Transcript Production Functions - Massachusetts Institute of Technology

Decision Analysis Cases
• Object of this Presentation
• Context for this discussion
• Overview of Cases
• Haneda Airport (Tokyo, Japan) in
Detail
• Overall Conclusions
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 1 of 39
Object of Presentation

Overall: to wrap up first half of the course

Detail: how technical elements fit together








Recognition of Uncertainty
Bayesian Updating
Modeling Approach
Economies of Scale
Deferral of Costs
Decision Analysis
Multiple Criteria
Illustration by Examples
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 2 of 39
Context of Discussion
• Decision Analysis is Simplest Approach to
identifying Valuable Flexibility
• Includes all necessary elements – provides
basis for investigating flexibility and
increasing value over standard design method
• Review appropriate
• 2nd half: advanced concepts and techniques
• Lattice Analysis
• Simulation
• Financial Analysis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 3 of 39
Sample Cases
• Abisoye Babajide “Real Options Analysis as a
Decision Tool in Oil Field Development,” Master in
SDM, MIT, 2007
• Katherine Dykes et al “Real Options for a Wind Farm
in Wapakoneta, Ohio: Incorporating Uncertainty into
Economic Feasibility Studies for Community Wind,”
MIT 2008
• Masahiro Kimura “Strategic Planning for Newcomer
in Silicon Wafer Industry,” Master in MOT, MIT, 1995
• Dai Ohama “Using Design Flexibility and Real
Options to Reduce Risk in Private Finance Initiatives:
the Case of Japan,” Master in TPP, MIT, 2008
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 4 of 39
Babajide: Gulf Oil
• Summary Perspective
Babajide: Gulf Oil Platform
Uncertainty Recognition
Bayesian Updating
Modeling Approach
Economies of Scale
Deferral of Costs
Decision Analysis
Multiple Criteria
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Derived from actual data
From first round of test wells
High level economic models
Yes, over 2 fields
Yes; vertical risers later, if at all
Fields 1 or 2, Both, and w/ Flexibility
Yes, VARG diagram and Capex
Richard de Neufville
©
Decision Analysis Cases Slide 5 of 39
The Gulf of Mexico Case





Oil recovery in Gulf of Mexico
“shallow water” ~ 100 m
Real case: numbers disguised for
company confidentiality
Two fields: “Sample” and “Rother”
Reservoirs ‘fractured’ – that is, they
consist of several smaller pools that each
require wells to suck out oil
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 6 of 39
0.60
0.70
0.50
0.60
0.50
0.40
Probability
Probability
Probability Mass Functions
(PMF)
0.30
0.20
0.40
0.30
0.20
0.10
0.10
0.00
0.00
80
150
200
100
150
220
Expected Ultimate Recov ery (MMBO)
Expected Ultimate Recovery (MMBO)
Sample Field PMF
Rother Field PMF
Note: “Most likely” scenarios are 150 and 100
Source: Babajide et al paper
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 7 of 39
Combined PMF
0 .4 0
0 .3 5
Probability
0 .3 0
0 .2 5
0 .2 0
0 .1 5
0 .1 0
0 .0 5
0 .0 0
180
230
250
300
350
370
420
Expect ed Ultimat e Recovery (MMBO)
C ombined Sample and Rother O il Res erve P M F
Source: Babajide et al paper
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 8 of 39
Flexible Combined Fields
Source:
Babajide
et al
paper
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 9 of 39
VARGs: Inflexible vs. Flexible
1 .0 0
Cumulative Probability
0 .9 0
0 .8 0
0 .7 0
0 .6 0
0 .5 0
0 .4 0
0 .3 0
0 .2 0
0 .1 0
0 .0 0
1 5 ,0 0 0 1 7 ,0 0 0 1 9 ,0 0 0 2 1 ,0 0 0 2 3 ,0 0 0 2 5 ,0 0 0 2 7 ,0 0 0 2 9 ,0 0 0 3 1 ,0 0 0 3 3 ,0 0 0
NPV ($, million)
I nflexible
Flexible
E N P V - I nflex.
E N P V - Flex.
Source: Babajide et al paper
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 10 of 39
Comparison of Economic Metrics
($, million)
ENPV
Minimum NPV
Maximum NPV
Initial CAPEX
NPV/CAPEX
Designs
Inflexible
Flexible
22,935
24,622
15,868
15,839
29,761
32,516
956
990
24
25
W hich better?
Flexible.
Same
Flexible.
Inflexible.
Flexible
No alternative best on all measures.
Source: Babajide et al paper
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 11 of 39
Benefit-Cost of Flexibility
Cost of flexibility ($, M)
Average Value of flexibility ($, M)
Cost-Benefit ratio
34
1,712
50
Note: Value of Flexibility very high (in this case)
Source: Babajide et al paper
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 12 of 39
Take-Aways: Gulf of Mexico Oil
• Economies of Scale large, so larger
(two field) development chosen
• Flexibility adds great value
• Benefit-cost ratio can be very large!
• Flexibility changes tails significantly
• Improves design over many criteria
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 13 of 39
Dykes: Wind Farm
• Summary Perspective
Dykes: Wind Farm
Uncertainty Recognition
Bayesian Updating
Modeling Approach
Economies of Scale
Deferral of Costs
Decision Analysis
Multiple Criteria
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Growth and Possible Step Change
Observation over time
Economic
Yes, for production
Yes, possible expansion later
Two Stage: immediate and later
not used
Richard de Neufville
©
Decision Analysis Cases Slide 14 of 39
The Wind Farm Case


Potential Wind Farm in Mid-West USA
Two development choices:



Build Big, Use Economies of Scale
Build Small, See how things develop
Two Uncertainties


Growth in demand
Regulatory Regime
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 15 of 39
Build Now or Delay?
Source:
Dykes
Wind
Farm
Paper
Large
Wind Farm?
EV: $9,227,184
EV: $8,102,250
40 years
2/9
1/9
none
$444795
10 years
1/3
287 bmt
$5341648
1/3
203 bmt
167 bmt
$11278992
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
1/2
Lax regulation
1/2
Strict regulation
$11306767
Richard de Neufville
©
Decision Analysis Cases Slide 16 of 39
Branch if “Strict Regulation”
Source:
Dykes
Wind
Farm
Paper
3 to 20 MW?
EV:
$10099523
EV: $(1182246)
40 years
40 years
1/18
none
287 bmt
2/9
4/9
203 bmt
5/18
167 bmt
$(2554647) $(1801285) $(887847) $(883574)
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
1/18
none
287 bmt
$3952478
2/9
4/9
203 bmt
5/18
167 bmt
$7314565 $11418944 $11445825
Richard de Neufville
©
Decision Analysis Cases Slide 17 of 39
Branch if “Lax Regulation”
Source:
Dykes
Wind
Farm
Paper
3 to 20 MW?
EV: $8354845
EV: $(1570907)
40 years
3/18
none
$(2554647)
287 bmt
$(1801285)
4/9
2/9
203 bmt
40 years
3/18
167 bmt
$(887847) $(883574)
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
3/18
none
$3952478
287 bmt
4/9
2/9
203 bmt
$7314565 $11418944
3/18
167 bmt
$11445825
Richard de Neufville
©
Decision Analysis Cases Slide 18 of 39
VARG for Wind Farm
VARG for Wind Farm
Cumulative Probability
1.2
1
0.8
0.6
Immediate
Investment
0.4
Deferred
Investment
0.2
0
0
2
4
6
8
10
12
Value of Outcome (Millions)
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 19 of 39
Take-Aways: Wind Farm Case
• Economies of Scale exist, but not
large enough to compensate for value
of deferring investment
• Case illustrates value of deferral –
valuable even though choice is same!
• Flexibility changes tail significantly
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 20 of 39
Kimura: Production of Si Wafers
• Summary Perspective
Kimura: Production of Silicon Wafers (1995)
Uncertainty Recognition
Bayesian Updating
Modeling Approach
Economies of Scale
Deferral of Costs
Decision Analysis
Multiple Criteria
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Estimated
After 1st period
Technical cost model for production
Yes, in wafer size
Yes, possible expansion later
Two stage, possible expansion later
not used
Richard de Neufville
©
Decision Analysis Cases Slide 21 of 39
The Silicon Wafer Production Case
• Japanese company faced with issue:
should they make 4” or 8” silicon wafers
• 8 inch wafers have significant economies
of scale – if sales reach capacity of plant!
• Classic trade-off between




Economies of Scale
Advantages of Deferred Expenses
Risk tilts balance toward smaller plant
Context of Design of Production Process
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 22 of 39
Take-Aways: Si Wafer Production
• Economies of Scale exist, but not large
enough to compensate for value reducing
exposure to possible big losses



Analysis leads to Strategy
Build smaller, expand (or not) depending
on market development, technical success
in production
Note: Company focused on Econ. of Scale,
went for big design, lost heavily…
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 23 of 39
Ohama: Tokyo/Haneda Airport
• Summary Perspective
Ohama: Design of Airport Runway
Uncertainty Recognition
Bayesian Updating
Modeling Approach
Economies of Scale
Deferral of Costs
Decision Analysis
Multiple Criteria
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
In traffic growth trends
simulation of future levels
detailed technical analysis
not applicable
Yes, possible taxiway later
Two stage
Yes, VARG diagram and Capex
Richard de Neufville
©
Decision Analysis Cases Slide 24 of 39
The Tokyo/Haneda Case
• Case is about an expansion of runway for
Haneda Airport, in Tokyo Bay.
• Issue: should design have conventional
parallel taxiway?
• This question “never” asked in practice
• However, value of runway (= contribution
to capacity) depends on level and
direction of use of runway
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 25 of 39
Location of Tokyo/Haneda Airport
Japan
Tokyo
Chiba
Yokohama
Kyoto
Haneda Airport
Tokyo
Tokyo Bay
Osaka
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 26 of 39
New Runway for Tokyo/Haneda
Source: Ohama MIT thesis; Japan Ministry of Land, Infrastructure and Transport
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 27 of 39
Possibility for Flexible Design
Runway
(R/W)
North
South
Departure
Arrival
Parallel Taxiway
(PT/W)
In the currently designed
operation, these areas are
not necessary.
South Edge
(S/E)
Existing
Airport
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 28 of 39
Technical Details
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 29 of 39
Directional Issues
North Wind
Departure: B-12 ac/hr
D-28
Landing : A-28 ac/hr
C-12
Total : 40 ac/hr
South Wind
Departure: A-22 ac/hr
C-18
Landing: B-28 ac/hr
D-12
Total: 40 ac/hr
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 30 of 39
The Capacity Issue
• Runway Capacity is not a fixed amount
(when demand exceeds “capacity”, this
means traffic delays, not that aircraft will
not be able to land…
• New runway will be inconvenient, and
thus used for “overflow”




Relatively little use
Mostly for landings (to left, or South)
These operations can taxi on runway
This reason taxiway not immediately needed
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 31 of 39
3 Scenarios for Analysis
Base Case
Scenario 1
Scenario 2
Initial Investment
R/W, PT/W&S/E
N/A
N/A
Future Uncertainty
R/W, PT/W&S/E
Recognizing
N/A
R/W
Recognizing
Future Expansion
(PT/W & S/E)
Flexibility
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 32 of 39
Details of Flexible Design
Runway
(R/W)
Parallel Taxiway
(Reclamation)
827,625m2
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
South Edge
(Piled Elevated Platform)
45,120m2
Existing
Airport
Richard de Neufville
©
Decision Analysis Cases Slide 33 of 39
Deterministic Forecasts
100.00
Demand of Passengers (million)
90.00
80.00
Actual Demand
Projected Demand (Deterministic)
Projected Demand (Government)
70.00
60.00
50.00
40.00
30.00
20.00
10.00
0.00
1985
1990
1995
2000
2005 2010
Year
2015
2020
2025
2030
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 34 of 39
Example Simulated Demand
Demand of Passengers (million)
140.00
120.00
100.00
Actual Demand
Projected Demand (Deterministic)
Projected Demand (Government)
Demand Scenario (Randomized)
80.00
60.00
40.00
20.00
0.00
1985
1990
1995
2000
2005 2010
Year
2015
2020
2025
2030
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 35 of 39
VARG diagram for Runway
100%
90%
80%
Probability
70%
60%
50%
40%
VARG (Scenario 2)
ENPV (Scenario 2)
30%
VARG (Scenario 1)
ENPV (Scenario 1)
20%
10%
0%
400,000
500,000
600,000
700,000
800,000
900,000
1,000,000
1,100,000
Net Present Value (Million Yen)
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 36 of 39
Comparison across Criteria
Design
Criteria
Scenario 1
(No Flexibility)
Expected Net Present Value (Million)
Initial Cost (Million)
Maximum Value (NPV) (Million)
Minimum Value (NPV) (Million)
Flexibility Value (Million)
¥713,642
¥570,000
¥919,795
¥452,671
-
Scenario 2
(Flexible)
¥785,120
¥505,019
¥1,077,365
¥532,827
¥71,478
Comparison
Flexibility Better
Flexibility Better
Flexibility Better
Flexibility Better
-
Source: Ohama MIT Thesis
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 37 of 39
Take-Aways: Tokyo/Haneda
• Technological Understanding, coupled with
appreciation of uncertainties, leads to
opportunities for flexible development
• Improvements


About 10% over conventional, on average
Almost 20% in terms of minimum results
• Flexibility creates stochastically dominant
outcomes
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 38 of 39
Decision Analysis
Take-Aways



Decision Analysis approach Can be used
Effectively in many situations
It gives insight into relative importance of
trade-offs between Economies of Scale and
Savings by deferring Investments
Provides a reasonably transparent view of



effect of uncertainties …
way flexibility increases average project value
and reduces downside and increases upside
Engineering Systems Analysis for Design
Massachusetts Institute of Technology
Richard de Neufville
©
Decision Analysis Cases Slide 39 of 39