoir quality and connected GIIP as key parameters: In-Place Volumes: In the initial assessment, ranges for each parameter affecting in-place volumes were assessed from variations in well values, analogues and a wide range of possible geological interpretations. Early volumetric analysis showed that the key uncertainties were net to gross and porosity (Fig. 3). These parameters are controlled by depositional environment, which became a focus area. To better quantify and constrain these parameters, three of the 1997/98 appraisal wells were fully cored across the reservoir. Detailed core description, wireline log analysis and the use of analogue data were used to construct multiple 3-dimensional geological realisations using object-based geo-cellular modelling [1]. These sophisticated models were used to reevaluate in-place volumes ranges and as input to dynamic modelling. Partial or total compartmentalisation had also been identified as a concern and recovery factor ranges from cross section simulations were appropriately discounted [3]. Dynamic Reservoir Behaviour: Dynamic performance was initially studied using a multi-layer 2 dimensional crosssection model. These models, based on the layering from individual wells, concluded that: • The drive mechanism will be largely depletion with limited, late-time, edge-water influx. • Recovery and performance are most sensitive to connected GIIP and reservoir quality (permeability-height product). • Minor vertical permeability (10-5 mD) will create effective vertical communication, although field performance is not overly sensitive to this. Recovery is also sensitive to the flowing well head pressure at abandonment, as in all depletion dominated gas reservoirs. However, this is a design parameter rather than an uncertainty. Though such laterally uniform layer-cake models are inherently simplistic, the approach permitted the identification of key dynamic sensitivities, upon which to focus further study and data gathering. Aquifer strength, gas-water relative permeability and vertical connectivity between layers were shown to be of lesser importance. Reservoir quality and distribution and in-place volumes became focus areas. Data gathering through appraisal drilling and testing was targeted at addressing these key uncertainties in detail.

SPE 64483

Fluid Properties: A single set of gas-condensate samples, collected when testing the 1974 Troubadour discovery well, exhibited an in-place CGR (condensate gas ratio) of 29 bbl/MMscf and a dew point close to initial reservoir conditions. An MDT (Modular Dynamic Tester) sample from the 1995 appraisal well on the Sunrise field exhibited almost twice the CGR and an unrealistic dew point 1700 psi above reservoir pressure. Fluid composition, which effects field performance, project revenue and facility design, was identified as another focus area for early appraisal. The three appraisal wells during 1997/98 were all tested, with fluid sampling being one of the main objectives. To aid representative sampling: • water based drilling mud was used to avoid hydrocarbon contamination, • well-bores were displaced to nitrogen prior to perforating to minimise clean-up and create adequate under-balance, • an initial sampling flow period was established, at minimum stable gas and condensate rates, prior to full clean up. • isokinetic fluid sampling was conducted both at the wellhead and test separator. • CGRs were determined from the test separator and a minilab at the wellhead. • multiple gas-condensate sample pairs were collected throughout the test and subsequently analysed in the laboratory. The CGRs and recombined reservoir compositions determined for multiple sample pairs from each well proved to be highly consistent (CGR +/- 5%). Because of this accuracy, trends in fluid composition across the reservoir were identified. Results from the fluid analysis are: • The reservoir is at or very close to the dew point. • The reservoir exhibits a steady reduction in CGR from the northeast to the southwest. This is supported by gas and condensate geochemical analysis, which indicates an increase in source maturity for fluid in the southwest. The fluid data raised several concerns regarding dynamic performance: • Will the produced CGR be significantly reduced by condensate dropout? • Will well performance be impaired by condensate banking around the well-bore? • Does the variation in CGR indicate compartmentalisation? Concerns of compartmentalisation are further increased by variations in Free Water Levels (FWL). Wireline pressure data from the gas bearing Upper Plover and water bearing Lower Plover, indicates a deepening FWL towards the southwest of Sunrise Field. However, the 30 metre variation across 30 kilometres equates to a FWL tilt of only 0.06 degrees. This is seen elsewhere in the region and believed to be due to a dynamic aquifer.

SPE 64483

IDENTIFYING, EVALUATING AND MODELLING KEY DYNAMIC PARAMETERS: SUNRISE-TROUBADOUR FIELD

The effect of condensate drop-out was addressed through detailed single well modelling discussed below. Compartmentalisation, a potential key factor, was addressed through structural modelling [3] and full-field 3-D dynamic simulation. Modelling Well Performance Appraisal well test results and the nature of retrograde gascondensate fluids raise concerns about rate dependent skin and condensate banking. However, detailed evaluation and fine grid single well modelling has alleviated these concerns, thereby simplifying the requirements for future full field modelling. Condensate Drop-Out: Laboratory PVT experiments show that as the reservoir depletes, condensate drops out and the condensate volume reaches a maximum of 1% of the total hydrocarbon pore volume. Such small condensate saturations in the bulk reservoir will barely flow and will not impede gas flow. This condensate drop-out causes the CGR of the remaining gas phase to be reduced by a maximum of 25% from the initial CGR (Fig. 4). Therefore, the produced CGR will remain reasonably high throughout depletion. Condensate banking around a vertical well-bore was evaluated using fine grid single well models with rate or capillary-number dependent relative permeability correlations [4]. This verified that the volume of condensate that would bank-up during field life was small and created a skin factor of only 5 in the worst case. The effect of condensate banking on sub-horizontal development wells will be even less. Rate Dependent Skin: Well tests on vertical appraisal wells indicated high rate dependent skins of 0.5 to 1.0 per MMscf/d. Such high skin factors would significantly impair the productivity of development wells initially expected to produce at over 150 MMscf/d. The extended completion interval and more efficient completion techniques should reduce rate-dependent skin in sub-horizontal development wells. However, this required further confirmation. Forchheimer's equation [5] was used to calculate rate dependent skin. Constants in the Beta correlation were adjusted to match the rate dependent skins measured over various well-test intervals. These constants represent the worst case due to the limited completion efficiency and limited clean-up in appraisal wells. However, using these constants, the rate dependent skins calculated for sub-horizontal development wells were confirmed to be small. Therefore, rate dependent skin was shown not to be a key parameter, despite large skins measured in appraisal well tests. Full Field Simulation As marketing efforts progressed, the next stage of dynamic studies became warranted. Full field dynamic modelling was needed to evaluate the two key uncertainties on field performance - reservoir quality / distribution and transmissibility across faults. The key dynamic parameters are

3

shown in the uncertainty tree in Figure 5. Vertical communication has been included because of its potential effect on compartmentalisation. Multiple realisations were run to determine the full range in field performance. Results are summarised in Figure 6, showing the effect of key parameters on the sustainable plateau length. Had all dynamic uncertainties been considered, the uncertainty tree would have over a thousand branches. This would be impractical to simulate and evaluate. Therefore, the uncertainty tree has been pruned to the key uncertainties identified from early studies. The number of branches simulated has been further pruned whilst adequately covering the uncertainty range [6]. The full distribution of results (Fig. 6) has been generated using a limited number of simulation model results plus mathematical combinations of simulation output. Six additional simulation models were used to verify the accuracy of the mathematical combination method. Reservoir Quality: Low, most likely and high case geological realisations were transferred from the 3-D geo-cellular models for dynamic simulation [1]. As described, variations in reservoir architecture, quality and distribution proved to have the largest effect on field performance - affecting both in-place volumes and dynamic performance. Capillary effects increase the range of in-place volumes. Laboratory capillary pressure data shows increased connate water saturation and increased transition zone in poorer quality reservoir (Fig. 7). Laboratory data was correlated with and modelled using Leverett-J functions, which were further grouped by permeability. Poro-perm correlations were developed by grouping core data by facies and geological unit. Figure 8 illustrates such a correlation. Statistical analysis of the data shows that uncertainty at individual points is significant, but field-wide the correlation on average, is accurate. However, this assumes that porosities are accurately known. When the field average porosity uncertainty (+/- 2 porosity percent) is included, uncertainties in the field wide permeability correlations are not insignificant (Table 1). Permeability uncertainties due to field-wide correlation uncertainty and porosity uncertainty have been incorporated as realisations. The relatively poor performance of realisations with low reservoir quality cannot be compensated for with additional wells. The incremental production is not sufficient to justify extra wells. Transmissibility Across Faults: A fault throw of 20 metres could juxtapose high quality reservoir against shale or low permeability sand. Detailed interpretation of existing 2-D seismic data allows mapping of faults with throws greater than 25 metres. Sub-seismic faults with throws below 25 metres were added stochastically [3]. The resulting structural pattern (Fig. 2) has a fault density typical of other fields in the region where structural configuration is defined by 3-D seismic data. The most likely structural model is not compartmentalised. Significant sand-sand juxtaposition occurs across all intrareservoir faults, when the fault map and 3-dimensional

4


geological models are combined. The large scale of the accumulation means that most faults terminate within the field and there is a flow path around the tips of the faults. Variations in fluid composition from well to well are believed to be caused by insufficent time for full composition equilibration - the structure being less than 1 million years old, and probably still filling today. It is unlikely that composition equilibruim has had time to occur across the 75 kilometer long field. To compartmentalise the field, faults would have to be totally sealing due to clay gouge or extreme cataclasis. The clay gouge potential in these reservoirs is low. However, although unlikely, full compartmentalisation cannot be dismissed given the variations in free water level and CGR. In a worst case realisation, all faults (Fig. 2) are assumed to be sealing irrespective of sand-sand juxtaposition. This results in the lowest recovery factors (Fig. 9). However, poor reservoir quality has a greater effect on ultimate recovery and plateau length (Fig. 6) because reservoir quality reduces both in-place volumes as well as recovery factor. Fault transmissibility primarily affects connectivity and not in-place volumes. To a degree, compartmentalisation can be offset through the drilling of additional wells. Pressure monitoring in development wells drilled after start-up will identify compartmentalisation. In the worst case compartmentalised model an additional 14 wells can be economically justified and assist plateau length (Table 2). Although full compartmentalisation is unlikely, reduced transmissibility across faults due to deformation and cataclasis is probable. This has been modelled using a fault transmissibility multiplier. Deformation and cataclasis causes a band of reduced permeablity either side of the faults. The thickness of this band is typically 1/170th of the fault throw [7]. Proprietery correlations have been used to estimate the permeability reduction within the damage zone. The fault transmissibility multiplier is then calculated from the damage zone thickness, permeability reduction and simulation gridblock size. The effects of pessimistic, most-likely and no fault transmissibility reduction on plateau length are shown in Table 2. Development Concepts Well Numbers, Layout and Design: Well numbers, layout and design have been evaluated for the most likely model and verified for alternative realisations. Initially simulations used only 4 wells at 6 km spacing around a central drilling location. Wells at 4km and then 3km spacing wells were then added and their incremental performance determined. Each group of infill wells was justified on their incremental production. Wells that could be justified were retained and wells from additional drilling centres were added and evaluated in the same way. For Sunrise, this results in 21 wells dispersed at 4 kilometre spacing across the thicker part of the field. Four wells are justified on the Troubadour Field. The dispersed well layout requires multiple drilling centres with the furthest 2 development wells being 40 km apart. High

SPE 64483

angle (85o) or sub-horizontal wells completed across all sand layers show advantages over horizontal wells restricted to individual sands, even in realisations with reasonable vertical communication. Figure 10 is a sketch of a possible gas development, which includes subsea wells, manifolds and flow lines tied back to a central processing facility in shallower water. Initially only a handful of wells are required to fulfil demand. Additional wells are required to maintain deliverability for a notional 30 year gas contract period, as the reservoir depletes. Field performance is not sensitive to the well drilling sequence. Therefore, to defer costs, the initial development can be clustered around 1 drilling center with additional drilling centers being developed as required. This spreading development also allows better monitoring of reservoir pressures by measuring the pressure profile in the new more distant wells. Optimising the Development Concept: Sweet spots, areas of high permeability and large equivalent gas column, can be identified in each dynamic model. However, their locations vary in different geological realisations. Positioning simulated wells on simulated sweet spots is unrealistic and would give optimistic results. Therefore, the simulated wells have been located on a uniform grid. However, simulations were repeated with wells located on sweet spots to determine the value of such information. This value of information [8] was used to justify the acquisition of 3D seismic. Encouraging results from the inversion of 2D data indicates that sweet spots may be directly mappable using seismic [9]. Multi-lateral and connector wells are currently being evaluated to further reduce development well numbers and costs. Current Status of Work 3-D geocellular modelling backed up with detailed core description, seismic inversion and dynamic modelling has significantly reduced reservoir uncertainty. Acquisition of quality data and the application of leading edge technology have saved several appraisal wells that might otherwise have been required. Identification and thorough evaluation of key dynamic uncertainties has provided the confidence that the full range of field performance has been captured in the multiple subsurface realisations. The multi-disciplinary approach requires integration and common understanding across the sub-surface, development engineering, marketing and commercial groups. Sub-surface issues and results are best explained using results from a suite of full field dynamic models built around the key uncertainties. Questions about concept selection, project robustness, the value of further appraisal and long term deliverability are readily answered using the suite of full field dynamic models. Development concepts and gas contract scenarios have been tested against the range of models. This has confirmed that the

SPE 64483


development concept and customer commitments are robust. The business case for acquisition of 3D seismic has been justified by it’s potential to optimise well locations across the range of models. Sub-surface confidence, development concepts and gas markets have now matured sufficiently to make a number of developments viable. A project has been initiated to deliver gas to domestic and industrial customers in and around Darwin from 2005. Although closer, the option of piping gas to Timor is not technically feasible due to the extreme water depth in the Timor Trench, as well as earthquake and slope instability hazards. Floating LNG (Liquified Natural Gas) is carried as an alternative development option. Conclusions

• • •

•

•

Early identification of the key dynamic uncertainties resulted in focussed appraisal and study.

4.

5. 6.

7.

8.

Several early concerns (rate dependent skin, condensate drop out) were dismissed through detailed analysis. Focused appraisal and leading edge technology (geocellular modelling, seismic inversion) has significantly reduced uncertainty and reduced the number of appraisal wells that might otherwise have been required. Questions about project robustness, the optimum development, the value of further appraisal and long term deliverability are quickly and confidently answered using a suite of full field dynamic models built around the key uncertainties. The value of additional appraisal (3D seismic) has been justified from it’s potential impact on development across the range of dynamic models.

Acknowledgements We thank the NT/RL2, NT/P55, ZOCA 95-19 and ZOCA 9620 Joint Venturers, Woodside Energy Limited, Shell Development (Australia) Proprietary Limited and Phillips Australasia Exploration Company for granting permission to publish this paper. References 1. Seggie R.J., Ainsworth R.B., Arditto P., Burns F., Johnson D.A., Koninx J.P.M., Stephenson P.M., and Thompson J.: “Capturing Depositional Uncertainty: Sunrise-Troubadour Giant Gas Condensate Fields” Paper SPE 59406, 2000 SPE Asia Pacific Conference, Yokohama, Japan. 2. Seggie R.J., Ainsworth R.B., Johnson D.A., Koninx J.P.M., Spaargaren B. and Stephenson P.M.: “Awakening of a Sleeping Giant: Sunrise-Troubadour Gas Condensate Field,” APPEA Journal, (2000), 40 (1), 1-20. 3. Ainsworth R.B., Stephenson P., Johnson D.A. and Seggie R.J.: “Predicting Connected Hydrocarbon Volume Uncertainty Ranges: From Static Fault Model to 3-D

9.

5

Dynamic Simulator,” Paper SPE 64429, 2000 Asia Pacific Oil & Gas Conference and Exhibition, Brisbane, Australia. van de Leemput L.E.C., Bertram D.A., Bentley M.R., Gelling R. “Full Field Reservoir Modelling of Central Oman Gas/Condensate Fields,” Paper SPE 30757, 1995 SPE Annual Technical Conference, Dallas, USA. Dake L.P. “Fundementals of Reservoir Engineering” Elsevier 1978. van Elk J.F., Guerrera L., Vijayan K., Gupta R. “Improved Uncertainty Management in Field Development Studies through the Application of the Experimental Design Method to the Multiple Realisations Approach,” Paper SPE 64454, , 2000 Asia Pacific Oil & Gas Conference and Exhibition, Brisbane, Australia. Manzocchi T., Walsh J.J., Nell P., Yielding G. “Faults Transmissibility Multipliers for Flow Simulation Studies” Petroleum Geoscience 5, 53-63 (1999). Koninx J.M.P.: “Value-of-Information; from cost cutting to value adding.” Paper SPE 64390, 2000 Asia Pacific Oil & Gas Conference and Exhibition, Brisbane, Australia. Ainsworth R.B., Spaargaren B., Seggie R.J., Stephenson P.M., Johnson D.A., Koninx J.P., and Mcnutt J.: “Modelling from Sunrise to Sunset and back again: Integrated 3-D reservoir modelling and seismic inversion studies, Sunrise-Troubadour Field, Timor Sea” Proceedings 2000 AAPG/IPA International Conference and Exhibition, Bali, Indonesia. Table 1. Porosity permeability correlation Correlation Permeability* (mD) Mid case 54 P90 individual points 14 P10 individual points 210 P90 field average 43 P10 field average 66 P90 field average inc. 32 porosity uncertainty P10 field average inc. 91 porosity uncertainty * Permeability of Upper Shoreface facies at 15% porosity

Table 2. Effect of key uncertainties on plateau length Realisation Mid Case High depositional case Low depositional Case Worst compartments Worst compartments with extra wells Pessimistic fault transmissibility multiplier No fault transmissibility multiplier

Plateau Length % different from Mid Case 0 +54% -32% -42% -21% -20% +11%

6


SS-1

SR-1

21 km

SPE 64483

LS-1

13 km

Datum

Figure 1: Stratigraphic Correlation showing Gamma Ray Log of Upper Plover (gas bearing) reservoir divided into 4 units separated from the underlying, water bearing Lower Plover sands by 80 metre thick D.Caddaense Shale. Line of cross section shown on map below. SS-1, SR-1 and LS-1 are exploration/appraisal wells Sunset-1, Sunrise-1 and Loxton Shoal-1.

N SR-2 LS-1

Sunrise Field SR-1

SSW-1

SS-1

TRB-1 BA-1 0

5

10

Troubadour Field

km Figure 2: Top reservoir outline with interpreted and stochastically infilled faults. Line of cross section in figure 1 shown. SR-1, SR-2, SS-1, SSW-1, LS-1, TRB-1 and BA-1 are exploration/appraisal wells Sunrise-1, Sunrise-2, Sunset-1, Sunset West-1, Loxton Shoal-1, Troubadour-1 and Bard-1.

SPE 64483


7

Porosity x Saturation

Net-To-Gross

Gas Water Contact

Gross Rock Volume

Reservoir thickness

p50 to p10 parameter values p50 to p90 parameter values Gas Expansion Factor

-45%

0%

45%

Effect on p50 GIIP Figure 3: Tornado diagram showing sensitivity of in-place volumes to parameter uncertainties

100%

CGR in Gas Phase (% of initial CGR)

Liquid Drop Out (% of Dew Pt. Volume)

1.2%

1.0%

0.8%

0.6%

0.4%

0.2%

Well A Well B Well C

95%

90%

85%

80%

75%

70%

0.0% 0

400

800

12 0 0

16 0 0

2000

2400

Reservoir Pressure

2800

3200

3600

0

400

800

1200 1600 2000 2400 2800 3200 3600

Reservoir Pressure

Figure 4: Liquid Drop Out and its effect on produced CGR. From laboratory CVD experiments using recombined reservoir fluid

8


Uncertainty

SPE 64483

Dynamic Uncertainty

Low

Porosity Permeability

ML

Low

Compartments

None

None

Multi

Fault transmissibility

ML

ML

ML

ML

Low

Shale Kv

No

No

No

No

No

No

Full Field Models

A

B

C

D

E

F

Most Likely (ML)

High

ML

Large

High

ML

None

None

High

ML

ML

Yes

No

No

No

G

H

I

J

None

ML

These branches not simulated

Depositional Environment (reservoir quality)

Figure 5: Uncertainty Tree for dynamic modelling showing the key parameters, and realisations selected for full field simulation. See Figure 6 for impact of uncertainties on plateau length.

100%

A) Poor reservoir quality

90%

Cumulative Probability

80% 70%

D) Large compartments C) Multiple compartments

60% 50%

G) Permeable silts/shales

40% F) Base model

30%

J) Good reservoir quality

20% 10% 0%

10

15

20

25

30

35

40

45

50

55

60

65

Plateau Length Figure 6: Effect of key uncertainties on plateau length from full field simulation. Modelled realisations shown on uncertainty tree in figure 5 above.

70

SPE 64483


9

0.06 mD

160

100

0.7 mD 1.7 mD 2.0 mD

140

2.8 mD 7.6 mD 9.7 mD

75

100 1 mD

80

50

60 10 mD

40

14 mD

Height above FWL (m)

Capilary Pressure (psi)

120

18 mD 22 mD 28 mD 54 mD 60 mD 64 mD 115 mD 160 mD 167 mD 189 mD

25

277 mD 317 mD

100 mD

20

322 mD 419 mD

1000 mD

714 mD 837 mD

0 0

20

40

60

0 100

80

845 mD 1032 mD 1302 mD

Sw (%)

1363 mD 2685 mD

Figure 7: Centrifuge Capillary Pressure data from 29 cores showing a clear correlation of capillary pressure with permeability. Normalised using Leverett-J function and further grouped by permeability

10000 Core Data Points Best Least Squares Fit Single point p90 fit Single point p10 fit Field average p10 fit Field average p90 fit

Permeability (mD)

1000

100

10

1

0.1 2

4

6

8

10

12

14

16

18

20

22

24

Porosity (%) Figure 8: Porosity Permeability correlation of upper shoreface facies of unit 2, showing the permeability uncertainty (p90 to p10) in individual points and the field average uncertainty.

26

10


100%

SPE 64483

C) Multiple compartments

90% Cumulative Probability

80% D) A few large compartments

70% A) Poor reservoir quality

60% 50% 40%

J) Good reservoir quality

30%

F) Base model

20% 10% 0%

G) Permeable shales 0.30

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.70

0.75

Gas RF Figure 9: Effect of key uncertainties on gas recovery factor from full field simulation. Modelled realisations shown on uncertainty tree in figure 5

Figure 10: Sketch of possible development with 18 sub-sea wells in 5 clusters tied back to a platform in shallower water with 7 platform wells.