Evaluation of Tight Gas Reservoirs

1
Evaluation of Tight Gas Reservoirs

• Victor Hein, P.E.
• Ryder Scott Company
• June 2009

2
Resource Pyramid
1000 md
Small Volumes
High Quality
100 md
Medium Quality
1 md
Increased Demand
Better Technology
Continued Drilling and Development
0.1 md
Tight Gas
0.001 md
Large Volumes
Low Quality
Coalbed Methane
Gas Shales
0.0001 md
3
Gas Consumption North America
4
Gas Consumption World
5
U. S. Net Natural Gas Imports
6
OECD Countries

• Australia Austria Belgium Canada Czech
  Republic Denmark Finland France Germany Greece
  Hungary Iceland Ireland Italy Japan Korea
  Luxembourg Mexico Netherlands New Zealand Norway
  Poland Portugal Slovak Republic Spain Sweden
  Switzerland Turkey United Kingdom United States

7
World Gas Reserves
8

• Excludes Russia

9
Overview of World Gas Reserves
10
Significant Gas Facts

• Tight Gas 20 of US Production
• Unconventional Gas 44 of 2005 US Production
• Unconventional Gas 49 of 2030 US Production
• 2005 OECD 38 Worlds Gas Production
• 2005 OECD 50 Worlds Gas Consumption
• 2030 OECD 27 Worlds Gas Production
• 2030 OECD 42 Worlds Gas Consumption

11
(No Transcript)
12
Tight Gas Resource Triangle

• 58 TCF produced through 2000
• 34 TCF Proven Reserves in 2001
• Technically Recoverable volume 185 TCF
• Undiscovered 350 TCF
• Additional GIP 5000 TCF

Source GTI 2001
13
Tight Gas Characteristics

• Low Permeability (lt0.1md)
• Frac Job or Horizontal Well Required
• Large Pressure Gradients across Reservoir
• High Transient Decline Rates
• Often Commingled Production
• Often Layered and Complex

14
Evaluation Methods

• Volumetrics
• Material Balance
• Decline Curves
• Production History Matching
• Advanced Production Analysis
• Simulation

15
Volumetrics
16
Archie EquationsClean Sand
For low porosity sandstones a is typically 1.0 m
initially thought to be 2.0, later used values of
1.8-2.0 for tight gas
17
Shaley Sand

• Fertl and Hammack reduction of Simandoux
• Original Simandoux (widely used) based on very
  limited samples
• Vsh is effective shale volume, fraction
• F is from shale corrected porosity
• Rsh is the resistivity of the shale in sand
  recommends 0.4R of shale beds

Hilchie 1982 Advanced Well Log Interpretation
18
Density PorosityCorrections for InvasionClean
Sand

• Density gets most of response from 3-4 from
  wellbore
• Solve equations by trial and error
• Initial guess for ?b is from ?mf 1 0.73P
  where P is Salinity in ppm divided by 1,000,000
• Calculate gas density or estimate from following
  figure
• When numbers converge you have answer

Hilchie 1982 Advanced Well Log Interpretation
19
Density PorosityGas Density
Hilchie 1982 Advanced Well Log Interpretation
20
Density Neutron PorosityType I Invasion Profile

• Very deep or very shallow invasion - logs read
  same fluid
• Quick approach below, best simultaneous eq using
  Rxo

Hilchie 1982 Advanced Well Log Interpretation
21
Density Neutron PorosityType II Invasion Profile

• Invasion on order of 3-4 but not beyond neutron
• Simultaneous equations for density with Rxo
  device best
• Use density porosity only will be somewhat high

Hilchie 1982 Advanced Well Log Interpretation
22
Combined Archie EquationClean Sand
23
(No Transcript)
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
Pickett Plot Example
29
Their New Procedure

• R1(T16.77) R2(T26.77)

30
Conclusions
31
Conclusions
32
What Is Pay?
0.5 Bscf Well
2.0 Bscf Well
33
High Resolution in Layered Sands
34
Net Pay

• Low Permeability Zones Often Have Extensive
  Transition Zones
• Calibrate Cutoffs with Cores
• Archie Equation Breaks Down at Very Low
  Porosities for m 2
• Watch Out for Laminated Sands Below Vertical
  Resolution of Logs

35
Data Integration
36
General Observations Log Analysis

• Tight gas systems often contain layered and/or
  dispersed clay (analysis methods different)
• Core data and FMI often critical to understanding
  type system and existence of fractures
• Must depth shift logs particularly in layered
  system
• Calibrate logs and k estimates to cores
• Consider special core analysis in new plays
• Full logging suite required for shaley sands

37
General Observations Log Analysis

• Rw often difficult to obtain or estimate
• Pickett plot can tend to overestimate Rw and SW
• Shale models are complex may not yield correct
  answers
• MRI can help with porosity, k, free water (Hamada
  et al SPE 114254, SPE 90740, Coates 1999,
  Prammer et al - 1996)
• Using Density and NMR porosities together can
  improve porosity estimates (Hamada et al SPE
  114254)
• Rock typing very important (Rushing et al SPE
  114164)
• Porosity determination difficult due to matrix
  changes, incomplete invasion, and clay (Kukal et
  al SPE 11620)
• GR generally better Vsh tool than D-N XP in older
  compacted rocks (Kukal et al SPE 11620)

38
What Is Effective Drainage Area?
0.1 md
0.01 md
0.1 md
0.01 md
0.001 md
0.0001 md
0.0001 md
0.001 md
1 Year
10 Years
39
Effective Drainage Area

• Drainage Area f(k, Xf, Porosity, Geometry)
• Have Dependent Relationship Between Effective
  Drainage Area and Recovery Factor
• Should Define Effective Drainage Area Relative to
  Time and Recovery Factor

40
Core Analysis

• Analysis critical to understanding of layered or
  complex reservoir
• Must measure rock properties under NOB (Jones and
  Owens, Soeder and Randolf)
• Can have ten fold or more reduction in
  permeability due to NOB at 0.01 md and below
• Lower permeability rocks - smaller pore throats

CER, Holditch Assoc 1991
41
Reduction of k with net overburden
CER, Holditch Assoc 1991
42
Reduction of k with net overburdenMesaverde
Formation
43
Estimation of Permeability From Logs

• Example Timurs Equation

• c 2 since k inversely proportional to surface
  area squared and Swi proportional to surface area

straight line with slope m on log log

• Plot

vs
and y axis intercept at Ka m related to
tortuosity of the rock
Kukal, Simons SPE 13880
44
Estimation of k in very tight rocks

• Timurs little data lt 1 md, also unstressed k
• Predicted k two orders of magnitude high for
  0.1µd lt k lt 0.1md
• Clay increases tortuosity, surface area
• Better method plot k(Swi2) vs. (F(1-Vcl) on
  log log plot

• where Swi and Vcl in fractions
• K includes effect of overburden at 1.0 psi/ft
• Swi must be at irreducible
• must correct k to Kg
• Derived from Mesaverde core date in Western
  Colorado

Kukal, Simons SPE 13880
45
Estimating k where Swi unknown

• Plot k vs. (F(1-Vcl)) on log log plot

• Porosity and Vcl in fractions
• Only slightly less accurate that equation with
  Swi
• Derived from Mesaverde also works in Travis Peak
  in East Texas
• Other examples of slot pore geometry are
  Albertas deep Spirit River, East Texas Cotton
  Valley, Piceance Cozzette, Wyoming
• Should be useful in many low k sandstones

Kukal, Simons SPE 13880
46
Material Balance P/Z
47
P/Z Based on Tank Model

• Constant Volume
• No Efflux or Influx
• Pressure Gradients Small
• Measured Pressures Representative of Average
  Pressure

48
Tight Gas P/Z Plot
49
Radius of Drainage Radial Flow
50
Transient vs. Boundary Flow

• Constant Rate Example

Boundary Dominated Well Performance
f(Volume, PI)



Fekete RTA Documentation
51
Transient vs. Boundary Flow

• Transient Flow
• Early time or low permeability.
• Flow that occurs when the pressure pulse is
  moving into an infinite or semi infinite acting
  reservoir.
• The fingerprint of the reservoir. Contains
  information about reservoir properties ie k

• Boundary Flow
• Late time flow behavior.
• Typically dominates long term production data.
• Reservoir is in a state of pseudo-equilibrium
  mass balance.
• Contains information about reservoir pore volume
  (OGIP).

Fekete RTA Documentation
52
Time to Pseudo Radial Flow
Lee, Wattenbarger SPE Textbook 5
53
Duration of Flow Periods HF Well

• ? 0.15, CrD 100, µ 0.03 cp, ct 0.0001
  1/psi

54
Tight Gas Pressure Buildup Tests

• very important to run prefrac BU for k and Pwsi
• prefrac results can insure on proper straight
  line after production period of significant
  length for pseudo radial
• common problem in PT tests is that shut in period
  is too short
• too short of flow period can also be a problem
  i.e wellbore unloading is incomplete (mass rate
  at surface exceeds mass rate out perforations)
• consider bottom hole shut-in (tubing plug) if
  afterflow exceeds maximum practical test time

55
Pressure Buildup Test Design

• make estimates of kg from guess then test data
• estimate end of WBS
• estimate beginning of pseudo radial flow
• estimate onset of BDF
• post frac estimates for time use type curves
• SPE 17088, Dr. W. J. Lee

56
(No Transcript)
57
Static Material Balance Problems

• Difficult to Analyze Due to Required SI Time,
  Heterogeneity, Large Pressure Gradients
• Common Curved Behavior
• Can Result in Large Errors for GIP which are
  Generally Low for Early Cum Values
• Increased SI Times Help but GIP Generally Low
• Scatter f(Pressure Gradients, Heterogeneity)

58
Static Material Balance Summary

• Get Pre Frac Initial Pressure
• Use Longer SI Times if Possible
• Existence of Straight Line Does Not Insure Tank
  Behavior
• If Curved or Two Slopes Look at Later Straight
  Line
• If Possible, QC Shut-in Times

59
Flowing Material Balance

• FTP converted to FBHP
• Pseudo Pressure and Pseudo Time to Correct for
  Viscosity and Compressibility Changes
• Uses BDF flow equation for Gas simplified to

• Iterative Process Dependent on guess for OGIP
• Plot Normalized Rate and Cum Prod

60
Flowing Material Balance
61
Flowing Material Balance

• Curve is concave up until PSS - generally
  yielding minimum OGIP
• Model is well in Center of Circle
• Liquid Loading, Condensate Ring, Scale, etc. can
  Affect Results
• Dont use where Water Drive or Abnormal Pressure
• Use with Caution for High Perm Variations
• Can be Useful but be Careful

62
Decline Curves
63
(No Transcript)
64
Decline Curve Equations - Arps
Exponential
Hyperbolic
65
Arps Traditional Analysis

• Based on Empirical Observations
• Exponential (D is constant)
• Hyperbolic (D changes with time, 0ltblt1)
• Harmonic (b1)
• Used to Calculate Future Rate, Remaining Reserves

66
Types of Decline Curves - Arps
67
Criteria for Arps Analysis

• Well Produced at or near Capacity
• Constant Flowing Bottom Hole Pressure
• Drainage Area Remains Constant (BDF has been
  Achieved)
• Same Completion

68
What About b gt 1 ?

• Wrong Interpretation
• Transient Flow instead of Boundary Dominated

69
Effects of b Factor
Cox, et al SPE 78695
70
Deep Tight Gas Example

• 0.0009 lt k lt 0.07 md
• Spacing 80 acres
• Pwsi 16,200 psia
• BHT 400 deg F
• Porosity 6.6
• Sw 36
• Kv/Kh 0.001
• Thickness 200

Xf 300 (unless specified) Fracture k 100md
Rushing, et al SPE 109625
71
Effects of Layers on b Exponents
72
Effects of Xf on b Exponents
73
Synthetic Single Reservoir80 acres
Cheng, Lee, McVay SPE 108176
74
Synthetic Single Reservoir80 acres
Cheng, Lee, McVay SPE 108176
75
Synthetic Single Reservoir80 acres
Cheng, Lee, McVay SPE 108176
76
Synthetic Single Reservoir80 acres all data
included

• All predictions from regression are too high
• All values of b gt1.0
• b is proportional to Di
• Percent error increases as b increases
• Correct values for b cannot be directly obtained
  from transient data

Cheng, Lee, McVay SPE 108176
77
Synthetic Single Reservoir80 acres

• Stabilization time was 4.4 years
• Discarded all data prior to 5.0 years
• b generally decreases with time and all b values
  lt 1.0
• Results indicate that using only stabilized data
  sufficiently accurate

Cheng, Lee, McVay SPE 108176
78
Decline Curve EstimateSingle Layer, b
constrained to 1.0
Cheng, Lee, McVay SPE 108176
79
Decline Curve EstimateSingle Layer, b
constrained to 1.0
Cheng, Lee, McVay SPE 108176
80
Decline Curve EstimateSingle Layer, b
constrained to 1.0

• b 1.0 results in either underestimates or
  overestimates
• At early times production generally
  underestimated
• Forecasts tend to be more stable as more late
  time data are included in the analysis

Cheng, Lee, McVay SPE 108176
81
Additional Observations

• b for stabilized flow related to reservoir drive
  mechanism, fluid properties and reservoir
  conditions Fetkovich et al 1996, Chen and Teufel
  2002
• b decreases as reservoir depletes
• Average b during entire depletion phase will be lt
  1 Chen and Teufel 2002

Cheng, Lee, McVay SPE 108176
82
Improved Analysis Technique

• Calculate bE
• Pi initial reservoir pressure
• Pp,n is normalized pseudo pressure
• Cgi is isothermal compressibility _at_ Pi
• Zi evaluated at Pp, Zwf evaluated at Pwf

Chen 2002
83
Improved Analysis Technique
Cheng, Lee, McVay SPE 108176
84
Improved Analysis Technique

• Back extrapolate to get qi at zero delta time

Cheng, Lee, McVay SPE 108176
85
Improved Analysis Technique

• Back extrapolate to get Di at zero delta time

Cheng, Lee, McVay SPE 108176
86
Improved Analysis Technique
Cheng, Lee, McVay SPE 108176
87
Improved Analysis TechniqueMulti Layer

• Estimate b at 0.6
• No theoretical basis, based on observed results
  from a few field and synthetic cases
• Other procedures are the same

Cheng, Lee, McVay SPE 108176
88
Decline Curve Analysis CBM

• Rushing, Perego, Blasingame studied CBM behavior
  using simulation (SPE 114514)
• Long term b exponents ranged from 0.20 to 0.80
• Early decline behavior for many wells was
  exponential becoming more hyperbolic with time
• None of the simulated cases exhibited long term
  exponential behavior due to non linear
  relationships between key coal properties and
  either pressure or saturation
• Wells with higher Pwf (all other factors being
  equal) exhibited higher b values for long term
  production

89
Improved Decline Curve Analysis

• Additional method proposed by Ilk, Rushing,
  Perego, and Blasingame (SPE116731)
• Presents power law lost ratio method
• Presents diagnostic curves to aid in decline type
  - hyperbolic, exponential
• Method has merit, however, industry probably slow
  to replace traditional decline curve methods

90
Summary Conventional Analysis

• Fit Hyperbolic when in BDF
• Research Terminal Declines!
• Research Time in Transient Flow!
• See if can Fit with b 1.0
• Use Improved or advanced methods
• Analogies Should Have
• Similar Completions
• Similar Reservoir Parameters
• Similar Spacing
• Sufficient Production History for Reliable
  Analysis

91
Type Curve Changes With Spacing
92
Production History Matching
93
Production History Matching

• Example PROMAT
• Single-Phase, Single-Layer Analytical Model
  combined with Regression Analysis
• Fast Can Provide Accurate k, S (Xf) and
  sometimes Drainage Area
• Accuracy Good where k Variation not big (Sergio
  Vera, MS Thesis, Texas AM 12/2006)

94
Production History Match Example
95
Production History Matching - Problems

• Non-Unique Solutions without Good Reservoir
  Description
• Pre-frac k can Over Estimate Reserves
• Inaccurate in Layered Systems with Large
  Permeability Variations
• Accuracy Increases with Production Data
• Multi-Layer Models often Require Detailed Data
  such as Production Logs, etc.

96
Advanced Production Analysis
97
Flow Periods for Fractured Well

• Most of flow due to expansion in fracture
• Generally too early to be of practical use
• Often masked by WBS

Cinco et al 1981
98
Flow Periods for Fractured Well

• Two types of linear flow simultaneously occur
• Most of flow comes from formation
• Cannot determine frac length just from bilinear
  flow
• Plot of Pwf vs t1/4 is straight line, plot of
  ?P or ?m(p) vs time is ¼ slope on log log plot

Cinco et al 1981
99
Flow Periods for Fractured Well

• Occurs only where high conductivity fracture, CrD
  100
• Continues approx tLfD 0.016
• Plot of Pwf vs. t1/2 is straight line, plot of
  ?P or ?m(p) vs time is ½ slope on log log plot

Cinco et al 1981
100
Flow Periods for Fractured Well

• Transition period between linear flow and radial
  flow
• Badazhkov et al (SPE 117023) contains method and
  references

Cinco et al 1981
101
Flow Periods for Fractured Well

• Fracture functions as extended wellbore
  consistent with effective wellbore radius concept
• Large Lf compared to drainage area can mask due
  to BDF
• Begins at tLfD of approx 3 for CrD 100, less
  for lower CrD
• Pwf vs log t is straight line

Cinco et al 1981
102
(No Transcript)
103
Tight Gas Flow

• General Behavior
• Small Xf compared to ROI
• Linear to Pseudo Radial less common
• Large Xf compared to ROI
• Long Term Linear Flow followed by BDF
• Infinite Acting Linear Systems
• Parallel Reservoir or No Flow Boundaries

104
Analysis of Linear Flow

• Plot of (m(pi)-m(pwf))/qg vs t1/2 yields
  straight lines of different slopes
• Slope departs from analytical value as flow rates
  or degree of drawdown become higher
• Dont see the same degree of departure from
  analytical solutions for pseudo radial flow

105
Analysis of Linear Flow

• Ibrahim and Wattenbarger SPE 100836

106
Analysis of Linear Flow

• Ibrahim and Wattenbarger SPE 100836

107
Analysis of Linear Flow

• Ibrahim and Wattenbarger SPE 100836

108
Analysis of Linear Flow

• Ibrahim and Wattenbarger SPE 100836

109
Analysis of Linear Flow

• Ibrahim and Wattenbarger SPE 100836

110
Analysis of Linear Flow

• Ibrahim and Wattenbarger SPE 100836

111
Analysis of Linear FlowCorrections to Constant
Pwf case for Drawdown

• Define Dimensionless Drawdown as

• Define correction factor as

• Solutions become

• Ibrahim and Wattenbarger SPE 100836

112
Analysis of Linear Flow Corrections to Constant
Pwf case for Drawdown

• Ibrahim and Wattenbarger SPE 100836

113
Analysis of Linear FlowSynopsis
from slope of (m(pi)m(pwf))/qg vs sqrt(t) plot

• Determine pore volume from slope and time to end
  of linear flow, OGIP by including rock and fluid
  properties
• Cannot determine k independently without Ac
• Slope of sqrt(t) plot affected by drawdown for
  constant pwf case
• Without correction factor can be in error by up
  to 22 at maximum drawdown
• No correction factors yet available for constant
  rate case

• Ibrahim and Wattenbarger
• SPE 100836

114
Advanced Production Analysis

• Combines Concepts from Pressure Transient
  Analysis with Production Data
• Allows for Determination of k and S or Xf in
  Transient Region, OGIP from BDF
• Newer Methods (Post Fetkovich) Allow Changes in
  Operating Conditions
• Can also use for Diagnostic Analysis (Transient,
  BDF, Interference)

115
(No Transcript)
116
Fetkovitch

• Combination of Analytical Transient Model and
  Traditional Arps
• Vertical Well, Center of Closed Circle, Single
  Phase Fluid
• Requires Constant Flowing Bottom Hole Pressure -
  less useful for gas wells
• Type Curve Match of Qd and Td against actual rate
  versus time

117
Blasingame, et al

• Utilizes Flowing Pressures
• 3 Different Rate Functions can be Plotted against
  Time Function (Material Balance Time)
• Normalized Rate

• Rate Integral (ave rate to production time)
• Rate Integral Derivative
• Material Balance Time make Solution look like
  Constant Rate, hence
• Depletion Stem of Normalized Rate is Harmonic

118
Blasingame, et al

• Select Model
• Radial
• Infinite Conductivity Fracture
• Finite Conductivity Fractures
• Elliptical Flow
• Horizontal Well
• Obtain match, multiple functions should fit same
  TC
• Possible to obtain k and S or Xf
• Can obtain OGIP if in BDF

119
Blasingame, et al
120
Advanced Production Analysis Blasingame and
others

• Tight gas reservoirs can suffer from non unique
  solutions
• Must have good idea of OGIP to reduce the
  non-uniqueness problem
• Even prefrac buildup for k and a post-frac
  buildup for Xf does not guarantee a unique
  solution
• Can use decline curves or FMB to estimate OGIP

121
Important Type Curve Techniques

• Palacio and Blasingame (SPE 18799, Fekete RTA
  documentation)
• Agarwal, Gardner, et al (SPE 57916, Fekete RTA
  documentation)

122
Important Other Methods
Crafton, Reciprocal Productivity Index
(SPE 37409, 49223)
Ozkan, et al Transient RPI for Horizontal Wells
(SPE 77690, 110848)
123
Reservoir Simulation
124
Reservoir Simulation

• Gold Standard for Evaluation
• Expensive, Time and Data Intensive
• Danger is Experience Level of Hands On Workers
  or Weakest Link
• Simulation Requires Knowledge of General
  Reservoir Engineering and Production Engineering
• Must have People in you Organization to
  Understand Work Process, Results and Integrate
  with Common Sense

125
Cox et al SPE 98035
126
re Generally Proportional to Lf, k
Cox et al SPE 98035
127
re Generally Proportional to Lf, k
Cox et al SPE 98035
128
Practical Limit Exists for Geometry and low k
Cox et al SPE 98035
129
Synopsis for Accurate Evaluation

• Industry Acceptance and Time Constraints Require
  Mostly Traditional Decline Analysis
• Volumetrics Must be Augmented with Core Data,
  Analogy, Information from More Advanced Methods
  or Insights from Performance particularly at
  Lower Permeability
• Advanced Methods are useful for Obtaining
• K, S, or Xf for Transient Flow
• OGIP for Boundary Dominated Flow and
  sometimes Linear Flow
• For OGIP, Confirm from Several Methods or with
  Reservoir Simulation or Analytical Model

130
Your Most Powerful Tools

• Peer Reviews!
• Integrated Methodology

131
(No Transcript)