You are on page 1of 6

SPE 101875

Quality Control of Static Reservoir Models


J.A. Spilsbury-Schakel, Woodside Energy Ltd.
Copyright 2006, Society of Petroleum Engineers
This paper was prepared for presentation at the 2006 SPE Asia Pacific Oil & Gas Conference
and Exhibition held in Adelaide, Australia, 1113 September 2006.
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
In the current climate of large numbers of projects, ever
shortening cycle times and increasing field complexity, there
is a need for rigourous and agreed Quality Control (QC)
guidelines for reviewing static reservoir models.
Within Woodside, the assurance process includes a
combination of checklists, guidelines, and milestone reviews
at predefined steps in the workflow, the latter being a
combination of project framing, 10%, 75%, and fully
integrated study reviews.
Although each field is different, and each study has its own
objectives, common checkpoints are easily identified and can
be grouped in general categories such as: study objectives;
geological setting and correlation; input data; structural model;
layering and log averaging; and facies and rock property
modelling (Fig. 1). There are supplemented by the export to
dynamic simulators, and comparisons between upscaled and
finely gridded properties, including volumetrics. Most
software packages provide some sort of audit trails whilst
manipulating the data that makes up the static reservoir model.
A set of pointers with respect to general model QC are given,
which are as handy as hints and tips for the beginner, and as
reminders (did you think about) for the experienced static
reservoir modeller.
Introduction
Faced with a 3D static reservoir model, a large number of
realisations and scenarios, a set of powerpoints and a deadline,
seeing the forest through the trees can be tricky when it comes
to checking static reservoir models. Where does one start?
What makes one model better than another? It would be good
if we could determine this prior to the drillbit entering the
ground.

Providing some clear guidelines on reservoir modelling


workflows, as well as key milestones at which the status of the
modelling study is reviewed, provides a focus for both the
geologist and the review panel.
Keep in mind that above all, we are dealing with a model, an
analogy used to help visualise something that cannot be
directly observed 1. The static model is an interpretation and
simplification of the real world and multiple sensitivities
should be run reflecting the uncertainties in the input data. The
resulting ranges should be realistic and reflect the maturity of
the project. A field with 20 years of production is likely to
have a tighter range of in-place volume estimates than a green
field. When using a static or dynamic model for well planning,
the final check should come from the seismic data, as fault and
surface data in the 3 static model is a simplified reflection of
the actual interpretation.
And, if it feels wrong, it probably is wrong!
Provide Clear Model Objectives
Prior to commencing any study, there should be a clear
understanding of the model objectives. Model geometries and
details are dependent on which deliverables are expected. For
instance, a static model constructed in a complex structural
setting for input into dynamic simulation would have a focus
on maintaining cell orthogonality and limiting model size to
facilitate dynamic fluid flow (maintenance of geologic detail
and model run times after upscaling?). However, in the same
setting, a model constructed for detailed well planning should
have as good as possible a match to the actual fault locations,
independent of cell orthogonality or model size.
Setting up a Terms of Reference document when
commencing the study can be of great help. This is generally a
one page document constructed by the subsurface team, which
clearly states the agreed model objectives and deliverables,
timing, available data and key uncertainties/issues. A model
framing workshop, whereby the study gets discussed with
input from all disciplines, is usually the best forum.
When reviewing a model it is important to keep these
objectives in mind, to ensure the model is fit for purpose.
Check the Geology
The first step in the QC, and building of any model is done
without the use of any computer aids, and is the most
important step. It involves setting the ground rules for the

geologist. All models should fit within the existing


understanding of the regional depositional system and
structural style. When reviewing a static reservoir model, this
is also where the review should start.
Talk to the exploration team, the structural geologist, the
biostratigrapher, the geochemist and sequence stratigrapher.
Create sketches of the paleo-depositional system, outline the
structural grain of the field, and make various cross sections
displaying the possible relationships between the different
flow units.
Is there core data available and has it been interpreted from a
sedimentological and stratigraphical point of view (preferably
by the same person)? Have core facies and flow units been
defined? And does the correlation define the flow unit
boundaries? How does the regional and locally available
seismic data help us in determining what is happening
between the wells? And, is there data that should be studied
further (like the geochemical analysis), or data that needs to be
created (like a seismic attribute volume)?
Develop the overall understanding of the field, and start
developing the uncertainty tree. Which models are possible?
Which uncertainties could affect the proposed field
development in a significant way?
Although critical, this step is often overlooked, forgotten, or
sacrificed to the gods of the timeline.
Data to be used
When we have defined the geological setting of the static
model, a review of the available data, with a view on the
impact on flow behaviour and hydrocarbon volumes, should
take place. We all know the phrase rubbish in rubbish out.
By importing both log and core plug data into the static
reservoir model database, visual checks can be made overlying
the porosity and permeability logs with the core data, and
creating crossplots of log vs core data.
At this stage, it is good to start comparing the permeability log
data with both the core data and the DST or equivalent
production information.
If cut-off values are used to define net vs non-net intervals, a
visual check of the net-to-gross logs with the actual core data
can reveal thin sand beds that fall below the log resolution.
Entire sections with thin sands in a predominantly shale
interval can be evaluated as non-net based on the log response.
Especially in gas reservoirs, significant movable volumes can
be present in these intervals2,3,4.
In a structural sense; which units are seismically mappable?
And how many flow units are considered to be within the
mappable zones. Normal practice is to model seismically
mappable surfaces as horizons, and derive the intermediate
flow unit boundaries as second order surfaces or zones from
these horizons. Erosive or onlapping boundaries should be
defined as such in the modelling process. Keeping the

SPE 101875

horizons limited to what can be defined on seismic allows for


quick updates for structural high or low case scenarios.
Structural model
After reviewing the geological background and the available
data, we can examine the 3D structural model. Building a
structural model comprises the process of surface and fault
modelling, and depth conversion if the interpretation that is
used was made in the time domain.
A first check should be the size of the model.
Review the number of cells in an aerial and vertical sense. Is
the cell size sufficiently small to capture key facies
heterogeneities? Is the cell size sufficiently large to allow for a
not too harsh upscaling to get the model to run in a dynamic
sense? Grids generally should run run parallel to the key
structural grain of the field, making it easy to attach an
analytical aquifer.
Check the fault model.
Check if the generated fault planes match the input data.
Review if the fault connections and truncations are honoured
in a way that makes sense based on the regional structural
context. Which simplifications have been made to facilitate
the (dynamic) modelling process? Straightening of listric
faults can be a very valid option if the hydrocarbon column is
relatively small, and a zigzag fault makes a more (more than
what?) orthogonal grid geometry without major effective
changes (to what?) if there are no wells close to the fault.
Finally, check if the fault compartments match the interpreted
fluid regions.
3D model geometry-check properties.
Geometry check properties can be generated, to QC grid
orthogonality. These properties identify twisted cells, or cells
with negative volumes. Although these cells do generally not
affect the static volumetrics, they can (and often will) cause
problems in the dynamic modelling phase.
Checking the constructed surfaces.
There is no better way than starting with a simple scroll
through the model, in mapview, and in intersections parallel
and perpendicular to the main grain of the structure. Special
attention should be given to checking around unconformities
and faults. How does the resulting model compare with the
original sketches that were made? A display of the constructed
surface with a seismic backdrop shows how the model
matches the original seismic data. A review should always
include a check of the horizons tie the wells, and the
adjustments that were required. In structurally complex fields,
or in fields with many wells, spending some time making sure
that the wells are located on the right side of the modelled
faults, especially after depth conversion, is essential.
An essential and easy check is to create isochore grids from
the seismic horizons, and from the resulting individual units.
Do the inferred trends make sense? Are there any bulls-eyes?
Model Layering and Log averaging
Part of the structural modelling process is the layering of the

SPE 101875

model. The layers provide the final vertical block size, and are
defined as parallel to the base of the unit, the top of the unit, a
fixed number of layers independent on the unit thickness, or
parallel to a predefined reference surface (e.g. a clinoform).
Although the layering process usually requires just a push of
the button, its impact on the actual data going into the model is
huge. Both the petrophysicist and the reservoir engineer
should be involved in assigning the optimum layer thickness.
One reason is that, in general, during the property population
of the 3D model, not the original well logs, but the blocked
logs are used. So it is paramount that the blocked logs
represent the actual well data correctly. It is therefore essential
to have the layer thickness of the model optimised so the
reservoir heterogeneity is preserved.
Bearing all that in mind, the reservoir engineer may prefer a
slightly coarser layering in order to avoid the requirement for
static to dynamic modelling upscaling, or the heterogeneity
may be considered so small or random that thicker layers are
sufficient to reflect the effective homogeneous flowunits
correctly.
There are a number of ways to check the effect of the layering.
1. Scroll through the layer model in X, Y and Z to
determine whether it makes sense.
2. Display the blocked logs next to the original logs.
3. Compare the minimum, maximum and mean values
of the original log data with the blocked logs.
4. Calculate the height*net-to-gross*por for each zone
from the original log and blocked logs.
5. Calculate the height*net-to-gross*perm for each zone
from the original log and blocked logs.
If the discrepancies are significant, adjust the layer thickness
(and method) until a suitable comparison is derived.
Prior to entering the phase of population of the created
geometries, it is prudent to check the gross rock volume
present in the model, and the split per flow unit, with the
volumes originally calculated by the geophysicist.
Facies model
The petrophysical property model should be built using a
facies model as a template. The geometry of the facies and
their areal and vertical relationships provide the framework in
which petrophysical properties can be populated. This is also
the phase where regional and/or analogue geological
knowledge can be introduced in the model.
Although the core analysis is likely to provide a large number
of facies divisions, the average 3D reservoir model is unlikely
to have more than 5-6 effective facies per depositional unit.
The grouping of reservoir facies should be based on the likely
geometry of each and the reservoir properties, as well as the
amount of occurrence, and the effect on fluid flow of each of
the facies. If the ultimate aim of the model is to provide input
into dynamic simulation, rationalisation (minimisation of
facies divisions?) of modelled facies should take place. For
example, a facies that is defined on 2% of the core, and is hard
to pick on logs in non-cored wells, and has similar properties

as other reservoir facies, is unlikely to have a major influence


on fluidflow. However, if it provides the one sealing shale in
the field, it should naturally be modelled separately.
QC of the facies model should include a review of the match
between modelled and core facies and the rationalisation
process. Ensure flow barriers are correctly defined by either
facies or surfaces, followed by a scroll in I, J and K direction
through the model, and a comparison with the original
depositional environment sketches.
The QC should also include a review of the input parameters
and trends used in the facies modelling process: do they make
sense given all available data? Seismic data can also be used
to provide a facies probability trend. And finally, a
comparison should be provided of the facies percentages per
zone for the original well log data versus the resulting 3D
model. Differences may be justified in terms of local and
regional geological trends.
Petrophysical Property Models
Petrophysical property models that are commonly generated
include: net-to-gross, porosity, horizontal and vertical
permeability. All of these parameters can (and should) be
generated from blocked petrophysical logs. The geostatistical
implications of modelling permeability by applying a
transform to a porosity property are such that, as a general
guideline, this technique is not recommended5. Permeability is
usually co-simulated or co-kriged with porosity. Modelling
normally occurs per facies per unit, but can in settings with
limited vertical variation or limited samples, be limited to a
population per facies groups.
Most geocellular modelling packages provide a data analysis
tool, allowing the modeller to review key statistical parameters
like minimum, mean, maximum and standard deviation for
each property per facies, per zone. As a base, these statistical
parameters should remain similar (to what?), or, when
selecting permeability, in the same order of magnitude as the
original well log data, when upscaling from well log to
blocked well log, and then populating the 3D model. There
may be valid reasons to move away from the original values,
as long as these are clearly given.
If the facies clearly separated the net-rock from the non-net
rock, no net-to-gross property is required. However, if (some
of) the facies include both net and non-net rock, like for
example a tidal flat facies, a net-to-gross property is usually
defined. When blocking and modelling Net-to-Gross, one
should consider what happens to the non-net rock. A cell with
a net-to-gross value of lets say 30 percent, should be populated
with the porosity and horizontal permeability value that
matches the average porosity / horizontal permeability of the
net rock only. To ensure this, it is advisable that prior to log
blocking, the intervals that are related to non-net rock are set
to undefined. This ensures that the values modelled for
porosity and horizontal permeability are those that are
applicable to the net rock only. Vertical permeability is always
applicable to the entire bulk-rock volume in a cell and is hence

not affected by the use of a net-to-gross property.


When reviewing net-to-gross, porosity and permeability
models, one keeps in mind the facies model. The easiest way
is to scroll in I, J, and K directions through the model, and
keep on changing the displayed property from facies to the
various property models. A close scrutiny of the cross sections
through the model at the well locations, can reveal a too high
or too low vertical heterogeneity if the blocked wells clearly
stand out from the surrounding 3D model. Generating crossplots (e.g. porosity vs depth or facies) can also be a good way
of visually checking the resulting properties, spotting outliers,
quickly.
Comparison of reservoir parameter histogram values between
the well logs, the blocked logs and the 3D static model is a
good start. A review of the variograms that were used to
model each property is recommended. The variograms used in
porosity and permeability modelling are normally expected to
be similar. Do the variogram ranges make sense when
keeping in mind the facies geometries? Review which
property population techniques have been used (sgs, kriging,
moving average), and if multiple realisations or scenarios
been run.
Often, vertical permeabilities are assigned using a multiplier
on the horizontal permeability at the dynamic model input
phase. A recommended practise involves gathering the entire
subsurface team around core (or even better, outcrop
analogue) data, and assigning KvKh relationships on a
geocellular model scale for each modelled facies, taking into
account the KvKh relationships derived from core plug data.
Volumetrics and Upscaling
When the objective of the static modelling is to provide input
into dynamic simulation, this is where the transfer of data
occurs. The porosity, net-to-gross and permeability models are
exported, and upscaling of the model prior to dynamic
simulation takes place.
To QC the upscaling process, pore volumes should be
compared between the dynamic and static model. Also, by
importing the upscaled dynamic model back into the static
modelling packing, synthetic logs can be created, continuing
the log QC as suggested above.
The creation of a hydrocarbon saturation property in the static
model, generally through means of a set of calculator
expressions that link the height above the fluid contacts and
the permeability (and porosity), allows for a quick and easy
calculation of the hydrocarbon in place. A good QC practise
here is to compare the calculated values at the well locations
with the petrophysical Sw or Sh curve, visually by overlying
curves and cross plot, and arithmetically by calculating the
EHPC at the well from the original well logs and the model
values.
Audit Trail
Documentation of reservoir models is important. Many
software packages allow for uncertainty analysis to take place

SPE 101875

on the models, so a clear tabulation of input and output


parameters that are used is essential.
Although the workflows used in the modelling packages are
generally saved, browsing through the input and output data is
cumbersome. A spreadsheet can be constructed to capture the
key data. Completing this while constructing the reservoir
model can be of great assistance in identification of key
discrepancies. The spreadsheet should contain, as a minimum,
the following:
1. Well names and locations, top and base of reservoir
interval to ensure all disciplines use the same data.
2. Biostratigraphy data providing the correlation
framework.
3. Core data where available (including core shifts
applied).
4. Structural data; the modelled horizons vs the seismic
and stratigraphic markers, keeping track of name and
version/date.
5. Results from the adjustments applied to the grid to tie
the wells in the horizon and zone modelling process.
6. List of all modelled flowunits, with their average
isopach and internal layering method applied, and the
resulting layer numbers for top and base of the zone.
7. Defined percentage of each modelled facies in each
zone, both original (welllog) and blocked percentage,
and a short description of the modelling method used
for these facies.
8. Facies modelling geometry, specifying geometries
used in object modelling and variograms used in
indicator and property simulation.
9. Basic property statistics, for each facies (in each
zone) the min, mean, max and standard deviation of
the net-to-gross, porosity and permeability for both
the original well data and the blocked (upscaled) logs.
Figure 2 shows an example of the various worksheets in such
a spreadsheet.
Conclusions
Quality control of static reservoir models can be a daunting
process. At times it can be hard to differentiate between a QC
of the underlying geological premises, and the resulting
geocellular model. It is a bit like checking a text for the
concept, idea and the spelling and grammar. However, with
clearly stated modelling objectives and deliverables, and a
systematic approach to the physics of building the model,
basic errors can be eliminated and the focus can move to
reviewing the geological soundness of a model.
Acknowledgements
The author would like to thank Woodside Energy Ltd for their
permission to publish this paper, and acknowledge Trey
(Lawrence D. III) Meckel and Steve Twarz for their valuable
input and comments in preparing this paper.
Reference

1.
2.

Merriam-Webster:
Merriam-Webster
Online.
http://www.m-w.com.
Sylvester, Ian F. et al.: Integrated Reservoir
Modelling Enhances the Understanding of Reservoir

SPE 101875

3.

4.

5.

Performance of the Dolphin Gas Field, Trinidad and


Tobago. SPE 94343 presented at the 14th Europe
Biennial Conference, Madrid, Spain, 13-16 June
2005.
Bramlett, Kenneth W. and Craig, Peter A.: Core
characterization of Slope-Channel and ChannelLevee Reservoirs in Ram Powell Field, Gulf of
Mexico. GCSSEPM Foundation Deep-Water Core
Workshop, Northern Gulf of Mexico, Houston,
Texas, March 10, 2002.
Meckel, Lawrence D. III: Core, Log, and Seismic
Characteristics of a High Rate Amalgamated
Channel Reservoir in a Salt-Withdrawal Minibasin:
The UIpper Miocene Above Magenta Sand, Ursa
Field, Northern Gulf of Mexico. GCSSEPM
Foundation Deep-Water Core Workshop, Northern
Gulf of Mexico, Houston, Texas, March 10, 2002.
Chambers, R.L. and Yarus, J.M.: Quantative use of
seismic attributes for reservoir characterization.
RECORDER, Canadian SEG, Vol. 27, June 2002.

SPE 101875

Build Structural

Interpret

Import and Upscale

TIOF-1 [SSTVD]

S TVD -0.10

VS HAL 1.20

Zonelog [Synthetic] v2.4

RCI Gas

RCI Oil

-0.1000 NET_TS 1. 1000

0. 0000

-0. 100 0

0 .0 000

NE T_ TS [S ynt het i c]

1. 10 00

PO R_TSS

POR_TSS [Synt het ic]

0. 4000

0. 40 00

0 KB_TS 10000 1.75 DENSITY 2.75


0

KB _TS [S ynt h etic]

10000

0.60

NEUTRON

-16 76
v2_
0. 18
4_B LAI _REP RO_ D EPTH4. seg y [Grid
20][732
M]

0.00

RCI Wat er
RC I U ndi ff ere nti at ed

RCI Tight

M-CN5-S4

2
6
4
0

621

M-CN5-S3

2
6
6
0

M-CN5-S2

2
7
0
0

2
6
8
0

M-CN4-M3

M-CN4-S3

M-CN4-S2

2
7
2
0

Create synthetic 3D
seismic

M-CN3-M4

2
7
4
0

M-CN3-S4
M-CN3-S3

2
7
6
0

M-CN3-M3

M-CN2-S3

2
7
8
0

M-CN3-S3

M-CN1-S3
MIS2
804

Facies

Re-Import & QC Dynamic


model in Static domain

Pr
ob

Petrophysical

Seismic

Export to
Figure 1: Example of a static
modelling workflow

Net-to-

well

Highest pick Lowest pick UTM E (m) UTM S (m)

Panel

Exploration and Appraisal wells:


Well-1
Well-2
4
Well
Well-3
1 2 3
Well-4
1 2 1
Well
Well-5
2
Well
14
Well-6
1

NF1
NE6
Marker
ND1
NC12
AA
ND22
BB
NC1

Development wells:
Well-7
Well-8
Well-9
Well-10

NE3
NE3
NE4
ND322

Well 1

Actual
Formations

3 4
3 4
3 4
2 3

Top Horizon Name


field x
AA

Fm A

Fm B

CC

CC

unit 2
Well
unit 3Missing
wellunit
14
No
unit 5
wellunit
26
No
wellunit
37
No
wellunit
48
No
unit 9
wellunit
5 10
No
unit 11
unit 12
unit 13

top 14
top 15
top 25
top 26

783xxxx

seismic marker

Md
3075.0
3140.7

X-value
413xxx
413xxx

DFE (m)

Figure 2: Example of the various


sheets in a static model QC
spreadsheet

6.3
28.5

Z10.8
MD
28.48
-yyyy
zzzz
28.3
-yyyy
zzzz
6.3
-yyyy Recovered
zzzz
core

Core number Core top Core base interval Recovery Core shift
783xxxx
yy
(d=down,
783xxxx
yy
1
2835.1
2852.3
17.2
97.2%
d
2.3
783xxxx
yy
2
2853.4
2870.8
17.4
95.1%
d
1.5
783xxxx
yy

number
of
layers

Facies Number

AAA

Y-value
784xxxx
784xxxx

Z-value Horizon before Diff before Horizon after Diff after


-3046.52
-3046.17
-0.35
-3046.52
0
-3133.21
-3137.87
4.66
-3133.21
0
-2752.45
-2755.27
2.82
-2752.45
0
CCC
layering
413xxx
784xxxx -2767.20
-2768.22
1.03
-2767.19
0
Zone Name
Fm top general description
zone no. Av Isop layering cell thickness
no.
413xxx
784xxxx -2738.17
-2744.02
5.85
-2738.52
metres
method 0.35
or # layers
layers

5732.0
5571.0

name_of_horizon_pick_or_grid_petrel

zone 1

zone 5
zone 6
zone 7

facies 2

zone 1
zone 2

1
3

100
0

zone 3
zone 4

8
Facies

33
8

zone 5
zone 6
zone 7

channel
facies
13
6

22

12
25

0
18

other facies
ellipsoid

facies % at wells
facies 3 facies 4 facies 5

94

72

facies 6

facies 7

blablabla
blablabla
blablabla
DDD
blablabla
blablabla
blablabla
blablabla

0
1
2
3
4
5
6

25
8
8
66
26
37
59

x
107

y
97

6
92

61
0

Orientation
Amplitude
10
51
Wavelength
width
thickness
orientation
minor width
major/minor ratio
thickness

205

Modelling
facies modelling approach

no cells total model

assign facies 0
facies transition from delta front to top delta
to tidal flat
0 Facies
sis Geometry data
0
facies transition from delta front to top delta
to tidal flat
240
20
0normal
as above
0uniform
as above
0
2000
0uniform
facies transition
500from tidal flat to coarse 2000
sands. Then object modelling of the bars
uniform
200
2000
and channels in the tidal flat facies.

prop
prop
|| base
prop
|| top
prop
prop

1
3
1
22
2
12
25

1
3
8
22
13
12
25

to

1
2
5
13
35
48
60

1
4
12
34
47
59
84

metres
z
84

876,472

0
0

94
28
30

from

CC fs

top 1
top 2
top 3
top 4
top 5
top 6
top 7

zone
2 part of
seismically traceable
across
region zone
only 3
picked, but not used
in the
zone
4 model

unit 14
unit 15
unit 25
unit 26

facies 1

Unit
AAA
BBB

4670.5
413xxx
784xxxx
name_of_horizon_pick_or_grid_petrel

total reservoir sequence


Formation

Log Depth
2837.4
2854.9

Seq Strat
Major events
>3th Order

name_of_horizon_pick_or_grid_petrel
name_of_horizon_pick_or_grid_petrel

DD
Fm C

408xxx

Well Name
NF3
409xxx
NG22
409xxx
Well 1
NG22
409xxx
Well 2
NF3
409xxx

underlying zone

top 1
top 2
top 3AAA
top 4
top 5
top 6
top 7
top 8
top 9
top 10
top 11
top 12
top 13

BB

NG22
409xxx
783xxxx
base NH
409xxx
783xxxx
X 409xxx Y783xxxx
NH12
NG22408xxx
409xxx783xxxx
783xxxx
NG22
409xxx
783xxxx
408xxx
783xxxx
NG22
409xxx
783xxxx

uniform

normal
uniform
uniform
uniform

Porosity

240
200 facies 1
1 facies 2
2 facies 3
facies 4
facies 5

10

20
2000
3
15

porosity/permeability
major direction
major range
minor range
vertical range
nugget
min

major direction
major range
0.09
minor range
0.07
vertical range
0.06
nugget 0.05
0.05

240
4000
1000
5

well logs
0.1
mean
max

stdev

min

0.05
0.05
0.04
0.05
0.04

0.08
0.08
0.11
0.07
0.07

Upscaled arithm fac bias


mean
max
stdev

240
0.17
0.18
0.22
0.19
0.17

500
0.30
500
0.30
8
0.30
0.25
0.30
0.30

0.15
0.17
0.22
0.19
0.16

0.26
0.24
0.30
0.28
0.27

0.04
0.04
0.04
0.04
0.03

You might also like