Strategic Engineering of Large Infrastructure Systems: Real Options, Staged Development, and the Role of Flexibility Jijun Lin Prof. Olivier de Weck, Prof. Richard de Neufville Dr. Afreen Siddiqi Dec. 4 th, 27 * This research work is sponsored through BP-MIT research program Outline Background about offshore oil/gas field development project Motivation and research questions Integrated Matlab model of offshore oil/gas project Applications of the integrated model Design optimization Reservoir and market uncertainty modeling and simulation Flexibility in oil/gas field development Simulation case study of flexible staged development Conclusion and Future work 2
Background Generic stages for oil/gas exploration and production project 2~3 years Exploration Appraisal Development Production Abandonment Update estimation More Exploration? Acceptable Uncertainty? Reservoir Type Reservoir? Facility constraints Facility: #, location, etc? Facility Operation: qmax? /change facility?? Abandon Reservoir? Sell the field to low cost operating company? Geographical location Reservoir identification (seismic) Local environment politic, economic conditions, etc Drill appraisal wells to learn more about reservoir and reduce uncertainty (reservoir volume, fluid properties, etc.) Based on reservoir properties, selects reservoir type, facility concept designs Reservoir and Facility Models capture link between physical dynamics and facility configuration Estimate the production profile, field development cost Select -> Define -> Execute Actively monitor and manage reservoir production Potential further field development Determines Abandonment Conditions Technical condition Economical condition 3 Background - cont. An illustration of offshore oil/gas field in deep water Some other types Mobile drilling unit of facility: Surface well head on Tension Leg Platform Production Facility Temporary crude storage facility Pipeline Subsea wells 4 2
Motivation Large-scale engineering systems, such as offshore oil gas platforms, are very complex in nature due to the interactions among multiple disciplines, such as reservoir engineering, facility designs, project economics, etc. Existing tools are very domain specific and a lot of integration effort is required from decision makers in the early field development stages. Furthermore, multi-billion dollars investment decisions on offshore infrastructure have to be made under a wide range of uncertainties such as reservoir, market conditions, and political environment uncertainties. Traditional practice of designing the optimal facility to the specifications potentially leads to undesired outcomes (e.g., opportunity loss, high risk). Therefore, there is a need for major oil companies to think more strategically in their offshore oil and gas development. 5 Research Questions The key research questions are: Modeling: How to model the essential interactions among multiple disciplines for strategic decision making in offshore field development? Flexibility: How can flexibility in design and operation improve projects expected value under uncertainties? Strategic Engineering Framework: In the context of offshore oil and gas projects, how to develop a strategic engineering framework for field development which is robust and adaptive to future environment? 6 3
Overview of Model Structure Conceptual structure Implementation structure Integrated Modeling and Simulation GUI Reservoir dynamics model Facilities model Economics model Five reservoir operation modes parafile.mat Standardization Flexibility reservoir.m q_rates q_capacities facilities.m Commonality Learning effect Bulk effect Configuration flexibility Operation flexibility Flexibility valuation Determine economics abandonment condition OPEX finance.m NPV.m CAPEX 7 Model Capabilities The integrated model intends to consider following the scenarios (with focus on B and D) number of facility D Staged Development Standardization Flexibility C A Tie back options Flexibility B number of reservoir # of Facilites # of Reservoirs 2 3 4 5 2 3 4 5 6 7 Staged development of a large reservoir Tie-back a small reservoir 3 To facility 5 as capacity becomes available 8 4
Multi-reservoir Multi-facility Integrated Model Different Simulation Options () Reservoir Dynamics Module Feedback (2) Facility Module Feedback (economic recovery rate) (3) Project Economics Model (4) Simulation Control 9 Reservoir dynamic model A simplified reservoir with water / gas / aquifer drive Aquifer support Reservoir Well rate q w, t STOOIP Water re-injection q w _ in Gas re-injection q g _ in Mixture of oil, gas, and water Sea water Water production q w Facility q t Gas Oil production production Five reservoir operation modes. Primary depletion 2. Water drive 3. Aquifer drive 4. Gas cap drive with gas injection 5. Combination drive q g q o 5
Parametric facility cost model Develop facility cost capacity parametric model using OGM software and Design of Experiment (DOE) method for Steel Pile Jacket substructure y r = Ax r + B A linear relationship seems to capture the facility cost - capacity model very well Prediction of fitted model (5 factors) Field total cost (CAPEX+decommissioning) USD: 8 7 6 5 4 3 2 Y Estimated cost by OGM Estimated cost by DOE model Compare predictions of facility model with OGM model outputs The average error for field development cost is less than 3% 2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9 2 Samples of random designs Typical flow rates and cash flow profile for oil/gas project Production and injection profiles, RF = 49.8% (total 22.75 years) Total NPV =.2427 Bn $ in 23 years Oil /gas /water production rates 8 6 4 2 8 6 4 total fluids production rate [mbd] oil production rate [mbd] gas production rate[mmscf/day] gas export rate[mmscf/day] water production rate [mbd] water inj. rate [mbd] aquifer influx [mbd] gas inj. rate [mmscf/day] cash flows.5.5 -.5 Gas Sale Oil Sale Tax Royalty OPEX CAPEX 2-5 5 2 25 Years of production Some outputs: 5 5 2 25 period CAPEX: equally spread out in project development phase Learning effect on CAPEX reduction for stage development : define learning factor internally (e.g., 9%) OPEX: fixed OPEX + variable OPEX 2 6
Applications of Integrated Model 3 Uncertainty and flexibility in oil field project Multi-stage strategy to take into account uncertainties Reservoir: OOIP Deterministic inputs: 5% OOIP Bespoke facilities design Fixed oil/gas market price + Bespoke Design + Oil / gas price Baseline NPV NPV distribution RU+MU Traditional Practice: Single number for NPV as decision Making criteria Reservoir: OOIP Reservoir: OOIP + Flexible, staged facilities +decision rules + + + Flexible facilities + decision rules + tie-back flexibility Oil / gas price Oil / gas price NPV distribution + RU + MU + flexible, staged facility NPV distribution + RU + MU + flexible facility + tie-back option RU: reservoir uncertainty MU: market uncertainty New Paradigm: Value-at-Risk-Gain Curve (VARG) Expected NPV Maximal Gain Maximal loss Reduced initial CAPEX Value of Flexibility 4 7
Reservoir uncertainty Estimation of mean for STOOIP This green line represents one instantiation of estimation of mean for STOOIP over time (many possible trajectories!) Initial estimation error of mean for STOOIP (critical for fixed design case) With flexibility, decision making is based on current estimation of STOOIP underlying STOOIP (unknown!) Start appraisal t Number of years for field development and operation Acceptable range to start development (execution) 5 Distribution of reservoir volume estimation Assume initial estimation of reservoir volume follows some distributions, e.g., 8 x -4 Normal distribution with negative tail cutoff Lognormal distribution (using in practice) Lognormal distribution p 7 6 5 4 sigma =.2462 sigma =.5 sigma = All curves have the same statistical mean: 24 Learning processes will narrow the distribution of STOOIP over time. 3 2 2 3 4 5 6 STOOIP (mmstb) The standard deviations for blue, green, and red curves are: [6 279 346] mmstb. 6 8
Market uncertainty Crude oil price constantly changes over time Field development strategy needs to adapt accordingly Some mathematical techniques are available to model price evolution (e.g., binomial tree model) Binomial tree model Binomial Tree Model, drift rate =3% per annum, volatility = % per annum 2 Historial Crude Oil Price 8 7 6 USD 6 5 4 3 Crude Oil Price ($) 4 2 8 6 2 949 952 955 958 96 964 967 97 973 976 Nomial Linear regression Drift rate = 2% per annum 979 982 985 988 99 994 997 2 23 26 Years Inflation adjusted(in Jan 27 dollar) Drift rate = % per annum 4 2 5 5 2 25 3 Years Initial oil price: $35 Δt μ =.2 σ =. Δt = 3 T = 3 max (per annum) (per annum) years years u = e σ d = u μδt e d p = u d p( scenario) k n = p ( p) k 7 Bespoke design Given P5 estimation of STOOIP, design a facility to optimize project NPV Bespoke designs for different STOOIP estimations (4, 8, 6 mmstb) Optimal Facility Size for Different STOOIP Estimates Project NPV (Bn $) 3.5 2.5.5.5 This figure shows that different STOOIP estimates can lead to different optimal facility sizes. -.5 2 3 4 5 -.5 Facility sizing factor: alpha 8 mmstb 4 mmstb 6 mmstb Single factor facility sizing: alpha* [2 5 5 2] mbd or mmscf/d for [Total fluid rate oil production rate gas production rate water production rate] 8 9
Bespoke designs + reservoir volume uncertainty Bespoke designs for three STOOIP estimates (8, 4, and 6 mmstb) Monte Carlo simulation (2 runs) for each bespoke design with OOIP uncertainty (normal distribution, mean 8, std 4, cutoff at 5) Cum. Prob. Value at Risk and Gain.9.8.7.6.5.4.3.2. -2. -... 2. 3. 4. 5. Net Present Value (Bn $) This figure shows that different bespoke designs will behave differently under reservoir uncertainty Bigger facility can take larger upside opportunity but will potentially lose big Smaller facility is a relative conservative strategy 8 4 6 9 Flexibility in oil/gas field development There are many possible ways to embed flexibility into oil/gas field development: Field level Postpone, abandon field development. new platforms, tie-back other reservoir fluids to platforms, etc. Platform level Capacity expansion (i.e., add extra water injection equipment) Operational level Well and production management (i.e., shut down wells, increase / decrease water injection volume) 2
Pre-determined vs. Flexible Staged Development Compare three field development strategies: ) Pre-determined staged development Three identical stages (33% capacity each) in year, 2 and 4 9% learning factor on CAPEX reduction 2) Flexible staged development Stage options: 33%, 5%, % capacity ~2 stages (33%, %, % + 5%, % + %) depending STOOIP estimates A decision rule has been coded in the program to determine when to add additional stage with how much capacity 3) One big monolithic facility development Single stage in year with % capacity 2 Decision rules for flexible staged development This is an example of the decision rule S (t ) S: estimation of STOOIP over time P: oil price over time* <2 mstb 2 mstb 33% capacity % capacity Initial stage No action S(t)<2 5% 2 S(t)<4 % 5% 5% % 4 S(t)<5 S(t) 5 4 S(t)<5 S(t) 5 the 2 nd stage *ing stage can be made only when P(t)>$35 22
Pre-determined three stages development Production and injection profiles, RF = 48.352% (total 2 years) Oil /gas /water production rates 5 4 3 2 5 5 2 Years of production Clone facilities total fluids production rate [mbd] oil production rate [mbd] gas production rate[mmscf/day] gas export rate[mmscf/day] w ater production rate [mbd] w ater inj. rate [mbd] aquifer influx [mbd] gas inj. rate [mmscf/day] Cash Flows 6 5 4 3 2 - -2 Total NPV = 3.6468 Bn $ in 2 years Gas Sale Oil Sale Tax Royalty OPEX CAPEX -3 st facility -4 2 nd facility 3 rd facility 5 5 2 25 Years Learning factor on CAPEX: 9% 23 stage development (one big monolithic facility) Production and injection profiles, RF = 48.2282% (total 2 years) Oil /gas /water production rates 5 4 3 2 total fluids production rate [mbd] oil production rate [mbd] gas production rate[mmscf/day] gas export rate[mmscf/day] w ater production rate [mbd] w ater inj. rate [mbd] aquifer influx [mbd] gas inj. rate [mmscf/day] 5 5 2 Years of production 5 years for drilling ramp up 9 wells! cash flows 6 5 4 3 2 Total NPV = 4.3884 Bn $ in 2 years Gas Sale Oil Sale Tax Royalty OPEX CAPEX - -2 Assume 3.25 years development time -3 5 5 2 25 period No learning benefit 24 2
Comparison results Value-at-Risk-Gain (VARG) Curve (with reservoir uncertainty) In this simulation case study, the flexible staged development strategy is better than the single stage big monolithic facility: Cummulative Probability.8.6.4.2-4 -2 2 4 6 8 2 4 Net Present Value [Billion $] Flexible staged deployment Pre-determined staged deployment stage big monolithic facility ENPV ENPV ENPV Increases ENPV by.79 Bn. Allows to increase capacity when it is needed, as a result, it can take significant amount of upside opportunity (6.7 Bn). Decreases maximal downside loss by.34 Bn. Decreases minimal initial CAPEX by.72 Bn. ENPV Min NPV Max NPV Minimal Initial Expected Initial CAPEX CAPEX Flexible staged development 7.87.5 3.65.4 3. Pre-determined staged development 5.28 -.82 6.5.4.4 stage big monolithic facility 6.8 -.29 6.94 3.3 3.3 Mean 45 mmstb STD 5 mmstb # of Runs 6 Learning factor for CAPEX: 9% Assume normal distritution for STOOIP %, 5%, 33.3% staging capacity for the flexible case Initial CAPEX is the CAPEX for the first stage without discounting 25 Key factors for Expected NPV Pareto Chart of The Main Effects for Expected NPV (alpha =.5) Market Uncert aint y Reservoir Uncertainty Discount Rate Timing of Staging Bulk Effect 2.262 The figure on the left shows the relative importance of several factors on project ENPV. The top three factors are: Market Uncertainty Reservoir Uncertainty Discount rate Learning Effect 2 3 4 5 Standardized Effect 6 7 8 A design of experiment is conducted to study the relative importance of these factors contributing to ENPV. The exact sequence or relative importance scale will depend on the various assumptions and range of levels for each factor. 26 3
Contributions and Future Work The main expected contributions are followings: Operationalize Strategic Engineering (system design for uncertainty environment, flexibility in design) in the context of oil/gas field development project. Demonstrate the integrated technical / economical modeling and simulation as an approach for Engineering Systems design. Future work Apply the models and approaches to a real offshore oil/gas field development. Propose a generic Strategic Engineering framework to be applicable to large-scale infrastructure systems 27 Back up slides 28 4
How does estimation of Recovery Factor (RF) change over time? Estimation of mean for RF This green line represents one instantiation of estimation of mean for RF over time (many possible trajectories!) e.g., discover new information about reservoir: fractured reservoir RF doesn t necessarily converge uniformly to a value! Start appraisal e.g., when oil price increase significantly, enhance recovery technique become economical, injection of miscible gas CO 2 to enhance oil recovery e.g, slightly increase of RF due to active reservoir well management t Number of years for field development and operation Acceptable range to start development (execution) 29 Decision rules for flexible staged development This is an example of the decision rule S (t ) S: estimation of STOOIP over time P: oil price over time* <2 mstb 2 mstb 33% capacity % capacity Initial stage No action S(t)<4 5% % 5% % 4 S(t)<5 S(t) 5 4 S(t)<5 S(t) 5 the 2 nd stage *ing stage can be made only when P(t)>$35 3 5