jared philpot principal reservoir engineer arrow energy · pdf filebest practices: coal bed...
TRANSCRIPT
BEST PRACTICES:
COAL BED METHANE MODELLING
Jared Philpot
Principal Reservoir Engineer
Arrow Energy
August 2013
ARROW ENERGY SUBSURFACE UNCERTAINTY
Typical project dimensions:
• How big is the average CSG-LNG development?
• QGC – 6,000 wells across 4,700 km2*
• GLNG – 2,650 wells across 6,900 km2*
• APLNG – 10,000 wells across 5,700 km2*
• Arrow’s Bowen Gas Project – 6,500 wells across 8,000 km2
• Arrow’s Surat Gas Project - 7,500 wells across 8,600 km2
• Somewhere between the State of Palestine (5,640 km2) and Puerto
Rico (8,870 km2)
* Taken from EIS documents of respective projects
4
THE CHALLENGE...
• Lots of development
wells drilled early
• Large geographic
area
• Limited data
• Significant uncertainty
16
ARROW ENERGY CHALLENGE – A COMPLICATED WORKFLOW
Reservoir
Description
Uncertainty
Range
Dynamic
Model
(10 or so)
Alternative
Reservoir
Descriptions
Well Spacing
Study x (10 or
so)
Well Spacing
RF S-Curve x
10 (or so)
Generate RF-
Depth Curves
Low, Mid, High
Low EUR
Mid EUR
High EUR
Pilot History
Screening
Exploration
Data
Wells, Logs,
Seismic
Experimental
Design
Appraisal Data
Static Model
Well Completions
Chevron, Quad,
MBL
Volumetric
Calculation
Low, Mid, High
Gas & Water Type
Curves
Low, Mid, High
17
ARROW ENERGY THE CHALLENGE – IN A NUTSHELL
• Sources of uncertainty
• data density – data distribution across wide area
• data models - accuracy, predictive capability
• pilot production – how to propagate across areas
• Understanding uncertainty
• uncertainty in model predictions given uncertainty in input data
• uncertainty in model itself (how good is the model)
• Living with uncertainty
• capturing uncertainty in development planning
• multiple deterministic cases
• probabilistic framework
18
ARROW ENERGY UNDERSTANDING UNCERTAINTY – SIMPLE PERMEABILITY MODEL
Simple permeability model
• K = f (depth)
• model can be defined by
taking a best fit line
through the data points
19
ARROW ENERGY
Uncertainty in input data:
• Calculate residuals
between trend and data
measurements
• Analyse residual
distribution
• From CDF choose
P(10), P(90) values and
use these to construct
corresponding trends
UNDERSTANDING UNCERTAINTY – SIMPLE PERMEABILITY MODEL
20
ARROW ENERGY
Confidence Intervals :
• Given sample distribution
and size of sample
calculate confidence
interval (95%).
• Function of number of
measurements.
• There is a 95% confidence
the true perm vs depth
trend lies within this
interval.
0.01
0.10
1.00
10.00
100.00
100 200 300 400 500 600 700
Perm
eab
ilit
y (
md
)
Depth (m)
Meas. Low High
CI CI Expon. (Meas.)
UNDERSTANDING UNCERTAINTY – SIMPLE PERMEABILITY MODEL
22
ARROW ENERGY
Impact of Heterogeneity
• Three different methods used to populate Permeability in static
model
• (1) Apply trends directly
• (2) Normal distribution about trend
• (3) SGS about trend
• Impact on EUR assessed using RF = f(Perm, Well Spacing)
b
a
k
WS
RFRF
1
max
UNDERSTANDING UNCERTAINTY – SIMPLE PERMEABILITY MODEL
23
ARROW ENERGY
Mid
Low
High
Impact of Heterogeneity
• Three different methods used to
populate permeability in static model:
1. Apply trends directly
2. Normal distribution about trend
3. SGS about trend
UNDERSTANDING UNCERTAINTY – SIMPLE PERMEABILITY MODEL
ARROW ENERGY PROPERTY POPULATION: PERMEABILITY
Normal
Distribution
About Trend
SGS Surface
About Trend
25
ARROW ENERGY
• The impact of the alternative permeability realizations is relatively
insignificant when compared to the impact of the low, base or high
permeability trend.
• However, this would not be the case if the ultimate recovery was
being estimated for a small sector or for an individual well.
Low Mid High
(1) Apply Trends Directly 0.69 1.00 1.59
(2) Normal Distribution About Trends 0.70-0.71 0.98-0.98 1.60-1.61
(3) SGS Surface About Trends 0.69 0.97 1.58
Normalised EUR
UNDERSTANDING UNCERTAINTY – SIMPLE PERMEABILITY MODEL
26
ARROW ENERGY
• The scatter in the data may be indicative of two different
types of uncertainty:
• the first is related to the heterogeneity of the coal itself
• the second type of uncertainty lies in the position of the trend and
is related to the fact that the current data set is only a limited
sample of the complete population
• Careful attention is required to capture the uncertainty
associated with various correlations developed during
property modeling
• Large range in EUR outcomes.
UNDERSTANDING UNCERTAINTY – SIMPLE PERMEABILITY MODEL
27
ARROW ENERGY
PILOT WELLS
COMPLEX GEOLOGY
CHALLENGING WELL TYPES
UNDERSTANDING UNCERTAINTY – APPRAISAL PILOT PRODUCTION DATA
28
ARROW ENERGY
Define Parametric Outcomes
• Reservoir performance outcomes usually assessed include:
• Historical pressure production
• Historical gas and water production
Optimal Case Selection
• Select cases with good pressure match that are within 10% cumulative gas production
and 20% cumulative water production.
• Extract the parameters used to achieve match for Scope of Recovery analysis
Run Multiple realizations using Experimental Design Sampling
• Parameters used in this example are :
• Permeability (anisotropy)
• Porosity
• Relative Permeability
• Coal compressibility
• Desorption time constant (tau)
UNDERSTANDING UNCERTAINTY – APPRAISAL PILOT PRODUCTION DATA
30
ARROW ENERGY
Base case history match
• single combination of input
parameters (permeability, gas
content, etc)
• single model matched to pilot
production.
Dynamic uncertainty workflow
• quickly find alternative
combinations of parameters
• multiple models matched to
pilot production.
UNDERSTANDING UNCERTAINTY – APPRAISAL PILOT PRODUCTION DATA
31
ARROW ENERGY
Case Perm Comp Corey
Gas
Corey
Water
Gas
Content
KvKh KyKx PL Poro Tau VL
Base 300 0.005 3 2 0 0.25 1 29 0.004 90 0
1 Mid Mid Low Mid Low Mid High Mid Low Mid Low
2 Mid Mid Mid High Low Low Low Low Mid Mid Mid
3 Mid Mid Mid Low Low High Low Low Mid Mid Mid
4 Mid Mid Mid Low High Low Low Low Mid Mid Mid
5 Mid Mid Mid High Low High Low High Mid Mid Mid
6 Mid Mid Mid High High High Low Low Mid Mid Mid
7 Mid Low High Mid Mid High Low Mid Mid High Mid
8 Mid Mid Mid Low Low Low Low High Mid Mid Mid
9 Mid Low Low Mid Mid Low Low Mid Mid High Mid
10 Mid High Mid Low High Mid Mid Mid Low High Mid
UNDERSTANDING UNCERTAINTY – APPRAISAL PILOT PRODUCTION DATA
32
ARROW ENERGY LIVING WITH UNCERTAINTY – WELL SPACING OPTIMISATION
Well configurations:
• three well types considered
• sensitivity of production and economics to various spacing, in-seam
lengths considered for each depth.
Well spacing optimisation example
Quad Lateral Chevron Multi-Branch Lateral
33
ARROW ENERGY
• a set of reservoir
simulation sector models
are generated to cover
depth range of interest.
• permeability, gas content
from depth trends
(different values in each
model)
• various well
configurations assessed
for each depth.
100 m
200 m
300 m
400 m
500 m
600 m
LIVING WITH UNCERTAINTY – WELL SPACING OPTIMISATION
34
ARROW ENERGY
(1) Start with lowest UTC case
(2) Find subsequent case
showing smallest
incremental UTC
Incremental
UTC > Limit?
(3) Found optimal case
No
Yes
(1)
(3)
Incremental UTC =
D ( PV Total Cost )
D ( PV Produced Gas )
For each depth and reservoir:
WELL TYPE A B C
LIVING WITH UNCERTAINTY – WELL SPACING OPTIMISATION
35
ARROW ENERGY CAPTURING UNCERTAINTY IN DEVELOPMENT PLANNING
36
• Builds GIIP distribution considering:
• Thickness
• Density
• Gas Content
• etc.
ARROW ENERGY CAPTURING UNCERTAINTY IN DEVELOPMENT PLANNING
Recovery factor curves
• possibility of estimating
probabilistic EUR’s
without running full field
simulation models
• estimate EUR’s for
various well spacing and
well type assumptions.
Development planning
• area wide P10/50/90
profiles for gas and water
for different well types.
37
ARROW ENERGY CAPTURING UNCERTAINTY IN DEVELOPMENT PLANNING
38
GIIP, EUR Distribution
Load into mapping software
to integrate with other data
and aid decision making Mean GIIP Standard Deviation GIIP
ARROW ENERGY SUMMARY
• Sources of uncertainty
• data density – data distribution across wide area
• data models - accuracy, predictive capability
• pilot production – how to propagate across areas
• Understanding uncertainty
• uncertainty in model predictions given uncertainty in input data
• uncertainty in model itself (how good is the model)
• Living with uncertainty
• capturing uncertainty in development planning
• multiple deterministic cases
• probabilistic framework
39
ARROW ENERGY
• Coauthors: Saikat Mazumder, Siddharta Naicker, Gladys
Chang, Mohammad Boostani, Miguel Tovar, Vikram
Sharma
• Arrow Energy for permission to give presentation
ACKNOWLEDGEMENTS
40
ARROW ENERGY
41
© Arrow Energy Pty Ltd August 2013
While Arrow Energy Pty Ltd has endeavoured to ensure that all information provided in
this publication is accurate and up to date at the time of publication, it takes no
responsibility for any error or omission relating to this information. Furthermore, the
information provided shall not constitute financial product advice pursuant to the
Australian Financial Services Licence held by Arrow Energy Pty Ltd’s related body
corporate. To the maximum extent permitted by law, Arrow Energy Pty Ltd will not be
liable for any cost, loss or damage (whether caused by negligence or otherwise)
suffered by you through your use of this publication.