SPE 93164
Experimental Design as a Framework for Multiple Realisation History Matching:
F6 Further Development Studies
L. Alessio, S. Coca, and L. Bourdon, Sarawak Shell Sdn. Bhd. Malaysia
Copyright 2005, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the 2005 Asia Pacific Oil & Gas Conference and
Exhibition held in Jakarta, Indonesia, 5 – 7 April 2005.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal 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 a proposal of not more than 300
words; illustrations may not be copied. The proposal 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
History matching is traditionally complex and time-
consuming: multiple parameters influence the match and their
inter-dependency produces effects that are difficult to predict.
Defining the match itself can be challenging, since various
indexes or responses can be used: water breakthrough timing,
pressures, layer contributions etc… Consequently, whilst
multiple realisations methodologies are routinely applied for
“green” field development planning, most of the time
incremental activity screening on “brown” fields is done on a
single matching realisation –“the” matched model - with little
confidence that the full range of uncertainties is captured.
Experimental design provides a well-suited framework to
tackle the challenge of multi-realisation history matching,
following these key steps:
• Selection of key parameters with variance analysis,
• Reduction of dimensionality by creating hybrid
parameters, using techniques related to principle
component analysis,
• Predicting matching domains: combination of
parameters levels (once discretised) that are likely to
generate a match. This greatly helps the likelihood of
finding multiple matching realisations, covering the range
of parameter variation.
This methodology was successfully applied in the F6
subsurface studies, aimed at screening field redevelopment
opportunities. F6 is the largest gas field in the Central Luconia
carbonate province, offshore Sarawak (Borneo), having a GIIP
of more than 7 Tscf. With over half the reserves produced,
well capacity is now threatened by the rising aquifer. In order
to safeguard and possibly increase the reserves, a field review
was undertaken to identify further development opportunities,
and a multi-realisation approach was chosen to capture the
effect of key subsurface uncertainties on those activities.
A total of 28 matching realisations were generated,
covering the variation range of the identified key seven
parameters whilst optimising the number of runs performed,
thus saving time. Key to the success of the method lies in the
integration of disciplines to allow the upfront identification of
parameters and their ranges.
The screening of redevelopment options against those
realisations allowed to establish the range of expected
incremental reserves, assess risks, and form a sound basis for
business decisions.
Introduction
F6 is the largest gas field within the Sarawak Shell
portfolio, with a GIIP of over 7 Tscf. The field covers a large
area of approx. 168 km2 and has a gas-bearing interval of over
850 feet thickness. It is an elongated carbonate build-up of
Miocen age, which has steep flanks and a generally flat crest.
Two main units make up the gas accumulation: the Upper
(Zone 1 and Zone 2) reservoir and Lower (Zone 3) reservoir,
separated by an extensive baffle (figure 1). The Lower
reservoir contains almost two-thirds of the gas in place.
Production started in 1987 and the initial field performance
indicated a weak aquifer drive, with a slow water rise. Water
breakthrough occurred in late 2001, and capacity went under
threat since then, and triggered the need to look at infill
drilling and redevelopment opportunities.
To this purpose a comprehensive subsurface review was
undertaken, using the latest information, notably a 3D seismic
dataset shot in 2002, but also innovative technologies such as
multi-attribute imaging to map the internal architecture, and
Experimental Design techniques to perform a multi-parameter
history match, aimed at delivering matching subsurface
realisations covering the whole range of uncertainties.
Experimental Design was mainly used to identify, screen
and rank key subsurface uncertainties, and to generate a match
response surface predicting the quality of the match. A match
response surface is a surrogate model for the numerical
simulation results of the HM. The response surface was then
2 SPE 93164
used to guide the History Match runs to eventually generate a
set of matching realisations.
Initial findings: importance of large-scale
Karstification
F6 production started in 1987 from a total of 11 deviated
wells placed in the central part of the field, drilled on a dataset
of multi-vintage, good quality 2D seismic. Little was known
about the detailed reservoir architecture away from well
penetrations. Following 15 years of production and half of the
reserves produced, a 3D survey was acquired in mid 2002 to
support an infill drilling campaign and a potential field re-
development. Several iterations of Jason Constrained Sparse
Spiked Inversion (CSSI) were performed, following the “3D
All-the-Way” ™ workflow 1.
The 3D seismic data showed strong lateral heterogeneity
within Zone 3 and complex progradational features could be
mapped within Zone 1. Interestingly, as the first 3D dynamic
simulation models were generated on the basis of this initial
static model, it became clear that a combination of enhanced
properties in the lower producing intervals (Zone 3) and
possibly higher volumes were required to match the
combination of contact rise and pressure data in the field 1.
Seismic multi-attribute extractions revealed an extensive
dendritic Karst network 2, 3 covering the vast majority of the
Lower reservoir. Because of their shear size and density
(figure 2), this imaged Karst system could explain the matrix
property enhancement, both in terms of volumes and effective
permeability. A Karst 3D-model was constructed based on the
extracted seismic attributes and used for dynamic history
matching. The open, property-enhancing nature of the
modeled Karst features was confirmed when a good match to
operational data such as total drilling losses encountered in
early development wells was possible 1,2.
Reservoir modelling included two facies: matrix and Karst.
Matrix properties were interpolated from wells (core and log
data available), guided by the Acoustic Impedance (AI) data
derived from seismic inversion. The effective Karst properties
- porosity enhancement and effective permeability - were
identified as variable to be dealt with at the dynamic stage.
Following the findings of the 3D seismic imaging studies,
a comprehensive, multi-parameter history match study was
initiated, aimed at improving our understanding of the range
of future reservoir performance, along with scouting for re-
development opportunities.
The challenge of History Matching
History matching (HM) is traditionally a painstaking
process, for several reasons. Firstly, multiple parameters are
generally found to influence the match, and their combination
at various levels is sometimes found to produce non-linear,
complex effects on the match. Secondly, defining precisely
what constitutes a “match” can be a complex task, because it
is multi-dimensional: the HM quality can be assessed in terms
of water breakthrough times, WGR/WOR trends, pressures,
rates, at different levels: reservoir, wells, layers etc, all this as
a function of time! A combination of these time-dependent
match responses (or indexes) make up the overall match.
History Matching is effectively minimizing a multi-
dimensional (as many as the match responses) time-dependent
function.
We must also recognize that, in the vast majority of cases,
there is an infinite number of solutions to the history-match.
However we most definitely only dispose of a finite amount of
time to complete the study! We must keep in mind why we are
performing a history-match: in most cases, the aim is to use
the HM model(s) to assess the attractiveness of a future
investment (such as infill drilling, redevelopment etc), or a do-
nothing case, in terms of reserves developed, production or
capacity forecast, etc.This is the history-match objective. It
can be expected that with different HM solutions, eg models
satisfying the history, different outcomes would be predicted,
providing therefore an appreciation for the risk or uncertainty
associated with the forecast.
The real challenge of the subsurface team is to produce a
number of scenarios covering the range of outcomes on the
objective (reserves, production forecast etc) that the residual
uncertainty range present at the time of the HM. For this to be
successful and achieved within a constrained time frame, the
authors propose the four conditions to fulfil:
1. The range of scenarios covers the range of
combination of the chosen HM parameters,
2. The number of matching parameters is kept to a
manageable size, to limit the number of possible
combinations,
3. The number of runs performed in search of a history-
match must be optimized: the success ratio of the HM
runs should be as high as possible,
4. Finally, it is important that the chosen HM parameters
have an impact on the objective, not just on the HM,
The first three conditions can be achieved using some
concepts borrowed from Experimental Design techniques. The
fourth one can be achieved by doing some forecasts tests in
the early stages of the HM, whilst selecting the HM
parameters. Such selection process is at the essence of what
can be called an Objective-Driven History Match (ODHM).
The next sections will describe the overall methodology in
more details.
Multi-parameters History Matching: Set up
General Set up methodology and results for F6
The general set up of the HM follows three main steps:
• Firstly identifying the HM parameters:
A total of 15 “matching” parameters were identified as
variables across the seismic, geological and dynamic domains.
SPE 93164
Ranges of possible variation were assigned, based on
(local/global) expert knowledge. In particular, at an early stage
of the study, through application of the 3D-All-The-WayTM
methodology, large-scale Karstification was identified as a
key feature to explain reservoir performance1 (reference 1).
Efforts focused on imaging and mapping the Karst from the
seismic domain into a static model 2. Karst volumes were
transferred and modelled with a numerical simulator (Shell
proprietary MoReS package), where their effective properties
such as permeability and porosity were treated as matching
parameters.
• Secondly defining the individual match responses
(match indexes)
In the case of F6 subsurface studies, three responses were used
to determine the quality of the match: the overall pressure
match, the field water cut timing and the pressure differential
across two main producing layers. Figure 4 shows these match
responses graphically. Score on each response are given to
each realisation, with a gradation from zero (match) to +/- 2
(poor match, either too much or too little). An overall match
response is defined as the geometric average of the 3
responses.
• Thirdly ranking and selecting the (OD)HM
parameters.
Each parameter is firstly tested against HM, using the match
responses as a way of quantifying the impact of each
parameter on the match. Note that this requires to start from a
matching realisation (a reference match), and testing the high
and low levels of each parameter. This is an ANOVA
(Analysis of Variance) technique, or single parameter test: one
parameter only is varied (high and low extremes), with all
other parameters set to the reference level. The impact (in
absolute terms) is measured against each match response and
recorded. Once all parameters have been tested, a Tornado-
type plot is created, see Figure 5.
Testing parameters on the HM is however insufficient, to
arrive to a value-adding multi-parameter HM:
The case for ODHM
Before starting the HM, a key step is to define which
parameters will be used for the History Match. To derive value
from the multi-parameter history match, we must ensure that
the HM parameters are selected not only from their impact on
the HM, but also on the objective or forecast (FC) –we will
use FC and Objective as analogues in the text. The rationale
behind this statement (condition 4 cited above) is that it would
be not only a waste of time but also dangerous to generate
multiple matching realisations that would all have a very
similar response to the FC: it could potentially wrongly trick
the subsurface team into perception narrowing, by under-
estimating the amount of risks associated with the forecast (of
a further development for instance).
In practice, the selection of parameters is made following a
traditional ANOVA (or Tornado plot) approach. Each
parameter is tested against the HM first and their impact on
the HM is evaluated (Figure 5). Theoretically, those
3
parameters should also be screened against the FC (objective)
and their impact plotted out. The figure 3 shows the
(theoretical) screening on parameters to obtain an ODHM
selection of parameters. The benefit of testing the impact on
FC is two-fold:
• Firstly, the HM parameters that do not have an impact on
the FC can be dropped (orange bar), despite the fact that they
influence the match, thereby reducing the complexity of the
run,
• Secondly, some parameters that did not have a large
impact on the HM (and would have been left out) but have a
large impact on the FC (red bar) are kept as part of the HM
process and will feature in the final set of matching
realisations. An example of this could be Sgr, tested on the
HM of a gas field subject to moderate aquifer influx, in the
early years of production.
The authors acknowledge that it is difficult in practice to do a
rigorous testing on FC for parameters that do have a large
impact on the HM but not on the FC (case 1 above), since it
requires to find matching realisations developed by varying
several parameters. Therefore, one must rely on analytical
techniques or general assessment to screen out some
parameters.
In the F6 case, the figure 6 shows the final selection of 7
parameters, through an integrated HM and FC series of tests.
The selected parameters are: Karst (Kkarst) and Matrix
(KhMult) permeability, Matrix porosity (PorAdd), residual gas
saturation (Sgr), Karst volume enhancement (KarstVolMult),
and vertical permeability of the sub-zone 2.3 (Kv11Stretch), a
low porosity interval which effective permeability is not well
known.
Conducting the HM using Experimental design
Experimental Design and Analysis (EDA) or also known
under Design of Experiments (DoE) techniques have started
apprearing in the industry literature as early as the 1970’s 15, 16.
More recently, their use has become more common. These
techniques have been successfully applied to subsurface
uncertainty assessment5, 9, 11, reserves and UR forecasting7, 9, 10
or GIIP evaluation problems4, with the aim of reducing the
number of physical runs done and models built. There is
however less published account of using EDA for HM
purposes, although a few publication offer interesting
examples of application12, 13.
In the F6 subsurface studies, EDA was used to:
1. Ensure that the whole range of combination and variation
of parameters were tested, and the key parameters
selected by providing a rigorous framework for testing the
impact of parameters. This task is essentially performed
by ANOVA testing of all parameters.
2. Predict with better accuracy where, in the multi-
dimensional space of matching parameters (7 here),
possible matches were likely to be found. For this
purpose, a response surface is generated as a surrogate
model to the numerical modelling result of the match. To
4 SPE 93164
ease the a spreadsheet-based visualization tool was
developed.
The HM was conducted following a number of key steps.
First step: Discretisation of the problem
With any number of selected parameters (here we have 7),
conducting a HM in a structured way relies on discretising the
parameters variation range. However, with too few levels, for
instance the traditional three level: Low-Medium-High, the
risk is to find too few actual combinations of parameters that
give a good match. With too many levels, the number of runs
increases dramatically: a problem with p parameters with n
levels per parameters, allows for n^p possible combinations.
For F6, p=7 and n (levels) was chosen to be 5, so there was 2
levels on each side of the reference level.
Having selected the number of parameters and their
number of levels, the next task is to fill the parameter table:
• Assign values to each parameter level
• Assign likelihood of occurrence to each parameter
level (with the sum = 100%).
These two tasks relied heavily on drawing the experience
from all disciplines within and outside the team, generally
refered as expert knowledge. This is an important step
since it allows to define boundaries (eg the maximum
permeability multiplier that will be used). The benefit of
doing this upfront is clear: it will prevent the reservoir
engineer from wandering into “non-geological” territories
and generate HM solutions that cannot be supported.
• Assign the impact on each match response, by using
the result of the ANOVA runs (where each parameter is
varied to its extreme High and extreme Low). As described
earlier, there are three matching responses: the overall
pressure match (P match), the water cut match (Water
match) and the differential pressure between the Upper and
Lower reservoir (dP match).
Figure 7 shows the EDA parameter table used to record the
impact of each parameter (result of the ANOVA tests) on
the three match responses and the probability of
occurrence at each level.
Note that with p=7 and n=5, we have n^p = 78125 possible
HM realisations! To hope scanning the entire range of possible
combination and creating a set of matching realisations
representative of the overall uncertainty range, the reservoir
engineer would require a guide to help select a much reduced
subset of possible parameter combinations: a tool to predict
which combinations are the most likely to provide a history
match.
Second step: Predicting HM solutions
Having performed an ANOVA on each parameter and
quantified the impact on the three match responses, we have
effectively described how the components of the match
(pressure, water cut, pressure differential) are affected by
varying one parameter at the time, all the other being set to the
reference value. This is shown in the table of Figure 7.
The basic concept being applying Experimental Design is
to use the results of the ANOVA runs to generate a “match
response surface”: that is a mathematical relationship between
parameters and the predicted quality of the match. To this
purpose, we have first to define an overall (also called
composite) match response: we chose for F6 the geometric
average of all the match responses (components of the match).
The Composite Match Response is defined as follows:
C.M.R = SQRT { (Pmatch)^2 + (Water match)^2 + (dP
match)^2 }
A perfect match is obtained when the overall match
response is equal to zero; the higher the value for Match
Response the poorer the match is predicted to be. Note that
this choice is not unique, alternatives can be chosen: one could
be to give weights to the different match responses (for
instance: if we felt that a pressure match was much more
important than the water cut match etc…)…
The basic method to generate a response surface is to
perform a multiple parameter regression. This was carried out
and a linear regression model was selected, with no interaction
between parameters:
MRi=1,2,3 = ∑ph=1 { Ah * Fh ( k, m ) }
k is the parameter (eg PorAdd) and varies from 1 to p
(p=7 in our case),
m is the level and varies from 1 to n (with n=5 in our case)
MRi=1,2,3 is the match response: Pmatch,
Watermatch…
Ah is a weighting factor, which can always be
set to 1 with the right choice of the function Fh (k, m)
Fh (k, m) is simply a lookup table that assigns the
relative impact on each match response of the parameter. The
lookup table is essentially the EDA table shown on Figure 7.
This is of course a simplification, and therefore one could
not expect a 100% exact prediction of the match, based on the
regression model.
However, the key objective is not to predict the exact
locations, in the 7-dimensional space of all parameters, where
the actual solution to the HM are, but merely to provide an
approximate solution. This is typically different from the use
of the response surface for UR prediction in a number of
published cases 7,10. The response surface generated is not
explicitely used, but rather implicitely, a guide for the
selection of the actual HM realisations that will be run.
A spreadsheet is set up to generate the predicted response
to the various combinations of the parameters of the match
responses (P match, Water match and dP match) and the
overall, composite match. The likelihood of the parameter
combination is also computed, based on the assignment in the
SPE 93164
EDA table shown on figure 7. The figure 8 shows an extract of
the prediction table (in reality a very long table!)
Step Three: Visualising the predicted matches
The results of the match prediction are used as a guide to
decide “where” to find matches. To aid the search,
visualization was found to add a lot of value.
With relatively simple spreadsheet functionalities, a
visualization device was constructed. The objective of the
visualization is to project the HM response surface (7
dimensions but discretised) onto a map. Taking advantage of
the discretisation of the parameters, we opted for a
representation that allowed to project 5 = 2 + 3 dimension:
Horizontally: 2 parameters combined over 5x5 = 25
columns
Vertically: 3 parameters combined over 5x5x5 = 75 rows
The spreadsheet was built such that parameter
arrangements could be interchanged. We are showing in this
paper the following arrangement:
• Horizontal axis is Kkarst and KhMult, respectively
Karst and matrix permeability variables.
• Vertical axis is PorAdd (matrix volume enhancement),
KarstVolMult (Karst volume enhancement) and Kaqf
(aquifer permeability), arranged along increasing GIIP and
increasing aquifer strength.
Figure 10 shows the detail of the construction.
The visualization tool has a colour coding, based on the
prediction of the Composite Match Response, as defined in the
table 1 below:
Colour Code Composite Match quality
Match
Response
Value
Green < 0.5 Excellent to Good
Yellow 0.5< C.M.R< 2 Moderate to poor
Red >2 Poor to very poor
Table 1. Colour coding of the vizualisation tool
The result is shown on figures 9a and 9b, with different
arrangement (ordering) of the parameter levels along the
horizontal axis. The obtained “5-dimensional map” from the
visualization tool is called a predicted match projection map.
Note that all three different responses can be visualized, along
with the overall, composite match response. In the figures 9a
and 9b, only the composite match is shown.
One observation that was made early on whilst
experimenting with the visualization tool is that, depending on
the way the horizontal and vertical axis were constructed, eg
how the different parameter levels were ordered along those
axises, trends or shapes could be seen.
Examine figure 9a:
5
• Horizontal axis is Kkarst and KhMult, respectively
Karst and matrix permeability variables, arranged
randomly
• Vertical axis is as described previously: PorAdd,
KarstVolMult and Kaqf, arranged in increasing GIIP and
aquifer strength.
The result of the predicted match projection map is a
patchy, trend less profile.
However, if one arranges the horizontal axis in the
following way, by defining an intermediate variable Keff:
Keff = { KmAvg * KhMult + Kkarst } / 2
Where:
KmAvg is the geometric average of the matrix
permeability
Keff is appoximatively the average effective permeability
of the whole reservoir, matrix and Karst contributions
combined. Note that the presented definition of Keff is
somewhat arbitrary choice, and alternatives can be tried.
Figure 14 shows the increasing Keff arrangement vs. a
sequential arrangement of the horizontal axis (note it is shown
vertically !).
If the horizontal axis of the predicted match projection
map is arranged along increasing Keff, then a trend appears, as
shown on figure 9b. We are looking at a simple case of
creating a principle component: the HM solution (defined as
Composite Match = 0) seems to align itself along a GIIP –
effective permeability (Keff) trend. This is physically
expectable in a large, elongated gas field such as F6: the water
rise will increase with a lower effective permeability (more
coning) but would be compensated by a larger available GIIP;
likewise, the P/z response is a function of the GIIP accessed,
which increases if the effective permeability is increased, or if
the overall available GIIP is greater. Our analysis confirms the
qualitative expectation, but also provides quantification.
Step Four: match density map and deciding which HM
cases to run
With p=7 parameters, and n=5 levels per parameters, a
total of n ^ (p-5) = 25 predicted match projection maps can be
generated.
Mathematically, those are projections of the match
function, our response surface, onto a 5-dimensions
hyperplane for Sgr=x and Kv11Stretch=y. On each of those, a
variable number of matches are predicted. To ensure that we
are both covering the full range of parameter variations but
also targeting the most realistic (or likely) matches, we created
a map of match “likelihood”. This is a 2-dimensional
representation of match density against Kv11Stretch (vertical
permeability of the baffle) and Sgr (residual gas saturation):
those parameters that are not shown on the visualization tool
map presented in figure 9a and 9b.
Each of the predicted match projection maps shows the
predicted response of the match (composite or overall match
6 SPE 93164
response) for 5^5 = 3125 different parameter combinations of
the 5 selected parameters: Kkarst, KhMult, PorAdd,
KarstVolMult, Kaqf. Because a likelihood of occurrence was
assigned to each parameter level in the EDA table (shown on
figure 7), each of those 3125 “points” carries a probability of
occurrence – this is simply, based on the team’s knowledge,
the chance that such geology could exist (matching or not the
history). By summing up the likelihoods of the predicted
matching “points” (shown in green), for each predicted match
projection map, and plotting the computed values of the 25
maps, we obtain the match density map.
For a (x,y) point, where x varies along Kv11Stretch and y
along Sgr
Match Density (x,y) =
∑match projection map { δ(M) * P(M) }
where
M is a given point within the match projection map
δ is such that δ (M)=0 if the point M corresponds to a
match, eg MI (M) = 0
P(M) is the probability of occurrence of the point M,
computed by multiplying the likelihood of each of the
parameter levels associated with M.
The match density map is used to further reduce the
number of tentative HM realisations that will be run, by
selecting only a subset out of the 25 predicted match
projection maps. Figure 11 shows the obtained match density
map and the selected “regions” i.e. predicted match
projections maps that will be scanned for actual matching
realisations.
We are now ready to start the HM.
Step Five: Running the HM and collecting matching
realisations
The HM is conducted by scanning for HM solutions
(matching models) across the 5 selected predicted match
projection maps. The idea is to run a number of scenarios,
around the areas of predicted solutions i.e. the green patches in
Figure 12a.
Since we are trying to generate a number of matching
realisations across the range of parameters, on a particular
map we will focus on a number of GIIP-effective
permeability-aquifer strength regions, starting from the
extremes (high and low ends of the spectrum).
The HM runs were done using the Shell proprietary
software MoReS, configured with the Scenario Manager
facility, which allows for multiple cases to be run sequentially
from a single source file. The results of the runs are analysed,
the quality of each HM realisation accessed against the three
match responses, and logged into the spreadsheet. The
spreadsheet allows for the results of the match to be
visualized, so that actual match projection maps can be
generated and compared to the predicted ones. Figure 12a
shows the full-scale predicted match projection map for the
region 1 and figure 12b shows the corresponding actual one,
after running a number of HM cases, and recording the result
in the spreadsheet. Note that the colour coding is the same in
both versions of the map, and that the grey shading is used for
untested cases (no HM run) in the actual map (figure 12b).
Since we know from EDA how each of responses are
affected by directional changes of a given parameter, not all
HM cases run need to be analysed in detail: we can deduce
that, for instance, if one HM run with a given horizontal
coordinate (Kkarst, KhMult) suffers early water breakthrough
at a given vertical position (PorAdd, KarstVolMult, Kaqf),
there is no chance of getting a match cases by going leftwards
–towards lower effective permeability. This further reduces
the amount of work required analyzing the runs.
Examining figure 12b vs. figure 12a, we can also make
the following remarks:
• The actual matches align along a similar trend as
predicted. However,
• The actual matches are not exactly where we expected
them to be, they are displaced, and
• The actual matching domain (span of the green
patches) is smaller than the predicted one. This is a
consequence of neglecting cross-parameter interactions
and over-estimating the compensation one parameter can
do over another (for instance, GIIP vs. effective
permeability): the proposed surface response analytical
formulation is only approximate.
Scanning the selected 5 predicted match projection maps, a
total of 28 matching HM realisations were collected (Figure
15). Interestingly, despite the relative maturity of the field,
these matching realisations could be found within a 20%
relative volume range, associated with combinations of Karst
and matrix permeabilities varying by a factor 1 to 4.
Those realisations were run in forecast mode, to validate
the range of recoveries that were carried using a more
conventional probabilistic range (RF range) and further
development opportunities testing was carried out, providing a
solid, scenario-based evaluation.
Remarks and discussion on the methodology
Efficiency of the method
To demonstrate the efficiency of the method, we propose
to look at the condition 3 cited earlier: “the number of runs
performed in search of ahistory-match must be optimized: the
success ratio of the HM runs should be as high as possible”.
In this study a total of some 430 scanning runs were
performed in about 30 packages (set of scenarios) to cover the
range of selected parameter variation. Whist a total of 78125
parameter combinations are possible, we know from the match
SPE 93164
density maps that most of them are in the 5 selected regions.
Hence, we actually restricted our search for HM within a
smaller set of possible combinations 5 x (5^5) = 15625. Out of
the 430 runs created, only ca 280 were analysed (refer
previous section). Out of those 280 runs, 28 were good
matches. This is a 10% accuracy: 1 in 10 runs is a match,
whist performing a 7 parameter history match and covering
most of the associated uncertainty range.
10% may not seem that accurate –that is largely the result
of the approximation of the HM surface response. However, if
one looks at the number of runs required to achieve the
objective of covering the full variation range on the 5 regions
the statistics are much more impressive: only 430 runs were
required out of a possible 15625, that is less than 3% !
The match density map is a very useful tool in determining
where to concentrate efforts in finding HM realisations. Two
approaches there are possible:
• The first one –presented here- is to focus efforts in
areas of highest match likelihood: computed likelihood of
occurrence of a particular parameter combination with
predicted instances of match.
• The second one is to simply compute the instances of
match over the total possible combination within a
predicted match projection map. The selection of regions
could be done on that basis, illustrated in Figure 13. The
merit of this approach is to not be potentially influenced by
an a-priori assessment of the likelihood of a parameter
variation away from the reference level.
Some improvements possible
The team experienced that running a large number of
realisations is not too cumbersome, provided that the HM
model is not too large –ours ran in approx 1hrs- and that
packages of realisations could be run in one go rather than one
at a time. However, the authors identified two areas of
improvements for the methodology and tools:
• Firstly, automating the match assessment process: the
analysis of the runs was done by visual inspection, using
simple metrics such as pressure off by more than 100psi
etc... At the time of the study, although an automated
method was tried, it was not found to be as reliable as
visual inspection. Nevertherless, reliably automating the
assessment of the match would save time.
• Secondly, implementing a learning loop into the
derivation of the match response surface. The regression
was essentially based on ANOVA runs (and a few 2-
parameter runs), and kept the same through the whole
progress of the project. However, as more scanning runs
are generated in search of history match solutions, the
regression could be improved by incorporating the results
of those runs, either via multiple parameter regression or a
neural network algorithm.
7
Conclusions
The method presented to conduct a multi-parameter history
match in a fast and efficient way was developed by borrowing
some basic concepts of Experimental Design and multiple
regression, and developing some innovative solutions to
visualisation of the match response surface.
Although an evident key benefit of this method is to
rapidly generate a series of matching realisations covering the
range of uncertainties, it is important to note the learning and
improved reservoir understanding value of following this
methodology when starting a HM study. In particular, going
through, with the whole subsurface team involved, the first
steps of: parameter identification, parameter range assessment,
and ANOVA runs, will go a long way towards improving the
team’s shared understanding of the reservoir.
This methodology and the associated spreadsheet tool are
now tested further within Sarawak Shell in a few ongoing
studies, where a similar HM objective is pursued.
Acknowledgments
The authors would like to thank Shell and Petronas for
authorizing the publication of this paper. Special thanks to the
members of the extended team that have been involved in the
various stages of the study, but also to the Sarawak asset
reservoir engineering group for providing a good sounding
board. Special Thanks to Mireille Toulekima and Olivier
Wambersie for proof-reading this paper and Gerco Janssen for
sharing some bright thoughts on the subject.
References
1. L.Alessio, S. Coca, L.Bourdon, 2004, "3D-All-The-Way"
In F6 Field Further Development. SPE 87014 presented at
the SPE Asia Pacific Conference on Integrated Modelling
for Asset Management held in Kuala Lumpur, Malaysia,
29-30 March 2004.
2. L. Bourdon, L. Alessio, S. Coca, 2004, Karst Identification
and Impact on Development Plan. SPE 88520, presented at
the SPE Asia Pacific Oil and Gas Conference and
Exhibition held in Perth, Australia, 18–20 October 2004.
3. L. Bourdon, 2004, Unravelling the reservoir architecture
from 3D seismic using multi-attribute Volume
interpretation: Central Luconia Karst Modelling.
RGC2004, internal Shell event.
4. Cheong, Y.P and R. Gupta, 2004, Experimental Design and
Analysis Methods in Multiple Deterministic Modelling for
Quantifying Hydrocarbon In-Place Probability Distribution
Curve, SPE 87002, presented at the SPE Asia Pacific
Conference on Integrated Modelling for Asset Management
held in Kuala Lumpur, Malaysia, 29-30 March 2004.
5. Cheong Y.P, R. Gupta, 2003, Experimental Design in
Deterministic Modelling: Assessing Significant
Uncertainties. SPE 80537.
6. Robert G. Loucks, 1999, Paleocave Carbonate Reservoirs:
Origins, Burial-Depth Modifications, Spatial Complexity,
8 SPE 93164

and Reservoir Implications. AAPG Bulletin, V. 83, No. 11,

P. 1795-1834.
7. Cheong Y.P, R. Gupta, Smith G. et al, Experimental
Design Methodology for Quantifying UR Distribution
Curve - Lessons learnt and still to be learnt, SPE 88585
8. White C, Royer S., 2003, Experimental Design as a
Framework for Reservoir Studies. SPE 79676.
9. Corre B., Thore P, V.de Feraudy, Vincent G., 2000,
Integrated Uncertainty assessment for project evaluation
and risk analysis. SPE 65205.
10. Van Elk J.F, Guerrera L., Vijayan K and Gupta R., 2000,
Improved Uncertainty management in field development
studies through the application of the experimental design
method to the multi-realisation approach. SPE 64462.
11. Charles T, Guemene J.M, Corre B, Vincent G, Dubrule O.,
2001, Experience with the quantification of subsurface
uncertainties. SPE 68703.
12. M. Feraille, F. Roggero,E. Manceau, L.Y. Hu, I. Zabalza-
Mezghani, L. Costa Reis, Application of Advanced History
Matching Techniques to an Integrated Field Case Study.
SPE 84463
13. A. Eide, L. Holden, E. Reiso, S. Aanonsen, 1994,
Automatic History Matching by use of Response Surfaces
and Experimental design, ECMOR 7-10 June 1994
14. T. Egeland, E. Hatlebakk, L. Holden, E.A Larsen,
Designing Better Decisions, 1992, SPE 24275.
15. Saxena Umesh, Vjekoslav Panelic, Factorial Designs as an
Effective Tool in Mining and Petroleum Engineering,1971,
SPE 3333.
16. Sawyer et al, Factorial Design of Wet combustion drive,
1974, SPE 4140
SPE 93164 9

Figures

Figure1: Cross section of the field (seismic attribute is CSS

Absolute Acoustic Impedance)

W E
Zone 1

Zone 2
Zone 3

Figure 2: Multi-attributes seismic interpretation showing Karst

features in the Zone 3
Top structure (Zone3) semblance mapping, highlighting
the large scale Karst dendritic features
10 SPE 93164

Figure 3: Objective Driven History Match

Highlighting the key differences between a traditional
history match, and a ODHM or HM conducted with an
objective. The horizontal bars represent the different
parameters.

ODHM workflow
Tornado Chart
Relative impact
HM parameters ODHM parameters
HM

FC parameters
FC

Binned parameters

Objective Driven History Match allows to develop HM scenarios with

focus on the uncertainties/parameters that have an impact on the forecast
(objective).

Figure 5: Ranking of HM parameters (against sum of all match

Figure 4: Match responses (also called Match indexes) responses)

Ranking of HM parameters (Pressure, Water and dP)

0 1 2 3 4 5 6 7

Kv11Stretch
Kkarst
KarstVolMult
PorAdd
SatMod
KaqfMult
KhMult
GBVaqfMult
Sgr
Nwi
Karst Geometry
Top Carbonate
Kzmin
KvMult
KvKhKarst
SPE 93164 11

Figure 6: Final ranking of ODHM parameters

Screen ODHM parameters: FC and HM Test

Further reduction (FC test):
Similar impact: select one only

Ranking of HM parameters (Pressure, Water and dP)

0 1 2 3 4 5 6 7

Kv11Stretch 1. Key HM
Kkarst Large parameters:
KarstVolMult
PorAdd impact on Reduce further nu mber
SatMod HM of parameters
KaqfMult
KhMult
GBVaqfMult
Sgr
Nwi 2. Marginal HM
Karst Geometry
Top Carbonate
Low parameters:
Kzmin impact on Forecast Test
KvMult
KvKhKarst
HM

Figure 7: Complete EDA parameter table (7 parameters)

param likelyho Water dP

param likelyho Water dP
level value od P match match match
Kkarst level value od P match match match
KarstVolMult 1 1 20% -1 1 0
1 100 5% -2 2.5 0
2 1.2 30% 0 0 0
2 200 25% -1 1.5 0
3 1.4 30% 1 -0.5 0
3 400 40% 0 0 0
4 1.6 15% 2 -1 0
4 800 25% 1 -1 0
5 2 5% 2 -2 0
5 1000 5% 2 -2 0
param likelyho Water dP
param likelyho Water dP
level value od P match match match level value od P match match match
KhMult 1 1 20% -1 0.5 0 KaqfMult 1 3 10% 0 0 0
2 1.5 25% 0 0 0 2 2 30% 0 0 0
3 2 30% 0 0 0 3 1 30% 0 0 0
4 3 15% 0.5 -0.5 0 4 0.1 20% -1 -1 0
5 4 10% 1 -1 0 5 0.01 10% -1 -2 0

param likelyho Water dP param likelyho Water dP

level value od P match match match level value od P match match match
PorAdd 1 -0.02 5% -2 1 0 Kv11Stretch 1 0.01 5% -2 -2 2
2 -0.01 10% -1 0.5 0 2 0.1 15% -1 -1 1
3 0.01 40% 0 0 0 3 0.5 30% 0 0 0
4 0.02 30% 1 -1 0 4 1 45% 0 0 0
5 0.03 15% 1 -2 0 5 10 5% 1 0 0

param likelyho Water dP

level value od P match match match
Sgr 1 0.1 20% 0 -1 0
2 0.15 25% 0 -0.5 0
3 0.2 30% 0 0 0
4 0.3 15% 0 1 0
5 0.4 10% 0 2 0
12 SPE 93164

Figure 8: Predicting the response matches and the overall

(composite) match

Predicted index matches

Total Match
Pressure Predicted (composite
Sgr Kv11Stretch KaqfMult KarstVolMult PorAdd KhMult Kkarst Match Water Match dP Match Prob P,W,dP)
0.1 10 3 1 -0.02 1 100 -5 4 0 1.0E-07 6
0.1 10 3 1 -0.02 1 200 -4 3 0 5.0E-07 5
0.1 10 3 1 -0.02 1 400 -3 1.5 0 8.0E-07 3
0.1 10 3 1 -0.02 1 800 -2 0.5 0 5.0E-07 2
0.1 10 3 1 -0.02 1 1000 -1 -0.5 0 1.0E-07 1
0.1 10 3 1 -0.02 1.5 100 -4 3.5 0 1.3E-07 5
0.1 10 3 1 -0.02 1.5 200 -3 2.5 0 6.3E-07 4
0.1 10 3 1 -0.02 1.5 400 -2 1 0 1.0E-06 2
0.1 10 3 1 -0.02 1.5 800 -1 0 0 6.3E-07 1
0.1 10 3 1 -0.02 1.5 1000 0 -1 0 1.3E-07 1
0.1 10 3 1 -0.02 2 100 -4 3.5 0 1.5E-07 5
0.1 10 3 1 -0.02 2 200 -3 2.5 0 7.5E-07 4
0.1 10 3 1 -0.02 2 400 -2 1 0 1.2E-06 2
0.1 10 3 1 -0.02 2 800 -1 0 0 7.5E-07 1
0.1 10 3 1 -0.02 2 1000 0 -1 0 1.5E-07 1
0.1 10 3 1 -0.02 3 100 -3.5 3 0 7.5E-08 5
0.1 10 3 1 -0.02 3 200 -2.5 2 0 3.8E-07 3
0.1 10 3 1 -0.02 3 400 -1.5 0.5 0 6.0E-07 2
0.1 10 3 1 -0.02 3 800 -0.5 -0.5 0 3.8E-07 1
0.1 10 3 1 -0.02 3 1000 0.5 -1.5 0 7.5E-08 2
0.1 10 3 1 -0.02 4 100 -3 2.5 0 5.0E-08 4
0.1 10 3 1 -0.02 4 200 -2 1.5 0 2.5E-07 3
0.1 10 3 1 -0.02 4 400 -1 0 0 4.0E-07 1
0.1 10 3 1 -0.02 4 800 0 -1 0 2.5E-07 1
0.1 10 3 1 -0.02 4 1000 1 -2 0 5.0E-08 2
0.1 10 3 1 -0.01 1 100 -4 3.5 0 2.0E-07 5
0.1 10 3 1 -0.01 1 200 -3 2.5 0 1.0E-06 4
0.1 10 3 1 -0.01 1 400 -2 1 0 1.6E-06 2

Figure 9a: Visualisation tool: predicted match projection map Figure 9b: Visualisation tool: predicted match projection map
(random horizontal arrangement) (horizontal arrangement along increasing average effective
reservoir permeability)

--------------> Increasing Heterogeneity K 33 KhMult --------------> Increasing Average K

33 KhMult
Kkarst
Kkarst
1 1.5 2 1 3 1.5 2 4 3 4 1 1.5 2 3 4 1 1.5
4 3 2 4 1.5 1 3 2 1.5 1 4 3 2 1.5 1 4 3
KaqfMult GIIP KarstVolMult PorAdd 100 100 100 200 100 200 200 100 200 200 400 400 400 400 400 800 800
KaqfMult GIIP KarstVolMult PorAdd 100 100 100 200 100 100 200 200 200 200 400 400 400 400 400 800 800 3 5.79 1 -0.02 6.4 5.3 5.3 5.0 4.6 3.9 3.9 3.9 3.2 2.5 3.4 2.2 2.2 1.6 1.0 2.1 1.0
3 5.79 1 -0.02 3.9 4.6 5.3 2.5 5.3 6.4 3.2 3.9 3.9 5.0 1.0 1.6 2.2 2.2 3.4 1.0 0.7 2 5.79 1 -0.02 6.4 5.3 5.3 5.0 4.6 3.9 3.9 3.9 3.2 2.5 3.4 2.2 2.2 1.6 1.0 2.1 1.0
2 5.79 1 -0.02 3.9 4.6 5.3 2.5 5.3 6.4 3.2 3.9 3.9 5.0 1.0 1.6 2.2 2.2 3.4 1.0 0.7 1 5.79 1 -0.02 6.4 5.3 5.3 5.0 4.6 3.9 3.9 3.9 3.2 2.5 3.4 2.2 2.2 1.6 1.0 2.1 1.0
1 5.79 1 -0.02 3.9 4.6 5.3 2.5 5.3 6.4 3.2 3.9 3.9 5.0 1.0 1.6 2.2 2.2 3.4 1.0 0.7 0.1 5.79 1 -0.02 6.7 5.6 5.6 5.4 4.9 4.3 4.3 4.3 3.6 3.0 4.0 3.0 3.0 2.5 2.2 3.0 2.2
0.1 5.79 1 -0.02 4.3 4.9 5.6 3.0 5.6 6.7 3.6 4.3 4.3 5.4 2.2 2.5 3.0 3.0 4.0 2.2 2.1 0.01 5.79 1 -0.02 6.3 5.2 5.2 5.1 4.6 4.0 4.0 4.0 3.5 3.0 4.0 3.2 3.2 2.9 2.8 3.4 2.8
0.01 5.79 1 -0.02 4.0 4.6 5.2 3.0 5.2 6.3 3.5 4.0 4.0 5.1 2.8 2.9 3.2 3.2 4.0 3.2 2.9 3 6.54 1 -0.01 5.3 4.2 4.2 3.9 3.5 2.8 2.8 2.8 2.1 1.4 2.2 1.1 1.1 0.5 0.5 1.0 0.5
3 6.54 1 -0.01 2.8 3.5 4.2 1.4 4.2 5.3 2.1 2.8 2.8 3.9 0.5 0.5 1.1 1.1 2.2 1.8 1.1 2 6.54 1 -0.01 5.3 4.2 4.2 3.9 3.5 2.8 2.8 2.8 2.1 1.4 2.2 1.1 1.1 0.5 0.5 1.0 0.5
2 6.54 1 -0.01 2.8 3.5 4.2 1.4 4.2 5.3 2.1 2.8 2.8 3.9 0.5 0.5 1.1 1.1 2.2 1.8 1.1 1 6.54 1 -0.01 5.3 4.2 4.2 3.9 3.5 2.8 2.8 2.8 2.1 1.4 2.2 1.1 1.1 0.5 0.5 1.0 0.5
1 6.54 1 -0.01 2.8 3.5 4.2 1.4 4.2 5.3 2.1 2.8 2.8 3.9 0.5 0.5 1.1 1.1 2.2 1.8 1.1
0.1 6.54 1 -0.01 5.6 4.5 4.5 4.3 3.8 3.2 3.2 3.2 2.5 2.0 3.0 2.1 2.1 1.8 1.8 2.2 1.8
0.1 6.54 1 -0.01 3.2 3.8 4.5 2.0 4.5 5.6 2.5 3.2 3.2 4.3 1.8 1.8 2.1 2.1 3.0 2.5 2.1 0.01 6.54 1 -0.01 5.2 4.1 4.1 4.0 3.5 3.0 3.0 3.0 2.5 2.2 3.2 2.5 2.5 2.5 2.7 2.8 2.7
0.01 6.54 1 -0.01 3.0 3.5 4.1 2.2 4.1 5.2 2.5 3.0 3.0 4.0 2.7 2.5 2.5 2.5 3.2 3.5 3.0 3 6.55 1.2 -0.02 5.0 3.9 3.9 3.6 3.2 2.5 2.5 2.5 1.8 1.1 2.1 1.0 1.0 0.7 1.0 1.1 1.0
3 6.55 1.2 -0.02 2.5 3.2 3.9 1.1 3.9 5.0 1.8 2.5 2.5 3.6 1.0 0.7 1.0 1.0 2.1 2.2 1.6 2 6.55 1.2 -0.02 5.0 3.9 3.9 3.6 3.2 2.5 2.5 2.5 1.8 1.1 2.1 1.0 1.0 0.7 1.0 1.1 1.0
2 6.55 1.2 -0.02 2.5 3.2 3.9 1.1 3.9 5.0 1.8 2.5 2.5 3.6 1.0 0.7 1.0 1.0 2.1 2.2 1.6
1 6.55 1.2 -0.02 5.0 3.9 3.9 3.6 3.2 2.5 2.5 2.5 1.8 1.1 2.1 1.0 1.0 0.7 1.0 1.1 1.0
1 6.55 1.2 -0.02 2.5 3.2 3.9 1.1 3.9 5.0 1.8 2.5 2.5 3.6 1.0 0.7 1.0 1.0 2.1 2.2 1.6
0.1 6.55 1.2 -0.02 5.4 4.3 4.3 4.1 3.6 3.0 3.0 3.0 2.5 2.1 3.0 2.2 2.2 2.1 2.2 2.5 2.2
0.1 6.55 1.2 -0.02 3.0 3.6 4.3 2.1 4.3 5.4 2.5 3.0 3.0 4.1 2.2 2.1 2.2 2.2 3.0 3.0 2.5 0.01 6.55 1.2 -0.02 5.1 4.0 4.0 4.0 3.5 3.0 3.0 3.0 2.7 2.5 3.4 2.8 2.8 2.9 3.2 3.2 3.2
0.01 6.55 1.2 -0.02 3.0 3.5 4.0 2.5 4.0 5.1 2.7 3.0 3.0 4.0 3.2 2.9 2.8 2.8 3.4 4.0 3.5
3 6.87 1.4 -0.02 1.4 2.1 2.8 0.0 2.8 3.9 0.7 1.4 1.4 2.5 1.8 1.1 0.5 0.5 1.0 3.2 2.5 3 6.87 1.4 -0.02 3.9 2.8 2.8 2.5 2.1 1.4 1.4 1.4 0.7 0.0 1.0 0.5 0.5 1.1 1.8 1.0 1.8
2 6.87 1.4 -0.02 1.4 2.1 2.8 0.0 2.8 3.9 0.7 1.4 1.4 2.5 1.8 1.1 0.5 0.5 1.0 3.2 2.5 2 6.87 1.4 -0.02 3.9 2.8 2.8 2.5 2.1 1.4 1.4 1.4 0.7 0.0 1.0 0.5 0.5 1.1 1.8 1.0 1.8
1 6.87 1.4 -0.02 1.4 2.1 2.8 0.0 2.8 3.9 0.7 1.4 1.4 2.5 1.8 1.1 0.5 0.5 1.0 3.2 2.5 1 6.87 1.4 -0.02 3.9 2.8 2.8 2.5 2.1 1.4 1.4 1.4 0.7 0.0 1.0 0.5 0.5 1.1 1.8 1.0 1.8
0.1 6.87 1.4 -0.02 2.0 2.5 3.2 1.4 3.2 4.3 1.6 2.0 2.0 3.0 2.5 2.1 1.8 1.8 2.2 3.6 3.0 0.1 6.87 1.4 -0.02 4.3 3.2 3.2 3.0 2.5 2.0 2.0 2.0 1.6 1.4 2.2 1.8 1.8 2.1 2.5 2.2 2.5
0.01 6.87 1.4 -0.02 2.2 2.5 3.0 2.2 3.0 4.0 2.1 2.2 2.2 3.0 3.5 3.0 2.7 2.7 2.8 4.6 4.0 0.01 6.87 1.4 -0.02 4.0 3.0 3.0 3.0 2.5 2.2 2.2 2.2 2.1 2.2 2.8 2.7 2.7 3.0 3.5 3.2 3.5
3 6.92 1.2 -0.01 1.4 2.1 2.8 0.0 2.8 3.9 0.7 1.4 1.4 2.5 1.8 1.1 0.5 0.5 1.0 n 2.5 3 6.92 1.2 -0.01 3.9 2.8 2.8 2.5 2.1 1.4 1.4 1.4 0.7 0.0 1.0 0.5 0.5 1.1 1.8 n 1.8
2 6.92 1.2 -0.01 1.4 2.1 2.8 0.0 2.8 3.9 0.7 1.4 1.4 2.5 1.8 1.1 0.5 0.5 1.0 3.2 2.5 2 6.92 1.2 -0.01 3.9 2.8 2.8 2.5 2.1 1.4 1.4 1.4 0.7 0.0 1.0 0.5 0.5 1.1 1.8 1.0 1.8
1 6.92 1.2 -0.01 1.4 2.1 2.8 0.0 2.8 3.9 0.7 1.4 1.4 2.5 1.8 1.1 0.5 0.5 1.0 3.2 2.5 1 6.92 1.2 -0.01 3.9 2.8 2.8 2.5 2.1 1.4 1.4 1.4 0.7 0.0 1.0 0.5 0.5 1.1 1.8 1.0 1.8
0.1 6.92 1.2 -0.01 2.0 2.5 3.2 1.4 3.2 4.3 1.6 2.0 2.0 3.0 2.5 2.1 1.8 1.8 2.2 3.6 3.0 0.1 6.92 1.2 -0.01 4.3 3.2 3.2 3.0 2.5 2.0 2.0 2.0 1.6 1.4 2.2 1.8 1.8 2.1 2.5 2.2 2.5
0.01 6.92 1.2 -0.01 2.2 2.5 3.0 2.2 3.0 4.0 2.1 2.2 2.2 3.0 3.5 3.0 2.7 2.7 2.8 4.6 4.0 0.01 6.92 1.2 -0.01 4.0 3.0 3.0 3.0 2.5 2.2 2.2 2.2 2.1 2.2 2.8 2.7 2.7 3.0 3.5 3.2 3.5
3 7.20 1.6 -0.02 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 3 7.20 1.6 -0.02 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
2 7.20 1.6 -0.02 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 2 7.20 1.6 -0.02 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
1 7.20 1.6 -0.02 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 1 7.20 1.6 -0.02 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
0.1 7.20 1.6 -0.02 1.1 1.5 2.1 1.5 2.1 3.2 1.1 1.1 1.1 2.0 3.2 2.5 2.0 2.0 1.8 4.5 3.8 0.1 7.20 1.6 -0.02 3.2 2.1 2.1 2.0 1.5 1.1 1.1 1.1 1.1 1.5 1.8 2.0 2.0 2.5 3.2 2.5 3.2
0.01 7.20 1.6 -0.02 1.8 1.8 2.1 2.5 2.1 3.0 2.1 1.8 1.8 2.2 4.1 3.5 3.0 3.0 2.7 5.4 4.7 0.01 7.20 1.6 -0.02 3.0 2.1 2.1 2.2 1.8 1.8 1.8 1.8 2.1 2.5 2.7 3.0 3.0 3.5 4.1 3.5 4.1
3 7.25 1.4 -0.01 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 3 7.25 1.4 -0.01 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
2 7.25 1.4 -0.01 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 2 7.25 1.4 -0.01 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
1 7.25 1.4 -0.01 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 1 7.25 1.4 -0.01 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
0.1 7.25 1.4 -0.01 1.1 1.5 2.1 1.5 2.1 3.2 1.1 1.1 1.1 2.0 3.2 2.5 2.0 2.0 1.8 4.5 3.8 0.1 7.25 1.4 -0.01 3.2 2.1 2.1 2.0 1.5 1.1 1.1 1.1 1.1 1.5 1.8 2.0 2.0 2.5 3.2 2.5 3.2
0.01 7.25 1.4 -0.01 1.8 1.8 2.1 2.5 2.1 3.0 2.1 1.8 1.8 2.2 4.1 3.5 3.0 3.0 2.7 5.4 4.7 0.01 7.25 1.4 -0.01 3.0 2.1 2.1 2.2 1.8 1.8 1.8 1.8 2.1 2.5 2.7 3.0 3.0 3.5 4.1 3.5 4.1
3 7.28 1 0.01 1.8 2.5 3.2 0.5 3.2 4.2 1.1 1.8 1.8 2.8 1.4 0.7 0.0 0.0 1.1 2.8 2.1 3 7.28 1 0.01 4.2 3.2 3.2 2.8 2.5 1.8 1.8 1.8 1.1 0.5 1.1 0.0 0.0 0.7 1.4 0.5 1.4
2 7.28 1 0.01 1.8 2.5 3.2 0.5 3.2 4.2 1.1 1.8 1.8 2.8 1.4 0.7 0.0 0.0 1.1 2.8 2.1 2 7.28 1 0.01 4.2 3.2 3.2 2.8 2.5 1.8 1.8 1.8 1.1 0.5 1.1 0.0 0.0 0.7 1.4 0.5 1.4
1 7.28 1 0.01 1.8 2.5 3.2 0.5 3.2 4.2 1.1 1.8 1.8 2.8 1.4 0.7 0.0 0.0 1.1 2.8 2.1 1 7.28 1 0.01 4.2 3.2 3.2 2.8 2.5 1.8 1.8 1.8 1.1 0.5 1.1 0.0 0.0 0.7 1.4 0.5 1.4
0.1 7.28 1 0.01 2.1 2.7 3.4 1.1 3.4 4.5 1.5 2.1 2.1 3.2 2.0 1.6 1.4 1.4 2.1 3.2 2.5 0.1 7.28 1 0.01 4.5 3.4 3.4 3.2 2.7 2.1 2.1 2.1 1.5 1.1 2.1 1.4 1.4 1.6 2.0 1.8 2.0
0.01 7.28 1 0.01 2.1 2.5 3.0 1.8 3.0 4.1 1.8 2.1 2.1 3.0 3.0 2.5 2.2 2.2 2.5 4.1 3.5 0.01 7.28 1 0.01 4.1 3.0 3.0 3.0 2.5 2.1 2.1 2.1 1.8 1.8 2.5 2.2 2.2 2.5 3.0 2.7 3.0
3 7.59 1.6 -0.01 1.0 0.7 1.0 2.2 1.0 1.8 1.6 1.0 1.0 0.5 3.9 3.2 2.5 2.5 1.4 5.3 4.6 3 7.59 1.6 -0.01 1.8 1.0 1.0 0.5 0.7 1.0 1.0 1.0 1.6 2.2 1.4 2.5 2.5 3.2 3.9 2.8 3.9
2 7.59 1.6 -0.01 1.0 0.7 1.0 2.2 1.0 1.8 1.6 1.0 1.0 0.5 3.9 3.2 2.5 2.5 1.4 5.3 4.6 2 7.59 1.6 -0.01 1.8 1.0 1.0 0.5 0.7 1.0 1.0 1.0 1.6 2.2 1.4 2.5 2.5 3.2 3.9 2.8 3.9
1 7.59 1.6 -0.01 1.0 0.7 1.0 2.2 1.0 1.8 1.6 1.0 1.0 0.5 3.9 3.2 2.5 2.5 1.4 5.3 4.6 1 7.59 1.6 -0.01 1.8 1.0 1.0 0.5 0.7 1.0 1.0 1.0 1.6 2.2 1.4 2.5 2.5 3.2 3.9 2.8 3.9
0.1 7.59 1.6 -0.01 1.0 0.7 1.0 2.2 1.0 2.1 1.6 1.0 1.0 1.1 4.0 3.4 2.7 2.7 2.0 5.4 4.7 0.1 7.59 1.6 -0.01 2.1 1.0 1.0 1.1 0.7 1.0 1.0 1.0 1.6 2.2 2.0 2.7 2.7 3.4 4.0 3.2 4.0
0.01 7.59 1.6 -0.01 2.0 1.6 1.4 3.2 1.4 2.1 2.5 2.0 2.0 1.8 4.9 4.3 3.6 3.6 3.0 6.3 5.6 0.01 7.59 1.6 -0.01 2.1 1.4 1.4 1.8 1.6 2.0 2.0 2.0 2.5 3.2 3.0 3.6 3.6 4.3 4.9 4.1 4.9
3 7.65 1.2 0.01 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 3 7.65 1.2 0.01 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
2 7.65 1.2 0.01 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 2 7.65 1.2 0.01 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
1 7.65 1.2 0.01 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 1 7.65 1.2 0.01 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
0.1 7.65 1.2 0.01 1.1 1.5 2.1 1.5 2.1 3.2 1.1 1.1 1.1 2.0 3.2 2.5 2.0 2.0 1.8 4.5 3.8 0.1 7.65 1.2 0.01 3.2 2.1 2.1 2.0 1.5 1.1 1.1 1.1 1.1 1.5 1.8 2.0 2.0 2.5 3.2 2.5 3.2
0.01 7.65 1.2 0.01 1.8 1.8 2.1 2.5 2.1 3.0 2.1 1.8 1.8 2.2 4.1 3.5 3.0 3.0 2.7 5.4 4.7 0.01 7.65 1.2 0.01 3.0 2.1 2.1 2.2 1.8 1.8 1.8 1.8 2.1 2.5 2.7 3.0 3.0 3.5 4.1 3.5 4.1
3 7.66 1 0.02 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 3 7.66 1 0.02 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
2 7.66 1 0.02 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 2 7.66 1 0.02 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
1 7.66 1 0.02 0.5 1.1 1.8 1.1 1.8 2.8 0.5 0.5 0.5 1.4 2.8 2.1 1.4 1.4 0.5 4.2 3.5 1 7.66 1 0.02 2.8 1.8 1.8 1.4 1.1 0.5 0.5 0.5 0.5 1.1 0.5 1.4 1.4 2.1 2.8 1.8 2.8
0.1 7.66 1 0.02 1.1 1.5 2.1 1.5 2.1 3.2 1.1 1.1 1.1 2.0 3.2 2.5 2.0 2.0 1.8 4.5 3.8 0.1 7.66 1 0.02 3.2 2.1 2.1 2.0 1.5 1.1 1.1 1.1 1.1 1.5 1.8 2.0 2.0 2.5 3.2 2.5 3.2
0.01 7.66 1 0.02 1.8 1.8 2.1 2.5 2.1 3.0 2.1 1.8 1.8 2.2 4.1 3.5 3.0 3.0 2.7 5.4 4.7 0.01 7.66 1 0.02 3.0 2.1 2.1 2.2 1.8 1.8 1.8 1.8 2.1 2.5 2.7 3.0 3.0 3.5 4.1 3.5 4.1
3 7.85 2 -0.02 0.5 0.5 1.1 1.8 1.1 2.2 1.1 0.5 0.5 1.0 3.6 2.9 2.2 2.2 1.5 5.0 4.3 3 7.85 2 -0.02 2.2 1.1 1.1 1.0 0.5 0.5 0.5 0.5 1.1 1.8 1.5 2.2 2.2 2.9 3.6 2.7 3.6
2 7.85 2 -0.02 0.5 0.5 1.1 1.8 1.1 2.2 1.1 0.5 0.5 1.0 3.6 2.9 2.2 2.2 1.5 5.0 4.3 2 7.85 2 -0.02 2.2 1.1 1.1 1.0 0.5 0.5 0.5 0.5 1.1 1.8 1.5 2.2 2.2 2.9 3.6 2.7 3.6
1 7.85 2 -0.02 0.5 0.5 1.1 1.8 1.1 2.2 1.1 0.5 0.5 1.0 3.6 2.9 2.2 2.2 1.5 5.0 4.3 1 7.85 2 -0.02 2.2 1.1 1.1 1.0 0.5 0.5 0.5 0.5 1.1 1.8 1.5 2.2 2.2 2.9 3.6 2.7 3.6
0.1 7.85 2 -0.02 1.8 1.8 2.1 2.5 2.1 3.0 2.1 1.8 1.8 2.2 4.1 3.5 3.0 3.0 2.7 5.4 4.7 0.1 7.85 2 -0.02 3.0 2.1 2.1 2.2 1.8 1.8 1.8 1.8 2.1 2.5 2.7 3.0 3.0 3.5 4.1 3.5 4.1
0.01 7.85 2 -0.02 2.7 2.5 2.5 3.5 2.5 3.2 3.0 2.7 2.7 2.8 5.1 4.5 4.0 4.0 3.6 6.3 5.7 0.01 7.85 2 -0.02 3.2 2.5 2.5 2.8 2.5 2.7 2.7 2.7 3.0 3.5 3.6 4.0 4.0 4.5 5.1 4.5 5.1
3 8.01 1.4 0.01 1.0 0.7 1.0 2.2 1.0 1.8 1.6 1.0 1.0 0.5 3.9 3.2 2.5 2.5 1.4 5.3 4.6 3 8.01 1.4 0.01 1.8 1.0 1.0 0.5 0.7 1.0 1.0 1.0 1.6 2.2 1.4 2.5 2.5 3.2 3.9 2.8 3.9
2 8.01 1.4 0.01 1.0 0.7 1.0 2.2 1.0 1.8 1.6 1.0 1.0 0.5 3.9 3.2 2.5 2.5 1.4 5.3 4.6 2 8.01 1.4 0.01 1.8 1.0 1.0 0.5 0.7 1.0 1.0 1.0 1.6 2.2 1.4 2.5 2.5 3.2 3.9 2.8 3.9
1 8.01 1.4 0.01 1.0 0.7 1.0 2.2 1.0 1.8 1.6 1.0 1.0 0.5 3.9 3.2 2.5 2.5 1.4 5.3 4.6 1 8.01 1.4 0.01 1.8 1.0 1.0 0.5 0.7 1.0 1.0 1.0 1.6 2.2 1.4 2.5 2.5 3.2 3.9 2.8 3.9
0.1 8.01 1.4 0.01 1.0 0.7 1.0 2.2 1.0 2.1 1.6 1.0 1.0 1.1 4.0 3.4 2.7 2.7 2.0 5.4 4.7 0.1 8.01 1.4 0.01 2.1 1.0 1.0 1.1 0.7 1.0 1.0 1.0 1.6 2.2 2.0 2.7 2.7 3.4 4.0 3.2 4.0
0.01 8.01 1.4 0.01 2.0 1.6 1.4 3.2 1.4 2.1 2.5 2.0 2.0 1.8 4.9 4.3 3.6 3.6 3.0 6.3 5.6 0.01 8.01 1.4 0.01 2.1 1.4 1.4 1.8 1.6 2.0 2.0 2.0 2.5 3.2 3.0 3.6 3.6 4.3 4.9 4.1 4.9
3 8.03 1 0.03 0.5 0.5 1.1 1.8 1.1 2.2 1.1 0.5 0.5 1.0 3.6 2.9 2.2 2.2 1.5 5.0 4.3 3 8.03 1 0.03 2.2 1.1 1.1 1.0 0.5 0.5 0.5 0.5 1.1 1.8 1.5 2.2 2.2 2.9 3.6 2.7 3.6
2 8.03 1 0.03 0.5 0.5 1.1 1.8 1.1 2.2 1.1 0.5 0.5 1.0 3.6 2.9 2.2 2.2 1.5 5.0 4.3 2 8.03 1 0.03 2.2 1.1 1.1 1.0 0.5 0.5 0.5 0.5 1.1 1.8 1.5 2.2 2.2 2.9 3.6 2.7 3.6
1 8.03 1 0.03 0.5 0.5 1.1 1.8 1.1 2.2 1.1 0.5 0.5 1.0 3.6 2.9 2.2 2.2 1.5 5.0 4.3 1 8.03 1 0.03 2.2 1.1 1.1 1.0 0.5 0.5 0.5 0.5 1.1 1.8 1.5 2.2 2.2 2.9 3.6 2.7 3.6
0.1 8.03 1 0.03 1.8 1.8 2.1 2.5 2.1 3.0 2.1 1.8 1.8 2.2 4.1 3.5 3.0 3.0 2.7 5.4 4.7 0.1 8.03 1 0.03 3.0 2.1 2.1 2.2 1.8 1.8 1.8 1.8 2.1 2.5 2.7 3.0 3.0 3.5 4.1 3.5 4.1
0.01 8.03 1 0.03 2.7 2.5 2.5 3.5 2.5 3.2 3.0 2.7 2.7 2.8 5.1 4.5 4.0 4.0 3.6 6.3 5.7 0.01 8.03 1 0.03 3.2 2.5 2.5 2.8 2.5 2.7 2.7 2.7 3.0 3.5 3.6 4.0 4.0 4.5 5.1 4.5 5.1
3 8.03 1.2 0.02 1.1 0.5 0.5 2.5 0.5 1.4 1.8 1.1 1.1 0.0 4.2 3.5 2.8 2.8 1.8 5.7 4.9 3 8.03 1.2 0.02 1.4 0.5 0.5 0.0 0.5 1.1 1.1 1.1 1.8 2.5 1.8 2.8 2.8 3.5 4.2 3.2 4.2
2 8.03 1.2 0.02 1.1 0.5 0.5 2.5 0.5 1.4 1.8 1.1 1.1 0.0 4.2 3.5 2.8 2.8 1.8 5.7 4.9 2 8.03 1.2 0.02 1.4 0.5 0.5 0.0 0.5 1.1 1.1 1.1 1.8 2.5 1.8 2.8 2.8 3.5 4.2 3.2 4.2
1 8.03 1.2 0.02 1.1 0.5 0.5 2.5 0.5 1.4 1.8 1.1 1.1 0.0 4.2 3.5 2.8 2.8 1.8 5.7 4.9 1 8.03 1.2 0.02 1.4 0.5 0.5 0.0 0.5 1.1 1.1 1.1 1.8 2.5 1.8 2.8 2.8 3.5 4.2 3.2 4.2
0.1 8.03 1.2 0.02 1.5 1.1 1.1 2.7 1.1 2.0 2.1 1.5 1.5 1.4 4.5 3.8 3.2 3.2 2.5 5.8 5.1 0.1 8.03 1.2 0.02 2.0 1.1 1.1 1.4 1.1 1.5 1.5 1.5 2.1 2.7 2.5 3.2 3.2 3.8 4.5 3.6 4.5
0.01 8.03 1.2 0.02 2.5 2.1 1.8 3.6 1.8 2.2 3.0 2.5 2.5 2.2 5.4 4.7 4.1 4.1 3.5 6.7 6.0 0.01 8.03 1.2 0.02 2.2 1.8 1.8 2.2 2.1 2.5 2.5 2.5 3.0 3.6 3.5 4.1 4.1 4.7 5.4 4.6 5.4
SPE 93164 13

Figure 10: Construction of the visualization tool (match map on 5 Figure 11: Match density map and 5 selected zones for match
dimensions, discretised) search

Horizontally: permeability
2 dimensions: Matrix and Karst Predicting “ zones” of highest match density

By summing the probability of

matches on each 5-dim
hyperplane
Vertically: Total Predicted Match Probability (within space)
Volume: 2 Dimensions: Matrix and Karst 0.350% -0.400%

Aquifer Strength: 1 Dimension 0.300% -0.350%

0.4
3 Sgr
0.250% -0.300%
0.3
0.200% -0.250%

0.150% -0.200%
1 2 0.2
Arranged by
0.100% -0.150%

0.050% -0.100%

Increasing 0.000% -0.050%

0.15

GIIP 5 4
0.1
0.01 0.1 0.5 1 10

Kv11Stretch
14 SPE 93164

Figure 12a: Predicted match projection map (Region 1) Figure 12b: Actual match projection map (Region 1)

Match Region 1 : PREDICTED Match Region 1 : ACTUAL

Mid Sgr, Mid Kv11Stretch Mid Sgr, Mid Kv11Stretch

1 1

Figure 13: alternative match density map, not weighted with Figure 14: different possible arrangements of the horizontal axis
probability of parameter combination of the (predicted) match projection map

Sequential arrangement Increasing Keff arrangement

Match density map (match instances likelyhood, not weighted
with parameter combination likelyhood) KhMult Kkarst KhMult Kkarst Keff =
1 100 1 100 75
0.4
4.5%-5.0% 1 200 1.5 100 87.5
Sgr 1 400 2 100 100
4.0%-4.5%
1 800 1 200 125
3.5%-4.0% 0.3 1 1000 3 100 125
3.0%-3.5% 1.5 100 1.5 200 137.5
2.5%-3.0% 1.5 200 2 200 150
0.2 1.5 400 4 100 150
2.0%-2.5%
1.5 800 3 200 175
1.5%-2.0% 1.5 1000 4 200 200
1.0%-1.5% 0.15 2 100 1 400 225
0.5%-1.0%
2 200 1.5 400 237.5
2 400 2 400 250
0.0%-0.5%
0.1
2 800 3 400 275
0.01 0.1 0.5 1 10 2 1000 4 400 300
Kv11Stretch 3 100 1 800 425
3 200 1.5 800 437.5
SPE 93164 15

Figure 15: List of generated matching realisations

Dynamic run name PorAdd KaqfMult Kv11Stretch Sgr KhMult Kkarst KarstVolMult
F06_ScParamTestv1#0 0.01 1 1 0.2 1 600 1.2
F06_ScParamTestv1#5 0.01 1 1 0.2 4 300 1.2
F06_ScTestv14bcbis#0 0.01 3 0.5 0.2 2 400 1.2
F06_ScTestv15#1 0.01 2 0.5 0.4 2 800 1.2
F06_ScMatchExts7#9 -0.01 3 0.5 0.2 4 200 2
F06_ScMatchExts7#10 -0.01 3 0.5 0.2 4 200 1.8
F06_ScMatchExts7#15 -0.01 2 0.5 0.2 4 200 2
F06_ScMatchExtv4#1 0.03 3 0.5 0.2 3 200 1.4
F06_ScMatchExtv7#5 -0.01 0.1 0.5 0.2 2 200 2
F06_ScMatchExtv7#9 0.01 0.1 0.5 0.2 1.5 200 1.6
F06_ScMatchExtv8#0 0.01 3 0.5 0.2 1 400 1.4
F06_ScMatchExtv8#1 0.01 3 0.5 0.2 1.5 400 1.4
F06_ScMatchExtv8#3 0.01 1 0.5 0.2 1 400 1.4
F06_ScMatchExtv8#4 0.01 1 0.5 0.2 1.5 400 1.4
F06_ScMatchMSgrHKv11v2#0 -0.01 1 10 0.2 2 200 2
F06_ScMatchMSgrHKv11v2#1 -0.01 1 10 0.2 3 200 2
F06_ScMatchBO4bis#11 0.02 1 0.5 0.1 2 200 1.4
F06_ScMatchBO3bis#1 0.01 3 10 0.2 1 400 1.4
F06_ScMatchBO3bis#4 0.01 3 10 0.2 2 400 1.2
F06_ScMatchLVol3#1 -0.01 1 0.5 0.2 4 400 1.4
F06_ScMatchHSgrHKv11v1#13 0.01 0.01 10 0.4 2 200 1.6
F06_ScMatchHSgrHKv11v1#21 0.01 0.01 10 0.4 4 200 1.6
F06_ScMatchHSgrHKv11v1#30 0.01 0.1 10 0.4 3 400 1.4
F06_ScMatchHSgrHKv11v1#35 0.01 0.1 10 0.4 4 400 1.6
F06_ScMatchLSgrMKv11v2#5 0.01 3 0.5 0.1 3 200 1.4
F06_ScMatchLSgrMKv11v2#7 0.01 3 0.5 0.1 4 200 1.4
F06_ScMatchLSgrMKv11v2#13 0.01 1 0.5 0.1 3 200 1.4
F06_ScMatchLSgrMKv11v2#15 0.01 1 0.5 0.1 4 200 1.4