5 Petronas Shale Gas Workshop - Engineering Module

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 59

Shale Reservoir Engineering

2011 Gaffney, Cline & Associates. All Rights Reserved.

Drilling and Development


The most important factor behind the rapid expansion in
shale gas development has been advances in drilling and
completions technology
Most notable among these are the use of:
(1) horizontal drilling,
(2) light-sand slickwater fracs, and
(3) microseismic
Typical shale wells have lateral well lengths of 5,000 ft or
more with 15 to 20 fracture stages
Typical patterns in a section (640 acres) range from 4
wells (160-acre spacing) to 8 wells (80-acre spacing) with
some pilot projects containing wells spaced at 40 acres
2

2011 Gaffney, Cline & Associates. All Rights Reserved.

Fractured Horizontal Well Schematic

Source: ESG Solutions


3

2011 Gaffney, Cline & Associates. All Rights Reserved.

Completions and Stimulation Design: Multi-Stage


Completion Systems

2010 Baker Hughes Incorporated. All Rights Reserved.

And this is What it Really Looks Like


52,000 HP in the Woodford Shale

2012 Gaffney, Cline & Associates. All Rights Reserved.

Shale Reserves (or Resources) Booking


Same definitional principles and guidelines apply as for
conventional reservoirs
To qualify as Reserves the key requirements must be met
Firm commitment to develop
Reasonable timeframe (five years)
Economic producibility
Etc.
Proved Reserves requires reasonable certainty of
recovery
For SEC purposes, use can be made of reliable
technology, but what does this mean in shales?

2011 Gaffney, Cline & Associates. All Rights Reserved.

Resources Classification
Reserves
Discovered, recoverable, commercial,
remaining
Proved, Probable, Possible
1P, 2P, 3P (now formalized)
Classification

Contingent Resources
Discovered, potentially recoverable, not
yet commercial, remaining
1C, 2C, 3C (new terms)
Equivalent to Low, Best and High Estimates

Prospective Resources
Undiscovered, potentially recoverable,
potentially commercial, remaining
Low, Best and High Estimates

Unrecoverable
Discovered or undiscovered, not
recoverable

Categorization

PRMS Reserves Defined


Reserves are those quantities of

petroleum anticipated to be
commercially recoverable by
application of development projects to
known accumulations from a given date
forward under defined conditions
Reserves must satisfy four criteria

discovered
recoverable
remaining (as of the evaluation date)
commercial
based on the development project(s)
applied

For unconventional formations,

Reserves are typically only attributed


after pilot programs have confirmed the
technical and economic producibility
and after capital is allocated for
development

Critical Factors
1. The extent of the resource (what has been discovered?)
Gas or oil in place
2. Number of wells in each reserves or resources category
How to demonstrate appropriate level of confidence
3. Volume recovery per well
Methods to estimate EUR
Decline analysis
Basic
Advanced

Simulation
Analytical
Numerical

Analogy
9

2011 Gaffney, Cline & Associates. All Rights Reserved.

How Much Has Been Discovered?


Consider spread of wells across acreage
Number, distance apart, types of information available
Information from neighboring leases
Has commerciality been established everywhere?
Short tests or longer term production
Estimates of gas or oil in place
Rock properties free gas
Adsorbed gas
Fluid properties and future behavior
Use as cross-check of recovery efficiency
10

2011 Gaffney, Cline & Associates. All Rights Reserved.

Projecting EUR
If you torture the data long enough, it will confess!
(Ronald Coase, Economist)

Decline analysis
Basic
Advanced

Simulation
Analytical
Numerical

Analogy

2011
2011 GAFFNEY,
GAFFNEY,CLINE
CLINE && ASSO
ASSOCIATES
CIATES (G
(GCA).
CA). ALL
ALL RIG
RIGHTS
HTS RESERVE
RESERVED.D. TE
TERMS
RMS AAND
ND CO
CONDITIO
NDITIONS
NS OF
OF USE:
USE: BY
BY ACCEPTING
ACCEPTINGTHIS
THIS DOCUMENT,
DOCUMENT, THE
THE RECIPIE
RECIPIENT
NTAGREES
AGREES THA
THATTTHE
THE DO
DOCUME
CUMENT
NTTOGETHE
TOGETHERRWITH
WITHALL
ALL
INFORMATIO
INFORMATIONN INCL
INCLUDE
UDEDD THE
THEREIN
REIN ISIS THE
THE CO
CONFIDE
NFIDENTIAL
NTIAL AND
AND PROP
PROPRIETARY
RIETARY PROPERTY
PROPERTY OF
OF GGCA
CA AND
AND INCLUDES
INCLUDES VALUABLE
VALUABLE TRA
TRADE
DE SECRE
SECRETS
TS AND/O
AND/ORR PROP
PROPRIETARY
RIETARY INFORMATIO
INFORMATIONN OF
OF GCA
GCA (COLLECTIVEL
(COLLECTIVELYY
"INFORMATION").
"INFORMATION"). GCA
GCARETAI
RETAINS
NSALL
ALLRIGHTS
RIGHTSUNDER
UNDERCO
COPYRIG
PYRIGHT
HTLAWS
LAWSAND
ANDTRA
TRADE
DESECRE
SECRETTLAWS
LAWSOF
OFTHE
THEUNITED
UNITEDSTA
STATES
TESOF
OFAMERI
AMERICA
CAAND
ANDOTHER
OTHERCO
COUNTRIES.
UNTRIES. THE
THERE
RECIPIENT
CIPIENTFURTHE
FURTHERRAGRE
AGREES
ESTHAT
THATTHE
THEDO
DOCUMENT
CUMENT
MAY
MAYNOT
NOTBE
BEDISTRIB
DISTRIBUTE
UTED,D,TRANS
TRANSMITTE
MITTED,D,COPIED
COPIEDOORRREPRO
REPRODUCE
DUCEDDIN
INWHOLE
WHOLEOR
ORIN
INPART
PARTBY
BYANY
ANYMEANS,
MEANS,ELECTRONI
ELECTRONIC,C,MECHANI
MECHANICAL,
CAL,OR
OROTHE
OTHERWISE,
RWISE,WITHOUT
WITHOUTTHE
THEEXPRESS
EXPRESSPRIOR
PRIORWRI
WRITTEN
TTENCONSENT
CONSENTOF
OFGCA,
GCA,AND
AND
MAY
MAYNOT
NOTBE
BEUSED
USEDDIRECTLY
DIRECTLY OR
ORINDIRECTLY
INDIRECTLY IN
INANY
ANYWAY
WAYDETRIMENTAL
DETRIMENTAL TO
TOGCAS
GCASINTEREST.
INTEREST.

11
2011 Gaffney, Cline & Associates. All Rights Reserved.

Shale Reservoir Analysis


Shale reservoirs require a new/unique approach for analysis
and forecasting as compared to conventional reservoirs
Nanodarcy permeability
Long transient periods
Short production histories mean decline curve analysis has large
uncertainty

Horizontal wells with multi-stage fracturing causes changes in


rock fabric and complicates aerial contribution
Draw down dependent recovery can have geomechanical
effects

An integrated approach is required to plan wells; plan


stimulation; and forecast production over the long term

12
2011 Gaffney, Cline & Associates. All Rights Reserved.

Shale Gas Well Performance


Wells in thermogenic shale gas reservoirs produce at very high
initial rates and decline rapidly
This is due to multiple factors, including a reduction in reservoir
pressure near the wellbore, a reduction in permeability as pore
pressure decreases, and reductions in fracture conductivity
resulting from proppant crushing, proppant embedment, and
fines migration
Because many wells produce more than half of their total gas
within the first two years, drilling must expand continuously to
increase the field gas rate
In shale gas reservoirs dominated by sorbed gas, such as the
Antrim shale, production may be delayed because of
dewatering and more closely-spaced wells may be needed to
accelerate this process

13

2011 Gaffney, Cline & Associates. All Rights Reserved.

Estimation of EUR
Conventional approaches do not work
Decline analysis is commonly used, but has its challenges
Flow regime is different and changes
b factor changes with time, but we dont know how
What is final value that b could reach?
0.5 in homogeneous reservoirs?
1.5 in very heterogeneous reservoirs?

Standard assumptions may be flawed

The well-known Arps plot is more a means of communication than

analysis
There may be alternatives that provide guidance
But perhaps look at several

Analytical and numerical simulation can help


Analogs so useful elsewhere are difficult to justify
The play is the analog to itself (PDPs for PUDs, etc.)

Take geo-mechanics into account


Numerical model with variable permeability function

14

2011 Gaffney, Cline & Associates. All Rights Reserved.

The Unconventional Challenges


How to predict performance for unconventional reservoirs,
especially shale gas wells
How and when to assign reserves
Assimilation of historical data and building of
type curves has drawbacks:

Does not involve natural constraints

Difficult to explain well design effects

Optimization is based on trial and error

Large investments for know-how

Geo-mechanical effects are ignored

Long term predictions are questionable

Not all shale formations are the same!

Type Well
Most critical assumption is often over-simplified

The Problem
Shale reservoir performance
prediction is more complex than more
traditional formations
Conventional theory does not work
Limited history can lead to forecasting
errors
Type curves are widely used and,
potentially, abused
The choice of which type curve to use
depends on a number of factors
including area, permeability-thickness,
lateral length, and completion
effectiveness
What other options are there?

Inconvenient Facts
Flow may be transient for many years
Initial best-fit b will almost always be greater than unity
It will decline with time, but this cannot (yet) be predicted
Some use a terminal decline (or Dmin) to try to prevent
over-estimation of EUR
Does this work?
What should the value be?

Simplified View of Flow Regimes


Shale Gas well performance is influenced by:
Matrix contribution
Extent and intensity of the artificial reservoir (fracking success)
Flow vectors that are perpendicular to fracture planes (linear flow)
Boundary conditions can be affected by fracture closure

Transient flow
Storage
Fractures dominate

Boundary Effects
Semi-Steady state
Geo-mechanical effects
Fluid effects?

Main Regime
Partially linear
Non-Darcy

Forecasting Methodologies
Extrapolation of decline curves
Basic
Arps: exponential, hyperbolic, harmonic, super-hyperbolic
Modified hyperbolic (very common in industry)

Advanced
Stretched Exponential Decline Model (SEDM)
Power-law
Long-duration linear flow
Duong model
We can only use DCA once a well has exhibited decline!

Analytical models
Numerical models
Analogs
Type curves

Arps Decline Formulae


Q = Qi * (1 + b*Di*t)^(-1/b)

Super Hyperbolic,
b = 1.2

Exponential,
b=0

Harmonic,
b=1

Plot from Fekete HarmonySoftware

Can b>1? Yes, during transient flow, infinite-acting linear flow b=2
b declines with time as flow regime changes
Ultimately it is expected that b<1

Modified Hyperbolic
Changes With Additional Data

From analogs?

EUR after 2 yrs = 8.8 Bcf


EUR after 4 yrs = 8.1 Bcf
EUR after 6 yrs = 7.9 Bcf

Plot from Fekete HarmonySoftware

D decreases with time: D = 1 / (1/Di + bt)


Requires constant b, which is not correct in transient flow

Examples of Minimum Terminal Decline

EUR = 22
EUR = 11
EUR = 6.7
EUR = 10
EUR = 7.9
Plot from Fekete Harmony Software

EUR = 8.6

Advanced Decline Methods

Prepare Bracketing Forecasts


Assume continuation of hyperbolic with best-fit b value
This will give a maximum value

Assume that decline continues as exponential with b=0


This will give a minimum value

Calculate decline using SEDM, Power Law and/or Duong


methods
Likely best estimate value

Calculate continued hyperbolic decline but with imposition of


minimum terminal decline
Use a range of Dmin values, based for preference on analogs
Range might be from 2% (optimistic) to 10+% (conservative)

DCA Fit Comparison


(based on average of 10 Barnett wells)
10000

Rate, Mcf/day

1000

100

Duong(a=2.71, m=1.22)
Exp
Hyp (b= 1.25)

10

Dmin = 5%
PLE (n = 0.7)
Historical data(5.5 yrs)

1
0

25

2500

5000
Time, Day

7500

10000

No. of days

30 Yr EUR (Bcf)

Method

b value

Residual

2273

2.79

Exp

b= 0

8.1

2273

4.65

Hyp

b= 1.25

2.3

2273

4.59

Mod Hyp

Dmin = 5%

2.3

2273

3.26

PLE

n= 0.7 Dinf= 1E-06

0.8

2273

3.84

Duong

a=2.71, m=1.22

0.68

New Change in EUR with Time


(based on average of 10 Barnett wells)
After 2 years

After 3 years

Summary Plot (Semi-Log)


q vs. Time, days - All Models

1.E+04

Summary Plot (Semi-Log)


q vs. Time, days - All Models

1.E+04

qg, MSCF/day or qo, STB/day

qg_MSCF/day_all
qg_MSCF/day_edited
qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

b= 1.75

1.E+02

2,000

After
4 years
1.E+04

4,000

6,000
8,000
Time, days

qg_MSCF/day_ ( Duong)

qg, MSCF/day or qo, STB/day

qg_MSCF/day_(Duong)

qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

b= 1.45

1.E+02

6,000
8,000
Time, days

Plots from Stephanie Curie Rate Time Relation spreadsheet


26

qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

b= 1.7

1.E+02

10,000

2,000

After
6.5 years
1.E+04
qg, MSCF/day or qo, STB/day

qg, MSCF/day or qo, STB/day

qg_MSCF/day_edited

4,000

qg_MSCF/day_(exp)

Summary Plot (Semi-Log)


q vs. Time, days - All Models
qg_MSCF/day_all

2,000

qg_MSCF/day_edited

10,000

qg_MSCF/day_(Duong)

qg_MSCF/day_all

4,000

6,000
8,000
Time, days

10,000

Summary Plot (Semi-Log)


q vs. Time, days - All Models
qg_MSCF/day_(Duong)
qg_MSCF/day_all
qg_MSCF/day_edited
qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)

1.E+03

qg_MSCF/day_(PLE)

b= 1.25
1.E+02

2,000

4,000

6,000
8,000
Time, days

10,000

Change in EUR with Time


(Time limit: 30 yr. or Zero Rate)
EUR vs Time
7
Exp

Hyp

Dmin=5%

PLE

Duong

Dmin = 10 %

EUR (Bcf)

1
500

1000

1500
Time (Days)

2000

2500

Method

No. of days

Years

30 Yr EUR(Bcf)

b value

Exp

730.08

2.31

b=0

Exp

1095.12

2.33

b=0

Exp

1460

2.72

b=0

Exp

2372.76

6.5

2.79

b=0

Hyp

730.08

5.6

b= 1.8

Hyp

1095.12

5.33

b=1.7

Hyp

1460

4.96

b= 1.45

Hyp

2372.76

6.5

4.65

b= 1.25

Mod Hyp 730.08

5.34

Dmin = 5%

Mod Hyp 1095.12

5.1

Dmin = 5%

Mod Hyp 1460

4.83

Dmin = 5%

Mod Hyp 2372.76

6.5

4.59

Dmin = 5%

PLE

730.08

3.07

n= 0.75 Dinf= 1E-06

PLE

1095.12

3.1

n= 0.75 Dinf= 1E-06

PLE

1460

3.2

n= 0.75 Dinf= 1E-06

PLE

2372.76

6.5

3.26

n= 0.75 Dinf= 1E-06

Duong

730.08
1095.02
1460
2372.76

2
3
4
6.5
2
3
4
6.5

2.68
2.9
3.35
3.84
4.45
4.28
4.2
4.13

a=1.65m=1.13
a=1.98m=1.17
a=2.14m=1.18
a=2.71m=1.22

Duong
Duong
Duong

Mod Hyp 730.08


Mod Hyp 1095.12
Mod Hyp 1460
Mod Hyp 2372.76

Dmin = 10%
Dmin = 10%
Dmin = 10%
Dmin = 10%

Barnett

Review of Several US Shales

Eagle Ford Shale


Dry Gas Window
69 wells

Barnett Shale
11,000 wells

Marcellus Shale
3 wells (high rate)
5 wells (low rate)

Compare different DCA methods


Historical Data from DIDesktop

Eagle Ford Wells

DCA Comparison After 2.3 Years


Summary Plot (Semi-Log)
q vs. Time, days - All Models
1.E+05
Duong

qg, MSCF/day or qo, STB/day

qg_MSCF/day_edited
qg_MSCF/day_(exp)

1.E+04

qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)

1.E+03

qg_MSCF/day_(PLE)

1.E+02

1.E+01
0

30

1,000

2,000

3,000
Time, days

4,000

5,000

No. of days

30 Yr EUR (Bcf)

Method

b value

851

1.96

Exp

b= 0

851

3.92

Hyp

b= 1.45

851

3.83

Mod Hyp

Dmin = 5%

851

3.6

PLE

n= 0.14, Dinf=9.84E-09

851

3.64

Duong

a=1.96, m=1.2

6,000

Forecast Based on Duration of Historical Data

qg, MSCF/day or qo,


STB/day

1.E+05

After 1.5 years

Summary Plot (Semi-Log)


q vs. Time, days - All Models
Duong

Duong

qg_MSCF/day_all

qg_MSCF/day_all

qg_MSCF/day_edited

qg_MSCF/day_edited

qg_MSCF/day_(exp)

1.E+04

Summary Plot (Semi-Log)


q vs. Time, days - All Models

1.E+05

qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

qg, MSCF/day or qo,


STB/day

After 1 years

qg_MSCF/day_(exp)

1.E+04

qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

b= 1.9

1.E+02

b= 1.65

1.E+02
0

500

1,000

1,500
2,000
Time, days

2,500

3,000

After 2.33 years


1.E+05

500

1,000

1,500
2,000
Time, days

Summary Plot (Semi-Log)


q vs. Time, days - All Models

qg, MSCF/day or qo, STB/day

Duong
qg_MSCF/day_edited
qg_MSCF/day_(exp)

1.E+04

qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

b= 1.45

1.E+02

Plots from Stephanie Curie Rate time relation spreadsheet


31

500

1,000

1,500
2,000
Time, days

2,500

3,000

2,500

3,000

Eagle Ford - Change in EUR with Data Availability


Time limit: 30 yrs

EUR vs. Time (Days)


6

EUR (Bcf)

4
Exp
3

Hyp
Mod Hyp
PLE

Duong

0
300

400

500

600
Time (Days)

700

800

900

Barnett Wells

DCA Fit Comparison After 6.4 Years


Summary Plot (Semi-Log)
q vs. Time, days - All Models
1.E+05
Duong

qg, MSCF/day or qo, STB/day

qg_MSCF/day_all
qg_MSCF/day_edited

1.E+04

qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

1.E+02

1.E+01
0

34

1,000

2,000

3,000
Time, days

4,000

5,000

No. of days

30 Yr EUR (Bcf)

Method

b value

2343

1.93

Exp

b= 0

2343

2.25

Hyp

b= 1.61

2343

2.17

Mod Hyp

Dmin = 5%

2343

2.19

PLE

n= 0.13, Dinf=9.84E-09

2343

2.16

Duong

a=0.88, m=1.08

6,000

Forecast Based on Duration of Historical Data


Summary Plot (Semi-Log)
q vs. Time, days - All Models

1.E+05

Duong
qg_MSCF/day_all
qg_MSCF/day_edited

1.E+04

qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

b= 1.65

1.E+02

b= 1.65

1.E+02

1.E+01

1.E+01
0

1,000

2,000

After 4.5 years q

1.E+05

qg, MSCF/day or qo, STB/day

Summary Plot (Semi-Log)


q vs. Time, days - All Models

qg_MSCF/day_all
qg_MSCF/day_edited

1.E+04

After 3 years

Duong

qg, MSCF/day or qo, STB/day

After 1 years

3,000
4,000
Time, days

5,000

6,000

1.E+05

2,000

3,000
4,000
Time, days

5,000

Duong
qg_MSCF/day_all
qg_MSCF/day_edited

1.E+04

qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

b= 1.65

1.E+02

6,000

Summary Plot (Semi-Log)


q vs. Time, days - All Models

qg_MSCF/day_all

qg_MSCF/day_(exp)

1.E+03

1,000

Duong

qg_MSCF/day_edited

1.E+04

After 6.4 years

Summary Plot (Semi-Log)


vs. Time, days - All Models

qg, MSCF/day or qo, STB/day

qg, MSCF/day or qo, STB/day

1.E+05

b= 1.61

1.E+02

1.E+01

1.E+01
0

1,000

2,000

3,000
4,000
Time, days

Plots from Stephanie Curie Rate time relation spreadsheet


35

5,000

6,000

1,000

2,000

3,000
4,000
Time, days

5,000

6,000

Barnett - Change in EUR with Data Availability


Time limit: 30 yrs
EUR vs. Time (Days)
2.5

EUR (Bcf)

1.5
Exp
Hyp
Mod Hyp
1

PLE
Duong

0.5

0
300

800

1300

1800
Time (Days)

2300

2800

A Smaller Set of Barnett Wells (30)


EUR vs Time
7
Exp

Hyp

Dmin=5%

PLE

Duong

Dmin = 10 %

EUR (Bcf)

1
500

1000

1500
Time (Days)

2000

2500

Marcellus

3 High-Rate Wells

5 Low-Rate Wells

DCA Fit Comparison- Marcellus


Summary Plot (Semi-Log)
q vs. Time, days - All Models

1.E+05

Duong

Three
High-Rate
Wells

qg, MSCF/day or qo, STB/day

qg_MSCF/day_all
qg_MSCF/day_edited
qg_MSCF/day_(exp)

1.E+04

qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

1.E+02
0

500

1,000

1,500
Time, days

2,000

2,500

3,000

Summary Plot (Semi-Log)


q vs. Time, days - All Models

1.E+05

Duong

Five LowRate
Wells
39

qg, MSCF/day or qo, STB/day

qg_MSCF/day_all
qg_MSCF/day_edited

1.E+04

qg_MSCF/day_(exp)
qg_MSCF/day_(hyper)
qg_MSCF/day_(MH)
qg_MSCF/day_(PLE)

1.E+03

1.E+02

1.E+01
0

500

1,000

1,500
Time, days

2,000

2,500

3,000

Marcellus - Change in EUR with Data Availability


Time limit: 30 yrs

Three High-Rate Wells

Five Low-Rate Wells

EUR VS. Time (Days)

EUR VS. Time (Days)

0.25

0.2

Exp

Hyp
3

Mod Hyp

PLE
Duong

EUR (Bcf)

EUR (Bcf)

5
0.15

Exp
Hyp
Mod Hyp

0.1

PLE
Duong

0.05

1
0

0
300

400

500
Time (Days)

600

700

300

500

700
900
Time (Days)

1100

Predicting Shale Performance is Challenging


Formation description requires special attention
Regional and local trends should be understood
Ro, rock mechanics, stresses, TOC, and mineralogy are key
factors in shale plays
Matrix properties are difficult to measure in logs
Coring is essential to understand shale reservoirs

Natural fractures are difficult to quantify

Conventional reservoir engineering tools dont always

apply
In-place and recoverable volumes do not correlate
Recoverable volumes more closely associated with exploitation
methods

Material balance approach does not apply


Flow is not directly related to pressure difference
The effect of natural fractures is often misunderstood
DCA especially in the short term is uncertain

Massive hydraulic stimulation creates an artificial reservoir; the


bounds and quality of which need to be understood

Most Common Approach: Analogy and Statistics


In-Place
Description

Statistical
Analysis and
Type Wells

Assimilation of historical data and building of


type curves has drawbacks:

Performance

Type Well
Most critical assumption is often over-simplified

Does not involve natural constraints


Difficult to explain well design effects
Optimization is based on trial and error
Large investments for know-how
Geo-mechanical effects are ignored
Long term predictions are questionable

All shale formations are NOT the same!


Source - 2010 Responsible Drilling Alliance

Reservoir Simulation
Model components

Geological description
Recovery process
Flow through matrix
Desorption
Geomechanics

Fracture hydraulics
Well hydraulics
Surface facility integration
Goals
Performance models
Development priorities
Adjustment to regional parameters
Integrated development plan

43

2009 Baker Hughes Incorporated. All Rights Reserved.

Model Process
Individual completions
Wells
Production modules
Field
Regional
Drawdown, psi

Shale Engineering
Advanced Reservoir Engineering for Shales
In-Place
Description

Shale
Engineering

Shale Engineering Modeling


Based on geo-mechanics
Implements physical shale properties
Describes properties of the induced
reservoir after stimulation
Matches observed performance
Interprets and implements micro-seismic
Optimizes well design and field
development
Provides early predictions of long term
production behavior

Performance

Fitting Challenges
Forecast Uncertainties
2,500

Surface pressure
restrictions
Interaction with other
wells or formations
through natural
fractures

Fluid
unloading

1,500

1,250

1,000

200

300

Line pressure
restriction

200

150

100

750

500

100

50

250

0
2010-1

2010-4

2010-7

2010-10
Time (Date)

45

2011-1

2011-4

2011-7

Water Rate SC (bbl/day)

Fluid unloading

1,750

400

Gas Rate SC, Model


Well Head Pressure, Model

2,000

Gas Rate SC (MSCF/day)

Initial trends can be


influenced by:

250

Well Head Pressure (psi)

Gas Rate SC, Historical


Well Head Pressure, Historical
Water Rate SC, Model

2,250

500

Shale Model Description

Proppant
concentration,
distance relation
Extension of the
wellbore

NATURAL
Dual Permeability formulation
Fracture spacing, stress
anisotropy
Proppant
placement

Natural fractures

Matrix

INDUCED
MINOR

NATURAL
Non-Darcy flow
Gas Desorption
46

Stimulated Rock
Volume (SRV)

INDUCED
MAJOR

Induced permeability
domain
Pore pressurepermeability relation

From Natural Shale to the Artificial Reservoir


Logs and core

In situ stress
determination

Microseismic Re-processing

Geomechanical

Natural Fracture
Permeability
Flow Model
Analysis

Benefits
Enhancing reservoir understanding
Exploiting modern technology

Modeling Objectives

Based on SPE 146912

SRV description related to hydrofrac (model A)


Long term well performance (model B)

Model A

Induced
permeability
Proppant
conductivity

Model B

Modeling Approach
Model Attributes
Flow tied to geomechanics through
dilation-compaction tables
Compositional formulation
Gas desorption
Non-Darcy matrix flow

A. Fracture propagation model

B. Full well model

Models fracture propagation

Models long term production

Fine grid (models Minutes)

Coarse grid (models Decades)

Stimulation and backflow timeframe

Uses SRV attributes from model A

Matches frac and microseismic surveillance

Adjusts frac stage contribution to PLT


Outputs stimulated rock volume (SRV) permeability measurements
Adjusted to historical production rates and
pressures
Provides Long production forecast

Model A Gridding

Stimulated Reservoir Volume (SRV)

s max

Calculated
Permeability

Welbore node
sH

Wellbore

50

Model B Gridding

Proppant
nodes

s max

Stimulated Reservoir Volume (SRV)

Welbore node
Full well model
sH

Wellbore

51

Full Model Results


Matched production history and production logging

Frac stage contribution


match
Proppant placement match

Pressure Drop, psi

2,500

1,000

2,000

800

1,500

600

1,000

400

500

200

0
2010-1

2010-7

2011-1
Time (Date)

2011-7

2012-1

2012-7

W ell Head Press ure (psi)

Gas Rate SC (M SCF/day)

Well History match

A Real Shale Engineering Application

Rates measured by PLT five months later

Micro-seismic events do not necessarily relate to production response


Do all micro-seismic events relate to fluid presence?
Production seems dependent on natural fractures

Modeling Workflow
Input Data
Petrophysical
analysis

Model A:
Fracture
propagation

Match Parameters
Frac Treatment
Data
Micro-seismic

Geomechanical
analysis

Model B:
Full Well Model

Flow-back

Geological
Model
Facility Design

Regional
Trends

54

2010 Baker Hughes Incorporated. All Rights Reserved.

Mini-frac

Multi-Well:
Pad Model

Analytical :
Play Evaluation

Production
Logging

Well rates and


Pressures

Can Statistics Help?


SPEE Monograph 3
Identify a resource play

Repeatable statistical distribution of EUR


Offset well performance is not a reliable predictor
Continuous hydrocarbon system exists
Free hydrocarbons not held in place by hydrodynamics

Select analogous wells and create statistical distribution for analysis


Work with randomly selected anchor wells and test set wells
Analyze increasing areas around anchor wells until test set wells show different
distribution
Determine number of opportunities (the project)
Carry out Monte Carlo simulation for undrilled locations and assign appropriate
category
But always consider the geology!

55

2011 Gaffney, Cline & Associates. All Rights Reserved.

A Final Check - Recovery Efficiency


What is Reasonable?
Recovery factors are typically much lower in shale formations
Each formation is different
Depends also on well spacing and completion
Lateral length and number, size and spacing of fracs
What is the drainage area?
Depends entirely on permeability
Nano-Darcy vs. micro-Darcy makes a large difference

Does micro-seismic provide quantitative or only qualitative information?

Use mini-frac to estimate initial pressure and permeability


1-2 bbl/min for 5 minutes, then fall-off (fracture closure in 3-12
hours)

Gas
Perhaps 25% to 50%
But what about NGLs and changing yield?
Liquid drop out considerations (e.g. Eagle Ford condensate
window)

56

2011 Gaffney, Cline & Associates. All Rights Reserved.

Recovery Factors
Calculating recoverable reserves requires understanding of recovery factors
How can we quantify a recovery factor?
The EIAs World Shale Gas Report divides plays into:

35%

30%

25%

20%

15%

Exceptional

Favourable

Average

Less Favourable

Least Favourable

Established high
rates of well
performance

Low clay content; low


to moderate geologic
complexity; overpressured shale
formation; high gasfilled porosity

Medium clay content;


moderate geologic
complexity; average
reservoir pressure
and properties

Medium to high clay


content; moderate to
high geologic
complexity; below
average reservoir
properties

Severe underpressure and


reservoir complexity

Higher RFs may be achievable, but realistic estimates must be based on

what is being achieved with current technology


But what about NGLs and changing yield?
Quantitative contribution and timing of adsorbed gas?

Shale Engineering Implementation


Integrated Team Deployment
Shale Engineering
From Rock to the Bank

Formation
Evaluation

Consulting

Appraisal

Fracturing

Fracture
Modeling

Monitoring

Geomechanics

Software

Planning
2010 Baker Hughes Incorporated. All Rights Reserved.

Completions

Microseismic
interpretation

Reservoir
Modeling

58

Drilling

Strategic
Planning,
Resource
Booking

Evaluation

Operations

Implementation

What we cover and where we can go


Formation evaluation
Microseismic processing
Frac point completions
Geomechanics
Reservoir modeling
Evaluation and Strategy
.other specialists

Reduce
Uncertainty

59

Increase
Production

I.
II.
III.
IV.
V.
VI.

Screening studies
Pilot program designs
Streamlined monitoring
Master plan evaluation
Integrated operations
Process optimization

Accelerate
Development

Reduce
Costs

You might also like