Volume Based Method (VBM) Horizon Modeler - Technical Reference

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 47

VBM technical reference

Volume Based Method (VBM) Horizon


Modeler - Technical reference
TABLE OF CONENTS

VOLUME BASED MODELING (VBM)..................................................................................................................6

EXAMPLES OF RESERVOIR MODELS CONSTRUCTED USING VOLUME BASED MODELING ..................................9

1.1. GEOMETRY DEFINITION.....................................................................................................................12

2. ACCESSING THE HORIZON MODELING ALGORITHM (VBM) ....................................................................12

2.1. VBM HORIZON MODELING SETTINGS ................................................................................................13

2.1.1. COMPUTE HORIZON TAB ...............................................................................................................13

2.1.2. SEQUENCE NUMBER COLUMN ......................................................................................................16

2.1.3. SMOOTHING VS. SMOOTHNESS ....................................................................................................19

2.1.4. NON–FILTERED DATA ....................................................................................................................21

2.1.5. ISOCHORES ....................................................................................................................................23

2.2. VBM VS. SURFACE-BASED MODELING ...............................................................................................23

2.2.1. VOLUME BASED MODELING ..........................................................................................................25

2.3. COMMON SETTINGS ..........................................................................................................................30

2.4. WELL ADJUSTMENT ...........................................................................................................................31

2.4.1. WELL TOP FIT ................................................................................................................................31

2.5. FAULT SETTINGS AND ACTIVITY .........................................................................................................32

2.6. ALGORITHM SETTINGS ......................................................................................................................34

2.6.1. STANDARD SETTINGS ....................................................................................................................35

2.6.2. EXPERT SETTINGS ..........................................................................................................................41

3. TROUBLESHOOTING ..............................................................................................................................45

4. KNOWN ISSUES .....................................................................................................................................47

Page | 1
VBM technical reference

TABLE OF FIGURES
FIGURE 1 – EXPLICIT, SURFACE-BASED HORIZON MODELING METHODOLOGY (MPI, FGD). INPUT DATA FOR EACH HORIZON DEFINE
THE TOPOLOGY OF EACH STRUCTURAL FRAMEWORK HORIZON REGARDLESS OF GEOLOGICAL CONFIGURATION. ....................... 4
FIGURE 2 - EXPLICIT, SURFACE-BASED HORIZON MODELING AS INPUT TO THE CREATION OF ZONES IN THE STRUCTURAL FRAMEWORK.. 4
FIGURE 3 - EFFECT OF INDIVIDUAL INPUT DATA CREATING TRENDS FOR INDIVIDUAL HORIZONS IN THE STRUCTURAL FRAMEWORK. AS
EACH HORIZON IN CREATED INDEPENDENTLY WHEN USING SURFACE-BASED TECHNIQUES, THEY ARE ALLOWED TO CROSS WHICH
IS GEOLOGICALLY UNREALISTIC ............................................................................................................................... 5
FIGURE 4 - RESULT OF HORIZON MODELING WHEN CROSSING DATA ARE USED.......................................................................... 5
FIGURE 5 - FAILURE TO CORRECTLY IDENTIFY ALL SEALING LOCATIONS BETWEEN HORIZONS AND FAULTS LEAD TO 'LEAK POINTS' WHERE
ZONES DO NOT GET CORRECTLY IDENTIFIED. .............................................................................................................. 5
FIGURE 6 - STAGES IN THE CONSTRUCTION OF A VOLUME BASED MODEL (VBM) ..................................................................... 6
FIGURE 7 - GALLERY OF COMPLEX MODELS BUILT USING VOLUME BASED MODELING ................................................................. 7
FIGURE 8 - CONFORMABLE MODELING USING LEGACY HORIZON MODELING TECHNIQUES. THE ALGORITHM IS FORCED TO USE A
SPECIFIC HORIZON TO MODEL FROM WHEN CREATING CONFORMABLE HORIZONS SO IT CANNOT MAKE A COMPLETE SEQUENCE. 8
FIGURE 9 - CONFORMABLE SEQUENCE MODELING USING VBM. AS ALL OF THE HORIZONS ARE CONTAINED WITHIN A SINGLE
'CONFORMABLE SEQUENCE' THEN THEY CAN ALL BE USED TO DEFINE THE CONFORMABILITY OF HORIZONS WITHIN THAT
SEQUENCE AND A COMPLETED MODEL CAN BE CONSTRUCTED....................................................................................... 8
FIGURE 10 - SHOWING PROPORTIONAL LAYERING CAPABILITIES OF VBM IN FAULTED AND UNFAULTED RESERVOIRS. ....................... 9
FIGURE 11 - VOLUME BASED METHOD (VBM) EXAMPLES OF COMPRESSIONAL AND EXTENSIONAL MODELS .................................. 9
FIGURE 12 – RESULTS OF USING GEOLOGICALLY INCONSISTENT/CONSISTENT DATA INPUTS FOR VBM ........................................ 10
FIGURE 13 - OUTLINING THE PROBLEMS THAT CAN OCCUR WHEN HORIZON INPUT DATA IS NOT GEOLOGICALLY CONSISTENT
(CROSSING) AND THE ERRORS IT CAN CREATE .......................................................................................................... 11
FIGURE 14 - GEOMETRY DEFINITION PROCESS WINDOW .................................................................................................... 12
FIGURE 15 - CHANGING THE HORIZON MODELING ALGORITHM ........................................................................................... 13
FIGURE 16 - HORIZON MODELING PROCESS DIALOGUE FOR VBM TECHNIQUES, IN RED SQUARES INDICATE CHANGES IN USER
INTERFACE IN COMPARISON TO THE LEGACY HORIZON MODELING TECHNIQUES (FGD & MPI) .......................................... 13
FIGURE 17 – UNREFINED VBM MODEL RESULTS SHOWING THAT THE ERODED HORIZONS ARE CORRECTLY CUT BY THE EROSION ...... 14
FIGURE 18 - UNREFINED VBM MODEL RESULTS (IN THE INITIAL MODE) SHOWING AN EXAMPLE OF HORIZON TRIANGLE MESH
RESOLUTION..................................................................................................................................................... 14
FIGURE 19 - REFINED VBM MODEL RESULTS (IN THE FINAL MODE) ..................................................................................... 15
FIGURE 20 - REFINED VBM MODEL RESULTS (IN THE FINAL MODE) SHOWING AN EXAMPLE OF HORIZON TRIANGLE MESH RESOLUTION
..................................................................................................................................................................... 15
FIGURE 21 - REFINED VBM MODEL RESULTS (IN THE FINAL MODE) SHOWING A GENERAL INTERSECTION PLANE THROUGH THE ZONE
AND HORIZON MODEL ........................................................................................................................................ 16
FIGURE 22 - REFINED VBM MODEL RESULTS (IN THE FINAL MODE) SHOWING A GENERAL INTERSECTION THROUGH THE ZONE AND
HORIZON MODEL............................................................................................................................................... 16
FIGURE 23 - REFINED VBM MODEL RESULTS (IN THE FINAL MODE) SHOWING THE FINAL ZONE MODEL ...................................... 16
FIGURE 24 - THE SEQUENCE NUMBERING RULES HIGHLIGHTING THE NUMBER OF CONFORMABLE SEQUENCES IN THIS MODEL. ......... 17
FIGURE 25 – TYPES OF SEQUENCES IN VOLUME BASED MODELING ..................................................................................... 18
FIGURE 26 - MODEL CONTAINING 8 CONFORMABLE SEQUENCES ......................................................................................... 18
FIGURE 27 - SMOOTHING EFFECT IN NOISY DATA SET (COLORED SOLID LINES- CREATED VBM HORIZONS, DASHED BLACK LINES- SEISMIC
DATA INPUT) .................................................................................................................................................... 21
FIGURE 28 - EXAMPLE OF MULTIPLE SEQUENCES WHERE SMOOTHNESS IS CONTROLLED INDEPENDENTLY ..................................... 21
FIGURE 29 - NON-FILTERED DATA COLUMN .................................................................................................................... 22
FIGURE 30 - UNSUCCESSFUL HORIZON MODELING DUE TO NON-FILTERED DATA BEING TOO CLOSE FROM THE FAULT PLANE ............ 22
FIGURE 31 – CORRECT USE OF ‘NON-FILTERED’ DATA CLOSE TO FAULTS ................................................................................ 23
FIGURE 32 - CONFORMABLE SEQUENCE, ALL HORIZONS CONFORMABLE TO EACH OTHER, REPRESENT ONE CONFORMAL SEQUENCE .. 23
FIGURE 33 - DATASET USED FOR HORIZON MODELING TECHNIQUES COMPARISON, VBM VS. SURFACE BASED MODELING .............. 24
FIGURE 34 - INPUT DATA USED IN VBM ......................................................................................................................... 25

Page | 2
VBM technical reference

FIGURE 35 - RESULTS OF VBM TECHNIQUE .................................................................................................................... 26


FIGURE 36 INPUT DATA USED FOR SURFACE BASED MODELING ............................................................................................ 26
FIGURE 37 - RESULTS OF SURFACE BASED MODELING TECHNIQUE ........................................................................................ 27
FIGURE 38 - EXAMPLE OF MODERATELY COMPLEX MODEL BUILT USING VBM; INTERMEDIATE HORIZONS GENERATED USING WELL TOP
INFORMATION ONLY .......................................................................................................................................... 28
FIGURE 39 - CALCULATION OF TRUE VERTICAL THICKNESS USING ISOCHORE THICKNESSES.......................................................... 29
FIGURE 40– VBM COMMON SETTINGS TAB ................................................................................................................... 30
FIGURE 41 - RECOMMENDED SETTINGS IN EXPERT SETTING TAB WHEN DEALING WITH THINLY LAYERED RESERVOIRS...................... 31
FIGURE 42 - STANDARD SETTING TAB WHEN DEALING WITH THINLY LAYERED RESERVOIRS ........................................................ 32
FIGURE 43 - FAULT ACTIVITY SETTINGS IN THE HORIZON MODELING DIALOGUE- FAULT ACTIVITY SETTINGS IN THE HORIZON MODELING
DIALOGUE........................................................................................................................................................ 33
FIGURE 44 - FAULT ACTIVITY, EXAMPLE BEHAVIOR............................................................................................................ 34
FIGURE 45 - HORIZON MODELING DIALOG, ALGORITHM SETTINGS, STANDARD SETTINGS TAB. .................................................. 35
FIGURE 46 – COMPLEXITY/SIZE PARAMETER CHOICE FOR EXAMPLE MODELS. ......................................................................... 37
FIGURE 47 – UNREFINED AND REFINED RESULTS OF THE HORIZONS...................................................................................... 38
FIGURE 48 – SMOOTHING EFFECT, LARGE DISTANCE FOR “IGNORE OUTLIERS, FURTHER THAN”.................................................. 39
FIGURE 49 – OPTION THE “IGNORE OUTLIERS, FURTHER THAN” EFFECT ON THE FINAL RESULTS ................................................. 39
FIGURE 50 – HALF-GRABEN STRUCTURE CREATED BY UPLIFT ON THE FAULT 1 ........................................................................ 40
FIGURE 51 – HORIZON MODELING DIALOG, ALGORITHM SETTINGS, EXPERT SETTINGS TAB...................................................... 41
FIGURE 52 – ALLOW LARGE THICKNESS VARIATIONS OPTION EFFECT ON THE FINAL RESULT ....................................................... 42
FIGURE 53 – ALLOW LARGE THICKNESS VARIATIONS OPTION EFFECT ON THE FINAL RESULT, EXAMPLE OF THE CASE WITH WELL TOP
INPUT WITH SOME THICKNESS VARIATION............................................................................................................... 43
FIGURE 54 – INCREASE OF “ VERTICAL RESOLUTION ENHANCEMENT” WILL EVEN MORE AMPLIFY STEEPNESS OF THE FOLD IN THE
COMPUTATIONAL SPACE FOR STEEPLY DIPPING FOLDS ............................................................................................... 44

Page | 3
VBM technical reference

Introduction
Petrel 2012.4 sees the introduction of a new algorithm for creating horizons in Structural Framework
models. This method called Volume Based Modeling or VBM complements the legacy methods
called FaultCentreGridDislocation (FGD) and Multi Patch Interpolator (MPI) introduced in Petrel
2010 and 2011 respectively.

The Volume based modeling (VBM) approach differs fundamentally when compared to the original
methods used to create the Structural framework. Horizons, in version 2010.1 – 2012.3, used a
surface-based methodology to create surfaces independently from each other as explicit objects
from a set of input data (figure 1).

Input data Building horizon 1 Building horizon 2

Figure 1 – Explicit, surface-based horizon modeling methodology (MPI, FGD). Input data for each horizon define the
topology of each Structural framework horizon regardless of geological configuration.

In detail, the input data is used to define the individual horizons to be modeled in the Structural
Framework. This data is then used to create a numerical representation of the surface network
formed by faults and horizon surfaces. Each surface is independent from all others and only when
the option to create zones is used does any surface-surface relationship calculations occur (the
specific cases of isochore maps and conformal modeling will be discussed later). The intersection
between faults and horizons are part of the calculation which defines the sealed zone model.

Figure 2 - Explicit, surface-based horizon modeling as input to the creation of zones in the Structural framework

Explicit, surface-based approaches are robust in concept, but increasing complexity which is now
commonly desirable in reservoir modeling means that an increasing number sealing calculations
need to occur to ensure model robustness. For each independent surface within the model to
have a calculated and sealed relationship to all other (where appropriate) requires a very complex
set of conditions to be honored. As a result, some complex or unusual modeling scenarios have
not successfully been accomplished using this approach.

Page | 4
VBM technical reference

Specific data scenarios, such as conformable horizons with thin layers (Figure 3), can highlight the
weakness of a surface based approach to horizon modeling. In this example, Horizon 2 has an
incomplete interpretation, but based on the trend in the input data the horizon modeller is expected
to create a continuous surface throughout the structural framework. This can be achieved,
however, this interpolation (horizon 2) does not account for any other horizon in the structural
framework. In this scenario the horizons come very close together, and in some areas actually
cross over which is geologically unrealistic; two geologically realistic scenarios could apply;
A. an erosion took place (so H1 should be identified as being an erosive event) or
B. the thickness of the H1-H2 layer is locally low, in which case H2 should be a continuous
surface that does not cross H1.
As surface-based techniques only use the input data on a ‘per horizon’ basis, these considerations
cannot be accounted for; some input data must be added to the area where there is none for
Horizon 2 to ensure the horizons do not cross if there is no erosion.

Figure 3 - Effect of individual input data creating trends for individual horizons in the Structural framework. As each
horizon in created independently when using surface-based techniques, they are allowed to cross which is geologically
unrealistic

Figure 4 - Result of horizon modeling when crossing data are used

These sorts of results in horizon modeling make it impossible to build a correct zone model. In
such cases or in any other situations when the horizon modeling process creates poor quality
surfaces, the algorithm is unable to correctly identify the relationships between the faults and
horizons to create the ‘sealing’ required to generate correct water tight zones.

Figure 5 - Failure to correctly identify all sealing locations between horizons and faults lead to 'leak points' where zones
do not get correctly identified.

Page | 5
VBM technical reference

Volume Based Modeling (VBM)


The VBM technology revolves around the concept of ‘implicit modeling’. This technique is very
different from ‘legacy’ approaches and relies on the calculation of surfaces as ‘iso-values’ of a
volume attribute which represents the gross stratigraphy of the model; usually denoted as an
“implicit function”.
In Volume Based Modeling (VBM), input data (faults/horizons) are used to create a background
tetrahedral mesh which represents the volume of interest (VOI) of the model. Onto this a iso-
stratigraphic function is calculated to represent all of the input data ‘stratigraphy’ in the model; this
property field is calculated from a watertight representation of the fault network (similar to fault
models constructed in legacy structural frameworks) and horizon input data. This property defines
interval boundaries which represents the ‘iso-stratigraphic’ age of these intervals; oldest at the base,
youngest at the top (note that only a relative age (and not the true geological age) is taken into
consideration during this calculation). The construction of this iso-stratigraphic property field is
such that it is continuous across the entire structural framework, discontinuous across faults and is
designed to minimize dip and thickness variations while accounting for all input data (seismic, well
tops, etc). Growth-fault / syn-rift depositional environments can also be handled thanks to an
optional control of the algorithm. Once this calculation is completed, iso-surfaces of the attribute
which correspond to the location of the original input data points are extracted as horizons in the
Structural framework and a zone model is computed; this yields a volume representation of
geological layers.

Input Data Create background Compute


volume model background iso-
stratigraphy property

Build zone model Extract horizons


from volume
Figure 6 - Stages in the construction of a Volume based model (VBM)

VBM horizon modeling method incorporates a set of robust and innovative algorithms which are
designed to approach the construction of complex structural frameworks from a new direction;
which will avoid the ‘surfaced-based’ modeling problems. This translates, for users, into the ability
to create realistic models without needing to be concerned by the structural complexity found in
many of the challenging reservoir modeling environments including, but not an exhaustive list;
fault configurations with crossing (X), synthetic/antithetic (Y), lambda (λ), reverse or low-angle
thrusts and overturned structures such as recumbent folds; non-conformable stratigraphy (presence
of multiple unconformities that form complex truncation patterns). The robustness of the

Page | 6
VBM technical reference

algorithm is also independent from the quality (presence of noise, misinterpreted reflectors) and
scarcity of data.
The main advantages of VBM method are:

 Ability to handle complex fault networks


o Being fully three-dimensional (3D), the VBM method is practically insensitive to the
complexity of the fault network and enables the accurate construction of very
complex structural reservoir models.

Figure 7 - Gallery of complex models built using Volume based modeling

 Modeling of geological sequences


o All conformable horizons belonging to a macro-layer are treated as a single
conformable sequence. These are then modeled simultaneously by VBM; as several
iso-values of the same implicit attribute. This approach prevents situations of
horizons crossing each other, which is a common problem with surface-based
modeling techniques especially in cases of thinly layered reservoir models. – Figure 4

Page | 7
VBM technical reference

Added points during horizon


modeling process

Figure 8 - Conformable modeling using legacy horizon modeling techniques. The algorithm is forced to use a specific
horizon to model from when creating conformable horizons so it cannot make a complete sequence.

H1

H2
H3

Figure 9 - Conformable sequence modeling using VBM. As all of the horizons are contained within a single
'conformable sequence' then they can all be used to define the conformability of horizons within that sequence and a
complete model can be constructed.

 Sparse data handling


o All horizons within a defined ‘geological sequence’ are conformable to each other
using the VBM algorithm. Therefore a single data-point in that sequence can
define the thickness between horizon interpretations, either above or below it (such
as a well top). Therefore, VBM presents the ability to construct consistent models
whose only input may be a single horizon interpretation and a sequence of sparse
data points; without the need for explicitly computing thickness maps. As
computation of the iso-stratigraphic property field is continuous across the
Structural framework then this property accounts for all the horizons that belong to
that sequence when computing the geometry of all other horizons.
 Proportional layering
o Using this technique, creating a new proportional horizon amounts to extracting an
additional iso-surface of the implicit function. Because the iso-stratigraphic
property is designed to be discontinuous across faults, it automatically calculates the
correct fault throw amount needed. This means that complex faulting is no longer
a problem when attempting to calculate consistent thicknesses across them.
Moreover, thickness discrepancies and/or variations that are commonly seen across
faults, often as a result of calculating isochores from well-tops, can now be avoided
as the calculation of an isochore map is no longer a mandatory requirement.
Instead calculation of consistent stratigraphic reservoir thickness this can be easily
carried out by adding a single well top (or a 2D seismic interpretation) to define the
new proportional layer.

Page | 8
VBM technical reference

Figure 10 - showing proportional layering capabilities of VBM in faulted and unfaulted reservoirs.

Examples of reservoir models constructed using Volume based modeling

Fault network Built horizon(s) Built zone(s)

Extensional

Extensional

Compressional

Compressional
Figure 11 - Volume Based Method (VBM) examples of compressional and extensional models

Page | 9
VBM technical reference

Important to know:

 VBM requires geologically consistent data

In particular VBM requires the fault extension to be consistent with their displacement (throw).
VBM also requires that geological type of interfaces (conformable, base, erosional, etc) is accurately
identified prior to the construction of the model.

The fault is too short to


accommodate the throw
represented in the horizon

Throw indicated
by horizon

Throw profile and fault extension are inconsistent: horizons tend to smear along faults

The consistency achieved by fault extension


was done to accommodate the throw Red line previous results
Black line the correctly
represented in the horizon rebuild horizon

Structurally consistent throw profile and fault extension; enough fault length to accommodate the through indicated by horizon,
accurate modeling results

Part of the fault surface is eroded. There is no strict


constraint on the extension of the non-eroded part
erosion
since faulting happened prior to the erosion

erosion

If the fault is eroded then


there is no need to extend the
fault

If the fault is eroded then there is no need to extend the fault and the application will compute the horizons properly.

Figure 12 – Results of using geologically inconsistent/consistent data inputs for VBM

 In VBM method; running the horizon modeling process twice with the same parameters may
create slightly different results.
 There is no guarantee that when re-running Volume Based Modeling on an existing model in
a future version of Petrel the exact same output will be obtained, even if using the same set
of parameters and data.

Page | 10
VBM technical reference

 The VBM method is very sensitive to overlapping or crossing input data (see figure 13 for
more detail explanation of prohibited configurations).

Seismic interpretations are locally crossing or


overlapping (geologically incorrect). Crossing seismic
horizons will not be handled properly by VBM and will
cause major artefacts in the surface geometries. If
such input has to be used in VBM, it is mandatory to
set one of the surfaces as an unconformity.

Miss pick

Well tops crossing each other (miss-pick), will cause


artefact in the created surface geometries

Miss picked well top AEO

Seismic horizon and well top crossing; this will cause


artefact in the created surface geometries

Seismic interpretation, LEM

Figure 13 - Outlining the problems that can occur when horizon input data is not geologically consistent (crossing) and
the errors it can create

Two important things when using Volume Based Modeling:

 The quality of interpretation data does not only affect the corresponding horizon, but also
the quality of all horizons that belong to the same sequence. Therefore it is important to
ensure that QC is carried out of all horizon input data for a geological sequence before
constructing the model.

 Wrong-sided data may create large pieces of model on the wrong side of faults, because
VBM will also attempt to re-create a correct stratigraphic column. The effect of wrong-
sided data is therefore significantly different from the one usually observed with FGD or
MPI.

Page | 11
VBM technical reference

1.1. Geometry definition

Figure 14 - Geometry definition process window

The Geometry definition process for defining the geometry of the new structural framework has the
same application in VBM as for MPI or FGD methods, with some minor limitations. These limitations
are as follow:

I increment Model resolution in I.


J increment Model resolution in J.

 I and J increments correspond to the edge length of the triangle in final horizon model. It is
recommended to set similar I and J increments when using Volume based Horizon Modeling
method.
 I and J increment will be applied only for final model, i.e. the input resolution given in the
“geometry definition” process will be met only if “Refine and create zone model” is checked
on.

If a boundary polygon is introduced to the geometry definition process, it will not be used for
modeling when the VBM method is selected.

2. Accessing the horizon modeling algorithm (VBM)

There is a choice of three algorithms in the horizon modeling process. The Volume Based Modeling
technique can be accessed in the Horizon Modeling tab by selecting it from the drop-down menu

Page | 12
VBM technical reference

named “Horizon modeling technique” (Fig. 20). Selecting this option will modify some of the
options available in the dialogue compared to the FGD/MPI methods.

Figure 15 - Changing the horizon modeling algorithm

2.1. VBM Horizon modeling settings

2.1.1. Compute horizon tab


The VBM horizon modeling algorithm behaves quite differently from the original FGD and MPI
approach and, as a result, the interface for this algorithm has been designed to reflect the required
algorithmic control available to users. Moreover, a series of ‘in UI’ warning messages have been
added to assist the user to understand the effect of their chosen options (Figure 21).

Figure 16 - Horizon modeling process dialogue for VBM techniques, in red squares indicate changes in user interface in
comparison to the legacy horizon modeling techniques (FGD & MPI)

Page | 13
VBM technical reference

VBM Horizon modeling workflow is divided into two steps;

 Unrefined/coarse version of structural framework horizons designed for quality control (QC).
This stage creates a coarse version of the structural framework which is not at the resolution
specified at the geometry defintion stage. This is for speed and rapid QC of the overall
model.
 Final refined modeling stage, where the user requested resolution of the horizons and zones
(from the geometry definition) are used to refine the initial (QC) horizons in the structual
framework.

The first step in VBM has a slightly different behavior when compared to the MPI and FGD horizon
modeling techniques. At this step the unrefined, coarse resolution horizons are generated but unlike
the legacy approaches all stratigraphic/geological rules are applied; meaning that horizons are cut
correctly with respect to the horizon type settings (i.e erosional, discontinuous, base). The
unrefined/coarse mode is designed to rapidly generate prelimenary results of horizon modeling (for
visualisation and QC the stratigraphy/geology) prior to creating the final resolution model and zones.

Note - The horizons created at a coarser resolution at this step, therefore they may not accurately fit
to their input data in this mode.

Figure 17 – Unrefined VBM model results showing that the eroded horizons are correctly cut by the erosion

Figure 18 - Unrefined VBM model results (in the Initial mode) showing an example of horizon triangle mesh resolution

Page | 14
VBM technical reference

The second step is consisting of:

1) Refinement of the horizon model to the resolution specified in the Geometry definition
process,
2) Zone generation.

The refinement and zone generation occurrs when the “Refine and create zone model” option is
toggled on in the horizon modeling dialouge. This is turned off by default so that rapid QC horizons
can be generated by default. Creation of refined horizons and zones should be made once all QC of
the model has been undertaken and completed; this process may be independently run once all
horizons have been generated and have undergone QC.

Selecting the “Refine and create zone model” option will finalize the structural framework model
and create the geological zones. Beware that when this option is turned on, building or updating the
model might take a long time, especially when dealing with a large number of horizons and/or when
the required resolution is very fine. Warning messages have been added in the horizon modeling UI
to inform users when this may occur. Im such case, the unrefined model will not be available for
preview, only the final refined model will be available.

Figure 19 - Refined VBM model results (in the Final mode)

Figure 20 - Refined VBM model results (in the Final mode) showing an example of horizon triangle mesh resolution

Page | 15
VBM technical reference

Figure 21 - Refined VBM model results (in the Final mode) showing a general intersection plane through the zone and
horizon model

Figure 22 - Refined VBM model results (in the Final mode) showing a general intersection through the zone and horizon
model

Figure 23 - Refined VBM model results (in the Final mode) showing the Final zone model

2.1.2. Sequence number column

Page | 16
VBM technical reference

A ‘Sequence’ column has been added to the horizon modeling dialogue when the VBM method is
selected (Fig. 43). This represents the collection of horizons which belong to a conformable
sequence; which are bound by sequence boundaries such as Erosions, Baselaps or Discontinuities (or
the top/base VOI of the structural framework). The Sequence number is assigned automatically
based on the selection made in the Horizon type column (Horizon type drop down menu); Erosions,
Baselaps or Discontinuity will set a sequence boundary. Initially all horizons are set to be
conformable, meaning that, by default, there is only a single conformable sequence in a structural
framework. By choosing a horizon to be a ‘sequence boundary’, two conformable sequences are
defined (above/below the sequence boundary horizon) which are represented by sequential
numbering in the new sequence column. The sequence column is not available for manual editing
and can only be controlled by changing horizon types.

Figure 24 - The sequence numbering rules highlighting the number of conformable sequences in this model.

Horizon types and their effects on the conformable sequence

 Conformable: all horizons belong to the same sequence and build conformable to each
other.
 Erosional: Erosions (erosional surfaces) belong to the sequence above, sequence below
erosion will be numbered differently from sequence above
 Base: Baselap belongs to the sequence below, sequence above will be numbered differently
from sequence below
 Discontinuous: Discontinuities do not belong to sequences below or above. They
correspond to a separate sequence which is collapsed into a single surface.

Page | 17
VBM technical reference

Figure 25 – Types of sequences in Volume Based Modeling

Horizon modeling settings dialogue Built model, 8 conformal sequences

Figure 26 - Model containing 8 conformable sequences

Important to know:

 Baselap-baselap and Baselap-erosion/discontinuity relationship is not working as expected.


It is not possible to explicitly define truncation rules between unconformities: the youngest
unconformity is always assumed to be truncating older unconformities.

Page | 18
VBM technical reference

Baselap-erosion/discontinuity case where


currently in this type of configuration the
youngest baselap will cut oldest discontinuities

Baselap-Baselap case where currently in this


type of configuration the youngest baselap will
cut oldest baselap

2.1.3. Smoothing vs. Smoothness


The VBM methodology uses a different concept of smoothing compared to the legacy horizon
modeling algorithms, therefore the smoothing column in the compute horizons table has been
replaced with a new smoothness option.

In surface-based modeling methods (MPI, FGD) the smoothing value represents a defined number of
smoothing iterations applied to the horizon(s) as a post-processing operation. In contrast to this,
the smoothness option in VBM controls how smooth the implicit function and structural framework
horizons are directly; the value is not a number of iterations, rather a linear scale (1-100) where the
higher the smoothness value the smoother the resultant iso-stratigraphic property and structural
framework horizons will be. In detail, this value is used as a balance between the global fit of the
computed property field to the input data points and the global smoothness of the produced
geometries.

A low smoothness value (e.g. 5) will force the iso-stratigraphy function to be fitted tightly to the
input data. This has the benefit of matching the input data as closely as possible; however, a side
effect of this is that any ‘noise’ (such as interpretation mis-ties/mis-picks) present in the input data
will then be honored by the VBM horizon modeler and translated to the structural framework. This
often results in horizons which look rough. It is not recommended to use a very low smoothness
value unless the input data is very clean and without mis-ties, as it can create some artifacts
especially if data is very noisy or has non filtered outliers.

Page | 19
VBM technical reference

A high global smoothness (e.g. 80) allows the iso-stratigraphic function to be created that broadly
matches the input data which produces horizons which are much smoother. This allows data
which is noisy or has some ‘mis-pick’ errors to be consumed by the VBM horizon modeler, however,
the result of using a high value means that the horizon may poorly fit to the input data points (Figure
32 - example of smoothing effect).

When using VBM we recommend to increase the default smoothness value (set to a value of 35)
when dealing with noisy data and to decrease it when dealing with pre-processed (cleaner) data
(e.g. with input data extracted from a 2D gridding process).

Noisy data example with Smoothing =5 Implicit function with Smoothing =5


Good fit to the input data even if it is noisy Rough implicit fnction

Noisy data example with Smoothing =35 Implicit function with Smoothing =35

Noisy data example with Smoothing =70 Implicit function with Smoothing =70
Bad fit to input data Smoother implicit fuction

Page | 20
VBM technical reference

Noisy data example with Smoothing =100 Implicit function with Smoothing =100
Bad fit to the input data Very smooth implicit function

Figure 27 - Smoothing effect in noisy data set (colored solid lines- created VBM horizons, dashed black lines- seismic
data input)

The Smoothness value can be controlled for each conformable sequence in the structural
framework. When a value is changed for one horizon of a sequence, it changes for all other horizons
of the same sequence (Fig. 47).

Figure 28 - Example of multiple sequences where smoothness is controlled independently

2.1.4. Non–Filtered data

VBM accepts ‘Non-filtered data’ in a similar way to the legacy structural framework horizon
modeling techniques (FGD, MPI). It is used when no fault filtering is desired in a particular location
and is usually desirable to achieve a specific fault-throw relationship.

Page | 21
VBM technical reference

Figure 29 - Non-Filtered data column

Input data (point sets or lines/polygons) used in this column will not be filtered out by the fault
filtering algorithm, during horizon modeling. This means that the final build horizons honor the
input data (regardless of its validity). VBM is very sensitive to ‘wrong sided data’, data located very
close to a fault or directly on the fault plane. Therefore, if non-filtered data is used, it is highly
recommended to apply stringent QC checks are applied; moreover we recommend that data
points located very close to, or directly on, fault planes are moved away (>5m) to ensure that errors
are avoided. It is essential that these hard data-points do not allow the structural framework
horizons to cross each other.
Figure 49 shows some of the problems that may be encountered by using Non-filtered data which
are located at the fault plane or very close to a fault.

Non-filtered data located on the fault plane causes Effect of wrong sided ‘hard data’ causing significant problems
problems during horizon modeling in the modeled horizons

Figure 30 - Unsuccessful horizon modeling due to non-filtered data being too close from the fault plane

To remove the artifacts the non-filtered data should be shifted away from the fault plane, the
recommended distance approximately should be equal to the half or one grid cell size indicated in
geometry definition (I, J increments).

Page | 22
VBM technical reference

Non filtered data was shifted away from fault plane The constructed horizon is correctly build and honoring the
given information

Figure 31 – Correct use of ‘non-filtered’ data close to faults

2.1.5. Isochores

VBM has been designed to minimize dip and thickness variations in the geological layers, while all
seismic and well data are properly honored. This means that typical isochore workflows used to
interpolate well-thicknesses maps, across reservoir models, in order to build correct conformable
successions are no longer necessary. Thicknesses can now be solely defined by well tops; the
computed thickness is then mapped across the entire model by the computation of the iso-
stratigraphic property. Create individual thickness maps from well tops is now not a requirement
for ‘isochore’ workflows. This is important for construction of complex faulted reservoirs as
traditionally the computation of correct isochore maps which correctly honor fault throw in these
types of areas proved very difficult to accurately construct; which means the thicknesses across
faults were often wrong. As VBM is designed to calculate the correct thicknesses of horizons while
accounting for fault discontinuities, we are now able to create consistent geological models in these
challenging areas. It should be pointed out that isochore modeling can still be used where
isochores provide conceptual geological insight, or where well tops are not present.

Figure 32 - Conformable sequence, all horizons conformable to each other, represent one conformal sequence

2.2. VBM vs. Surface-based modeling

A comparison of VBM vs. Surface-based methodology was carried out using a data set which consists
of 3 conformable horizons (H1, H2 and H3) which are faulted by a moderately complex fault

Page | 23
VBM technical reference

network. The seismic interpretation is such that the horizon interpretations only partially cover the
area of interest (Fig. 52). The input data was used ‘as is’ for VBM; isochore maps were created for
use with Surface based modeling, to better constrain the modeling process.

Horizon 3 exists only in the


North

Horizon 2 exists only in the


central part

Horizon 1 exists only in the


South

Cross section AA’


through the data

3D view of final dataset


for modeling

Figure 33 - Dataset used for horizon modeling techniques comparison, VBM vs. Surface based modeling

Page | 24
VBM technical reference

2.2.1. Volume Based Modeling

No isochore maps are used and all horizons are set to a conformable horizon type, therefore
belonging to a single conformable sequence (Fig. 53).

Figure 34 - input data used in VBM

Results of the VBM method are shown below, default horizon modeling parameters settings (Fig.
54).

Thick discontinuous lines represent the


seismic input data. Solid thin lines
represent the results of VBM horizon
modeling.
The results show consistent horizons
have been created using only the
partial seismic interpretation

Verification of the VBM model against


the seismic input data from various
locations in the reservoir. Thick
discontinuous lines show the original
seismic interpretation. Solid thin lines
show the VBM horizon modeling
results.

Computed iso-stratigraphic property


field created using the partial
interpretation

Page | 25
VBM technical reference

VBM - Map of Horizon 1 from the


structural framework. The horizon
result shows a consistent, correctly
faulted result from a partial
interpretation

3D view of Horizon 1. QC of the VBM


model vs. original seismic input
(present in all modeled area). The
initial (used for modeling) partially
existing input highlighted in red circle.

Conclusion: Very good and


geologically/structurally consistent
results achieved with minimum data
input and without computing any
Isochore maps.

Figure 35 - Results of VBM technique

2.2.2 Surface based modeling


All horizons still have conformable horizon type with horizon 2 is set conformable to horizon 1;
horizon 3 is set conformable to horizon 2. Two isochore maps are provided to constrain the
modeling between horizons 1 and 2 and horizons 2 and 3 (see settings below).

Figure 36 Input data used for surface based modeling

Page | 26
VBM technical reference

Results of Surface based method:

Thick discontinuous lines show the


seismic input data. Solid thin lines
show the FGD horizon modeling
results.
In the south part of the model the
horizons are acceptable, however the
results are poor at the north end as the
input data is under constrained in this
area.

Verification of the FGD model vs.


original seismic input data which
present all over the field. Thick
discontinuous lines represent the
original seismic input and solid thin
lines represent the FGD horizon
modeling results

Map view of Horizon 1. The shape of


the horizon in the north part was
extrapolated and doesn’t represent the
expected result.

Verification of the FGD model vs.


original seismic input data. The initial
partially existing input highlighted in
red circle.

Conclusion: It is impossible to build a


good model without extra
workarounds. First the shape of the
horizon 1 should be restored; only then
isochore maps could be used to build
the model.

Figure 37 - Results of surface based modeling technique

Page | 27
VBM technical reference

When dealing with VBM method the recommended workflow for modeling intermediate horizons in
between main seismic interpretations is simply to add data points which can be represented by well
tops (recommended option) or 2D seismic interpretations.

Modeling of intermediate horizons using


well top inputs.
The intermediate horizons are extracted
from the iso-stratigraphic property which is
discontinuous across faults. The correct
fault throws are automatically calculated
and use of isochore maps is not required.

Zone model results

Figure 38 - Example of moderately complex model built using VBM; intermediate horizons generated using well top
information only

The use of isochore maps is recommended only when dealing with sedimentary features (channels,
eroded surfaces, prograding, retrograding sedimentary bodies) which thickness cannot be
interpolated consistently from well data or seismic interpretation and when other “soft” data (e.g.
coming from “geological knowledge”: conceptual model, analogues, etc.) must be captured into the
structural model.

Important to know:

 Only True Vertical Thickness maps can be used with VBM method, TST maps will not be
accepted. When using isochore maps generated from isochore points (well tops), make sure
that the correct TVT attribute is used.

Page | 28
VBM technical reference

Figure 42 shows a map that shows the varying


interval thickness of a designated Stratigraphic
unit by means of contours. This is achieved by
measuring the True Vertical Thickness of the unit
prepared directly from two structure-contour
maps by subtracting their elevations.

Figure 39 - calculation of true vertical thickness using isochore thicknesses

 Isochore maps should not contain zero or negative values unless the reference surface has
been set as an unconformity in the horizon modeling dialog. In such a case, it is mandatory
to use negative values or undefined values rather than zero values for the eroded part of the
surface.
 Isochore maps should not contain any thickness values that have been measured across
fault surfaces.
 In compressional environments, isochore maps cannot be used to model layers which
thickness varies abruptly across fault blocks.

Thickness variation across the faults. This case would ideally


require one isochore map per fault block. This option is
currently not available in VBM

 In the example case shown above, only one isochore map per horizon can be used while
modeling, which may lead to incorrect results.
 When dealing with low angle faults, (see figures below) if isochore maps are used for
modeling intermediate horizons, the horizon geometry will be extrapolated in areas where
data is missing. In such situations, it is recommended to use well top data directly with the
VBM method to extract intermediate horizons instead of using isochore maps.

Page | 29
VBM technical reference

In case isochore maps are used for


both scenarios, the area highlighted
in red circles will be extrapolated to
Scenario 1, normal fault fill the gaps.

Such models will be built without


any problems if constant isochore
maps are provided

Scenario 2, reverse fault

2.3. Common settings

Well adjustment options are similar to those of the MPI or FGD algorithms; these options ensure
that the well tops are properly accounted for and control the horizons generation.

Figure 40– VBM Common settings tab

In the common settings tab for VBM the following options will be available:

Page | 30
VBM technical reference

2.4. Well adjustment


 Well adjustment toggled off
o Well tops are ignored (unless the considered horizon is represented only by well
data) and the horizons will only honor main input (i.e. seismic interpretation, etc.)
 Well adjustment toggled on
o If the influence radius is not specified, the radius is set to infinity and global shift of
the horizons to well tops will be performed.
o If the influence radius is specified, the well fit will be focused around wells within
the specified influence radius, beyond this radius data will honor the main input (i.e.
seismic interpretation, etc.)
 Iconize tied interpretation option - when toggled, will generate a folder containing a
modified version (tied to well tops) of interpretation data (e.g. Input #N), as point sets in
the Input pane of Petrel.

2.4.1. Well top fit

In order to maintain good geological consistency, well adjustment accounts for the subdivision of the
stratigraphic pile in several conformal sequences. Although the well adjustment is computed by the
mismatch between interpretation and well data, the geometry of the horizons that are controlled
only by well tops will also be automatically corrected during the well adjustment. This ensures all
conformable layers keep consistent fault displacements and thickness variations across the
model. As a result, the well adjustment will never produce any crossing horizons or pinch-outs
between conformal horizons; even when dealing with very thin layers.

In specific cases when dealing with dense lateral and vertical well top data input it is necessary to
select “Vertical resolution enhancement” as “Very thin layers” and keep “Dense well top
regularization” option toggled on (Expert settings tab).

Figure 41 - Recommended settings in Expert setting tab when dealing with thinly layered reservoirs

Page | 31
VBM technical reference

When generating final geological zone by applying “Refine and create zone model” for thinly layered
models it is highly recommended to check on “Enforce non-zero layer thickness” in Standard setting
of horizon modeling dialogue.

Figure 42 - Standard setting tab when dealing with thinly layered reservoirs

Important to know:

 In the unrefined model (when option “Refine and create zone model” is unchecked) the fit to
the well tops may not be exact, even though the well adjustment option has been activated.
This is true in particular when the density of well markers is high (locally or globally).
 The exact fit is only guaranteed at the "refine and create zone model" step
 Although it is possible to specify the location where a modeled horizon is expected to cross a
well by using well tops, constraining horizon geometries by zone logs is not currently supported.
 In models containing a large number of wells, modeling can take longer time when the well
adjustment option is checked.

Generate filtering attributes options

 Computation of filter attributes for each input horizon in Horizon modeling process dialog.
It has the same behavior as for the MPI and FGD methods.

2.5. Fault settings and activity

The static and non-static filtering distance and fault activity are set on the Fault Settings tab.

Page | 32
VBM technical reference

Figure 43 - Fault activity settings in the Horizon modeling dialogue- Fault activity settings in the Horizon modeling
dialogue

VBM fault activation logic is different compared to the legacy (FGD/MPI) horizon modeling
algorithms. Activity is now consistent with respect to the geological time and enforces geologically
consistent settings before modeling, which was not the the case for FGD and MPI methods or Pillar
griding.

In FGD, MPI and Pillar grid it was possible to activate and deactivate faults on a per horizon basis,
not taking in to consideration the geological time. For example: a major fault set as active in older
sequences can be set inactive in middle horizons and activated back in the shallowest (youngest)
horizons again, which doesn’t make any geological sense.

Fault activation in VBM is managed at the "sequence" level, setting a fault active/inactive for one
horizon sets it active/inactive for all horizons of the same sequence.

The general rule enforced in the VBM method when activating or deactivating a fault in the model is
that if the fault is "active" for a younger sequence it has to be active for all older sequences.

In other terms, when a fault is set active for a sequence, it should be set active for all older
sequences, when a fault is set inactive for a sequence; it should be set inactive for all younger
sequences.

Page | 33
VBM technical reference

Activation managed sequence per sequence (2


sequences example)

BCU is Sequence 1 (youngest sequence) and all


faults can be set as inactive in this sequence
without influencing the older sequences.

T_tarbet and T_Ness are belonging to the same


sequence, if any fault is set as inactive for one
horizon it is set as inactive for all horizons of the
same sequence

Figure 44 - Fault activity, example behavior

Other settings such as “Distance to fault” have a similar behavior in VBM as in any other method.

The “Displacement” control is not available in the VBM method.

Important to know:

 A known issue is that horizons located above a baselap and/or a disconformity will have a
fault cut with zero displacement (geometrically) for all faults which are set as active for the
baselap/disconformities located immediately underneath and set inactive in the considered
sequence.

2.6. Algorithm settings

Since the volume based method is very different form surface based methods (FGD, MPI) the
Algorithm setting tab is altered to the specifics of the VBM algorithm. The VBM Algorithm settings
tab is split in to two tabs;

 standard settings,
 expert settings tabs.

Page | 34
VBM technical reference

2.6.1. Standard settings

Figure 45 - Horizon modeling dialog, Algorithm Settings, Standard settings tab.

Complexity/Size parameter

The complexity /size parameter controls the level of detail captured in the unrefined model and the
accuracy of the final horizon-fault intersection lines. The correct selection depends on the model
complexity, the following information should be considered when selecting this parameter:

 How many horizons in the model?


 How thick or thin are they?
 Total area of the model?
 How many faults in the model?
 How dense the fault network?
 How many stratigraphic sequences in the model?
 Does the thickness of the zones vary a lot?
 How many wells in the field?

For example, models consisting of few horizons, relatively small in size (e.g. between 5x5 km to
10x10km) and containing less than a dozen of faults with simple contact scan be modeled using
“Very simple” or “Simple” settings for the “Complexity/size” parameter.

Model consisting of many horizons (15 to 20, including those represented by well tops only), not
very thinly layered, relatively big in size (10x10km to 20x20 km) and in which the fault network is not
very complex (not many X and Y connections) can be modeled using the “Normal” option.

Models consisting of many horizons (30 to 150), very thinly layered, relatively big in size (10x10km to
20x20 km) and containing many complex faults (100-150) can be modeled using the “Complex” or
“Very Complex” options.

Page | 35
VBM technical reference

The “Huge” option can be used for regional-scale models. It requires a computer with at least 32GB
of memory.

1 faults
7x7km
1 sequence
2 horizons not very closely spaced

Built using “Very Simple”

10 faults
10x10km
1 sequence
5 horizons not very closely spaced

Built using “Simple”

16 faults
10x10km
1 sequence
10 horizons not very closely spaced

Built using “Normal”

Page | 36
VBM technical reference

161 faults
20x20km
4 sequences
12 horizons not very closely spaced (>
40m from each other)

Built using “Complex”

118 faults (very complex connections ,


densely spaced)
10x10km
1 sequence
30 horizons

Built using “Very complex”

Figure 46 – Complexity/size parameter choice for example models.

Important to know:
The amount of CPU time and of memory required to build the model increases with the selected
level of complexity. A model built using the “Huge” option will require approximately 10 times more
CPU time and memory than a model built using the “Normal” option.

Post-processing (refined model)

The parameters related to “Post-processing” are only used when the “Refine and create zone
model” option is checked. During the post-processing, the final horizons will be created at the
resolution specified in the “geometry definition” setting by the ‘I and J’ increment parameters.

Page | 37
VBM technical reference

Unrefined model built with complexity size Final results created after applying “Refine and
parameter “Simple” create zone model”

Figure 47 – Unrefined and refined results of the horizons.

Smoothing

This parameter controls the smoothing of the final surfaces during the post-processing, a higher
number will produce a smoother result (Range of values: [0-100]).

Unrefined modeled horizons, normal


Unrefined horizons complexity/size. This cross section will be used
20 m distance

as a reference for understanding the effects of


the “smoothing” and of the “ignore outliers”
parameters. As the unrefined model is built at a
coarse resolution it does not capture the fine
Outliers/noise details contained in the input data. In some
situations these details are actually due to noise
or bad picks and should be discarded. In the
following figures the post processing effect is
illustrated and its parameters are explained.

Low smoothness with large specified distance for


the “ignore outliers” option will produce results
which will honor all noise in the data, to
eliminate the noise or outliers it is recommended
to increase the smoothness and/or decrease the
distance to outliers.

Page | 38
VBM technical reference

Very high smoothness with large specified


distance for the “ignore outliers” option will not
fit data accurately and will smooth out the
horizon geometry

Figure 48 – Smoothing effect, large distance for “ignore outliers, further than”

Ignore outliers

Points located further than the specified distance from the “unrefined” version of the horizon will be
ignored. This allows filtering out localized outliers.

Average smoothness with small specified distance for


the “ignore outliers” option will produce results which
will not fit to input data as the outliers beyond the
distance specified will not be used during post
processing. This option allows to eliminate some noise
in the data and to get smooth horizon geometries.

Average smoothness with large distance for the


“ignore outliers” option will produce results which
honor noisy input data.

Figure 49 – Option the “ignore outliers, further than” effect on the final results

Keep only blocks with input data

Due to the specifics of VBM algorithm, when working with sparse/partially present data the created
implicit function will be existing in all of the volume. It means that extracted horizons will cover all
the area of interest, even in the parts where data was not provided. In some particular cases the
absent part of the input data can be a syn-tectonic depositional feature and has to be accurately

Page | 39
VBM technical reference

represented. One of the classical examples is a half-graben structure where a major fault along only
one of the boundaries and most of the sediment will enter the half-graben down the unfaulted
hanging wall side.

Basin fill
Older sediments

Basement
Fault 1

Figure 50 – Half-graben structure created by uplift on the Fault 1

When modeling such type of environments in VBM in the unrefined model it is expected to see
some parts of the horizons which are not associated to any input data. Using the “Keep only blocks
with input data” option will allow eliminating unnecessary pieces and keeping only parts of the
horizon which are associated with Input data (seismic horizon, point sets, polygons, surfaces, well
tops, etc.).

Another example is unwanted pieces of eroded surfaces that could have been created below the
erosion.

Important to know:

This option is only available in refined mode and has no effect in the unrefined model.

This option is only working if part of the horizons which has to be removed is located on separate
fault block.

This option will enforce the geological integrity of the stratigraphic column. Pieces of horizons that
would create zones that are inconsistent with the input stratigraphic column will not be removed,
even if they are not associated with any input data point.

In some cases there is a need to preserve some piece created during modeling and some to be
removed. In such situations to preserve certain parts of the created horizon, it is recommended to
provide at least one data point per fault block when this option is selected.

Page | 40
VBM technical reference

Enforce non-zero layer thickness

The Enforce non-zero layer thickness options should be used when dealing with thinly layered
models. Using this option will ensure minimum thickness (usually a few centimeters) between
horizons to avoid any crossing in the “Refined” model .

Important to know:

This option is only available in refined mode and has no effect in the unrefined model.

Enforce non-zero layer thickness option may not always work, especially near horizon borders (VOI,
faults)

2.6.2. Expert settings

Figure 51 – Horizon Modeling dialog, Algorithm Settings, Expert settings tab

Allow large thickness variations

The “Allow large thickness variations” option is set on by default. The “Volume Based Modeling”
method tries to minimize dip and thickness variations while fitting the input data. In cases where
there are large variations of dips and thicknesses in the input data, it is recommended to toggle on
this option. This will account for global thickening/thinning trends in the input data (see some
examples below).

Page | 41
VBM technical reference

Before applying “Allow large thickness variations” After applying “Allow large thickness variations”

Figure 52 – Allow large thickness variations option effect on the final result

Another example of thickness variation effect: building a model with well data which indicates some
thickness variations.

Page | 42
VBM technical reference

Seismic horizon + well top


data available for Top horizon

For Base horizon only well tops


available, thickness of the
zone is increasing from right
to left.

After applying “Allow large


thickness variations”. Horizon
built accounting the variation
of thicknesses detected in
well data.

Before applying “Allow large


thickness variations”. Layers
forced to be paralele to each
other and do not account for
the thickening trend from well
data.

Figure 53 – Allow large thickness variations option effect on the final result, example of the case with well top input with
some thickness variation

Vertical resolution enhancement

This option controls the ratio between the vertical resolution of the model and its horizontal
resolution. Selecting the Very thin layers option means that the vertical resolution will increase but
concurrently with decrease of horizontal resolution.

• The Single horizon value is recommended to be used only when modeling a single surface.

• The Thick layers parameter should be used only when dealing with very thick zones (e.g.
overburden model with few horizons).

• The Average layers parameter is set as default and recommended to use when dealing with
several (up to 10) horizons which are not closely spaced.

Page | 43
VBM technical reference

• The Very thin layers parameter should be selected when dealing with thinly layered
reservoirs, in particular when the thickness of the layers varies a lot laterally.

Important to know:

It is not advised to use large “vertical resolution enhancement” factors when dealing with steeply
dipping folded surfaces.

 For steep folds, it leads


to very steep dip
variations
 Large dips in
computational space
may cause inaccurate
thickness estimates
 Combination of large dip
variations and poor
thickness estimates may
cause artifacts
 Moreover, increasing
the vertical resolution is
performed at the cost of
a decrease of the
horizontal resolution.

Figure 54 – Increase of “ Vertical resolution enhancement” will even more amplify steepness of the fold in the
computational space for steeply dipping folds

Optimize calculation for dense dataset

If this option is toggled, the input data will be smartly decimated when computing the “Unrefined”
(coarse) version of the surfaces. This will improve the run-time when dealing with dense (auto-
tracked seismic interpretation) data.

Note: All data points will always be accounted for when creating the final (refined) version of the
horizons, independently from this option.

Page | 44
VBM technical reference

3. Troubleshooting

Use the checklist below to verify that the input data and parameters have been prepared and set
correctly when trying to solve issues in a model created using the Volume Based Modeling
technique.

1. All input data that belongs to a defined sequence can influence the construction of all
other horizons that belong to the same sequence. As a consequence, artifacts or errors
observed on a built horizon may be due to the data of any other horizon belonging to the
same sequence.
2. Remember to always QC the “unrefined model” before checking the “refine and build
zone model” box, and edit the unrefined model if needed.

Check that… Known effects

 Large artifacts
 Bubbles in the modeled

 Input data are sorted into the correct


stratigraphic order. 
horizons
Extra pieces of modeled
horizons
 Very long run time
 Bubbles in the modeled
horizons
Horizons have the proper stratigraphic type
 (erosion, base, etc.), especially if they are
expected to truncate layers above/below.


Rough horizons
Horizons not matching input
data
 Incorrect truncation patterns
 Rough horizons
Input data for conformable horizons do not
 cross or touch each other, even in the
neighborhood of fault surfaces.
 Small bubbles located at the
crossing points
 Very long run time
 Rough horizons
 Small, extra, pieces of horizon
Stratigraphic ordering of well tops is correct at the well top locations
 along each well. Erosions/truncations are not
represented by collocated well tops.
 Bubbles around incorrectly
ordered well tops
 Very long run time

 Rough horizons
Vertical position of well tops and seismic data
 Poorly constrained horizon
 are consistent w.r.t. each other (e.g. no
younger well top is located below an older

around the well tops
Very long run time
horizon).


 Well tops should not share the same MD along Poorly constrained horizon
the same well path (i.e. avoid very closely around the well tops

Page | 45
VBM technical reference

spaced well tops).  Some well tops may not be


honored in the final model
 Poorly constrained horizon

 No horizon top is defined at an intersection


between a fault and a well. 
around the well tops
Artifacts, including bubbles on
the wrong side of the fault
The vertical extension of faults is consistent  Horizons may not honor input
with their throw: faults are large enough data at the fault locations
 (above/below the reservoir area) to
accommodate the displacement observed on
 Computed displacement may
be lower than interpreted
horizons.
 Horizons may not honor input
data at the fault locations
 ‘Wrong sided’ input data have been
removed/filtered out nearby faults.
 Computed displacement may
be lower than interpreted

The ‘Smoothness’ parameter has a high-  Rough horizons


 enough value (typically between 10 and 65 for
VBM).
 Bubbles in the modeled
horizons
 Errors will pop-up when
attempting to run horizon

 Validation
out.
of the fault model has been carried

modeling
Inaccurate horizons (throw,
fault connections)

 Incorrect model geometry


o Too big/small
VOI size and the “Model complexity/size” and o Too many cells
 “vertical resolution enhancement” parameters
have been set properly.


Inaccurate horizon geometry
Very slow run time
 Rough horizons

Erosions/truncations are not defined by null  Improper truncation of eroded


thicknesses in the input data (especially when surfaces by erosions
 using isochore maps as input). They should be
defined either by negative thicknesses or by
 Small patches of eroded
surface just below the erosion
undefined values.
The dense well tops regularization option is  Artifacts (spikes, bubbles) in

 used in presence of densely spaced wells or


when many horizons are based on well top
the vicinity of wells.

information.

Page | 46
VBM technical reference

4. Known Issues and limitations


As of version 13.3, Petrel Structural Framework Volume Based Modeling has the following known
limitations:

 Sometimes after a constant use of a 3D window it is not possible to display error points in
that window.
 Baselap-Baselap relationship not working as expected.
 Stair-step faulting process disabled.
 Eroded surfaces are not always handled properly in pillar gridding. They may be
extrapolated beyond erosion line.
 Well tops are not fully honored in “unrefined” model.
 Dense well top regularization option is required in some cases, otherwise artifacts, bubbles
and/or spikes.
 Non-reproducible results. 2 different run generate 2 slightly different results (should be less
than 1% volume difference, unless large extrapolations)
 Enforce non-zero layer thickness option may not always work, especially near horizon
borders (VOI, faults)
 Refine and create zone model option could be very slow and memory consuming.
 User interface could be slow when dealing with many faults.
 Although it is possible to specify the location where a modeled horizon is expected to cross a
well by using well tops, constraining horizon geometries by zone logs is not currently
supported.
 In models containing a large number of wells, modeling can take longer time when Well
adjustment option is checked.
 The amount of CPU time and of memory required to build the model increases with the
selected level of complexity. A model built using the “Huge” option will require
approximately 10 times more CPU time and memory than a model built using the “Normal”
option.
 VBM requires the fault extension to be consistent with their displacement (throw). VBM
also requires that geological types of interfaces (conformable, base, erosional, etc) are
accurately identified prior to the construction of the model.

Page | 47

You might also like