PT264 Decision Completion Type

Making the Decision
Decision
DecisionCriteria
Criteria
••Maximize
MaximizeNPV
NPV
••Acceptable
AcceptableReturn
Returnon
onInvestment
Investment(ROACE)
(ROACE)
••Done
Doneas
asprevious
previousprojects
projectswere
were
••Gut
Gutfeel/Warm
feel/Warmfeeling/Comfort
feeling/Comfortzone
zone
Now,
Now,your
yourtask:
task:
••Pick
Pickthe
thedecision
decisioncriteria
criteriato
tobe
beused.
used.
••Identify
Identifythe
thedecision
decisioncriteria
criteriaused
usedon
onpast
past
projects.
projects.
PT3542-Decision.ppt 1
Making the Decision
Included
Includedin
inthese
thesecriteria
criteriamust
mustbe:
be:
••Safety
Safetyto
topersonnel
personnel
••Safety
Safetyto
toenvironment
environment
Plus
Plusconsiderations
considerationsof:
of:
••Money
Money(Capital
(Capitallimitations)
limitations)
••Personnel
Personnel(Limitations
(Limitationsof/on
of/onqualified
qualifiedstaff)
staff)
••Time
Time(Equipment
(Equipmentavailability,
availability,lease
leaseexpiry,
expiry,
etc.)
etc.)
Making the Decision
The
Theoptimum
optimumwell
wellwill
willbe
beone
onethat
thatmaximizes
maximizesNPV
NPVatat
the
thefield/business
field/businessunit
unitlevel.
level. What’s
What’sthe
theprocess
processtoto
make
makethis
thishappen?
happen?
Select Reservoir
From “Making the Most of Well Planning” EP92 2500
Management Option
Design Inflow System
Design Outflow System

•Well Initials
•Number of Wells
Define Facilities
Define Complet’n Equip
Define Well
Define Rig/Equipment
•Production Profile
Optimize •Capex
•Opex
Max NPV
Yes
Making the Decision
What
Whatare
arethe
thekey
keyissues?
issues?
Key Issues:
Select Reservoir
Reservoir Parameters
Management Option
Pressure Maintenance
Fluid Composition
Reservoir/well monitoring
Drainhole Configuration
Design Inflow System
Impaiment Prevention
Productivity Improvement
Artificial Lift, boosting

Design Outflow System
Conduit size
Satellite technolgoy
Define Facilities
Operations Philosophy
Computer Aided Operations
philosophy
Making the Decision
What
Whatare
arethe
thekey
keyissues?
issues?
Tubing design
Define Complet’n Equip
Flow controls
Corrosion/Wax/Scale
Downhole Well Maintenance
(remedial) Philosophy
Downhole Monitoring
Philosophy
‘Slimline’ design to minimize

Define Well costs
Well evaluation
Minimum environmental impact
Define Rig/Equipment Match with well requirements

Manning
WO requirements
Making the Decision
What
Whatare
arethe
thekey
keyissues?
issues?
In
Inappraisal
appraisalwells,
wells,the
theprocess
processstarts
startsoff
offaabit
bit
differently.
differently. Here
Herethethekey
keygoal
goalisisto
toacquire
acquire
data
datathat
thatmanages
managesthe therisk
riskfor
forthe
thewhole
whole
project
projectsuch
suchthat
thatthe
theexpected
expectedNPVNPVisishighest.
highest.
Prove Hydrocarbon charge

Define Key
Prove Commerciality
Well Objectives
Prove Development parameters
Define evaluluation Sample

methods Core
Log
Production Test
Abandon/Suspend
Define future use of well Early Production
Obaservation
Making the Decision
Decision
DecisionProcess
Process
The
Thethree
threetechniques
techniquesemployed
employedby
bythe
the
industry
industryare:
are:
••Analogy
Analogy
••Qualitative
Qualitative
••Quantitative
Quantitative
Making the Decision - Analogy
This
Thisisisalso
alsocalled
calleddesigning
designingon
on‘historical
‘historical
basis’.
basis’.
“Most
“Mostfrequently,
frequently,well
wellcompletion
completiondesigns
designshave
havesome
some
historical
historicalbasis.
basis. Specifically,
Specifically,they
theyare
arebased
basedon on
previous
previousdesigns
designsthat
thatwere
wereproven
provenacceptable
acceptableininaa
different
differentfield
fieldor
orwell
welland
andappear
appeartotosatisfy
satisfy
requirements
requirementssimilar
similartotothe
thecurrent
currentdesign
designproblem.
problem.
The
Theengineer
engineeruses
usesthis
thishistorical
historicaldesign
designand
andadapts
adaptsitit
totomeet
meetspecific
specificcriteria
criteriaand
andaccommodate
accommodatelocal local
operating
operatingexperience.”
experience.”
“Petroleum Well Construction”, Economides, et al., 1998, Wiley&Sons, Ltd, U.K.
Advantages
Advantageswith
withthis
thisapproach:
approach:
••Simple
Simple
••Quick
Quick
••Works
Workswell
wellififanalog
analogisisvery
veryclose
closeand
and
options
optionslimited
limited
Example:
Shell Offshore 1980-1990
Problems
Problemswith
withthis
thisapproach:
approach:
••Conditions
Conditionsand
andrationale
rationaleused
usedto
todevelop
developaa
design
designare
arenot
notreadily
readilydiscernable
discernable
••Requires
Requiresaahigh
highlevel
levelof
ofexperience
experienceand
and
operations
operationsknowledge
knowledge[with
[withthe
theanalog]
analog]
••AAcoherent
coherentdesign
designphilosophy
philosophymay
maynot
notbe
be
applied
applied
••“This
“Thisapproach
approachcan
canlead
leadto
toinexplicable
inexplicable
geographical
geographicalcompletion
completiondesign
designpractices.”
practices.”
••Does
Doesnot
notstimulate
stimulateconsideration
considerationof ofaawide
wide
range
rangeof
ofalternatives
alternatives(i.e.
(i.e.stifles
stiflesinnovation)
innovation)
••“Although
“Althoughthe
thedesign
designselected
selectedcould
couldbe
bethe
the
best,
best,no
nobasis
basiswill
willbebeavailable
availablefor
for
demonstrating
demonstratingit.”
it.”
Making the Decision - Qualitative
Gather
GatherData
Data
Answer
AnswerQuestions
Questions
Use
Uselogic
logicto
to
discard
discardleast
least
likely
likely
techniques.
techniques.
Rank
Rankremaining
remainingtechniques
techniquesbased
basedon
on
rules
rulesand
andanalogs,
analogs,some
some
quantification
quantification
Advantages
Advantageswith
withthis
thisapproach:
approach:
Requires
Requiresless
lesstime,
time,effort
effort
Possible
Possibleto
toincorporate
incorporate‘soft’
‘soft’data,
data,such
suchas
as
expert
expertopinions
opinions
Example: SC2OOP
Halliburton’s Sand Control COmpletion
Optimization Program
In 1998 Shell had a plan to co-develop
with Halliburton, but retain geological
database as proprietary.
Note that this program was to contains
significant quantitative aspects (e.g.
simple economics, production rate
estimation)
Problems
Problemswith
withthis
thisapproach:
approach:
[Typical
[Typicalproblems
problemswith
withselection
selection
guidelines/matrices
guidelines/matricesand
andrule-based
rule-basedsystems]
systems]
••May
Maybe bedifficult
difficultto
todetermine
determinedecision
decisionpath,
path,
sensitivities,
sensitivities,or
orcompare
comparewithin
withinranking
ranking
••More
Moredifficult
difficultto
toincorporate
incorporatequantitative
quantitative
information
information
••Conflicting
Conflictingopinions
opinionsamong
amongexperts
expertscan’t
can’tbe
be
compared
comparedin inthe
theresult
resultatatthe
theend
end
In the end the development of the SC2OOP

tool failed because it still required too much
judgment on the part of the user and the
developers could not agree on the outcomes.
Making the Decision
Quantitative
QuantitativeDecision
DecisionAnalysis
Analysis
Here’s
Here’sthe
theprocess:
process:
Geological
Geological Engineering
Engineering Commercial
Commercial
Risk
Risk Risk
Risk Risk
Risk
Type of HC Production Cash flow
Profile
Reserve size Profitability
Cost Profile
Statistical Reservoir Studies Economic Studies

Data
Engineering Premises
Volumetrics Studies Feasibility
Geochemistry and design
Cost Estimation
Hydrocarbon Fiscal Regime
Generation Recovery factor
Political
Migration Path Well Performance
Markets
Reservoir Trap Trouble costs
Cost of Capital
Hydrocarbon Seal Delay in startup
Products
Reservoir Rock Feasibility
Project Cost
From PT2702 “Decision Analysis”

Making the Decision - Quantitative
Here’s
Here’sthe
the
completion
completion
Gather
GatherData
Data selection-
selection-
centric
centricview:
view:
For
ForEach
EachCompletion
CompletionTechnique:
Technique:
Estimate
Estimate
Performance
Performance Estimate
EstimateCosts
Costs
Estimate
Estimate
Reliability
Reliability
Compile
CompileDecision
DecisionEconomics
Economics
Consider
ConsiderSensitivities
Sensitivities
Decide
Decide
Quantitative
DecisionAnalysis
Analysis
One
Onething
thingthat
thatquantitative
quantitativeanalysis
analysisdoes
doesisislet
letus
usdeal
deal
with
withrisk
riskininaarigourous
rigourousand
andopen
openmanner.
manner.
In
Inorder
ordertotodo
doso,
so,we
weneed
needtotobe
beable
abletotodetermine
determine
probabilities.
probabilities.
There
Thereare
arethree
threemain
mainmethods
methods
-A
-Apriori
priorimeaning
meaningthe
theprobabilities
probabilitiesare
areknown
known(e.g.
(e.g.
odds
oddsfor
forcards,
cards,dice,
dice,etc.)
etc.)
-Empirical
-Empiricalprobabilities
probabilitiesthat
thatare
arebased
basedononaccumulated
accumulated
data
dataor
orexperience
experience(e.g.
(e.g.probability
probabilitydistributions
distributionsfrom
from
lab
labtests
testsor
orfield
fieldfailure
failuredata
datacollected
collectedininaadatabase
databaseover
over
time).
time).
-Subjective
-Subjectiveprobabilities
probabilitiesreflect
reflectthe
theopinion
opinionof
ofaaperson
person
or
orgroup
groupofofpeople.
people. This
Thiscan
canbe
beused
usedwhere
wheredata
dataare
aretoo
too
sparse
sparse(e.g.
(e.g.technology
technologyandandprocesses
processesused
usedonly
onlyaafew
few
times
timeswith
withfew
fewfailures).
failures).
Quantitative
DecisionAnalysis
Analysis
These
Thesecategories
categoriesfit
fittogether
togethersomething
somethinglike
likethis:
this:
A Priori
Probabilities
Empirical
Probabilities
100
Personal/Subjective
Knowledge, %
Probabilities
0
0 Opinion, % 100
Quantitative
DecisionAnalysis
Analysis
Process
Processisisthis:
this:
Initial Situation
Structure Problem
Deterministic Analysis
Probabilistic Analysis
Appraisal
Decision
Quantitative
DecisionAnalysis
Analysis
-Structuring
-Structuringthe
theproblem
problem
Start
Startby
bylisting
listingthe
thedecisions
decisionsareas
areasand
andpossible
possible
options
optionsin
ineach:
each:
Zones Orientation Casing SC Technique Tubing Size

Single Horiz Openhole None 3.5
Screen only 4.5
Multiple GP 5.5
Vert FP 7
Cased Frac only
Quantitative
DecisionAnalysis
Analysis
Then
Thenselect
selectdifferent
differentcombinations,
combinations,“paths”,
“paths”,
through
throughthe
thetable
tablefor
forfurther
furtherevaluation.
evaluation.
Zones Orientation Casing SC Technique Tubing Size

Single Horiz Openhole None 3.5
Screen only 4.5
Multiple GP 5.5
Vert FP 7
Cased Frac only
Quantitative
DecisionAnalysis
Analysis
Influence
Influencediagrams
diagramscancanbe
beused
usedtotoidentify
identifyimportant
important
variables
variablesand
andrelationships
relationshipsininthe
thedecision
decisionprocess.
process.
They
Theyhelp
helpcapture
captureimportant
importantvariables
variablesand
andrelationships
relationships
within
withinteam.
team.
Completion Profit
Type
Initial Costs
Production Forecast
Reserves Ps
Initial Rate
Result
Perm/Porosity Net/Gross
Decision
Skin Uncertainty
Influence
Quantitative
DecisionAnalysis
Analysis
This
Thisinfluence
influencediagram
diagramforforURSA
URSAshows
showsthe
the
relationship
relationshipbetween
betweeninitial
initialcompletion
completionselection
selectionand
and
completion
completiontype
typeon
ondrilling
drillingsuccess
successand
andtubing
tubingsize.
size.
There
Thereisisno
noprobability-based
probability-basedrelationship
relationshipbetween
between
PVT
PVTPVPAT
PVPATant antthe
therest
restof
ofthe
theitems.
items. That
Thatisisnot
notto
to
say
saythere
thereisisno
nofunctional
functionalrelationship
relationshipatatall.
all.
Quantitative
DecisionAnalysis
Analysis
Influence
Influencediagrams
diagramsare
areprimarily
primarily
communication
communicationtools,
tools,used
usedto
tostructure
structurethe
the
decision
decisionproblem.
problem.
••Decisions
Decisions
••Uncertainties
Uncertainties
••Relationships
Relationships
Later,
Later,influence
influencediagrams
diagramscan
canhelp
helpdevelop
developthe
the
computer
computermodel
modelfor
foranalyzing
analyzingthethedecision
decision
problem.
problem. They
Theycapture
capturerelationships
relationshipsbetween
between
data.
data. AAcommon
commonmistake
mistakeisisto
toignore
ignore
relationships
relationshipsbetween
betweenvarious
variousdata
dataitems.
items.
Quantitative
DecisionAnalysis
Analysis
Next
Nextwork
workon
onthe
thedeterministic
deterministicanalysis.
analysis.
Use
Usebest
bestguess
guessor
orexpected
expectedprobabilities
probabilitiesin
inaa
decision
decisiontree:
tree:
Success
IGP
Fail
Completion FP
Type
Fail
Hz
Fail
Quantitative
DecisionAnalysis
Analysis
For
ForUrsa
UrsaIGP
IGP
Expected
0.99
S PVPAT
Compl.
-7.4
S
Drill 0 prod
.99 0.01
-11.5
S .01
.99 F -7.4
.01 0.99
S Expected
Compl. PVPAT
F -11.5
-7.1
S F
S/T 0.01 0 prod
.99
-7.4 .01
S
.99 F -7.1
.01
F -7.4
Result of this is an expected PVPAT for the IGP

case.
Quantitative
DecisionAnalysis
Analysis
Decision
Decisiontree
treefrom
fromURSA
URSAstudy.
study.
Quantitative
DecisionAnalysis
Analysis
After
Afterthe
thedecision
decisiontree
treeisisconstructed,
constructed,ititmay
may
be
behelpful
helpfulto
tocreate
createaatornado
tornadograph.
graph.
Create
Createthis
thisby
byestimating
estimatinglow
lowand
andhigh
high
numbers
numbersfor
foreach
eachmajor
majorvariable.
variable.
Low
Low--There
Thereisis~10%
~10%probability
probabilitythat
thatthe
the
variable
variable will
willbebeless
lessthan
thanthis
thisvalue.
value.
High
High--There
Thereisis~10%
~10%probability
probabilitythat
thatthe
the
variable
variable will
willbe
bemore
morethan
thanthis
thisvalue.
value.
Low High
K (md) 10 1000
Ps Drilling .4 .8
Skin -6 6
etc.
Quantitative
DecisionAnalysis
Analysis
Use
Usethese
theselow
lowand
andhigh
highvalues
valuesto
tointerrogate
interrogate
the
thedecision
decisiontree.
tree. Rank
Rankfrom
fromthe
thewidest
widestband
band
to
tothe
thelowest.
lowest.
Drill Ps
GP Ps
Inst. Ps
Kv
Kh
PVT
Skin
-100 -50 0 50 100 150 200 250

NPV ($MM)
The
Thevariables
variablesthat
thatcontribute
contributeto
tothe
themost
most
variation
variationare
areselected
selectedfor
forfurther
furtherstudy
study
including
includingprobabilistic
probabilisticanalysis.
analysis.
Quantitative
DecisionAnalysis
Analysis
Probabilistic
Probabilisticanalysis
analysisstarts
startsby
bydefining
definingthe
the
probability
probabilitydistribution.
distribution.
Here is an example of a distribution,

an a priori probability distribution
using a throw of 2 dice:
2 3 4 5 6 7 8 9 10 11 12
There are 36 possible cases. If you add the cases in each grouping
you get
1/36+2/36+3/36+4/36+5/36+6/36+5/36+4/36+3/36+2/36+1/36
= 36/36
“Risk Analysis in Capital Investment” - DB Hertz, Harvard Business Review

Jan/Feb 1964 -- From P2702.
Quantitative
DecisionAnalysis
Analysis
The
Theprobability
probabilityfunction
functionlooks
lookslike
likethis,
this,in
infact
fact
it’s
it’saatriangular
triangulardistribution.
distribution.
Probability Function for Rolling Dice
0.2
Probability
0.15
0.1
0.05
0
0 5 10 15
Value of 2 Dice
The sum of all of the probabilities must equal 1.0.
Summing the product of the value V times the

probability p(V) gives the mean or expected value.
For this case 2* 1/36 + 3* 2/36 … + 12 * 1/36 = 7
This is not always so apparent from inspection as it
seems to be here.
Quantitative
DecisionAnalysis
Analysis
And
Andhere
hereisisthe
thesame
sameinformation
informationdisplayed
displayedas
asaa
cummulative
cummulativeprobability
probabilityfunction.
function.
Cummulative Probability Function for

Rolling Dice
1.2
1
Probability
0.8
0.6
0.4
0.2
0
0 5 10 15
Value of 2 Dice
The cummulative probability curve answers

the question, What is the probability of a roll
of dice adding up to less than the value on the
graph?
The curve must peak at 1.0.
Quantitative
DecisionAnalysis
Analysis
Some
Someof ofthe
thevariables
variablesturned
turnedout
outtotobe
becritical
criticaltotothe
the
decision
decisionprocess
processasasfound
foundbybythe
thetornado
tornadograph
graphof of
sensitivities.
sensitivities.
These
Thesevalues
valuesare
areworthy
worthyof
ofaacloser
closerlook
lookusing
using
probabilistic
probabilisticanalysis.
analysis.
Build
Buildthe
theprobability
probabilitydistributions
distributionsusing
usingthe
theavailable
available
data
data(empirical
(empiricalor
orsubjective).
subjective).
Use
Usethese
thesedistributions
distributionsininsoftware
softwaretotosolve
solvefor
forexpected
expected
values
valuesand
andalso
alsofor
forcumulative
cumulativedistributions.
distributions.
0.99
Compl. S
-7.4
S
Drill 0 prod
.99 0.01
-11.5
S .01
.99 F -7.4
.01 0.99
S
Compl.
F -11.5
-7.1
S F
S/T 0.01 0 prod
.99
-7.4 .01
S
.99 F -7.1
.01
F -7.4
Quantitative
DecisionAnalysis
Analysis
Two kinds of answers result (E.g. from
URSA study):
Expected Values:
IGP FP Horiz
Expected PVPAT($MM) 790 788 782
and Probability Distributions:
Ursa used probability distributions both empirical

(relying on geostatistics with explor. wells) and
subjective (relying on SMEs).
Quantitative
DecisionAnalysis
Analysis
Software
Softwarerequired
requiredfor
forquantitative
quantitativeanalysis:
analysis:
Excel
Excelfor
forbuilding
building‘Simple’
‘Simple’decision
decisiontrees
treesand
and
summing
summingup upeach
eachscenario.
scenario.
Crystal
CrystalBall
Ballor
or@Risk
@Risk(addins
(addinsto
toExcel)
Excel)for
for
handling
handlingprobabilistic
probabilisticfunctions.
functions.
Xana
Xana or
orother
othereconomics
economicscalculation
calculationpackage
package
[Also,
[Also,there
thereisissomething
somethingcalled
calledDPL
DPL
[Decision
[DecisionProgramming
ProgrammingLanuage]
Lanuage]which
whichisis
helpful
helpfulfor
forbuilding
buildingand
andmanaging
managingcomplex
complex
decision
decisiontrees.]
trees.]
Making the Decision
Advantages
Advantageswith
withthis
thisapproach:
approach:
••Decision
Decisionprocess
processisisclear
clearto
tothe
thewhole
wholeteam
team
(everyone
(everyonesees
seesthe
theimpact
impactofoftheir
theirown
own
information)
information)
••Decision
Decisionprocess
processisisfair
fair(as
(asisispossible)
possible)
••Creation
Creationof
ofaaframework
frameworkallows
allowsfast
fast
decisions
decisionsas
asnew
newdata
dataarrives
arrives(core/log
(core/logdata,
data,
new
newtechnologies,
technologies,changing
changingcosts,
costs,
accumulating
accumulatingfailure
failurerates,
rates,etc.)
etc.)
Example:
“Results of the URSA Yellow Sand
Recompletion Study”, TIR BTC-3513,
8/98
•Used reservoir simulation with Prosper
lift tables.
•Used geostatistical methods to generate
distributions for all variables.
•Used Decision Programming Language,
DPL, to create decision tree from influence
diagrams (>250 end branches).
•Linked XANA (economics tool) with
DPL for fast solution times.
•Extensive sensitivities run.
Problems
Problemswith
withthis
thisapproach:
approach:
••Relatively
Relatively*expensive
*
expensive(calendar
(calendardays,
days,man-
man-
days,
days,data
datacosts)
costs)
**
Relative
Relativetotothe
theother
otherdecision
decisionprocesses,
processes,but
but
NOT
NOTrelative
relativetotothe
thecost
costof
ofaapoor
poordecision
decision
••Size
Sizeand
andcomplexity
complexityofofprocess
processcan
can
(purposefully
(purposefullyor
oraccidentally)
accidentally)obscure
obscure
uncertainty
uncertainty
••Soft
Softinformation
information(expert
(expertopinions
opinionsand
and
experience)
experience)must
mustbe
betranslated
translatedaccurately
accurately
into
intoperformance/reliability/cost
performance/reliability/costbefore
before
incorporating
incorporating
Making the Decision
In
Inthe
theend,
end,most
mostprojects
projectshave
haveused
usedaa
combination
combinationof
ofthese
theseprocesses
processesto
toreach
reachaa
decision.
decision.
Suggested Process
1. Use analogs to select a likely
completion type as potential
solution. This becomes the base
case for comparison.
2. Use qualitative process to
create a short list of potential
solutions.
3. Use quantitative process to
analyze the potential solutions,
rank the list and examine
sensitivities of decision.
4. Return to quantitative
process as new data is found.
Making the Decision
Let’s
Let’slook
lookatatthe
thethree
threemain
maincomponents
componentsin
inaa
decision:
decision:
••Performance
Performance
••Reliability
Reliability
••Cost
Cost
Making the Decision
Performance
Performance
This
Thisisisdone
doneatatvarious
variouslevels
levelsof
ofrigor,
rigor,
depending
dependingon onthe
theproject
projectsize,
size,staff
stafftime,
time,data
data
available,
available,etc.
etc.
••Initial
Initialrate
rateprediction
prediction
••PV
PVproduction
productionusing
usingdecline
declinecurves
curves
••PV
PVproduction
productionusing
usingmaterial
materialbalance
balance
••PV
PVproduction
productionusing
usingreservoir
reservoirsimulation
simulation
Making the Decision
Performance
Performance--Initial
Initialrate
rateprediction
prediction
Many
Many(most?)
(most?)times
timesthe
the‘answer’
‘answer’to tothe
the
question
questionof ofbest
bestcompletion
completiontype
typeisisthe
thesame
same
answer
answerto tothe
thequestion
questionof
ofwhich
whichgives
givesthethe
best
bestinitial
initialrate.
rate.
Obviously,
Obviously,this
thisdoesn’t
doesn’tgive
givePV
PVor
orrecovery
recovery
effects.
effects.
Still
Stillneed
needgood
goodanalog
analogand
anddata
datato
toestimate
estimate
initial
initialrate.
rate.
FBHP
Making the Decision
Performance
Performance--Decline
DeclineCurve
Curve
Start
Startwith
withcalculated
calculatedinitial
initialrate
rateand
anddecline
decline
in
inorder
orderto
tointroduce
introducePV
PVeffect
effectand
andrecovery.
recovery.
‘Poor-boy’
‘Poor-boy’technique
techniquefor
forquick
quickPV
PVeffect.
effect.
Need
Needgood
goodanalog
analogwith
withdecline
declinecharacteristic.
characteristic.
Initial rate
Decline
rate
Q Reserves
Econ.
Limit
T
Decline
Declinecurve
curvemay
maybebeused
useddirectly
directlyby
byeconomics
economics
package
packageinstead
insteadof
oftable
tableof
ofrates
ratesand
andtime.
time.(?)
(?)
Making the Decision
Performance
Performance--Material
MaterialBalance
Balance
Introduce
IntroducePVT
PVTandandreservoir
reservoireffects
effectswith
with
tank
tankmodel
modelof
ofreservoir.
reservoir.
Decline
Declineisisbased
basedon
onreservoir
reservoirpressure
pressuredecline
decline
with
witheffects
effectsof
ofGOR
GOR& &WC
WCchanges.
changes.
Q
Off production
or econ. limit
T
MBAL
MBAL&&Prosper
Prosper(for
(forlift
lifttables).
tables).
Making the Decision
Performance
Performance--Reservoir
ReservoirSimulation
Simulation
Introduction
Introductionofofmost
mostaccurate
accuratedescription
descriptionof
of
reservoir
reservoir(dip,
(dip,interaction
interactionbetween
betweenwells,
wells,
moving
movingcontacts,
contacts,layering,
layering,etc.).
etc.).
Initial rate
T
Prosper
Prosper[lift
[lifttables
tablesand
andcalculation
calculation(‘just
(‘justaa
check’)
check’)of
ofinitial
initialrate]
rate]then
thentied
tiedto
toMultiSim.
MultiSim.
Making the Decision
Note
Notethat
thatall
allthe
thedecision
decisionprocesses
processesstarted
started
with
withthis:
this:
The
Theoutflow
outflowcurve
curveneeds
needsthe thesensitivity
sensitivityof
of
various
varioustubing
tubingsizes,
sizes,but
butisisotherwise
otherwise
straightforward.
straightforward.
Modeling
Modelingthetheinflow
inflowcurve
curveisiscritical
criticalto
to
selecting
selectingthe
theright
rightcompletion
completiontype.
type.
How
Howto todo
dothis
thismodeling
modelingwill
willbe
becovered
covered
later.
later.
Making the Decision
Reliability
Reliability
Aka
Aka--Probabilities
Probabilities
IfIfthere
thereisisaa‘fly
‘flyin
inthe
theointment’
ointment’ininmaking
making
decisions
decisionsin inthe
theoilfield
oilfield(and
(andelsewhere)
elsewhere)ititisis
in
inthe
thereliability
reliabilityof ofthe
theoptions.
options.
What’s
What’sthe
thechance
chancethat
thataacompletion
completiontype
typewill
will
succeed?
succeed?
To
Toanswer
answerthis
thisitithelps
helpsto
toask:
ask:
What
Whatare
arethe
thefailure
failuremodes?
modes?
What
Whatisisthe
thelikelihood
likelihoodof
ofthe
thefailure
failure in
inthe
the
various
variousmodes?
modes?
Making the Decision
Reliability
Reliability
The
Themethod
methodemployed
employedwhere
wheresufficient
sufficientdata
data
exists
existslooks
lookslike
likethis:
this:
Chance of hole collapse: 0.05

Chance of screen not reaching bottom: 0.2
Chance of screen damage: 0.2 These
Theseare
arejust
just
example
example
Chance of fatal voids in pack: 0.1 numbers!!
numbers!!
If this represents all the possible failure

modes which can occur independently
from one another or together:
So success is the situation that the hole
won’t collapse and the screen will reach
bottom and the screen won’t be damaged
and their won’t be fatal voids in the pack.
Ps = (1-.05)(1-.2)(1-.2)(1-.1) = 55%
Making the Decision
Reliability
Reliability
In
Inpreparation
preparationfor
forthis
thiscourse
courseititbecame
becameobvious
obvious
that
thatShell
Shelland
andthe
theindustry
industryneed
needto tocollect
collectsuch
such
component
componentlevel
levelfailure
failuredata.
data.
BTC
BTCwill
willbe
becollecting
collectingand
andstudying
studyingadditional
additional
data.
data.
The
TheWellMaster
WellMasterdatabase
database(by
(bySINTEF)
SINTEF)contains
contains
some
somecompletion
completionfailure
failuredata
datain
inother
otherareas.
areas.
This
Thisdata
datamay
maybebeof
ofhelp
helpin
inyour
yourdecision
decision
process.
process.
The
Thedatabase
databasealso
alsoneeds
needstotobe
bepopulated
populatedwith
with
SOI
SOIdata
dataas
asfailures
failuresare
areunderstood.
understood.
Making the Decision
Reliability
Reliability
Where
Wherecomponent
componentlevel
levelfailure
failuredata
dataexists,
exists,failure
failure
rates
ratesof
ofnew
newcombination
combinationof ofvarious
variouscomponents
componentscan can
be
betreated
treatedproperly.
properly.
When
Whenthethedata
dataisisvery
veryrecent,
recent,the
theonly
onlyPsPsavailable
available
will
willbe
bePs
Psof
ofthe
thewhole
wholesystem,
system,i.e.
i.e.did
didthe
thesystem
system
failed
failedor
ornot.
not.
Extrapolating
Extrapolatingthe thetotal
totalsystem
systemfailure
failurerate
rateto
toother
other
situations
situationsisiswrong.
wrong. This
Thisholds
holdstrue
truewhether
whetherthe
the
situation
situationisisaanew
newfield
fieldwith
withdifferent
differentattributes,
attributes,or
or
new
newcombinations
combinationsof ofcomponents
componentsand andtechniques.
techniques.
ItItisiseasy
easy(too
(tooeasy)
easy)totokill
killoff
offaacompletion
completionoption
optionby
by
estimating
estimatingthat thatthe
thesuccess
successrate
rateisistoo
toolow.
low.
What
Whatisismuch
muchharder
harderisisto
tocollect
collectand
andanalyze
analyzefailure
failure
data
dataatatthe
thecomponent
componentlevel
leveland
andto totrack
trackchanges
changesinin
components
componentsand andtechniques
techniquesoverovertime.
time.
Making the Decision
Reliability
Reliability
Some
Somedata
datafrom
fromthe
theShelf
ShelfHorizontal
HorizontalWell
Well
Lookback
Lookback(1997)
(1997)
Pf(10%)
Drilling:
Rebuilding ‘build’ section 10
Hole collapse/stuck pipe/liner 10
Geologic (Requiring Sidetrack) 5
Sand Control: 25.3*
Pstotal = (.9)(.9)(.95)(.75) = 60%, thus Pftotal = 40%
*Note that failure rate for wells near the end of the
study period had declined to less than 10%.
Making the Decision
Reliability
Reliability
Some
Somedata
datafrom
fromthe
theShelf
ShelfHorizontal
HorizontalWell
Well
Lookback
Lookback(1997)
(1997)
Sand Control:
Total Pf 25.3
Catastrophic 16.9
Curtailed 8.4
OBM 40
WBM 17
Higher failures in wells with screen-running problems
and if screen was positioned off bottom.
Failures occurred much more often in Net to gross .55
and lower.
Success varied by screen type, but data was not
conclusive -- 5-10% chance of failure with most
popular pre-packed screens (Baker Fat pack and Hall
EX Lo Pro).
Making the Decision
Reliability
Reliability
BSP
BSPhas
hasmade
madesome
someestimates
estimatesof
ofPs
Psfor
forgravel
gravel
pack
packcompletions:
completions:
Parameter Description Value
Hole Angle 0-15 0
16-45 1
45-65 2
>65 4
Blank Length (m) >30 0
18-30 .5
17-12 2
11-9 3
<9 4
Perm Contrast <200 0
200-700 1
710-1500 3
>1500 5
Add the values up for a given installation and
look up the total in the last table for Ps.
Sand Management Guide SIEP 97-5254
Making the Decision
Reliability
Reliability
BSP
BSPestimate
estimateof
ofPs
Ps

Non-Perfed
Interval <2 0
within gross 2-5 1
interval (m) 6-8 2
8-10 3
10-16 4
>16 5
Csg /Screen Size 9 5/8”/5” .5
9 5/8”/6 5/8” 1
7”/4” 0
Perfed Net/Gross% 100-85 0
84-75 1
74-65 2
64-50 3
49-30 4
<30 6
Making the Decision
Reliability
Reliability
BSP
BSPestimate
estimateof
ofPs
Ps

Gross Perforation <20m 0
Length 21-40m 1
41-75 2
76-100 3
>100 5
Type of Pack IGP 0
EGP 1
Previous Sand None 0
Produciton Detected 1
Sanded Up 3
Making the Decision
Reliability
Reliability
Add
Addthe
thevalues
valuesto
toget
getaa‘design
‘designindex’
index’and
andlook
look
up
upPs
Psbelow.
below.
Fluid Losses 0 l/min 0
0-100 l/min 1
100-200 l/min 6
200-500 l/min 12
500-750 l/min 18
> 750 l/min 24
______________
Design Index Ps %
0 99
1-7 95
8-12 90
13-17 80
18-23 70
24-29 50
>30 <30
Making the Decision
Reliability
Reliability
Example:
Example:URSA
URSA
IGP
IGP FP
FP Hz
Hz
Drilling
Drilling .99
.99 .99
.99 .5.5
Completion
Completion .99
.99 .95
.95 .7.7
‘Infant
‘InfantMortality’
Mortality’ .99
.99 .99
.99 .9.9
Total
TotalPs
Ps .97
.97 .93
.93 .32
.32
Ursa N/G expected to be 0.7-.75.

Data for this study did not include subsequent experience
from Ram Powell with has had no failures in 4 Hz wells
(with GPs).
Making the Decision
Reliability
Reliability
So
Sowhat
whattotodo
dowith
withthe
thedata
dataavailable
availablenow?
now?
Some
Somefailure
failurerisks
risksare
areeither
eitherinherent
inherentororpoorly
poorly
understood
understoodand
anddata
datamust
mustbebeused
usedas
asthey
theyare.
are.
Some
Somefailure
failuremodes
modesareareunderstood
understoodbetter
betterand
and
have
haveresulted
resultedinindesign
designand
andoperational
operational
recommendations
recommendationsthat thatare
arelikely
likelytotoreduce
reducethe
the
risk
riskof
offailure.
failure.
E.g.
E.g.from
fromShelf
ShelfHz
HzLookback:
Lookback:
-centralize
-centralizescreens
screens
-don’t
-don’twork
workscreens
screens
-use
-useWBM
WBM
-avoid
-avoidhorizontal
horizontalininlow
lownet
nettotogross
gross
-run
-runscreens
screensininclean
cleanfluids
fluids
-acidize
-acidizecarbonate
carbonatesystems
systemsand
anduse
usemutual
mutual
solvent
solventon
onoil
oilbased
basedsystems
systems
-run
-runwashcups
washcupsand
andwash
washacid
acidaway
away
Making the Decision
Reliability
Reliability
Q:
Q:Is
Isitita)
a)foolish
foolishor
orb)
b)presumptuous
presumptuoustotouse
use
lower
lowerfailure
failurerates
ratesthan
thansuggested
suggestedby
by
historical
historicaldata
datawhen
whennewnewdesigns
designsand
and
procedures
proceduresare areavailable,
available,(if
(ifunproven)
unproven)to
to
mitigate
mitigaterisk?
risk?
Making the Decision
Cost
Cost
Good
Goodnews:
news: Cost
Costdata
datacan
canbe
bedetermined
determinedwith
with
reasonable
reasonableaccuracy
accuracyfor
forany
anycompletion
completiontype
type
Bad
Badnews:
news: NoNoshortcuts
shortcutsavailable.
available. Requires
Requires
derivation
derivationfrom
fromcomponent
componentcosts,
costs,installation
installation
times,
times,rig
rigrates,
rates,etc.
etc.
Good
Goodnews:
news:Costs
Costshave
havebeen
beenrelatively
relativelyconstant
constant
for
forfew
fewyears
yearsso
socomponent
componentdata
dataisisavailable
availablein
in
well
wellservicing
servicinggroups.
groups.
Bad
Badnews:
news: Quick
Quickcost
costestimation
estimationfor
foruse
usein
infirst
first
pass
passscreening
screeningcriteria
criterianot
notavailable.
available.
Good
Goodnews:
news: Compared
Comparedwith
withPerformance
Performanceand
and
Reliability,
Reliability,Costs
Costswill
willgenerally
generallyhave
havemuch
muchless
less
impact.
impact.
Making the Decision
Cost
Cost
Wouldn’t
Wouldn’tititbe
benice
niceto
tohave
haveballpark
ballpark
numbers?
numbers?
Here
Hereare
areaafew
fewexamples.
examples.
Making the Decision
Cost
Cost
Ram
RamPowell
Powell--OH
OHHorizontal
HorizontalGP
GPselected
selected
Ram-Powell (US$ millions)

OH Hz GP Vert. IGP
Completion 9.75 9
Drilling 9.5 8
Total 19.25 17
Ram Powell
Cost Factor (OH Hz GP / Vert IGP)
1.2
1.18
1.16
1.14
1.12
1.1
1.08
1.06
1.04
1.02
Completion Drilling Total
Scott Lester, SOI, 8/98

Making the Decision
Cost
Cost(data
(datafrom
from1993
1993study
studyalso
alsoin
inRP)
RP)
Showing
Showingcomparison
comparisonfor
forvarious
varioustubular
tubularsizes.
sizes.
Cost Comparison Ram Powell (1993)
1.6
1.5
Cost Multiplier
1.4
1.3 Vert
1.2 Horiz
1.1
1
0.9
3.5 4.5 5.5 7 3.5 4.5 5.5 7
Tubing Size
331/2”
1/2”Tubing
TubingIGP
IGP($5.2
($5.2MM)
MM)as
asbasis
basis
Horizontal
Horizontalwells
wellsassumed
assumedpre-pack
pre-packscreens
screens
Costs
Costs==drilling
drilling++completion
completion
“Large Tubular Completion Study”, P. Fair, R. Gonzales, D. Hebert, R.

Wattenberger, SOI & Shell Development, 10/93 for the 1993 Production
Engineering Conference.
Making the Decision
Cost
Cost--URSA
URSA
IGP FP Hz (no GP)
Drilling 11.5 11.5 12.7
Completion 7.4 7.4 10.5
Total 18.9 18.9 23.2
Side Track 1.5 1.5 2.3

Workover 4 4 4
URSA Cost Factor (relative to IGP)
1.6
1.4
1.2
1
Drilling
0.8
Completion
0.6 Total
0.4
0.2
0
IGP FP Hz (no GP)
Identical
Identicalcosts
costsfor
forIGP
IGPand
andFP FPare
arebecause
becauserigrigcosts
costsdominate
dominateand
andIGP
IGP
time
time equals FP rig time (relatively slow to pump HRWP and = timefor
equals FP rig time (relatively slow to pump HRWP and = time for
rig up and pumping F&P if include some risk for weather time
rig up and pumping F&P if include some risk for weather time with FP with FP
boat).
boat). Some
Someextra
extracosts
costsininDW
DWFP FPfor
formore
moreexpensive
expensiveceramic
ceramicproppant.
proppant.
URSA TIR BTC-3513 8/98

Making the Decision
Cost
Cost--Shelf
Shelf
FP/IGP
FP/IGPcost
costfactor
factor~1.5
~1.5(on
(onjust
justthe
the
completion),
completion),or
oradd
add~$100K
~$100Kto toIGP
IGPcost.
cost.
Equip Pump Perf Total ($K)
IGP
20' 76 59 35 170
50' 79 69 45 193
80' 81 79 60 220
FP
75' 86 160 60 306
150' 96 200 90 386
IGP vs FP on Shelf
Compl. Cost (e xcl. rig
500
400
cost)
300 IGP
200 FP
100
0
0 50 100 150 200
Zone thickne ss
Ronnie Peeler, Halliburton, 8/98


PT264 Decision Completion Type

Uploaded by

Copyright:

Available Formats

PT264 Decision Completion Type

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PT264 Decision Completion Type

Uploaded by

Copyright:

Available Formats

Making the Decision

Design Outflow System

Define Complet’n Equip

Artificial Lift, boosting

‘Slimline’ design to minimize

Define Rig/Equipment Match with well requirements

Prove Hydrocarbon charge

Define evaluluation Sample

In the end the development of the SC2OOP

Statistical Reservoir Studies Economic Studies

From PT2702 “Decision Analysis”

Zones Orientation Casing SC Technique Tubing Size

Zones Orientation Casing SC Technique Tubing Size

Result of this is an expected PVPAT for the IGP

-100 -50 0 50 100 150 200 250

Here is an example of a distribution,

“Risk Analysis in Capital Investment” - DB Hertz, Harvard Business Review

Probability Function for Rolling Dice

The sum of all of the probabilities must equal 1.0.

Summing the product of the value V times the

Cummulative Probability Function for

The cummulative probability curve answers

and Probability Distributions:

Ursa used probability distributions both empirical

Chance of hole collapse: 0.05

If this represents all the possible failure

Parameter Description Value

Parameter Description Value

Ursa N/G expected to be 0.7-.75.

Ram-Powell (US$ millions)

Scott Lester, SOI, 8/98

Cost Comparison Ram Powell (1993)

“Large Tubular Completion Study”, P. Fair, R. Gonzales, D. Hebert, R.

Side Track 1.5 1.5 2.3

URSA TIR BTC-3513 8/98

Ronnie Peeler, Halliburton, 8/98

You might also like