2016
List of Contents
Table of Contents
List of Contents ....................................................................................................................................... 1
List of Figures ......................................................................................................................................... 3
List of Tables .......................................................................................................................................... 4
List of Graphs ......................................................................................................................................... 4
List of Equations ..................................................................................................................................... 4
Abstract ................................................................................................................................................... 5
Declaration .............................................................................................................................................. 6
Copyright ................................................................................................................................................ 6
Dedication ............................................................................................................................................... 7
Acknowledgement .................................................................................................................................. 7
1 INTRODUCTION .......................................................................................................................... 8
1.1 Problem Definition.................................................................................................................. 8
1.1.1 Objectives ....................................................................................................................... 9
2 LITERATURE REVIEW ............................................................................................................. 10
2.1 Inverse Distance Weighting .................................................................................................. 10
2.2 Kriging .................................................................................................................................. 11
3 GEOLOGICAL SETTING ........................................................................................................... 12
3.1 Location ................................................................................................................................ 12
3.2 Background ........................................................................................................................... 12
3.3 Regional Geology ................................................................................................................. 13
3.4 Local Geology ....................................................................................................................... 13
3.5 Petroleum System ................................................................................................................. 14
3.5.1 Source ........................................................................................................................... 14
3.5.2 Reservoir ....................................................................................................................... 15
3.5.3 Seal and Trap ................................................................................................................ 15
4 MATERIALS AND METHODS .................................................................................................. 16
4.1 Structural Modelling ............................................................................................................. 16
4.1.1 Fault modelling ............................................................................................................. 16
4.1.2 Pillar gridding and boundary definition ........................................................................ 16
4.1.3 Horizon modelling and depth conversion ..................................................................... 16
1|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
2|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
APPENDIX ........................................................................................................................................... 55
A. Materials and Methods .............................................................................................................. 55
B. Results and Interpretations ........................................................................................................ 60
C. Workflow .................................................................................................................................. 71
List of Figures
Figure 1 Geographical block location and structural framework of the study area (Gullfaks
field) modified after (NPD, 2016). .......................................................................................... 12
Figure 2 Stratigraphic log of the Gullfaks fields indicating the lithology and quality of the
two main reservoir zones of the Brent Group study area modified after (Fossen &
Hesthammer, 2000). ................................................................................................................. 14
Figure 3 Fluvial facie model displaying (A) Plan view of Top Tarbert zone multi-dimensional
flow pattern and (B) Side view showing the facie variations within the Tarbert and Ness
zones. ....................................................................................................................................... 23
Figure 4 Horizontal Variogram maps of upscaled porosity and permeability data ................. 24
Figure 5 Vertical semivariance and sample variogram of upscaled porosity data .................. 25
Figure 6 Horizontal semivariance and sample variogram of upscaled porosity data .............. 26
Figure 7 Cross-plot relationships between porosity and permeability for (A) Upscaled
properties, (B) Kriging Realisation 1 properties, (C) Kriging Realisation 2 properties and (D)
Kriging Realisation 3 properties. ............................................................................................. 28
Figure 8 Porosity distributions through the model indicating a tri-modal histogram
distribution relating to the facies.............................................................................................. 29
Figure 9 Permeability distributions through the model indicating a tri-modal histogram
distribution relating to the facies.............................................................................................. 30
Figure 10 Cross-sections of (A) Realisation 2 IDW porosity model, (B) Realisation 2 Kriging
porosity model, (C) Realisation 3 IDW porosity model, (D) Realisation 3 Kriging porosity,
(E) IDW permeability model, (F) Kriging permeability model and (G) Facie model ............. 31
Figure 11 Spider charts showing porosity comparison of normalised summary statistics of (A)
Realisation 2 and (B) Realisation 3 ......................................................................................... 33
Figure 12 Spider charts showing permeability comparison of normalised summary statistics
of (A) Realisation 2 and (B) Realisation 3............................................................................... 34
Figure 13 Cross-validation analyses between IDW and Kriging petrophysical properties in
the Gullfaks fields. (A) Realisation 2 porosity analysis, (B) Realisation 3 porosity analysis
and (C) Permeability analysis .................................................................................................. 35
Figure 14 Volumetric comparison per reservoir zones for realisation 3 showing (A) Kriging
recoverable oil, (B) IDW recoverable oil, (C) Kriging recoverable gas and (D) IDW
recoverable gas......................................................................................................................... 36
Figure 15 Cross-validation analysis between IDW and Kriging Volumetric results in the
Gullfaks fields for Realisation 2. (A) Segment by segment analysis for recoverable oil and
(B) Segment by segment analysis for recoverable gas ............................................................ 40
3|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure 16 Cross-validation analysis between IDW and Kriging Volumetric results in the
Gullfaks fields for Realisation 3. (A) Segment by segment analysis for recoverable oil and
(B) Segment by segment analysis for recoverable gas ............................................................ 41
Figure 17 Comparison of reservoir zone in (A) Kriging fine grid, (B) IDW fine grid, (C)
Kriging coarse grid, (D) IDW coarse grid for recoverable oil and (E) Kriging fine grid, (F)
IDW fine grid, (G) Kriging coarse grid, (H) IDW coarse grid for recoverable gas. ............... 42
List of Tables
Table 1 Variogram parameters used in porosity and permeability modelling ......................... 19
Table 2 Volumetric Analysis of the Two (2) Realisations based on Kriging and Inverse
Distance Weighting algorithms................................................................................................ 21
List of Graphs
Graph 1 Histogram graph of recoverable oil for Realisation 2 and 3 indication percentage
error bars .................................................................................................................................. 37
Graph 2 Histogram graph of recoverable gas for Realisation 2 and 3 indication percentage
error bars .................................................................................................................................. 37
Graph 3 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable
STOIIP for Realisation 2 ......................................................................................................... 38
Graph 4 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable
GIIP for Realisation 2 .............................................................................................................. 38
Graph 6 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable
STOIIP for Realisation 3 ......................................................................................................... 39
Graph 7 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable
GIIP for Realisation 3 .............................................................................................................. 39
List of Equations
Equation 1 Inverse Distance Weighting method ..................................................................... 10
Equation 2 Collocated cokriging method ................................................................................ 11
Equation 3 Volumetric Calculation ......................................................................................... 20
4|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Abstract
Being able to understand and characterise the behaviour of the subsurface and combining it
with a suitable statistical method gives a higher level of confidence in the model produced.
Interpolation of porosity and permeability data with minimum error and high accuracy is,
therefore, essential in reservoir modelling. The most widely used algorithm for interpolation
is kriging which with enough well data is the best linear unbiased estimator. This research
sought to compare the applicability and competitiveness of inverse distance weighting (IDW)
method using power index of 1, 2 and 4 to kriging when there is sparse data (14 wells) due to
time and budget constraints to calculate hydrocarbon volumes in a fluvial-deltaic reservoir.
After defining the reservoir extent and fluvial facies, different variograms were computed to
adequately quantify the variations in the distribution of porosity and permeability due to
reservoir heterogeneity. Interpolation results, estimated from descriptive statistics, showed
that IDW with a power index of 1 outperformed kriging. The differences in the interpolated
values were, however, insignificant but IDW with power index of 1 indicated the least error
estimation. Volume calculations also showed a marginal difference of 0.08 between IDW
power index of 1 and kriging in the reservoir zones. Cross-validation of hydrocarbon volumes
based on dividing the reservoir into fault segments resulted in an RMSE of ±2.228 MMbbls
and 0.273 BCF for recoverable oil and gas respectively. With inverse distance weight
exhibiting least errors and higher accuracy, the volumetric and statistical results confirms that
when there is less well data in a fluvial-deltaic reservoir, the best porosity and permeability
interpolation choice should be inverse distance weighting method with power index of 1.
5|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Declaration
This dissertation is the student’s original work unless referenced clearly to the contrary, and
that no portion of the work referred to in the dissertation has been submitted in support of an
application for another degree or qualification of this or any other university or other institute
of learning;
Copyright
The author of this dissertation (including any appendices and/or schedules to this dissertation)
owns certain copyright or related rights in it (the “Copyright”) and he has given The
University of Manchester certain rights to use such Copyright, including for administrative
purposes.
ii. Copies of this dissertation, either in full or in extracts and whether in hard or electronic
copy, may be made only in accordance with the Copyright, Designs and Patents Act 1988 (as
amended) and regulations issued under it or, where appropriate, in accordance with licensing
agreements which the University has entered into. This page must form part of any such
copies made.
iii. The ownership of certain Copyright, patents, designs, trademarks and other intellectual
property (the “Intellectual Property”) and any reproductions of copyright works in the
dissertation, for example graphs and tables (“Reproductions”), which may be described in
this dissertation, may not be owned by the author and may be owned by third parties. Such
Intellectual Property and Reproductions cannot and must not be made available for use
without the prior written permission of the owner(s) of the relevant Intellectual Property
and/or Reproductions.
iv. Further information on the conditions under which disclosure, publication and
commercialisation of this dissertation, the Copyright and any Intellectual Property and/or
Reproductions described in it may take place is available in the University IP Policy, in any
relevant Dissertation restriction declarations deposited in the University Library, and The
University Library’s regulations.
6|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Dedication
I dedicate this project work to my parents for having faith enough to invest in me. I pray and
hope that their labour will not be in vain and that they will live long to enjoy the fruits of
their labour.
Acknowledgement
All praise and thanks to the Almighty God for His blessings and mercies. I would like to
show my gratitude to the Otchere and Hammond families for their continued love and
support.
7|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
1 INTRODUCTION
The Inverse Distance Weighting method (IDW) on the other hand, is a simple and time
sufficient method of estimating which does not require as much data in comparison with
Kriging (Al-Hassan & Adjei, 2015). It has been proven to be effective and reliable in ‘some’
aspects. IDW gives you explicit control over the influence that distance has over
petrophysical properties; an advantage you don’t have with Kriging (Naoum & Tsanis, 2004).
This method also has its shortcomings such as; the assumption of interpolated result quality
can decrease, no account of clustering, variogram parameters and scale ignored and can over-
predict for positively skewed or high nugget effects. In view of its shortcomings, Inverse
Distance Weighting method is not reliable in all deposits hence the emphasis on ‘some’
aspects.
8|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Due to the time factor and data inadequacy, Inverse Distance Weighting method and Kriging
method are compared in the Gullfaks field for the Tarbert and Ness reservoir zones in
Norway to verify if Inverse Distance Weighting method can be used as a suitable alternative
algorithm when there is sparse well data and time constraint to drill new wells.
1.1.1 Objectives
The objectives of this project are to;
9|Page
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
2 LITERATURE REVIEW
There have been gluts of studies in many disciplines on Inverse Distance Weighting method
and Kriging in interpolating values (Isaaks & Srivastava, 1989; Matheron, 1989; Goovaerts,
1997; Webster & Oliver, 2001; Rocha, et al., 2012). Both methods are used in many
disciplines with some preferring one or the other. There have been claims that IDW is better
in sedimentary deposits (Al-Hassan & Adjei, 2015) whiles others also attribute Kriging to
being the Best Linear Unbiased Estimator (Skøien, et al., 2006). All interpolation methods
are based on Tobler’s first law of geography, which states “Everything is related to
everything else, but near things are more related than distant things” (Tobler, 1970) hence
choosing an interpolation method should usually be based on knowledge of the depositional
environment of the reservoir as such there should not be only 1 acceptable way of
interpolating results. As such the existence of uncertainty in a reservoir model is as a result of
our lack of subsurface understanding and the knowledge that there is that which we do not
know. This report seeks to exhausts all geological understanding and effects before
comparing a deterministic approach to a statistical method to establish the best interpolator.
1
⁄𝑑𝑃
𝜆𝑖 = 𝑖
∑𝑖=11⁄ 𝑝
𝑛
𝑑𝑖
Where di is the distance between the sample and estimated points, p is a power parameter,
and n represents the number of sampled points used for the estimation.
10 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
2.2 Kriging
The type of stochastic method employed in this research is the Kriging (Gslib) in
combination with collocated cokriging which uses secondary variables and explicitly
accounts for spatial cross-correlation between measured and interpolated variables
(Goovaerts, 1997). This method requires the development of a cross-covariance that is quite
difficult as the relationship between variables needs to be well understood. Co-kriging was
used in defining the cross relationship between porosity and acoustic impedance from seismic
(Doyen, 1988), hence the use of this method in the upscaling and distribution of porosity in
reservoirs. The collocated co-kriging can be mathematically expressed as (Xu, et al., 1992):
Where σ1 and σ2 are stationary standard deviations associated with means m1 and m2 for
primary and secondary variables Z1 and Z2 respectively whiles λ is the weight assigned and
Z1 (xo) is the primary variable estimate.
11 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
3 GEOLOGICAL SETTING
3.1 Location
The Gullfaks field is situated within the Norwegian borders of the North Sea. The discovery
of this field was made in 1978 in the North Eastern part of block 34/10 in the offshore licence
PL050. Its exact location from the town of Bergen is approximately 175 km north-west. The
offshore block which is mainly within a complicated 10-25 km wide NNE-SSW trending
fault blocks covers the eastern half of a total acreage of 55 km2 (Petterson, et al., 1990). See
Figure 1 below.
Gullfaks Shetland
Basin
Figure 1 Geographical block location and structural framework of the study area (Gullfaks field) modified after
(NPD, 2016).
3.2 Background
The block licence owned by Statoil (85%) as major operators, Norsk Hydro (9%) and Saga
Petroleum (6%) as minor operators had its first exploration well spudding on June 20, 1978.
The Jurassic section of the Brent sandstone of the Brent Group was intersected by the first
well and recorded an oil column height of about 162 metres. This initiated the drilling of four
12 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
new exploration wells, mainly in the western part of the field, which was hugely successful.
Three of the wells (34/10-3, -4 and -5) intersected more oil filled Brent sandstone. The other
well (34/10-6) was dry because it intersected 60 metres of thick water bearing Cook and
Statfjord regressive sandstone Formations (Petterson, et al., 1990).
After five (5) years of drilling 14 wells within the Gullfaks field, the successful exploration
was concluded. To summarise, the majority of wells (10) proved economic discoveries of
hydrocarbon with 3 wells having intersected water-bearing sandstones and 1 well abandoned
with reasons not published (Petterson, et al., 1990).
1. Many seismic surveys were not able to identify many of the smaller but important
faults within the basin due to its resolution.
2. The faults that were able to be interpreted on the seismic were either sealing (having
sand-to-sand contact) or not thereby impacting fluid flow across them.
3. Seismic remapping and improved resolution of the extremely faulted locations
identified the Brent sandstones to be much deeper than initially interpreted, with the
Tarbert and Ness Formations being the main reservoir zones in the western part of the
Gullfaks field.
4. New wells identified that the initial water bearing Cook Formation had oil filled sands
in two fault blocks.
13 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
5. Additional well control reports identified that the Tarbert has good lateral correlation
whiles Ness, slightly eroded to the west of the fault blocks, has moderate lateral
continuity over wide areas. A brief stratigraphy is shown in Figure 2 below.
Figure 2 Stratigraphic log of the Gullfaks fields indicating the lithology and quality of the two main reservoir zones of
the Brent Group study area modified after (Fossen & Hesthammer, 2000).
14 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
"Oseberg kitchen" and "Troll kitchen" respectively. The accumulation of the hydrocarbons
migrated from the North of the Snorre field (Petterson, et al., 1990).
3.5.2 Reservoir
The Brent sandstone of the Brent Group constitutes the main oil bearing reservoir of the
Gullfaks field which is of middle Jurassic age. Older sandstones of Lower Jurassic and Upper
Triassic age from the other reservoir Formations. The additional reservoir zones come from
the Cook, Statfjord and Lunde Formations. The two reservoir zones relevant to this research
are the Tarbert Formation, formed in a marginal marine environment by reworked delta plain
deposits and Ness Formation, floodplain deposits with isolated fluvial sandstone. The
reservoir situated at a relatively shallow depth exhibits excellent reservoir properties with
porosity being as high as 0.3 v/v and permeability values around 4000 mD. These excellent
petrophysical properties are as a result of poor cementation resulting in loose unconsolidated
sands. The bad side of this is the probable production of sand with oil (Petterson, et al.,
1990).
15 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
A Domain conversion of the 3D grid and fault model was done using a velocity model to
bridge the gap between time and depth using the seismic reference datum to create a
realisation of the subsurface independent of time. The layer cake method for depth
conversion was used for this process with the velocity model V=V0+KZ used due to the
relative relationship between depth and time, velocity values for K becomes negative. The
resulting 3D grid was used throughout the project.
16 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
created between well tops to define the zones. Layers were then created to define the
thickness, geological deposition and orientation between horizons with an appropriate
division applied to create a geologically reasonable termination of the layers.
17 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
acceptable range of -5 to 5 was applied. An output truncation and cox-box process were then
applied to the wells that fell out of the hypothetically acceptable range.
18 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Major Minor
FLUVIAL Vertical
dir. dir. Azimuth
FACIES Range
Range Range
Floodplain 5000 5000 2 19
Levee 1000 500 12 90
Porosity Channel 5000 3000 2.5 90
Crevasse
850 350 15 45
Realisation Splay
1 Floodplain 1500 1500 15 19
Levee 400 250 5 45
Permeability Channel 5000 3000 2.5 45
Crevasse
1000 850 11 45
Splay
Floodplain 4000 4000 3 19
Levee 1200 500 5 90
Porosity Channel 5000 3000 2.5 90
Crevasse
850 350 10 45
Realisation Splay
2 Floodplain 1500 1500 15 19
Levee 400 250 5 45
Permeability Channel 5000 3000 2.5 45
Crevasse
1000 850 11 45
Splay
Floodplain 5000 5000 2 19
Levee 1000 500 12 90
Porosity Channel 5000 2000 3 90
Crevasse
850 350 5 40
Realisation Splay
3 Floodplain 1500 1500 15 19
Levee 400 250 5 45
Permeability Channel 5000 3000 2.5 45
Crevasse
1000 850 11 45
Splay
4.3.2 Modelling
The porosity model was used as a correlation coefficient for the input secondary variable for
the permeability model. Since primarily porosity and permeability is a factor of facie
distribution, a different porosity model was created using the variogram parameters for the
fluvial facies model to define its anisotropy.
19 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Interpolation process was by the Kriging (Gslib) and Inverse Distance Weighting (IDW)
methods of power index 1, 2 and 4. Figure A 3 in Appendix A illustrates the histogram
comparison of the realisations. Realisation 2 and 3 were selected for further analysis due to
their distribution being more geologically reasonable based on their spatial continuity and
variations from porosity-permeability cross plots.
The property calculator was used to define the water saturation which was expressed as a
function of height above fluid contact, porosity and permeability due to the absence of core
data. The Net to Gross property was also defined using the property calculator as 0.8 based
on Gullfaks literature and limited well logs used in this model.
𝐺𝑅𝑉×𝑁/𝐺×𝑆𝑜 ×∅
𝑆𝑇𝑂𝐼𝐼𝑃 (𝐺𝐼𝐼𝑃) = 𝐵𝑜 (𝐵𝑔)
With GRV the gross volume of rock, N/G the net: gross, So the initial oil saturation, the
porosity and Bo (Bg) formation volume factor. The Bo, Bg and recovery factor was given a
constant value of 1.656, 0.006 and 62% for oil and 74% for gas respectively (Amirbayov, et
al., 2013). Table 2 below shows the total hydrocarbon volumes of the Gullfaks field for all
algorithms and realisations.
4.5 Upscaling
Using a finer and coarser grid size of 50 by 50 and 500 by 500 respectively, the structures
and properties in the original grid were upscaled. Most-of averaging volume weighted was
used to select the most common facie for the upscaled cell. Arithmetic averaging volume
weighted was used to upscale the porosity models for realisation three (3) whiles flow based
20 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
averaging method was used for upscaling the permeability models for realisation three (3) to
define permeability tensors.
Table 2 Volumetric Analysis of the Two (2) Realisations based on Kriging and Inverse Distance Weighting algorithms
REALISATION 2 REALISATION 3
VOLUMETRICS VOLUMETRICS
ALGORITHMS
STOIIP GIIP STOIIP GIIP
(MMBBLS) (BCF) (MMBBLS) (BCF)
Kriging 446.59 89.91 478.04 103.40
Inverse
Distance 496.91 103.65 496.91 102.84
Weighting
Inverse
Distance
515.78 110.18 515.78 109.55
squared
Weighting
Inverse
Distance
522.07 114.17 522.07 113.36
quadruple
Weighting
21 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
22 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure 3 Fluvial facie model displaying (A) Plan view of Top Tarbert zone multi-dimensional flow pattern and (B)
Side view showing the facie variations within the Tarbert and Ness zones.
23 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
The vertical semivariance, shown in Figure 5 , indicates a likely high nugget effect at a 2.1 m
range and a fining upward trend. The two (2) horizontal semivariances for porosity in Figure
6 below indicate a nugget value of zero (0) and an increasing variability. The horizontal
semivariances for the upscaled permeability data in Appendix B Figure B 3 allude to three (3)
cycles, a zero (0) nugget value and an increasing variability. The vertical semivariance,
Appendix B Figure B 4, also gave high nugget value of 0.48 at a range of 2.1 m.
The observed cyclicity is attributed mainly to the sparse data but can also be caused by the
underlying geological periodicity that is the different depositional periods of the channel
sands, floodplain shales and non-channel levee and crevasse splay (hole effect). The
increasing variability is also due to the direction used, with the vertical range being shorter
than the horizontal range due to a much larger horizontal separation between data points.
These results, not divided into facies, provided additional knowledge to the analogues for the
variogram parameters conditioned to facies.
Horizontal Variogram Map of Upscaled Porosity Data Horizontal Variogram Map of Upscaled Permeability Data
24 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
25 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
26 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
5.3.1 Anisotropy
The results of the above variograms in both the major and minor directions exhibit different
ranges in both porosity and permeability upscaled data at approximately the same sill. This is
usually typical of a fluvial meandering deposit where properties such as primary porosity and
permeability have larger ranges along flow direction as against directions perpendicular to the
flow. For the porosity data, the vertical variogram reaches a lower sill than the horizontal
variogram due to areal trends that cause the vertical variogram not to reach full sill of the
porosity property as the horizontal variogram did. The vertical sill of the permeability data
reaches a higher sill value than the horizontal sill as a result of variance increase from
stratification or layering. There is also a decrease in the spatial correlation of the horizontal
variograms for both porosity and permeability with distance. In general, the porosity and
permeability data exhibits both zonal and geometric anisotropy behaviour.
27 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
the cross plot relationship for the three (3) IDW algorithms illustrated in APPENDIX B
Figure B 5.
Figure 7 Cross-plot relationships between porosity and permeability for (A) Upscaled properties, (B) Kriging
Realisation 1 properties, (C) Kriging Realisation 2 properties and (D) Kriging Realisation 3 properties.
28 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure 8 Porosity distributions through the model indicating a tri-modal histogram distribution relating to the facies.
29 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
k
j
i
k
j
i
Figure 9 Permeability distributions through the model indicating a tri-modal histogram distribution relating to the
facies.
30 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
A B
C D
E F
LEGEND
G
Figure 10 Cross-sections of (A) Realisation 2 IDW porosity model, (B) Realisation 2 Kriging porosity model, (C)
Realisation 3 IDW porosity model, (D) Realisation 3 Kriging porosity, (E) IDW permeability model, (F) Kriging
permeability model and (G) Facie model
31 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
32 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure 11 Spider charts showing porosity comparison of normalised summary statistics of (A) Realisation 2 and (B)
Realisation 3
33 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure 12 Spider charts showing permeability comparison of normalised summary statistics of (A) Realisation 2 and
(B) Realisation 3
34 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
A B
Figure 13 Cross-validation analyses between IDW and Kriging petrophysical properties in the Gullfaks fields. (A)
Realisation 2 porosity analysis, (B) Realisation 3 porosity analysis and (C) Permeability analysis
35 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Kriging and IDW of approximately 8% in both recoverable oil and gas and this is illustrated
in Figure 14 below. The Tarbert zone, being classified as a very good reservoir, has the
highest amount of oil and gas. The Gullfaks, proving profitable hydrocarbons, will then be
divided into segments based on the fault model hence further analysis done to quantify the
volumes in each segment and test the comparable results of IDW, ID2W, ID4W and Kriging
are shown in the Graph 3–Graph 6 below. The results also confirm that IDW is the feasible
alternative algorithm to use with realisation 3 still proving better results.
Figure 14 Volumetric comparison per reservoir zones for realisation 3 showing (A) Kriging recoverable oil, (B) IDW
recoverable oil, (C) Kriging recoverable gas and (D) IDW recoverable gas.
36 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
140
RECOVERABLE OIL (MMBBLS)
120
0
Top Tarbert Tarbert2 - Tarbert1 - Top Ness - Ness1 - Top
- Tarbert 2 Tarbert1 Top Ness Ness1 Etive
RESERVOIR ZONES
Graph 1 Histogram graph of recoverable oil for Realisation 2 and 3 indication percentage error bars
35
30
RECOVERABLE GAS (BCF)
25
0
Top Tarbert Tarbert2 - Tarbert1 - Top Ness - Ness1 - Top
- Tarbert 2 Tarbert1 Top Ness Ness1 Etive
RESERVOIR ZONES
Graph 2 Histogram graph of recoverable gas for Realisation 2 and 3 indication percentage error bars
37 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
130
120
110
100
90 KRIGING [2]
80
70 IDW [2]
60 ID2W [2]
50
40 ID4W[2]
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
SEGMENTS
Graph 3 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable STOIIP for Realisation
2
40
35
RECOVERABLE GIIP (BCF)
30
25 KRIGING [2]
20 IDW [2]
15 ID2W [2]
ID4W[2]
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
SEGMENTS
Graph 4 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable GIIP for Realisation 2
38 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
130
120
110
100
90 KRIGING [3]
80
70 IDW [3]
60 ID2W [3]
50
40 ID4W [3]
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
SEGMENTS
Graph 5 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable STOIIP for Realisation
3
40
35
RECOVERABLE GIIP (BCF)
30
25 KRIGING [3]
20 IDW [3]
15 ID2W [3]
ID4W [3]
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
SEGMENTS
Graph 6 Segment by segment comparison of IDW, ID2W, ID4W and Kriging of recoverable GIIP for Realisation 3
39 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure 15 Cross-validation analysis between IDW and Kriging Volumetric results in the Gullfaks fields for
Realisation 2. (A) Segment by segment analysis for recoverable oil and (B) Segment by segment analysis for
recoverable gas
40 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure 16 Cross-validation analysis between IDW and Kriging Volumetric results in the Gullfaks fields for
Realisation 3. (A) Segment by segment analysis for recoverable oil and (B) Segment by segment analysis for
recoverable gas
5.8 Upscaling
Upscaling results in a loss of information and changes in the geological model. There were
slight variations in the statistical comparison of IDW and Kriging in both the fine and coarse
upscaled grids. The differences were negligible as shown in Appendix B Table B 4.
Appendix A Table B 5 shows the comparison of volumetrics of Realisation 3 fine and coarse
upscaled grid between IDW and kriging algorithms. The difference between the total field
recoverable hydrocarbons of the original grid was the same for fine grid upscale. The coarse-
scale grid reduced the original grid total recoverable hydrocarbon difference from 3.9% to
1.3% for oil and 0.5% to 0.4% for gas. For the reservoir zones, Figure 17, the fine-scale grid
41 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
did not impact the original grid recoverable oil for both Kriging and IDW but it rather
decreased the average difference between recoverable gas of the two algorithms from 8% to
6%. The use of a coarser grid for the reservoir zones instead reduced the average percentage
difference of the original grid recoverable oil between kriging and IDW from 8% to 7% but
did not impact significantly the volumes of recoverable gas. Upscaling results further show
the strong correlation between IDW and Kriging regardless of the grid size used hence
enhancing its competitiveness as an alternative to Kriging.
Figure 17 Comparison of reservoir zone in (A) Kriging fine grid, (B) IDW fine grid, (C) Kriging coarse grid, (D) IDW
coarse grid for recoverable oil and (E) Kriging fine grid, (F) IDW fine grid, (G) Kriging coarse grid, (H) IDW coarse
grid for recoverable gas.
42 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
6 Discussion
The discussion is mainly biased to the porosity interpolated values because it was used in the
reservoir volumetric analysis. Permeability results were however moderately discussed so as
to give an indication of what to expect during flow simulation.
6.1 Variogram
A detailed variogram capturing the underlying spatial variation of the reservoir is not only an
important prerequisite for static modelling but dynamic flow simulation. Variograms define
the tortuosity and connectivity between data points clearly impacting the flow behaviour
within the reservoir. This is illustrated in many texts (Armstrong, 1984; Journel & Huijbregts,
1991; Olea, 1995; Goovaerts, 1997) although none has been able to give a robust and
systematic workflow to produce a geologically legitimate variogram model. The different
variogram parameters used to determine the 3-D variations within the model was based on
the summary statistics of the upscale properties. In characterising the petrophysical properties
data set, geological knowledge on flow barriers and diagenetic processes was considered.
Prior to variogram modelling, the data analysis tool was used to transform the data points to a
standardised normal distribution that were slightly skewed to make the variogram
interpretation easier.
The variogram, conditioned to facies from the conceptual fluvial facie model, depicted a lot
of variogram properties that indicated geologic process at different scales. The vertical
variograms for both porosity and permeability exhibited areal trends indicating zonal
anisotropy which is attributed to the zonations within the reservoir. The major and minor
horizontal variogram parameters were inferred from the variogram used to model the fluvial
facie as primarily porosity and permeability are controlled by facie distribution. The
moderations to the parameters were to capture diagenetic processes that vary within the
reservoir. The experimental variogram identified interfacies variations hence indicating zonal
anisotropy in the reservoir due to these stratifications. The differences in the range of spatial
correlation also suggest geometric anisotropy behaviour within the model. The different
anisotropic behaviour corresponds to a specific geological interpretation which is typical of a
fluvial-deltaic reservoir. To define all these behaviours, large directional ranges were used
due to sparse data and this was combined with analogue data (Kupfersberger & Deutsch,
1999) and published variogram parameters (Gringarten & Deutsch, 2001). The parameters
used in modelling the variogram were somehow different due to the number of data points.
43 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
To compute a better variogram to get a better understanding of the spatial variations, outcrop
data or simulation results of a similar field should be used as qualitative inputs. A better
documentation of simulation and outcrop data of a similar producing field will go a long way
in accurately characterising a variogram consistently thereby reducing uncertainty. This
somehow makes an inference that to get the best kriging results more data (wells) is
paramount. The anisotropic ratio used in the Inverse Distance Weighting modelling was also
based on the ranges used in the kriging variograms so as to make a good comparison of the
interpolation methods that characterizes the same porosity and permeability variations.
This project clearly defined anisotropic trends and directions which were used as an informed
input in defining variograms for each reservoir facie with the initial semi-variance model
suggesting that the valid model type to best fit the porosity and permeability data is
exponential. The non-shale facie ranges were kept relatively smaller due to a much larger
internal variation with the opposite applying to the floodplain facie. In the variograms models
for the reservoir petrophysical properties, the majority had a nugget to sill ratio of less than
0.1 indicating a strong spatial structure with a few showing a weak spatial structure with a
nugget to sill ratio of about 0.6. The weak spatial structure is generally associated to the
vertical variograms that mean the majority of the data variations are as a result of the short
lag distance.
44 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
power index was increased as illustrated in Appendix B Table B 1 and Table B 2. IDW
recorded the relatively lesser standard error and deviation values than kriging. This does not
subsequently prove IDW to be better but rather shows its competitive nature when few input
data points (14 wells) are used. The bulls-eye and smoothing effect caused by Inverse
Distance Weighting and Kriging respectively does not represent the raw data geologically
and this is as a result of the polynomial regression of the two algorithms (Myers, 1990).
45 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
unsystematic errors into RMSE (Willmott, 1981), the recorded small errors of the porosity
values indicate a good performance of the model. The proportion of realisation 3 RMSE was
less (2.2%) than that of realisation 2 (4.5%) suggesting the superiority of realisation 3. The
almost similar standard error values for realisation 3 porosity gave the inference that the
deviation will be similar resulting in equal precision in the interpolated values. The equal
sample variance results of 0.002 between IDW and Kriging for realisation 3 porosity
indicates reasonably precise interpolated values. This research paper clarifies that every
geoscience parameter should be taken as uncertain as even the reliable and best interpolation
algorithms do not accurately capture subsurface variability hence the error parameters
recorded. This difference in RMSE and absolute error between IDW and Kriging were
insignificant hence IDW is not prone to biased interpolation estimates. The equal standard
deviation of IDW and Kriging porosity values also reveals the consistency of the two
methods. The low but equal standard errors and deviations of the two methods for both
realisations indicate a slight difference between them and they gave comparable, consistent,
precise and acceptable interpolated values with less than 5% difference in both cases. Both
methods are acceptable for interpolation as the 95% confidence intervals are less than 10% of
the predicted mean. The recorded 95% confidence intervals for both porosity and
permeability realisations prove that IDW shows less uncertainty than kriging.
6.3 Volumetrics
Both visual assessment and statistical comparisons showed that IDW was most likely to give
the best and consistent volumetric results to kriging as compared to ID2W and ID4W which
contradicts a lot of papers (Al-Hassan & Adjei, 2015) but agrees with many as well
(Robinson & Metternicht, 2005; Sajid, et al., 2013). This research does not explicitly put this
contradiction to rest but the findings based on cross-validation results shows compelling
consistency and precision of IDW over ID2W and ID4W. From the zonal hydrocarbon
volumes of realisation 3, the differences between kriging, IDW, ID2W and ID4W is as a result
of smoothing caused by increasing the polynomial. Further comparison as a result of dividing
the reservoir into segments produced slight disparity between the performance of kriging and
the three power polynomials of Inverse Distance Weighting. Whilst the volumetric results
were consistent with the statistics, it was somehow at odds with results of (Al-Hassan &
Adjei, 2015), who reported that ID2W was more accurate to kriging than the other power
polynomials. A possible explanation for this discrepancy is that his work is based on using
ordinary kriging to model gold mineralisation that has been transported and deposited in a
46 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
6.4 Limitations
The whole point of this project is the use of fewer well data hence the true heterogeneity of
the reservoir was not captured. This was, however, defined by the variogram used for facie
modelling. The project had very sparse well log data with the only logs available being GR,
facies, porosity and permeability logs. This clearly made it difficult to determine several
factors like fluid contacts. Although depths from literature were used as fluid contacts depth,
the sealing capacity of the faults would have caused the fluid contacts to differ from each
reservoir segments. The pressure data, if provided, could have made it possible to determine
the different contacts and also know the segments that are in communication. This would
have changed the hydrocarbon volumes but not affected the results achieved.
47 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
algorithms. The use of one (1) realisation for the fluvial facie model causes variability of
possible outcomes. In quantifying the spatial correlation, inadequate data led to the
introduction of substantial uncertainty. Analogue information from published works was used
to address these uncertainties and measure adequately the spatial variability within the
reservoir. The use of three different variogram parameters represented statistically as close as
possible the heterogeneity nature of the reservoir with the exception of realisation 1 which
was not used in further analysis. The use of at least one (1) horizontal well will provide
invaluable data input for variogram modelling of horizontal ranges.
48 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
7 CONCLUSION
The choice of a suitable interpolation method with high accuracy, precision, consistency and
minimum errors is of chief importance in reservoir modelling. This research was divided into
three (3) important parts; quantification of spatial variation within the reservoir, assessing the
accuracy of the interpolated porosity and permeability properties using Kriging and IDW of
various power index algorithms and hydrocarbon volumes calculation in the reservoir.
The general reservoir variogram accounted for two (2) types of anisotropic behaviour
between the horizontal and vertical semivariances:
Inferences drawn from the statistical results of the interpolated properties of this research are
that:
1. Any of the four algorithms can be used as a suitable algorithm for modelling porosity
and permeability properties due to marginally insignificant differences with respect to
accuracy, consistency, and precision.
2. From the statistical parameters, realisation 3 recorded the least interpolation errors
and the most consistency and accuracy with IDW outperformed ID2W and ID4W by
recording the most comparable results to kriging.
3. Cross-validation of realisation 3 IDW and kriging relationship proved to be the best
realisation by recording the least average errors and highest correlation with both
having absolute errors of less than 1%.
4. This study, however, concludes that statistically in a fluvial-deltaic reservoir, Inverse
Distance Weighting algorithm with a power index of 1 is a suitable and better
alternative to kriging when there is sparse well data.
Similar results were achieved in the assessment of hydrocarbon calculations with the
following inferences drawn from the results:
49 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
2. Realisation 2 volumetric comparison between IDW and kriging were about 14% for
recoverable oil and 17% for recoverable gas which is above the accepted threshold of
10% for this research. However, realisation 3 results fell below the acceptable cut-off
for differences recording about 8% for both recoverable oil and gas.
3. Reservoir segments cross-validation and correlation analysis results indicate IDW to
have no significant difference to kriging with absolute errors of 3% for recoverable oil
and 0.7% for recoverable gas.
4. IDW results exhibited a correlational difference to kriging of ±2.2 MMbbls for
recoverable oil and ±0.3 BCF of recoverable gas. With IDW showing statistically
fewer interpolation errors and deviation, it exhibits a higher accuracy than kriging
volumetric results.
5. Grid upscaling, which normally affects models by causing a loss of geological
features and extreme porosity values, did not impact the results. It rather
complemented the robustness of IDW in both fine and coarse grid upscale.
50 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
8 RECOMMENDATIONS
1. More time needs to be spent on variogram models with different parameters not only
for facies but zones as well to adequately capture zonal anisotropy.
2. With enough additional well data, object based modelling should be used for facie
modelling for a more realistic representation of the channel pattern and connectivity.
3. When there are less than 15 wells, Inverse Distance Weighting method should be
employed in porosity and permeability modelling in a fluvial-deltaic shallow marine
reservoir. When more wells are drilled and information is attained, variogram model
can be improved and kriging can be used to attain very accurate result. This will give
better value for money.
4. Inverse Distance Weighting method could be used to estimate hydrocarbon volumes
when there is sparse data, time and budget constraints.
51 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Bibliography
Ahmadov, M. et al., 2012. Technical Report - Part A and B Improved Oil Recovery from Gullfaks -
IOR Challenge 2. [Online]
Available at: http://www.ipt.nynu.no/~kleppe/pub/Gullfaks-Reports-2013/G1-technical.pdf
[Accessed 3 August 2016].
Al-Hassan, S. & Adjei, D., 2015. Competitiveness of Inverse Distance Weighting Method for the
Evaluation of Gold Resources in Fluvial Sedimentary Deposits: A Case Study. Journal of
Geosciences and Geomatics, 3(5), pp. 122-127.
Amirbayov, T. et al., 2013. Subsea compression Gullfaks Sør satellite field EiT - Gullfaks village
2013 group 1. [Online]
Available at: http://www.ipt.ntnu.no/~kleppe/pub/Gullfaks-Reports-2013/G1-technical.pdf
[Accessed 18 July 2016].
Armstrong, M., 1984. Improving the estimation and modelling of the variogram. In: Geostatistics for
Natural Resources Characterization: Reidel. Dordrecht, Holland: s.n., pp. 1-20.
Doyen, P. M., 1988. Porosity from Seismic Data: A Geostatistical Approach. Geophysics 53.
Fossen, H. & Hesthammer, J., 2000. Possible absence of small faults in the Gullfaks Field, northern
North Sea: implications for downscaling of faults in some porous Sandstone. Journal of Structural
Geology, Volume 22, pp. 851-863.
Goovaerts, P., 1997. Geostatistics for natural resources evaluation, New York: Oxford University
Press.
Gringarten, E. & Deutsch, C. V., 2001. Variogram Interpretation and Modeling. Mathematical
Geology, 33(4), pp. 507-534.
Husanovic, E. & Malvic, T., 2014. Review of deterministic geostatistical mapping methods in
Croatian hydrocarbon reservoirs and advantages of such approach. NAFTA, 65(1), pp. 57-63.
Isaaks, E. H. & Srivastava, R. M., 1989. An introduction to applied geostatistics, New York: Oxford
University Press.
Journel, A. G. & Huijbregts, C. J., 1991. Mining Geostatistics. London: Academic Press.
Kupfersberger, H. & Deutsch, C. V., 1999. Methodology for Integrating Analog Geologic Data in 3-D
Variogram Modelling. AAPG Bulletin, 83(8), pp. 1262-1278.
Li, J. & Heap, A. D., 2014. Spatial interpolation methods applied in the environmental sciences: A
review. Environmental Modelling & Software, Volume 53, pp. 173-189.
MacEachren, A. M. & Davidson, J. V., 1987. Sampling and Isometric Mapping of Continuous
Geographic Surfaces. The American Cartographer, 14(4), pp. 299-320.
52 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Matheron, G., 1989. Estimating and Choosing: An Essay on Probability in Practice. 1st ed. Berlin:
Springer-Verlag.
Myers, R. H., 1990. Classical and Modern Regression with Applications. 2nd ed. Boston: PWS-Kent
Publishing Company.
Naoum, S. & Tsanis, I. K., 2004. Ranking Spatial Interpolation Techniques Using a GIS-Based DSS.
Global Nest: the Int. J., 6(1), pp. 1-20.
Olea, R. A., 1995. Fundamentals of semivariogram estimation, modeling and usage. In: Stochastic
modeling and geostatistics: Principles, methods and case studies. s.l.:AAPG Computer
Appliancations in Geology, pp. 27-36.
Petterson, O. et al., 1990. The Gullfaks field: geology and reservoir development, Netherlands:
Springer.
Prosvirnov, M., 2010. Evidence for tectonic control deposition of the Brent Group, northern North
Sea, Stavanger: University of Stavenger, Norway.
Robinson, T. P. & Metternicht, G., 2005. Testing the performance of spatial interpolation techniques
for mapping soil properties. Computer and Electronics in Agriculture, Volume 50, pp. 97-108.
Rocha, M. M., Yamamoto, J. K., Watanable, J. & Fonseca, P. P., 2012. Studying the influence of a
secondary variable in Collocated Cokriging estimates. Anais da Academia Brasileira de Ciéncias,
84(2), pp. 335-346.
Sajid, A. H., Rudra, R. P. & Parkin, G., 2013. Systematic evaluation of kriging and inverse distance
weighting methods for spatial analysis of soil bulk density. Canadian Biosystems Engineering, 55(1),
pp. 1.1-1.3.
Skøien, J. O., Merz, R. & Bloschl , G., 2006. Top-Kriging - geostatistics on stream networks.
Hydrology and Earth System Sciences, Volume 10, pp. 277-287.
Tobler, W. R., 1970. A computer movie simulating urban growth in the Detroit region. Economic
Geography, Volume 46, pp. 234-240.
Webster, R. & Oliver, M. A., 2001. Geostatistics for Environmental Scientists. West Sussex: John
Wiley & Sons Ltd.
Willmott, C. J., 1981. On the validation of models. Physical Geography, 2(2), pp. 184-194.
53 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Xu, W., Tran, T. T., Srivastava, R. M. & Journel, A. G., 1992. Integrating Seismic Data in Reservoir
Modelling: The Collocated Cokriging Alternative. In: Annual Technical Conference of the SPE, 67,
Washington, Proceedings, pp. 833-842.
54 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
APPENDIX
A. Materials and Methods
Figure A 1 Mid-Skeleton Pillar Grid of the Gullfaks field showing active fault models within the reservoir zone.
55 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure A 2 Well section showing original logs and upscaled logs for wells C1 and C2.
56 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure A 3 Porosity-Permeability Cross plot showing facie distribution and relationship for Reservoir zones Tarbert
and Ness.
Figure A 4 Histogram comparison of IDW, ID2W, ID4W and Kriging for (A) Porosity Realisation 1, (B) Porosity
Realisation 2, (C) Porosity Realisation 3 and Permeability models
57 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Major Minor
Fluvial Facies dir. dir. Vertical Range Azimuth
Range Range
Floodplain 5000 700 10 19
Levee 1000 500 10 25
Channel 3500 1500 10 25
Crevasse Splay 850 500 10 25
Table A 2 Summary statistics of porosity data for wells A10, A15, B2, B4, C1 and C2
58 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Table A 3 Summary statistics of permeability data for wells A10, A15, B2, B4, C1 and C2
Normality Test:
0 1.60E-13 0 0 2.07E-32 0
Kolmogorov-Smirnov
59 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
7010 m
BCU
Tarbert
Ness
Figure B 1 Horizons and faults interpretation showing the lateral extensivity of the Gullfaks field and intersected
wells at seismic Inline 550
60 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure B 2 Well section showing the reservoir zonal division, fluvial facies, porosity and permeability logs of wells C 6
and C7
61 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
62 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
63 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Figure B 5 Porosity/Permeability cross plots comparison for (A) Realisation 1, (B) Realisation 2, (C) Realisation 3 and
(D) Upscaled data
64 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
65 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Table B 1 Summary descriptive statistics of Kriging, IDW, ID2W and ID4W interpolated porosity and permeability
properties for realisation 2
REALISATION 2 POROSITY
STATISTICS
IDW ID2W ID4W KRIGING
Mean 0.127 0.125 0.124 0.124
Standard Error 8.958E-05 10.097E-05 11.078E-05 8.419E-05
Median 0.116 0.115 0.115 0.116
Mode 0.119 0.119 0.119 0.207
Standard Deviation 0.041 0.046 0.051 0.037
Sample Variance 0.002 0.002 0.003 0.001
Kurtosis 0.409 0.220 0.069 -0.070
Skewness 0.868 0.820 0.779 0.577
Range 0.292 0.292 0.292 0.292
Minimum 0.003 0.003 0.003 0.003
Maximum 0.295 0.295 0.295 0.295
Sum 26763 26295 26053 24520
Count 210388 210388 210388 197741
Confidence Level
1.756E-04 1.979E-04 2.171E-04 1.650E-04
(95.0%)
REALISATION 2 PERMEABILITY
IDW ID2W ID4W KRIGING
Mean 73.060 75.705 78.498 69.854
Standard Error 0.234 0.252 0.273 0.237
Median 34.487 34.438 35.204 30.650
Mode 2.059 4.609 2.059 88.091
Standard Deviation 104.005 112.338 121.631 105.104
Sample Variance 10816.967 12619.850 14794.220 11046.863
Kurtosis 18.751 22.743 27.707 21.102
Skewness 3.391 3.779 4.192 3.754
Range 1508.177 1508.177 1508.177 1508.177
Minimum 1.362 1.362 1.362 1.362
Maximum 1509.539 1509.539 1509.539 1509.539
Sum 14463443 14986985 15540011 13762557
Count 197966 197966 197966 197020
Confidence Level
0.458 0.495 0.536 0.464
(95.0%)
IDW gives comparable values to Kriging than ID2W and
INFERENCE
ID4W
66 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Table B 2 Summary descriptive statistics of Kriging, IDW, ID2W and ID4W interpolated porosity and permeability
properties for realisation 3
REALISATION 3 POROSITY
STATISTICS
IDW ID2W ID4W KRIGING
Mean 0.127 0.125 0.124 0.123
Standard Error 9.187E-05 10.211E-05 11.167E-05 9.296E-05
Median 0.115 0.115 0.115 0.116
Mode 0.119 0.119 0.119 0.207
Standard Deviation 0.042 0.047 0.051 0.042
Sample Variance 0.002 0.002 0.003 0.002
Kurtosis 0.377 0.214 0.053 -0.388
Skewness 0.901 0.827 0.780 0.519
Range 0.292 0.293 0.293 0.293
Minimum 0.003 0.003 0.003 0.003
Maximum 0.295 0.295 0.295 0.295
Sum 26619 26337 26106 25034
Count 210245 210245 210245 203513
Confidence Level
1.801E-04 2.001E-04 2.189E-04 1.822E-04
(95.0%)
REALISATION 3 PERMEABILITY
IDW ID2W ID4W KRIGING
Mean 73.060 75.705 78.498 69.854
Standard Error 0.234 0.252 0.273 0.237
Median 34.487 34.438 35.204 30.650
Mode 2.059 4.609 2.059 88.091
Standard Deviation 104.005 112.338 121.631 105.104
Sample Variance 10816.967 12619.850 14794.220 11046.863
Kurtosis 18.751 22.743 27.707 21.102
Skewness 3.391 3.779 4.192 3.754
Range 1508.177 1508.177 1508.177 1508.177
Minimum 1.362 1.362 1.362 1.362
Maximum 1509.539 1509.539 1509.539 1509.539
Sum 14463443 14986985 15540011 13762557
Count 197966 197966 197966 197020
Confidence Level
0.458 0.495 0.536 0.464
(95.0%)
IDW gives comparable values to Kriging than ID2W and
INFERENCE
ID4W
67 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Table B 3 Volumetric results of the Gullfaks field for Realisation 2 and 3 showing the comparative results of IDW to
Kriging
REALISATION 2
Algorithm KG IDW Marginal diff
STOIIP GIIP STOIIP GIIP STOIIP GIIP
Volumes
(MMBBLS) (BCF) (MMBBLS) (BCF) (MMBBLS) (BCF)
Field Recoverable 446.590 89.912 496.910 103.600 0.113 0.153
Zone
Top Tarbert -
50.320 12.749 69.190 16.100 0.375 0.263
Tarbert 2
Tarbert2 -
100.640 16.139 113.220 21.370 0.125 0.324
Tarbert1
Tarbert1 - Top
94.350 17.234 94.350 17.690 0 0.027
Ness
Top Ness - Ness1 113.220 17.481 125.800 19.320 0.111 0.105
Ness1 - Top Etive 88.060 26.310 94.350 29.170 0.071 0.109
Average 0.137 0.165
REALISATION 3
STOIIP GIIP STOIIP GIIP STOIIP GIIP
Volumes
(MMBBLS) (BCF) (MMBBLS) (BCF) (MMBBLS) (BCF)
Field Recoverable 478.040 103.400 496.910 102.800 0.039 0.005
Zone
Top Tarbert -
56.610 14.656 69.190 15.790 0.222 0.077
Tarbert 2
Tarbert2 -
106.930 18.788 113.220 21.190 0.059 0.128
Tarbert1
Tarbert1 - Top
100.640 18.964 94.350 17.900 0.063 0.056
Ness
Top Ness - Ness1 119.510 18.894 125.800 19.280 0.053 0.021
Ness1 - Top Etive 94.350 32.101 94.350 28.680 0 0.107
Average 0.079 0.078
68 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
69 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
Table B 5 Volumetric results of the Gullfaks field for Fine and Coarse grid upscaling showing the comparative results
of IDW to Kriging
70 | P a g e
Static Reservoir Modelling Comparing Inverse Distance Weighting to Kriging Interpolation Algorithm in
Volumetric Estimation. Case Study: Gullfaks Field
C. Workflow
Structural Modelling
Reservoir Grid & Boundary Make Horizons, Zones & Domain Conversion
Definition Layering
Quality Check Interpreted Data
Property Modelling
Statistical Analysis
Volume Calculation
Grid Upscaling
71 | P a g e