stochastic handling of uncertainties in the decision making process spe london, 26 th october 2010...
TRANSCRIPT
Stochastic Handling of Uncertainties in the Decision Making Process
SPE London, 26th October 2010
Dag Ryen Ofstad, Senior Consultant, IPRES Norway
Setting the sceneProduction Prognosis
0.0
150.0
300.0
2007 2030
MS
m3
o.e
. / Y
ea
r
Undiscoveredresources
Improvedrecovery
Discoveries
Reserves
Mature areas: Production decline and marginal discoveries
New areas: Risks and uncertainties may be high
• offshore ultra deep water• unconventional resources• use of new technology
Average Volume / Discovery
0
50
100
1969-78 1979-88 1989-98 1999-08
MS
m3
o.e
.
Averagevolume /discoveryMSm3 o.e.
NPD 2009NPD 2009
Increasing Need for Proper Decision Analyses
Technical Disciplines
Project Managers
Economic Analysts
Portfolio Management
Top Management
DECISIONS
DECISIONSITUATIONS
Basic Economics
•Drill exploration wells•Choose field development concepts•Choose drainage strategies•Rank and drill production wells •Buy/sell assets•Include/exclude projects from portfolio
Basic Probabilistics
Decision Theory
Quantifying Uncertainty
METHODOLO
GY
WO
RK PRO
CESSES
Geology, geophysicsproduction, drainagedrilling, facilities, timing
Monte Carlo simulationMean, Mode, P10, P50, P90Correlations
Systematic, unsystematic riskNPV, discount rateTax systems, price simulation
Decision parametersProject optimizationDecision treesPortfolio management
SO
FTWARE T
OO
LS
Decision Basis for Management
Decision Basis for Management
Decision analysis
Structure Problem
Structure Problem
Quantify Key Measures
Quantify Key Measures
Capture Uncertainties
Capture Uncertainties
Buy licence?
Sell? At which
price?
Develop discovery? Area Plan?
How?Drill exploration
well?
Negotiations -Licensees -Government
Strategy and
planning processes
LIF
EC
YC
LE
Production, EORRe-development projects
Project Execution
Concept Screening
ConceptOptimization
Exploration / Early
feasibilityLIF
EC
YC
LE
FE
ED
DECISION GATE 1 DG2 DG3 DG4
PDO
Facts• One existing platform• Exploration well, discovered gas with a thin oil column (>10 m)• Enough gas for development, but uncertain for oil development• Total of three discoveries and 3 prospects in the area
Discovery A
Export route A
Export route B
Prospect C
Field AWith oil rim
Discovery B
Prospect B
Prospect A
Decision Analyses - Project Examples
Area Development
& Concept Selection
Facts• 3 exploration wells• Gas-condensate + Oil leg• 3 development scenarios
Well CWell AA’A
Oil Leg ?
Well B
Field B Field A
?
• Produce oil leg?• Additional appraisal well? • Drainage strategy?
Decision Analyses - Project Examples
?
Facts• Oil + Associated gas• 2 segments, one proven• 6 development scenarios
Decision Analyses - Project Examples
Tie-in to A
Tie-in to B
FPSO1
FPSO2
FPSO3
FPSO4
2010
Differences in:Production start dateBuild-upCAPEX / OPEXLease / TariffsLiquid CapacityContract Period
Which option to choose given the uncertainty in reserves and productivity
20122014
Pro
babi
lity
NPV (10^6 USD)
Concept 1, 2, 3, 4, 5
Concept 1
2
34
5
Highest NPV, but also largest uncertainty
Decision Analyses - Methodology
Success criteria
CONSISTENCY
DECISION-MAKING PROCESS DG1 DG2 DG3 DG4 DG5
• Decision tools • Integrated work approach• Methodology
=> Need all!
DATA DECISIONSANALYSES
Method x
DECISION-MAKING PROCESS
Method y
Analysis 1
Analysis 2
Analysis 3
Analysis 4
Analysis 6
Method z
EXPERTS PROJECTS
CONSISTENCY
Tools, Work Approach and Methodology
PORTFOLIO
Economic Parameters
Semi-Deterministic work approachSub-Surface, Production, Drilling Parameters
Decision?
CAPEX / OPEX and Schedule
SENSITIVITES
CAPEX, OPEX and Schedule
Economic Uncertainties
Integrated and Stochastic work approach
Sub-Surface ProductionDrilling
MONTE-CARLO
SIMULATION
UNCERTAINTIES
AND RISKS
Portfolio risk
Portfolio effects on risk
Systematicrisk
Unsystematicrisk
Size of portfolio
Relevantrisk
Portfolio x
Unsystematic riskSystematic risk
Can be reduced in a portfolio of assets through diversification.
Exploration risks, reserves,recovery, production, drilling and operations.
Cannot be reduced by diversification.
Price, currency, inflation, material cost.
oil
gas
oil
gas
oil
gas
oil
gas
Nr. & type of production/Injection wells
ProducingReserves
Production & Transport Facilities
Process capacity
Production profiles
CAPEX schedule
CAPEX
Well CAPEX & OPEX
Well CAPEX schedule
Process & Transport EPCI timeDrill rate
OPEX
Well uptime
Productionbuild up
Well/Process Capacities
Processuptime
Oil priceTariffs Revenue,oil & gas
CO2 fee
Gas price
Tax
Market considerations
Inflation &Discount
rate
Project cash flow
Economic indicators:
EMV,NPV,IRR, etc.
Prospect(s)
Oil/gas priceforecast
CAPEX & OPEX Market prognosis
Discovery?
Field development planning
Provide clear insight into complex projects
Res
ult
sCapturing the Uncertainties
Capacity ConstraintsFacilities & Wells, Schedule
Oil and Gas Reserves / Resources
Production Profiles
CAPEX OPEX Tariff
P&AAbandonment
Cut off
Cash flow
Rock & Fluid Characteristics
Rock VolumeParameters
RecoveryFactor
Revenue
Fiscal Regime
Probability Plots
Decision Trees
Summary Tables
Tornado Plots
Time Plots
NPV
Cash Flow
PR
OB
AB
ILIT
Y
RESERVES
PR
OD
UC
TIO
N
TIME
Prod.start
Project descriptionResponsibilitiesChange Records
Model initialisationSystem set-up
Exploration risks
Reserves calculationsMay include different:-Geological scenarios-Seismic interpretations-Several sediment.models etc.
Production profilesProduction constraintsAvailable capacityProfile preview
Economics input(Oil price, gas price, discount rate, fiscal regime)
Run simulation
Inspect resultsComparisonsExport to STEA
Generate reports
New / Open / CloseSave / Save As / Exit
Drilling cost and timingRisk factors and cost implications
CAPEX / OPEXPhasingTransportation and tariffsLogistics and insurance
Separate analyses of field projects, concepts and sensitivites
Analysis A Analysis B Analysis C Analysis D Analysis E
Integrated Field Development Model
Integrated Field Development Model
HIGHESTEMV
E EE’
Compare and rank
Optimize andupdate
H
GF
BC
DE
A
CONCEPTS
Analyses
Optimum path basis for decisions
Compare and rank
BACK-UP SLIDES
Deterministic vs. probabilistic approach
How can input risk and uncertainty be quantified?
DETERMINISTIC PROBABILISTIC
• Full range of possible outcomes
• True expected NPV
• True P90
• True P10
• Correct comparison and ranking of options
PARAMETER 1 ’high’ ’base’ ’low’PARAMETER 2 ’high’ ’base’ ’low’PARAMETER 3 ’high’ ’base’ ’low’PARAMETER 4 ’high’ ’base’ ’low’PARAMETER 5 ’high’ ’base’ ’low’
• Three discrete outcomes
• Base Case Expected for the project
• High case and low case are extremely unlikely to occur
PARAMETER 1 DistributionPARAMETER 2 DistributionPARAMETER 3 DistributionPARAMETER 4 DistributionPARAMETER 5 Distribution
SimulationBase caseHigh case
Low case
0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
-1000 -500 0 500 1000 1500 2000 2500 3000
NPV (10^6 USD)
Why use "Mean" for decision-making ?
PRO: The mean:• Performs right "in the long run"
– Decisions based on the mean
has the lowest expected error• Caters for occasional large
surprises• Is additive across reservoirs,
fields and portfolios• Maximises the value of the portfolio
The mean is most companies’ preferred basis for decisions !
CON: The mean:
• Is possibly more complicated tocomprehend and explain
• May give "infeasible" values
– Mean number of eyes of a dice is 3.5
– Sum of 100 dice: Makes sense
Statistical Measures
Mode
P50Mean
Mean The same as expected value. Arithmetic average of all the values in the distribution. The preferred decision parameter.
Mode Most likely value. The peak of the frequency distribution. Base case?
P50 Equal probability to have a higher or lower value than the P50 value. Often referred to as the Median.
PR
OB
AB
ILIT
Y
DETERMINISTIC BASE
STOCHASTIC MEAN
DRILLING TIMEPER WELL
n EQUAL WELLS
DETERMINISTIC BASE
STOCHASTIC MEAN
P90
P10
# WELLS
TIM
E
n
Deterministic base: Underestimates drilling costOverestimates # wells drilled per yearOverestimates production first years
Drilling campaign example
Courtesy of IPRES
PRODUCTIONDEV.COST
RESERVESDRILLING
GRV
Ø
N/G
Rc
Sw
Bo
NEX
T T
AR
GET
Probabilistic approach SIMULATION
Presents full range of possible outcomes
Key factors contributing to overall uncertainty
SIMULATION
Presents full range of possible outcomes
Key factors contributing to overall uncertainty
Example Contact Uncertainties - Cases
2731
2577
26472625
2688
2800
Non-communication
2731
2577
26472625
Communication
OPTIMISTICPESSIMISTIC EXPECTED CASE???
Monte Carlo - Principle
Probability ofGas-Cap
GRV N/G Ø Sw Bg RfRandom Number Generator
Probability forCommunication
Fault location adjustment
Depth conversion adjustment
GOC OWC
Development scenarios
(1) Pure depletion
– Long curved horizontal producer
(2) Water injection
– Short horizontal producer
– Vertical injector
(3) Gas injection
– Long horizontal producer
– Vertical gas injector
(4) WAG injection
– Short horizontal producer
– WAG injector
Reserves
P