You are on page 1of 49

Alex Novlesky

Sr. Reservoir Simulation Engineer

Agenda
Shale Oil & Gas Production
Why use Reservoir Simulation for modelling Tight reservoirs, including Shales?
What Physics are being modelled in Tight & Shale plays?
New Advances in Modelling Hydraulic Fractures
How has simulation helped in understanding the physics & production/recovery
mechanisms of these plays?
Tight & Shale Reservoir Modelling: Challenges, Opportunities & Lessons
Learned?
Why use CMG for Modelling Tight & Shale plays?

North America Shale Plays

USA Shale & Tight Oil & Gas Production


(2000-2013)
USA Shale & Tight Oil Production (mmbpd)

USA Dry Shale Gas Production (bcfd)


2.8

Eagle Ford (TX)


Bakken (MT & ND)

2.4

Rest of US
Marcellus (PA and WV)

2.0

Haynesville (LA and TX)

Granite Wash (OK & TX)


Bonespring (TX Permian)

35
30
25

Eagle Ford (TX)

Wolfcamp (TX Permian)


Spraberry (TX Permian)
Niobrara-Codell (CO)

1.6

Bakken (ND)

1.2

Woodford (OK)

20
15

Fayetteville (AR)

Woodford (OK)
Monterey (CA)
Austin Chalk (LA & TX)

0.8

Barnett (TX)

0.4

Antrim (MI, IN, and OH)

10
5

0.0
2000

2002

2004

2006

2008

2010

2012

Source: EIA based on DrillingInfo and LCI Energy Insight

0
2000

2002

2004

2006

2008

2010

2012

USA Gas Production (1990-2040)


History

2012

Projections

40
100

35
30

80

25

70
60

Shale gas

20

50

15
10
5

40

Tight gas

Non-associated onshore

20

Non-associated offshore

1990

1995

2000

30

2005

2010

Source: EIA, Annual Energy Outlook 2014 Early Release

Associated with oil


Coalbed methane
2015
2020
2025
2030

Alaska
2035

10

2040

Bcf/d

Tcf/y

90

USA Oil Production (1990-2040)


History

2012

Projections

10

U.S. maximum production level of


8 9.6 million barrels per day in 1970

mmbpd

Tight oil
6

Lower 48 offshore
Alaska

1990

Other lower 48 onshore


1995

2000

2005

Source: EIA, Annual Energy Outlook 2014 Early Release

2010

2015

2020

2025

2030

2035

2040

Why Use Reservoir Simulation?


For Physics-based EURs & Optimization

Long time to pseudo-steady-state


Multi-phase flow
Non-darcy (turbulent) flow
Multi-component phase behavior,
adsorption & diffusion
Compaction of fractures
Heterogeneous rock properties
Heterogeneous fractures
Geomechanics
Geochemistry

Why Use Reservoir Simulation?


To Represent Current Development Practices
Analyze & Forecast multi-well pad models
exhibiting interference
Model re-fracs & infill drilling
Interpret production
surveillance data
Simultaneously account for
many uncertain parameters

Commonly Modelled Physics


Reservoir Description

Matrix porosity & permeability


Natural & propped fractures
Pore volume compaction/dilation
Non-darcy (turbulent) flow

PVT
Black Oil
Primary production

EoS
Miscible gas injection EOR & near-critical fluids

Commonly Modelled Physics


Adsorbed components

Gas phase only, dry tight/shale gas


Multi-component gases & liquids

Diffusion
Multi-component gas
Miscible gas injection EOR

Rock Physics
Tight rock Rel Perm & Cap Press in
matrix
Straight line Rel Perm & no Cap
Press for fractures

Source: SPE 164132

Commonly Modelled Physics


Simulation Model Gridding
LS-LR-DK or Tartan Grids surrounding the propped
fractures
Transient multiphase fluid flow from matrix to natural
fractures & from matrix to propped fracs
Non-darcy flow in propped fracs near laterals

Simulation Model Initialization


Initialize propped & natural fracture network with
water
Flowback of injected fracture fluid

CMGs LS-LR-DK Tartan Grids

The key to modelling transient


flow from matrix to fractures!

Modelling Planar & Complex Geometry


Propped Fractures

Planar Fractures in SRV

Complex Fractures in SRV

Product Suite
Advanced Processes & Thermal Simulator
Compositional & Unconventional Reservoir Simulator
Three-Phase, Black-Oil Reservoir Simulator
Sensitivity Analysis, History Matching, Optimization & Uncertainty
Analysis Tool
Integrated Production & Reservoir Simulation
Intelligent Segmented Wells
Phase Behaviour and Fluid Property Application
Pre-Processing: Simulation Model Building Application
Post-Processing: Visualization and Analysis Application

CMG has the Right Physics


Physics

IMEX

GEM

BO, VO, GC, WG

EOS

Gas Phase

Multi-Comp

Multi-Comp/OWG Phases

Natural Fracs (NF)

Dual Perm

Dual Perm

Propped Fracs (PF)

LS-LR in Matrix (MT)

LS-LR in Matrix (MT)

MT, NF & PF

MT, NF & PF

MT

Krel & Pc

MT, NF, PF & time

MT, NF, PF & time

Press-dependent Compaction

MT, NF, PF & time

MT, NF, PF & time

Stress-dependent Compaction

Geomechanics-based

Chemical Reactions

Ion Exchange & Geochemistry

PVT
Adsorbed Components
Molecular Diffusion w/ Dispersion

Non-Darcy (turbulent) Flow


Non-Darcy (slip) Flow

Primary Production

Primary Production & EOR

CMG Milestones in Unconventional Reservoir


Modelling Capabilities & Workflows

Microseismic Data
Can use to estimate the extent of
the unpropped SRV during
pumping & the geometry of its
fractures
Acquired to monitor or even
control the treatment*
Easily incorporated into Builders
workflow using the Microseismic
import wizard

* Reference: George Kings SPE course

Geomechanics
Model permeability change, with hysteresis, as a function of stress
change during production and shut-in periods
Fracture opening during hydraulic fracturing treatments
using GEOMECHs Barton-Bandis feature

New Advancements In
Hydraulic Fracture Modelling

Existing Situation
Dataset keywords:
**$ Fracture
RESULTS FRACTURE BEGIN
RESULTS FRACTURE WELLNAME

Well 1'

REFINE 303,343,7 INTO 2 5 1


CORNERS RG 303,343,7
7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000
2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000
7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000
7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000
2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000
7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000
7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000
2*7562.5000 7575.0000 7550.0000 2*7562.5000 7575.0000 7550.0000 2*7562.5000
7575.0000 7550.0000 2*7562.5000 7575.0000
4*8550.0000 8*8560.5481 8*8562.1952 8*8562.8048 8*8564.4519 4*8575.0000
4*8550.0000 8*8560.5481 8*8562.1952 8*8562.8048 8*8564.4519 4*8575.0000
472.5700 2*472.5335 472.4970 472.5770 2*472.5417 472.5065 472.5770
2*472.5417 472.5065 472.5780 2*472.5430 472.5080 472.5780 2*472.5430
472.5080 472.5785 2*472.5435 472.5086 472.5785 2*472.5435 472.5086
472.5795 2*472.5448 472.5101 472.5795 2*472.5448 472.5101 472.5865
2*472.5530 472.5196 474.5700 2*474.5335 474.4970 474.5770 2*474.5417
474.5065 474.5770 2*474.5417 474.5065 474.5780 2*474.5430 474.5080
474.5780 2*474.5430 474.5080 474.5785 2*474.5435 474.5086 474.5785
2*474.5435 474.5086 474.5795 2*474.5448 474.5101 474.5795 2*474.5448
474.5101 474.5865 2*474.5530 474.5196

Refinements:
17 wells
117 stages
8,129 refined blocks
203,225 refinement cells
32,516 property specs
~ 720,000 lines of input deck

Solution?

Concise Fracture Definitions

Remove the refinements keywords from the datasets

Builder and Simulator share the same code

Fractures created upon simulator initialization


What you see in Builder is exactly what the
simulator will create

While Were At It
Fracture Templates
Contain refinement definitions
Re-use multiple fractures or wells
Single place to parameterize in dataset

Make Hydraulic Fractures a simulator keyword


Apply different fracture templates
Fracture properties recognizable in dataset
Parameterization of fractures available outside Builder

Fractures defined as an Object


Assign properties by fracture name
Block Groups allow for quick & easy defining/editing

New Setup- Fracture Template


RESULTS PLNRTEMPLATE NAME 'Template_I_Direction'
RESULTS PLNRTEMPLATE PRIMFRACWIDTH 0.0018
RESULTS PLNRTEMPLATE PRIMFRACPERM 100000
RESULTS PLNRTEMPLATE PRIMFRACTIP 100
RESULTS PLNRTEMPLATE END
*PLNRFRAC_TEMPLATE 'Template_I_Direction'
*PLNR REFINE *INTO 5 5 1
*BWHLEN 65
*IDIR
*INNERWIDTH 0.6096
*LAYERSUP 0
*LAYERSDOWN 0
*PERMI MATRIX *FZ 295.3 0.2953
*PERMJ MATRIX *FZ 295.3 0.2953
*PERMK MATRIX *FZ 295.3 0.2953
*END_TEMPLATE

Primary Width (Intrinsic)


Fracture Perm (Intrinsic)
Fracture Tip Perm
Half-Length
Direction
Height (via Layers)
Fracture Perm (Effective)

New Setup- Fracture Definition


RESULTS PLNRSTAGE NAME 'Planar Stage 8'
RESULTS PLNRSTAGE WELL Well 1'
RESULTS PLNRSTAGE DATE 2006-08-14
RESULTS PLNRSTAGE BASENAME Well 1 - Frac'
RESULTS PLNRSTAGE FRACS 'Well 1 - Frac 1' 'Well 1 - Frac 2'
RESULTS PLNRSTAGE FRACS 'Well 1 - Frac 3' 'Well 1 - Frac 4'
RESULTS PLNRSTAGE SLABS '262, 268, 275, 281'
RESULTS PLNRSTAGE PERFOPTION 1
RESULTS PLNRSTAGE LAYERMIN 4
RESULTS PLNRSTAGE LAYERMAX 4
RESULTS PLNRSTAGE END
*PLNRFRAC 'Template_I_Direction' 298,262,4 *BG_NAME 'Well 1 - Frac 1'
*PLNRFRAC 'Template_I_Direction' 298,268,4 *BG_NAME 'Well 1 - Frac 2'
*PLNRFRAC 'Template_I_Direction' 298,275,4 *BG_NAME 'Well 1 - Frac 3'
*PLNRFRAC 'Template_I_Direction' 298,281,4 *BG_NAME 'Well 1 - Frac 4'

Fracture Name
Well
# of Fractures
Template Application
Block Group

Setup Comparison
What does this imply?
Well with 4 stages:

Old:
New:

~ 9500 lines of refinements


~ 5600 lines of property specif

31 Lines

Fast
Saving
Fast
Loading

Fast
Generation

Block Groups Make Life


Easier
Refinements, permeability alterations, and non-Darcy flow
corrections done automatically by simulator
With Block Group definitions, apply additional properties to
fractures:

Relative Permeability Tables


Rock Types / Compaction Tables
Initial Saturations
Etc.

Define Block Groups by


Dual Permeability Systems

Matrix
Natural Fractures

Hydraulic Fractures

Main Fracture Conduit (Fractured Zone)


Enhanced Near-Fracture Region (Non-Fractured Zone)

Converting Old Datasets


Builder and Results 3D views are the same as
before
Old datasets run with new simulator
No Conversion Required

Old datasets can be converted to new syntax


using Builder (automatically when saved)
May be easier and faster to work with

Workflow Demo

What is CMOST?

Better
understanding
Identify important
parameters

Calibrate
simulation model
with field data
Obtain multiple
history-matched
models

Improve NPV,
recovery, etc.
Reduce cost

Quantify
uncertainty
Understand and
reduce risk

Easily Vary Propped Frac


Properties & SRV Size
Propped Frac Properties:
Half-length, Width, Perm, Spacing,
Height & Perm Gradient
Stimulated Natural Frac Properties:
Width, Perm

SRV Size & Shape:

# MS events per gridblock


MS Moment Magnitude
MS Confidence Value
Etc.

How is it Done?
CMOST uses Master Datasets to specify parameters
to be altered
Datasets with CMOST keyword strings

Files can be created:


Manually
Through CMOST (CMM Editor)
Through Builder

Parameterization
With CMOST

Physics-based EURs
History-Match Run Progress Plot

Engineer only has to monitor HistoryMatch progressand so is free to work


on other projects while CMOST does the
rest!

Physics-based Optimization
Cum Oil & NPV after 30 years vs # of Wells
6.00e+6
Cumulative
Cumulative
Cumulative
Cumulative
Cumulative

OPT_1
OPT_3
OPT_5
OPT_7
OPT_9

Well
Wells
Wells
Wells
Wells

# of Wells
1
3
5
7
9

4.00e+6

3.00e+6

2.00e+6

NPV
(MMUSD)
13.0
39.0
64.6
85.3
80.7

100
1.00e+6

0.00e+0
2015

2020

2025
2030
Time (Date)

2035

2040

2045

NPV, MMUSD

Cumulative Oil SC (bbl)

5.00e+6

Oil SC
Oil SC
Oil SC
Oil SC
Oil SC

80
60
40
20
0
1

5
7
# of Wells

Benefits of Reservoir Simulation


Understand and predict tight & shale well production

Reservoir heterogeneity
Well complexity
Physics of fluid flow & heat flow
Geomechanics
Geochemistry

Enable physics-based analysis and optimization of tight & shale


plays in an efficient manner, when using CMOST:
EUR Calculation & Validation
Well Completion Design Optimization
Well Spacing Optimization

Tight & Shale Reservoir Modelling:


Challenges
Lack of PVT data in Shale Liquids plays
Lack of BHP data
Shale reservoir property measurement is uncertain,
costly & time-consuming
Microseismic data acquisition and analysis is
not well understood or accepted
Frac Treatment design software lacks proper
modelling initiation and propagation of naturally
fractured rocks

Tight & Shale Reservoir Modelling:


Challenges
Costly to acquire reservoir rock geomechanical properties
and initial stress states
Not enough Reservoir Engineers:
To conduct physics-based reservoir modelling work
Are cross-trained in Production/Well Completions Technology
and/or Geomechanics

Technology discipline silos inhibit learning between


companies and even within companies

Tight & Shale Reservoir Modelling:


Opportunities
Constrain reservoir parameters using known relationships
between natural frac geometry, width, perm & density
These should not be independent variables

Constrain rock-physics relationships


Rel perm & cap pressure should not be independent functions

Natural fracture characterization via Discrete Fracture


Network (DFN) modelling

Tight & Shale Reservoir Modelling:


Opportunities
Correlate Seismic Attributes & Microseismic analysis with
Fracability
Monitor production using Distributed Temperature
Sensors & Tracer Surveys
Incorporate production logging data into reservoir simulation
history matching

Predict optimum well locations and design multiple


coincident well treatments using Geomechanics
E.g. Simultaneous Fracs like Zipper Fracs

Tight & Shale Reservoir Modelling:


Lessons Learned
Statistical Analysis of early time rates and unqualified EURs
can lead to new oilfield myths that incorrectly become rule of
thumb
30-day, 90-day, 180-day rate versus cumulative well plots that arent
normalized for flowing pressure (BHP or WHP) and for effective
propped fracture parameters are very misleading
EUR versus cumulative production plots can be even more misleading
given the uncertainty with which EURs are generally being determined
using analytical-solution based production decline analysis methods

Tight & Shale Reservoir Modelling:


Lessons Learned
Reservoir Simulation can also be misleading if model design and
physics is not appropriate for the problem at hand
Shale well models that dont use Logarithmically-Spaced grids
yield misleading results
Similar to models that dont use radial grids around wells to model pressure
transient tests
Those models cannot properly model transient inflow performance behavior
(IPRs)

Effect of Not Using LS-LR-DK Grids


Simple DK approach cannot model the initial transient
correctly because the grid blocks are too large!
Well-1 shale gas model constant perm fcd 60.irf
3,000

Well Bottom-hole Pressure shale gas model_constant perm_fcd_60.irf


Well Bottom-hole Pressure Shale Gas Model_Simple DK.irf

Well Bottom-hole Pressure (psi)

2,000

1,000

-1,000
2000-2

2000-3

2000-4

2000-5
Time (Date)

2000-6

2000-7

2000-8

2000-9

2000-10

Companies using CMG to


Model Unconventional Reservoirs
90

16

Canada

USA

80

ROW

70

12

NEW CUSTOMERS

60
10
50
8
40
6
30
4

20

10

0
2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

Cumulative Customer Growth

14

2014 SPE Papers featuring CMG


Reservoir Simulation Technology
216 papers
54 Unconventional, Tight or Shale, including
6 on Gas Injection EOR

Why use CMG for Modelling Tight &


Shale Plays?
1. CMG has the physics required to understand and forecast
production from Unconventional Wells & Reservoirs
2. Import geologic models from geologic modelling software
to jump-start your modelling workflows
3. Add planar, complex or mixed geometry propped and
stimulated natural fractures to your models
4. Use microseismic data in the model building process

Why use CMG for Modelling Tight &


Shale Plays?
5. Add only the LGR required to model transient flow from
matrix to fractures
6. Easily and efficiently build single and multi-well models
7. Parameterize matrix & fracture properties & dimensions
when doing history-matching & optimization,
No limitations to only a few half-lengths, spacings, etc.
No need to manually pre-create

8. CMGs track record of continually enhancing our


capabilities and workflows for Unconventional Wells &
Reservoirs

Training

Register for courses on


www.cmgl.ca/training
Available at worldwide CMG
offices or on-site
All skill levels
Contact: training@cmgl.ca

For more information:


Please contact sales@cmgl.ca

Vision: To be the leading developer and supplier of


dynamic reservoir technologies in the WORLD

www.cmgl.ca

You might also like