GeologicallyBased Upscaling: What Matters

1
Geologically-Based Upscaling What Matters?

• Gillian Pickup
• Karl Stephen, Julian Clark, Jingsheng Ma and
  Mauricio Silva
• Institute of Petroleum Engineering
• Heriot-Watt University

2
Outline

• Introduction
• What is geologically-based upscaling?
• Application to a Turbidite Reservoir
• Summary and Discussion

3
Introduction

• Upscaling
• two separate stages

4
Introduction

• Upscaling
• two separate stages

5
Upscaling Stages

• Stage 1
• large scale-factor (1012)
• most people use averaging for this stage
• Geopseudo method developed at Heriot-Watt
• Stage 2
• usually only single-phase carried out
• two-phase upscaling more challenging

6
The Geopseudo Method

• Create models of sedimentary structures
• Upscale at geologically significant length scales

7
The Geopseudo Method

• Create models of sedimentary structures
• Upscale at geologically significant length scales

8
Geopseudo Method contd.

• Integrated approach
• geologists, petrophysicists and engineers
• Geologists
• study well logs and cores
• determine the geological environment
• study analogue outcrops
• decide what kinds of sedimentary structures to
  model

9
Geopseudo Method contd.

• Integrated approach
• geologists, petrophysicists and engineers
• Petrophysicists
• take probe permeability measurements in cores
• perform SCAL measurements

10
Geopseudo Method contd.

• Integrated approach
• geologists, petrophysicists and engineers
• Engineers
• create models of sedimentary structures
• single-phase upscaling
• two-phase upscaling
• sensitivity studies

11
Geologically-Based Upscaling

• Be aware of geological structures at all scales
• Understand which are important

12
Aim of this Talk

• Demonstrate how to carry out multi-level
  upscaling
• with a practical application
• Provide guidelines
• when are small-scale structures important?
• what else matters?

13
Application to a Turbidite Field
14
Application to a Turbidite Field

• Introduction
• Geological Modelling
• Upscaling
• Sensitivity Studies
• Summary

15
Field Study

• North Sea field
• Deep marine environment
• turbidites
• Collaboration between the Genetic Units Project
  (GUP)
• Aims
• create geological models at various scales
• carry out sensitivity studies

16
Channelised Turbidite System
17
Data Set

• Seismic surfaces and attribute data
• Well logs
• Core data
• including probe data and SCAL
• Production data
• PVT, formation testing and history
• Existing model
• from sponsoring company

18
Geological Modelling
19
Genetic Units
20
Fine-Scale Geological Model

• Created using Roxar RMS

21
Multi-Scale Modelling
22
Length Scales

• Small-scale
• account for structure within beds
• upscaling from about 1 cm to 1 m
• RMS model
• 25 m x 25 m x 1.5 m
• Full-field model
• 100 m x 100 m x 6 m

23
Small-Scale Structures

• Genetic Units
• GU1
• mainly massive sandstone
• some argillaceous andinterbedded sandstone
• GU2
• argillaceous sandstones
• mudstone intra-clasts and laminae
• GU3
• interbedded
• horizontal layers

24
Small-Scale Structures, contd.

• Genetic units
• GU4
• injected intervals
• investigated in a separate study
• GU5
• slumps and debris flows
• GU6
• mudstone

25
GU2 - Argillaceous
26
GU3 - Inter-bedded Sandstones
27
GU4 - Injected Sand Intervals

• Sand injection structures are common in turbidite
  fields
• formed when unconsolidated sand is forced into
  mudstone
• In the present study, sand injection intervals
  made up 25 of core studied
• May affect flow through the reservoir by
  connecting sand bodies

28
GU4 - Injected Sand Intervals
29
Heterogeneous GU1

• Separate channel models created using outcrop data

30
Singe-Phase Upscaling
31
Single-Phase Upscaling

• GU2 - Intraclasts
• pressure solution method
• no-flow boundary conditions
• Results
• slightly anisotropic, kv/kh 0.8
• results not sensitive to mudstone permeability

32
Single-Phase Upscaling

• GU3 - Interbedded
• analytical
• arithmetic and harmonicaverages
• 800 realisations
• Results
• kv/kh ratio very small
• kh and 1/kv normally distributed

33
Application of Single-Phase Upscaling

• How do we use this in the RMS model?
• Effective perms should be conditioned to well
  values
• calculated from k-f relationships
• Compared distributions of model perms and well
  perms
• particularly kv/kh

34
Application of Single-Phase Upscaling

• GU2
• kv/kh of models was consistent with kv/kh of
  upscaled well data
• using arithmetic average for well kh
• using harmonic average for well kv

35
Application of Single-Phase Upscaling

• GU3
• kv/kh of models was much lower than kv/kh at
  wells
• well logs cannot resolve thin beds
• used model results for kv and kh to condition RMS
  model

36
Single-Phase Upscaling

• GU4 - Sand injection structures
• small models
• grid the model directly and upscaleusing
  pressure solve method
• large-scale models
• upscale in stages
• Sand injections increase mudstone permeability

37
Two-Phase Upscaling
38
Two-Phase Upscaling

• 3 levels

39
Two-Phase Upscaling

• Three levels of upscaling
• cm-scale to bed-scale
• to cells of about 1 m cubed
• bed-scale to RMS model cell
• to cells of 25 m x 25 m x 1.5 m
• RMS model to full-field model
• to cells of 100m x 100 m x 6 m

40
Which Methods?

• Steady-state
• quick and easy
• do not compensate for numerical dispersion
• Dynamic
• time consuming
• do compensate for numerical dispersion

41
Steady-State Upscaling

• Capillary Equilibrium Method
• assume Pc constant across model

42
Steady-State Upscaling

• Viscous-Dominated Steady-State
• assume Pc 0, and fractional flow constant

43
Results of Multi-Level Study

• Level 1
• capillary equilibrium
• Level 2
• Kyte and Berry (dynamic)
• Level 3
• viscous-dominated steady-state
• These methods provide a practical way of
  multi-level upscaling

44
Small-Scale
45
Rel Perm and Pc Data

• Ignored SCAL data
• unreliable
• Used simple formulae for rel perms
• Swor const 0.75

46
Initial Water Saturation

• Scale Swi for different GUs

47
Capillary Pressure Curves

• Simple formulae with scaling

48
Results for GU2 and GU3

• GU2
• upscaled rel perms similar to rockcurves
• GU3
• upscaled rel perms anisotropic

49
Intermediate Scale
50
Intermediate Scale

• Upscaling from bed-scale to RMS model
• scale-up in horizontal only
• from 1 m to 25 m
• Use Kyte and Berry (1975)
• take account of numerical dispersion

51
Results for GU1

• Similar results for other GUs

52
Large Scale
53
Large-Scale Upscaling

• Upscale by factor of 4 in each direction for
  full-field simulation
• 25 m x 25 m 1.5 m to 100 m x 100 m x 6 m
• Viscous-dominated steady-state
• tested on sector model
• tended to increase watercut
• more coning
• Sensitivity studies used RMS grid

54
Scale-up Summary
55
Summary
56
Sensitivity Studies
57
Sensitivity Studies

• Used sector models with RMS grid
• Simulation details
• aquifer support
• horizontal well
• Three sets of sensitivity studies
• absolute permeability
• relative permeability
• geological realisation

58
The Sector Model
59
Sensitivity studies, contd.

• In each case compare BHP and water cut to base
  case
• single rel perm and no Pc
• Combined response function

m is base case property, m is model property, p
BHP or watercut, i time step
60
Absolute Permeability Tests

• Base case perms used geometric average of the
  well perms
• Sensitivity studies used
• GU1 upscaled for channel heterogeneities
• GU2 from well perms upscaled using arith. and
  harm. averages
• GU3 perms from models
• highly anisotropic

61
Rel Perm and Pc Tests

• Base case used single rel perm and no Pc
• Sensitivity studies used
• end point saturation modified for each GU
• water-wet Pc
• mixed-wet Pc
• small-scale pseudos
• Kyte and Berry pseudos
• pseudos for heterogeneous channels

62
Note on Channel Heterogeneities

• Base case
• channels treated as massive sandstone
• Sensitivity study
• channels composed of several facies
• Pseudos were estimated from models of massive and
  interbedded sandstones

63
Base Case Absolute Perms
End point mod Water-wet Pc Mixed-wet Pc Pc equil
upsc KB upsc
Response Function
64
Modified Abs Perms (GU1, 2, 3)
Base case rel k End point mod Water-wet
Pc Mixed-wet Pc Pc equil upsc KB upsc Het GU1
Response Function
65
Realisation 2
Base case rel k Mixed-wet KB upsc
Response Function
66
All Sensitivities
End point mod Water-wet Pc Mixed-wet Pc Pc equil
upsc KB upsc
base case absolute perms
Base case rel k End point mod Water-wet
Pc Mixed-wet Pc Pc equil upsc KB upsc Het GU1
altered absolute perms
Base case rel k Mixed-wet KB upsc
2nd realisation
Response Function
67
Summary of Sensitivities

• Significant effects
• changing Pc from zero to water-wet of mixed wet
• affects size of transition zone
• upscaling rel perms for channel sand
  heterogeneities
• oil trapping in interbedded facies

68
Summary of Sensitivities

• Moderate Effects
• large-scale geological structure
• different Swc for different GUs

69
Summary of Sensitivities

• Negligible effects
• upscaling GU2 and GU3
• occurred as isolated regions within high perm
  sandstone
• Kyte and Berry upscaling
• accounted for numerical dispersion in the
  horizontal but flow was mainly vertical
• sand injection structures
• although they have been shown to have a large
  impact in other turbidite wells

70
Field Study Summary
71
Field Study Summary

• We carried out a thorough study of a turbidite
  field
• There are still many uncertainties
• geological structure
• petrophysical properties
• Sensitivity studies indicate important features

72
Discussion
73
Discussion

• Are the results of this study applicable to other
  fields?
• different geological environments
• different fluid properties
• different production scenarios
• What matters in general?

74
Important Features

• Connectivity
• Petrophysics
• capillary pressure
• Geological Realisation

75
Connectivity

• Different geological environments have different
  types of heterogeneities
• The underlying factor of importance is the
  connectivity of different units
• especially connectivity with wells
• It is important to study the connectivity
• start at the large-scale

76
Example

• Models with
• GU1, massive sand, k 1000 mD
• GU3, interbedded, kh 500 mD, kv 2 mD
• anisotropic pseudo rel perms

b)
a)
77
Recoveries

• Compared recovery for models with and without
  small-scale upscaling

78
Recoveries

• Compared recovery for models with and without
  small-scale upscaling

79
Connectivity Conclusions

• The importance of small-scale heterogeneity
  depends on large-scale connectivity
• case a)
• GU3 in isolated patches
• fluid flows around
• small-scale structure had little effect

80
Connectivity Conclusions

• The importance of small-scale heterogeneity
  depends on large-scale connectivity
• case b)
• GU3 incontinuous layer
• fluid had to flow through
• small-scale structure had a significant effect

81
Connectivity Conclusions

• Two-phase upscaling is important
• not just single-phase

82
Petrophysics

• Including Pc had a significant effect on our
  results
• at the large-scale affected the size of the
  transition zone
• at the small-scale gave rise to trapping in the
  interbedded units
• Petrophysical measurements are important

83
Large Scale Structure

• There is always a large uncertainty in the
  geological structure
• It is important to capture the effect of this
  uncertainty by generating a number of models with
  a range of parameters

84
Guidelines

• Start at the large scale
• Consider large-scale connectivity of different
  facies
• which facies does fluid have to flow through to
  get to the wells?
• are these facies heterogeneous on small scales?
• If so, carry out Geopseudo upscaling
• if not, you may be able to ignore them!

85
Acknowledgements

• Most of this work was carried out as part of the
  Heterogeneity Project, which was sponsored by
• BG, Conoco, JNOC, Petrobras, Shell, Statoil and
  Veba (now Petrocanada) and the UK DTI
• We should aslo like to thank Roxar for the use of
  RMS and Schlumberger for the use of Eclipse