SBED meeting October 16th 2006

1
Use of SBED as a tool for permeability modelling
in heterolithic tidal reservoirs a test study
from the Njord Field

• SBED meeting October 16th 2006
• Mike Young

2
Outline

• Introduction
• Motivation behind the study
• Use of SBED in Hydro
• Tilje Formation, Njord Field - SBED test study
• Challenge of modelling the Tilje Fm.
• Why SBED?
• Data set used in the study
• Methodology/workflow
• Results
• Summary/comments on the SBED approach

3
Introduction

• Motivation
• Theoretically SBED is a good concept
• Hydro has supported SBED for some time
• Test out in practice can the tool be
  implemented in the BUs ?
• Good test case Heterolithic tidal facies of the
  Tilje Fm., Njord Field
• Difficult to characterize using conventional
  petrophysical/property modelling
• SBED/TBED designed to model tidal heterogeneity
• Internal use in Hydro
• Limited to a research activity
• Low priority activity
• At present, it is difficult and time consuming
  (expensive) to get results !
• Uncertainty in SBED is a key issue that is poorly
  addressed

4
Challenge of modelling the Tilje Fm., Njord
Thin intercalations of mudstone and sandstone
layers will have a strong influence on the flow
properties!
5
Why use SBED?

• Question
• Can we derive petrophysical properties from these
  data (core plugs) that are representative at the
  grid-cell scale of a reservoir model ?
• Problems
• Kv and Kh are NOT well characterized by core
  plugs or wireline data.
• Permeability data measured from core plugs that
  have a sample volume below the REV will be
  unrepresentative.
• Therefore we see extreme variability of plug
  permeabilities (Kv, Kh) even inside small
  vertical intervals.
• Answer
• No! But using SBED to model the heterogeneity at
  a more suitable volume (REV) and using flow-based
  upscaling to determine properties could be a more
  realistic solution

Core plug data Tilje 3A Formation
Thin intercalations of mudstone and sandstone
layers will have a strong influence on the flow
properties!
6
Data set

• Cored well 6407/7-4 (Njord East Flank)
• Used as a test case
• Thick Tilje 3A
• Good petrophysical data set
• Core plug mini-perm. data
• Wireline and synthetic poro-perm data
• Tidal heterolithics (Tilje 3A)
• Vertically aggraded tidal flat deposits
• Composed of tidal bundles
• Wavy, flaser and lenticular beds

7
Core plug poro. perm. data (Tilje 3A, Njord)
Correlation 0,67
8
SBED Methodology/Workflow used

• Generate bedding-scale sub-models/SBED templates
• Petrophysical data analysis (core
  plugs/mini-perm. data)
• Populate sub-models with petrophysical values
• Calibrate SBED petrophysical input values
• Moving window upscaling to generate the SBED
  output results

9
Generate SBED sub-models
6407/7- 4 Tilje 3A

• Split up core into intervals modelled with
  specific SBED templates
• Different sub-models needed to capture variations
  in the sandshale ratio (NTG)
• Intervals of approx. 5 sandshale (SBED NTG)
• i.e. 10, 15, 20 .. 100 sand
• Key parameters
• Sand to shale ratio (NTG in SBED)
• Geometry, thickness variation and frequency of
  the mud layers
• Mean STD for porosity and permeability
• Values needed for each lithotype in each submodel

50 m
26 different SBED models needed to capture
variation in sandshale (NTG) and sedimentary
architecture
Model size 30x30x30 cm
10
Sandshale ratio key in these tidal facies
Key modelling parameter - it will have a strong
influence on vertical and horizontal
permeability. Predictable relationship between
sandshale ratio and geometry/continuity of the
bedforms.
11
Petrophysical input data

• Porosity and permeability for each of the
  lithological components of the model (i.e. ebb
  sand, flood sand, mud)
• Mean and STD values
• Variogram value
• Poro-perm correlation (e.g. 0,67)

12
Petrophysical data analysis

• We need to find permeability/porosity values for
  each lithotype (sand 1 2, mud)
• Not an easy task
• There is typically a biased data base sampling at
  the wrong volume (core plugs)
• Mini-permeameter data are better, but rarely
  taken as standard
• Especially difficult to get permeability values
  for the mud layers
• Key steps Filter out biased plugs that contain
  multiple lithotypes
• Asses porosity and permeability distributions
  for the entire dataset and for subsets
• e.g. specific intervals of NTG, depth intervals
  
• Use these results as a starting point !

Mini-perm Plug data
Few data points below 50
sandshale ratio (essentially SBED submodel
divisions)
13
Porosity (sand)
Shapes of the distributions sketched to
highlight their nature

• Porosity distribution for the entire data set
  (sand layers)
• Need to make sense of this and break it down into
  subsets
• Porosity distribution for specific intervals of
  NTG (sandshale ratios)
• Divisions based on visual inspection of the core
  and determination of intervals with similar
  sand type
• Based on splitting up the data set into various
  different intervals of NTG
• Based on this data it is possible to determine
  mean and STD values
• May not be simple Gaussian distributions !

14
Permeability (sand)

• Distributions for the NTG intervals
• Core plug data and mini-perm
• Complex distributions, commonly with at least two
  sand types in each of the NTG intervals !

• Permeability distribution for the entire
• data set

Shapes of the distributions sketched to
highlight their nature
Shapes of the distributions sketched to
highlight their nature

• Core plug data are biased, so mini-perm data are
  better at capturing values for individual sand
  layers
• However, mini-perm data show bias towards the
  higher perm. layers !

15
Effect of varying input data (petrophysics)

• Results from upscaling of all realisations of
  SBED submodels for the Tilje 3A, 6407/7-4
• 5 different model versions - same geometric
  input, but different petrophysical input
• i.e. the variation is related almost exclusively
  to the petrophysics
• Key observation The petrophysical input data
  have a significant impact on the results!

16
Calibration of the SBED model input

• An important step is to calibrate the
  petrophysical values in the models
• Key question Have we captured the true variation
  in petrophysical values ?
• Answer Almost certainly not at the first attempt
  !
• Need to generate pseudo core plugs from the
  SBED models
• Extract volumes from the models that are the same
  as the actual core plugs
• These need to be upscaled (flow based, fixed
  boundary) and compared to the actual core plug
  data
• Compare the porosity-permeability cross-plot
• Compare the porosity and permeability frequency
  distributions

90 Sand model populated with permeability
17

• Input petrophysics should be adjusted until an
  acceptable match is obtained between pseudo and
  real core plugs
• It is likely that several iterations of this
  process will be needed !

18
Results

• Several different options here
• Chosen to build a stacked model for the cored
  interval i.e. a direct representation of the
  core
• Moving window upscaling enables upscaled results
  at a log scale (e.g. every 12.5 cm) for Kh, Kv,
  Porosity
• Upscaling method is dynamic (flow-based) method
  with fixed boundary conditions
• Why log scale ?
• Consistent with wireline data (same scale)
• Can be used together with wireline data to
  predict properties (esp. Kh, Kv) in non-cored
  wells
• The software Facimage was used to generate
  electrofacies and predict permeability in
  non-cored wells/intervals SBED results used as
  training data

19
(No Transcript)
20
Summary

• In tidal heterolithic facies (e.g. Tilje Fm.) it
  is difficult to determine representative
  permeability values using conventional core plugs
  
• SBED method was used to generate log-scale
  permeability based on process based modeling of
  small-scale geology (bedding scale)
• Moving window upscalling enabled Kh and Kv values
  to be generated at the log scale (ca. 12,5 cm),
  consistent with other wireline data
• Kh and Kv logs from SBED can be used as
  training data for permeability and facies
  prediction in non-cored wells
• E.g. Using the NNI method in Paradigm Facimage
• SBED/Facimage can provide a more realistic (Kh
  and Kv) and consistent data input to the
  geo-model (RMS)
• All input data with a similar sample volume
• Bedding scale, but not necessarily representative
  at the grid block !
• Additional modelling and upscaling step may be
  necessary

21
Comments on the use of SBED

• Calibrating the input data is an
  important/critical step
• Otherwise it is difficult to determine whether
  the results can be trusted
• Potentially a large range in uncertainty
• SBED could be equally as uncertain as
  conventional plug-based methods
• Manual and very time consuming, but critical step
  in the workflow !!
• SBED projects are labour intensive and thus
  expensive
• Possibly hundreds of man hours for a relatively
  small project !
• Especially with the manual calibration technique
• SBED is a specialist tool and still somewhat
  immature
• Not recommended for wider use in Hydro BUs yet !

22
(No Transcript)