11/5/2020

Geomodelling for Reservoir Engineers

Adrian Thomas and Alex Novlesky

November 10, 2020

1
 11/5/2020

Online Training
❑ Certificates will be sent after the course to those who attend 75%
of the course or higher
❑ Will take a few days to process
❑ “Raw” unedited video will be sent 2 days after the course via e-mail
❑ Fully edited video will be available on the website
❑ Can take over a week to process

2
 11/5/2020

Online Training
❑ Previous online training courses are available here:
https://www.cmgl.ca/training/online/previous/

❑ Upcoming live online training courses are available here:

https://www.cmgl.ca/training/online/

Online Training
❑ Duration of the course:
❑ 9:00am to 4:00pm MDT
❑ 1 hour break ~12:00pm to 1:00pm MDT
❑ Two 15 minute breaks

3
 11/5/2020

Online Training
❑ Online Agenda
❑ PowerPoint Presentation
❑ Tutorials

Other Courses of Interest

❑ Introduction to CMG's Modelling Workflows:
https://www.cmgl.ca/training/introduction-cmg-modelling-workflows

❑ Fundamentals of Reservoir Simulation:

https://www.cmgl.ca/training/fundamentals-reservoir-simulation

4
 11/5/2020

Course Agenda
❑ Basic Concepts
▪ Geological modelling
▪ Geostatistics
▪ Upscaling
▪ Uncertainties in reservoir modelling

❑ Geomodelling Capabilities in Builder

▪ Mapping
▪ Property distribution though geostatistics
▪ Upscaling/up-layering
▪ Rescue Import/Export

❑ Exercise

CMG Workflow

5
 11/5/2020

Geological Modelling
Basic Concepts

What is Geomodelling?
❑ All mathematical methods used to produce a geological model of a subsurface
object

❑ Object modelled can describe:

▪ Topology
❖ history of a region
▪ Geometry
❖ size, shape, and position in space
▪ Physical properties
❖ static formation properties

6
 11/5/2020

What is Geomodelling?
Primary Objective:
▪ Predict the spatial variation of geological variables

Variable
▪ Properties of the geological subsurface that exhibit spatial variability and can be
measured in terms of numerical values

Spatial Variation
▪ When a quantity that is measured at different spatial locations exhibits values that differ
across the locations

Why Build a Geomodel?

❑ Handles large amount of data
❑ Analysis in three dimensions
❑ Test/visualize multiple geologic interpretations
❑ Estimate in-place volumes of hydrocarbons in a subsurface reservoir
❑ Gain an understanding of physical properties of the reservoir
❑ Input framework for a numerical reservoir simulation
❑ Development drilling and well planning scenarios
❑ Exploration
❑ Surface mine planning
❑ Assess uncertainty

7
 11/5/2020

Background
❑ Modern reservoir characterization started due to:
▪ Deficiency of oil recovery techniques
❖ inadequate reservoir description
▪ Aim → predict inter-well property distributions (𝜑, K)

❑ Inter-well heterogeneity cannot be measured:

▪ Seismic data (large support, low resolution)
▪ Well data (small support, high resolution)

❑ Complementary sources of information:

▪ Geological models
▪ Statistical models
❑ Need to combine all info → static reservoir model

Basic Concepts (Geological Model)

Static Data

Dynamic Data

8
 11/5/2020

Basic Concepts (Geological Model)

❑ Typical Data Information
▪ Hard or Soft Data

Outcrop & Well Data Seismic Dynamic Data

Core Data Data

▪ In geological model construction, soft data can be used as trend or co-variance data to
decrease the geological uncertainty

Basic Concepts (Geological Model)

❑ How can data with different resolutions be merged?
▪ Data integration and consistent interpretation
▪ Make geological sense of all interpreted data
▪ Regional trends and local variations
▪ Use hard data to constrain soft data
▪ Use seismic for constraining the spatial distribution
▪ Layer geometry honours stratigraphic bedding structure

9
 11/5/2020

Basic Concepts (Geological Model)

Outcrop/Analogue Fields
Flow
Simulation

Seismic
Interpretation

Upscaling Typical Geomodelling Structure Model

Workflow

Petrophysical Model

Petrophysical &
Well Correlation

Facies/Lithotype Model Stratigraphy Model

Geostatistical Study

Outcrop/Analogue
❑ Outcrop is a visible exposure of bedrock on the surface of the earth
❑ Does not cover the majority of the Earth’s surface
▪ In most places the bedrock is covered by soil and vegetation
❑ Rock may be exposed due to erosion or tectonic uplift
❑ Outcrops are visually inspected and analyzed to get an idea of the subsurface
formation and bedding

10
 11/5/2020

Outcrop/Analogue
❑ Analogue data is made up from nearby fields with similar rock
characteristics and distributions

❑ Outcrop and analogue field data may provide:

▪ Trends and stacking patterns
▪ Size distributions
▪ Variogram measures of lateral continuity
▪ Other translatable spatial statistics
▪ Knowledge of geological process/principles established through
accepted theories in the area of interest

Seismic Data
❑ Found by estimating the properties of the Earth's subsurface from seismic
testing
❑ Shock waves are sent into the earth and the data is generated by the sound
waves
❑ Sound waves reflect off the rock formations and are captured by geophones
on the surface
▪ A geophone is a device that converts ground movement (velocity) into voltage
▪ Deviation of the measured voltage from the base line is called the seismic
response
❑ Acoustic properties of the rock can help determine the type of rock and
geological structure
❑ Seismic Interpretation is the extraction of subsurface geologic information
from seismic data

11
 11/5/2020

Seismic Data
❑ Aim of seismic interpretation is to produce structural maps that reflect
spatial variation in depth of certain geological layers
❑ Seismic data can be used to infer:
▪ Structure
❖ Faults
❖ Fractures

▪ Depositional Settings
❖ Channels
❖ Turbidite system

▪ Sand Continuity/Variation
▪ Bed Thickness
▪ Depositional Facies
▪ Reservoir Properties (Φ)
▪ Fluid Type Identification

Basic Concepts (Seismic Data)

Vertical Seismic Slice Showing Faulted Stratigraphic Features

12
 11/5/2020

Structure Model
❑ Horizons picked on seismic interpretation and well data
▪ In absence of seismic data, possible to be based off well data only

❑ Only faults with sufficient throw that may impact reservoir performance continuity, fluid
distribution etc. should be incorporated

❑ Four main sections of structure modelling:

▪ Fault Modelling
▪ Pillar Gridding
❖ Faults used as the “skeleton” for generating the 3D grid
❖ Skeleton is grid consisting of a Top, Mid, and Base
▪ Surfaces
❖ Full surfaces added to model
▪ Depth Conversion
❖ Converts the acoustic wave travel time to actual depth
❖ Based on seismic and well tops

Petrophysical and Well Correlation

❑ Petrophysics
▪ Study of physical and chemical properties of rocks and their
contained fluids
▪ Emphasizes properties relating to the pore system and its fluid
distribution and flow characteristics
▪ Highlights the integration of core data with log data; adjustment of
core data to reservoir conditions; and the calibration and regression
line-fitting of log data to core data
❖ Statistical correlation and calibration of core and log data requires that the data
is properly depth aligned, have outliers deleted, and can be mathematically
transformed
❖ A variety of line-fitting techniques are available

13
 11/5/2020

Petrophysical and Well Correlation

❑ Properties to be determined:
▪ Thickness (bed boundaries)
▪ Lithology (type of rock)
❖ Determined by working with cores and rock cuttings
❖ Can be combined with log characteristics to identify depositional environments and
characterize how these change throughout the reservoir
▪ Porosity
❖ Computed from well logs in combination with routine-core data adjusted to reservoir conditions
▪ Fluid saturations
❖ Core analysis and well logs (sw found from resistivity logs)
▪ Fluid identification and characterization
▪ Permeability (absolute)
❖ Calculated from Φ logs through a permeability/porosity transform
▪ Fractional flow (oil, gas, water)
❖ Core and SCAL analysis

Petrophysical and Well Correlation

14
 11/5/2020

Stratigraphy Model
❑ Use info from well log and core correlation along with seismic depth conversion to determine
stratigraphy
❑ Subdivide reservoir in zones based on sub-surface horizons and seams

Geostatistical Study
❑ Statistical method accounting for spatial variability
❑ Goal:
▪ Utilize wide variety of data, in different scales and accuracies, to construct reservoir models
▪ Models are able to represent geological heterogeneities
▪ Quantifying uncertainties by producing many equiprobable models

15
 11/5/2020

Facies Model
❑ Facies:
▪ A mass of rock that can be recognized by its composition, structures or fossil content
and mapped on the basis of those characteristics
▪ Within a given facies the reservoir properties can vary significantly
❖ Leads to a further subdivision known as Flow Units

Flow Units
▪ Regions in the sedimentary sequence that are judged to control the movement of
injected and produced fluids within the reservoir

❑ Facies Model Integrates

▪ Geological Study
▪ Geostatistical Study
▪ Geophysical Interpretation

Facies Model (Integration)

❑ Three main ways to integrate interpretations:
▪ Flow Units:
▪ Differentiate flow units using relative permeability/SCAL data
▪ Model flow units will hopefully align with log properties
❖ Possible correlation to non-cored wells across the field
▪ Model network of rock using flow units
❖ Production characteristics of hydrocarbons or water vary by flow unit

▪ Log Facies:
▪ Separate the reservoir into Φ classes possibly linked to Vshl or other logs curves
❖ Calibrated to core properties or depositional facies
❖ This model may reflect production characteristics

▪ Depositional Facies:
▪ Rarely link to logs perfectly
▪ Dependant on core interpretations → arguable
▪ Too many facies = a complicated answer

16
 11/5/2020

Petrophysical Model
❑ Petrophysical data needs to be upscaled to the geomodel
❑ Properties from log interpretation:
▪ Volume shale
▪ Effective Porosity
▪ NTG
▪ SW
❑ Log data should honor core analysis
❑ Geomodel should honor log data
❑ Fluid contacts (GOC, WOC) also need to be identified

Upscaling
❑ Geological characterizations typically contain on the order of 107 - 108 cells
▪ Impractical for flow simulation
▪ Long simulation run times
❑ Upscaling involves reducing the amount of cells by amalgamating the data
❑ Smaller sets of characteristics to represent the most significant aspect of the
reservoir

17
 11/5/2020

Flow Simulation
❑ Grid block model of petroleum reservoirs where each block represents a
local part of the reservoir
❑ Within a grid block the properties are uniform
▪ Porosity, permeability, relative permeability, etc.
❑ Model incorporates:
▪ Reservoir fluids (PVT)
▪ Reservoir description (porosities, permeability, etc.)
▪ Property distribution in space
▪ Individual well pressure and flowrates with time
❑ Purpose is estimation of field performance (e.g., oil recovery) under one
or more producing schemes

Geological Model Summary

▪ Reservoir geology is the science of building predictive reservoir models
from the foundation of geological knowledge
▪ data, interpretations, and models

▪ Reservoir models represent spatial variation of lithology

▪ 𝜑, K distributions → static model

▪ Flow simulations (dynamic models) require high-quality static reservoir

models

▪ Static reservoir models are improved through analysis of dynamic data

→ iterative process

18
 11/5/2020

Geostatistics
Basic Concepts

Geostatistics
❑ What is Geostatistics?

▪ Statistical methods accounting for spatial variation

▪ Spatial correlation among sample data

▪ Science of estimation and simulation for spatial data

▪ Data integration

19
 11/5/2020

Geostatistics
❑ Why Geostatistics?

▪ Integrate geological knowledge and depositional environment interpretation

▪ Predict petrophysical parameters away from well location

▪ Identify geological trends

▪ Uncertainty analysis

Geostatistic Concepts
❑ Deterministic
▪ All data known beforehand
▪ No representation of uncertainty
▪ Always reproducible given same starting conditions

❑ Probabilistic (Stochastic)
▪ Element of chance involved
▪ Know likelihood something will happen, but don’t know when
▪ A random probability distribution or pattern that may be analyzed statistically, but may
not be predicted precisely

20
 11/5/2020

Geostatistics (Deterministic or Probabilistic?)

❑ Oil reservoirs are physical systems that are perfectly deterministic

▪ There is a finite value for all properties in the reservoir
▪ However, the problem is sampling

❑ Geological systems were created by many processes at different scales

❑ Only have few measurements
▪ Lack a great deal of information to fully determine those systems
❑ Impossible to write the deterministic equations that will calculate the property
value observed at any given location in the system
▪ e.g. porosity in the reservoir

Geostatistics (Deterministic or Probabilistic?)

❑ We do NOT know the behaviour of the system in terms of mathematical equations

❑ Based on sampling, we can still describe its behavior in terms of statistics and
probabilities

❑ Stochastic = Deterministic + Random

▪ Noise is random by definition, most data are stochastic
▪ Apparent randomness implies sensitivity to initial conditions

❑ Stochastic simulation:
▪ Generation of hypothetical data (realizations) with certain probabilities by
providing (pseudo) random input values

21
 11/5/2020

Basic Concepts (Geostatistics)

In general, geostatistical
estimation consists of 3 steps:

1. Examine the similarity between a set of sample

(known) data points via an experimental
variogram analysis

2. Fit a permissible mathematical function to the

experimental variogram

3. Conduct an interpolation method based on this

function

Basic Concepts (Geostatistics)

❑ Geostatistical methods are optimal when data are:
▪ Normally distributed
▪ Stationary (mean and variance do not vary significantly in space)

❑ Significant deviations from normality and stationarity can cause problems

▪ Always best to begin by looking at histogram or similar plot to check for normality

22
 11/5/2020

Basic Concepts (Geostatistics)

❑ Histogram
▪ Graphical representation of the distribution of numerical data

Skewed Normal

Geostatistic Terms
❑ Variance
▪ Numerical value indicating how widely individuals in a group vary
▪ The average of the squared differences from the mean

𝑁
1
σ² = ෍(𝑥𝑖 − µ)²
𝑁
𝑖=1

N = number of samples
𝑥𝑖 = sample value
µ = mean

23
 11/5/2020

Geostatistic Terms
❑ Standard Deviation
▪ Used to quantify amount of variation/dispersion for a set of data values
▪ Square root of the variance

𝑁
1
σ= σ² = ෍(𝑥𝑖 − µ)²
𝑁
𝑖=1

N = number of data points

𝑥𝑖 = data value
µ = mean

Geostatistic Terms
❑ De-clustering
▪ Assignment of weights whereby redundant data are given less weight in the
construction of a representative probability distribution or the calculation of summary of
statistics
❖ Used to reduce possible bias in the data histogram
❖ Reduces the influence of over sampling in sweet spots (wells drill in good quality reservoir)
❖ Clustered data will have lesser weights than isolated data points
❖ If not used, the data are all equally weighted

❑ Normal Scores Transformation

▪ Way to normalize data which may otherwise yield a very erratic variogram
▪ Useful with any interpolation method when the data histogram is highly skewed
▪ Results are transformed back into the original histogram or into the declustered
histogram after estimation

24
 11/5/2020

Geostatistic Terms
❑ Normal Scores Transformation

Variogram
▪ Is a function of the data variance against
the distance between the data locations

▪ Describes the spatial structure and used to

model spatial correlation in the data

▪ May vary with direction

▪ Each dataset is characterized by its own

spatial structure, thus has its own
variogram

25
 11/5/2020

Variogram (terms)
❑ Sill
▪ Semivariance value at which the variogram levels off

❑ Range
▪ Lag distance at which the semivariogram reaches the sill value
▪ No correlation of data beyond the range

❑ Lag
▪ Separation distance when comparing pairs of data value

❑ Nugget
▪ Semivariogram value at the origin
▪ Variability at distances smaller than typical sample spacing, including measurement error
▪ Noise level

Variogram

Sill
Semivariance

Range

Experimental Variogram Data

Variogram Model

Nugget

Lag Distance

26
 11/5/2020

Variogram – Lag
▪ Pairs of values with respect to point 55
113 99 90
103 for Lag 0:
97
112
101 92 55 62
106 91
55 53
86 Lag 2
100
104 99 74 55 78
67
96 Lag 1 111 55 98
62 Lag 0 105
109
63

107 97 82
53
55
102 ▪ Pairs for Lag 1:
109 98 94

78 88
55 99
77 55 86
95 102
68 69
87 55 67
76
78 55 74
79 84
77 55 …
Precipitation

Variogram and Uncertainty

27
 11/5/2020

Variogram and Uncertainty

All possible versions have in common:

▪ Same Data
▪ Same Histogram
❖ Represent the global uncertainty
(all locations together)
▪ Same Variogram
❖ Represent the spatial correlation
between any two points

Variogram and Spatial Heterogeneity

28
 11/5/2020

Variogram Models

Variogram (Isotropic vs Anisotropic)

Isotropic Anisotropic

29
 11/5/2020

Variogram and Trend

❑ Variogram does not reach a sill, instead it increases above the data variance level

▪ If possible, remove trend before computing variogram

▪ Compute variogram in a direction where the trend is not present
▪ Use only the beginning of the variogram

Geostatistical Modelling Methods

❑ Estimation
▪ Use when a smooth interpolation across the data is needed
▪ Can generate one result
❖ At least two wells needed for data input

❑ Simulation
▪ Use when variability observed in the data needs to be reproduced
▪ Can generate multiple results
▪ Important for uncertainty analysis
ESTIMATIONS
A. Inverse Distance Estimation
B. Ordinary Kriging Estimation
C. Ordinary Kriging with secondary variable
D. Trend Estimation

SIMULATIONS
E. Gaussian Geostatistical Simulation.
F. Gaussian Geostatistical with secondary variable

30
 11/5/2020

Estimation - Inverse Distance Estimation

❑ Interpolation estimates made based on values at nearby locations weighted
only by distance from the interpolation location
▪ Assume that sample values closest to the prediction location will be similar
▪ As move farther away from prediction location, influence of points will decrease

Estimation - Kriging
❑ Weighted-average mapping technique that uses the variogram as the
weighting function
❑ Locally accurate and smooth (underestimates variability)
❑ Appropriate for visualizing trends
❑ Inappropriate for flow simulation where extreme values are important
❑ Does not assess for global uncertainty
▪ Produces one map –> Deterministic
❑ Good for Tops and Thickness

31
 11/5/2020

Estimation - Kriging
❑ Basic Principles
▪ Search made around the area of the point being estimated
▪ Samples are assigned weights that reflect their spatial variability using the variogram
model
▪ A weighted average is calculated to estimate the value of the point
▪ Kriging minimizes the variance of the estimation error
▪ Provides the best average distribution
❖ High values will be underestimated
❖ Low values will be overestimated
❖ Average will be as predicted
❖ Honors data points, but does not reproduce global statistics
➢ Histogram, variance, and covariance

Estimation - Inverse Distance vs Kriging

λ2=1/3 λ2=1/4

u u
λ3=1/3 λ3=1/2
λ1=1/3 λ1=1/4

Inverse Kriging
Distance
▪ ID gives same weight for ▪ Kriging gives shared weight to the two clustered
equal distances data points

32
 11/5/2020

Geostatistical Simulation
❑ Gaussian Geostatistical Simulation
▪ Used to generate multiple versions (realizations) of grid values
▪ All realizations reproduce:
❖ Data
❖ Histogram
❖ Variogram of the data
▪ Appropriate for flow simulation
▪ Contrary to Kriging, adds an element of randomness
▪ Accounts for the uncertainty associated with some reservoir properties
▪ Small scale variability may mask large-scale trends

Geostatistical Simulation
❑ Gaussian Geostatistical Simulation
▪ Averaging multiple realizations will generate results very close to Kriging
▪ However, the average behavior of multiple realizations may be different
from the kriged model

33
 11/5/2020

Estimation vs Simulation

Kriged curve = Average of all simulated curves

Simulated curves = All curves found from Gaussian method

Estimation vs Simulation (Variogram)

34
 11/5/2020

Estimation vs Simulation (Histogram)

Estimation vs Simulation (Flow Simulation)

One Answer Multiple Answers

35
 11/5/2020

Choosing a Modeling Method

Need to
Investigate
No Uncertainties? Yes

Deterministic Stochastic
Model Model

Continuous Discrete Continuous Discrete

Sequential Indictor Simulation

Kriging Sequential Gaussian
Simulation Truncated Gaussian Simulation
Co-kriging Indicator Kriging Multipoint Simulation
Random Gaussian
Inverse Distance Function Object Modeling

Upscaling

36
 11/5/2020

Upscaling
❑ What is upscaling?
▪ Process in which a very heterogeneous region of reservoir rock
described with a large amount of fine grid cells is replaced by an
equivalent less heterogeneous region (coarse gird cells)
▪ Assign effective static properties to coarse scale cells
▪ Essentially an averaging procedure
▪ Coarse grid model should maintain the same storage and transport
properties of the reservoir rock of the fine scale model

Upscaling
❑ Why Upscale?
▪ Infeasible to run flow simulations directly on the geological model
▪ Reduce CPU time for flow simulation
▪ Make fine grid simulation practical
❖ Geological model ~ 10-100 million grid cells
▪ Reduce number of cells, but preserve accuracy in prediction
▪ Will be the starting point for prediction in the absence of historical
data
▪ Will be the starting point for history matching if historical data is
available

37
 11/5/2020

Upscaling

Source: Upscaling of grid properties in reservoir simulation, PetroWiki, SPE International

Upscaling
❑ How QC and Verify:
▪ Visual QC
▪ Initial Volume check
❖ Pore Volume (POVO)
❖ Hydrocarbon Pore Volume (HC POVO)
▪ Spatial Check
❖ Histogram
❖ Mean
❖ Variance and Standard Deviation
▪ History Matching through flow simulation

38
 11/5/2020

Upscaling
❑ Difficulties of upscaling:
▪ Bottleneck in the workflow
▪ Loss of details
❖ Dominated by loss of vertical resolution
▪ Lack of robustness
❖ Applicability to models with different global boundary conditions and well
locations
▪ How well does the upscaled model represent the fine scale
geological model?
❖ Won’t know fully until flow simulation is preformed
❖ Can be time consuming

Uncertainties in Reservoir Modelling

39
 11/5/2020

Uncertainty
❑ Includes:
▪ Measurement accuracy
▪ Incomplete or missing data
▪ Computation approximations
▪ Stochastic systems

Geophysical Uncertainties
❑ Acquisition
❑ Processing and interpretation
❑ Migration
❑ Time to depth conversion
❑ Horizon picking
❑ Fault position
❑ Well ties

40
 11/5/2020

Geological Uncertainties
❑ Sedimentary depositional environments
❑ Rock type and associated heterogeneities
❑ Spatial distribution
❑ Sediment supply and particle size
❑ Geological markers

Petrophysical Uncertainties
❑ Net reservoir thickness
❑ Volume of shale
❑ Porosity
❑ Permeability
❑ Water saturation
❑ Fluid contact locations (GOC & WOC)

41
 11/5/2020

Geological Modelling Capabilities in Builder

Geological Modelling: Typical Workflow

Geological Modelling Software

Grid (unit system)

Properties Simulation output
Geological Units RESCUE Model properties at one or
Fault Surfaces (Binary files) more times
Well Paths

Builder Results

Simulator

42
 11/5/2020

Builder Geomodelling Capabilities

❑ Import well data
▪ Trajectories, well log information, and top markers
❑ Create horizons
▪ Create 2D maps based on well trajectories and top markers
❑ Create 3D grid
▪ Create radial, cartesian, and corner point grid systems
❑ Create and distribute static properties
▪ User input
▪ Predesigned property map
▪ Geostatistics

Builder Geomodelling Capabilities

❑ Formula based property distribution based on correlations
❑ Property statistics
❑ Property cross-plots
❑ Ruler Tool
❑ Up-layering grid for flow simulation
▪ Builder can combine layers in the vertical direction
▪ Builder can also split layers if required
❑ Rescue/RESQML import/export
▪ 3rd party geomodelling software
❑ Scripting Tool

43
 11/5/2020

Import Well Data

❑ Can Import:
▪ Well trajectories
▪ Well log information
▪ Top markers

Maps

❑ Import Maps and display map over simulation model

❑ Create maps with well top and log information

44
 11/5/2020

Creating Maps in Builder

Geostatistical Methods Available

Tools for importing data

Imported Data Values for

Geostatistical Methods

Geostatistics in Builder
❑ Geostatistics on 3D Properties (Reservoir Menu)
▪ Object Modeling (Geological Bodies)
▪ Gaussian Geostatistical Simulation

Porosity 2009-01-01 K layer: 1

File: CMGBuilder1
460,000

1,520,000 1,530,000 1,540,000 User: JuanV

Date: 5/26/2009
Scale: 1:68599
Y/X: 1.00:1
Axis Units: ft
450,000
450,000

Results are saved as Builder’s variables for

440,000
440,000

Geostatistics. Selected property will be

created or replaced. 0.00 0.50 1.00 miles
0.00 1.00 2.00 km
1,520,000 1,530,000 1,540,000
0.00

45
 11/5/2020

Property Statistics

Histogram based on
Color Scale intervals

Reports for each

sector and region

Cross-Plots
❑ Scatter point plotting tool
❑ Property vs. property filtered plotting

❑ QC the input received from geostatistical data

❑ Revise correlations between important
petrophysical properties
▪ ex: K, 𝜑 , and Sw

46
 11/5/2020

Cross-Plots

Scatter point plot

Y axis frequency
of chosen X & Y
or density plot properties based
on chosen filters

Filter items by X axis

search term frequency or
density plot

X-axis Y-axis Filter by Data type 1st data 2nd data

property property date options type filter type filter

Ruler Tool

47
 11/5/2020

Array Calculator (Formulas)

▪ Builder/RESULTS has an array calculator
▪ Can calculate new properties based on other properties

Enter new property name

Choose independent variables

Enter formula

Up-layering in Builder
❑ Options in CMG
▪ Can be done manually or automatically in Builder
❖ Note: Builder can only upscale in the vertical direction (up-layer)
▪ Manual:
❖ Builder provides mean and standard deviation of selected property
❖ Helps decide which layers to combine
❖ Guild property is typically porosity, permeability K, or combination
❖ Manually choose number of layers after upscale
❖ Manually choose which layers to combine
▪ Automatic:
❖ Builder can suggest optimal layers to combine given the desired number of layers and property
❖ Method based on “residue optimization” (SPE paper #57273)
❖ A table of residues is produced from analysis of all possible layering combinations
❖ Can determine number of simulation layers needed based on table

48
 11/5/2020

Up-layering in Builder
❑ Can amalgamate layers
▪ Property variability calculations allow Builder to aid in up-layering process
▪ Builder can suggest the optimal combination of the layers
❖ Method based on “residue optimization”
❖ SPE paper #57273

Additional Grid Modifications

❑ Splitting Layers, Grid Refinement, and Submodel Extraction
▪ Layers can be split or combined
▪ Can also divide a grid in I or J planes
▪ Similar approach for sub-models or refinement

49
 11/5/2020

Rescue/RESQML Format

Import Rescue File Format

Export Properties to Rescue Model

❑ Builder can read grid from Rescue model

❑ RESULTS can add simulation output properties to Rescue model

▪ Grid in Rescue model must match grid in simulation run

❑ RESULTS can write new Rescue model

▪ With more detail than previous model

50
 11/5/2020

Builder: Scripting Tool

❑ Powerful tool that supports automation of workflows and calculations

❑ Used to create multiple geostatistical realizations

❑ Automatically re-calculate predefined property formulas

❑ Save multiple datasets

❑ Initialize and run simulation datasets

Builder: Geostatistics

51
 11/5/2020

Builder: Geostatistics with Scripting Tool

Exercise
Geological Modelling using Builder

52