strategic engineering nov 2006 v2strategic.mit.edu/docs/designunderuncertaintyv2.pdfwhat is...

© Olivier de Weck, November 2006

Strategic EngineeringStrategic Engineering

Olivier L. de Weck, Ph.D.

[email protected]

Associate Professor of Aeronautics & Astronautics and Engineering Systems

Designing Systems for an Uncertain FutureDesigning Systems for an Uncertain Future

November 7, 2006


Strategic Engineering Topics

Three main research areas:

System Design under Uncertainty

Product Platforms and Commonality

Interplanetary Space Logistics

Acknowledgments

Dr. Afreen Siddiqi, Matt Silver, Monica Giffin, Gergana Bounova

today’s topic


Outline

Motivation

Personal Experience (F/A-18 Program 1991-1997)

Framing the Research

Meta-Controls Framework

(Potential) Unifying Methods

Non-Homogenous Markov Chains

Time-Expanded Decision Networks

Ongoing and Future Work


What is Strategic Engineering?

Strategic Engineering is the process of architecting and designing complex systems and products in a way that deliberately accounts for future uncertainty and context in order to avoid lock-in and maximize lifecycle value.

The extension of strategic thinking to the design and operation of complex engineering systems

Warfare [Sun Tzu 500 A.D., Carl von Clausewitz (1780-1831)]

Business [Michael Porter 1979, Arnoldo Hax 1996, others]

Engineering [now]


Why is it important?Many large-scale complex systems suffer from “lock-in”,the inability to change/switch to a better configuration (or technology) despite superior solutions being known

Nuclear reactor technology [Cowan 1990]Economic aspects of lock-in [Arthur 1989, Arrow 2000]Impediment to technological evolution [Utterback 1974+]Political economic context [Nelson & Winter 1982]Impact on innovation [Henderson 1990; Christenson 1997]Causes are still being debated [Liebowitz 1985]Network externalities are important [Katz 1997, Witt 1997]Trade-off between operating slightly inefficient fielded technologies and developing new technologies [Sarsfield 2001]Political and organizational inertia [Puffert 2003]Recent AA/TPP S.M. Thesis on lock-in [Silver 2005]*

* incoming ESD PhD candidate, Jan 2007


Examples and ConsequencesCommunications Satellite Constellations [Chaize et al. 2003]

Iridium Bankruptcy [1999, $5B], Globalstar similar fateOversized System based on optimistic market forecasts, could not easily adapt capacity, service, footprint …We’ve studied previously how to deploy them in stages

Automotive Platforms [Suh et al. 2005]

General Motors, e.g. Epsilon Platform [2003-2013+] Challenge in adapting platform to changing requirements over 10-15 year life: stretching chassis, incorporate new powertrain technologies, rigid tooling, too many constraints O($10B) commitments for design, factory layout, tooling,…

NASA Launch Vehicles [Silver et al. 2005]

NASA Space Shuttle Program [1972-present]Original traffic model called for ~ 50 flights/yearActual flight rate much smaller, technical issues, long turnaround times, high fixed cost, ~$4 billion/year

gasoline

Iridium780 km

hybrid

STS


Key Concept: Switching Cost & Risk

Switching Cost summarizes the difficulty of changing a system’s configuration or state after it has been designed or fielded for operational use

Switching time

Switching risk


Personal History: F/A-18 ExperienceU.S. Navy Version C/D

fighter and attackaircraft carrier based3000 flight hours90 min average sortiemax 7.5g positive

(1987) Swiss Versioninterceptorland based5000 flight hours40 min average sortiemax 9.0g positivedifferent avionics

(1993)

“Redesign”

(Switch)Approach

Apply new operational usage spectrum to existing configurationFind those locations in the system which do not complyApply selective, prioritized local redesign one-element-at-a-timeExample: Center barrel, wing-carry-though bulkheads Al Ti


F/A-18 Center Barrel SectionY488

Y470.5Y453

WingAttachment

74A324001


F/A-18 Complex System ChangeF/A-18 System Level Drawing

OriginalChangeFuselage

Stiffened

Manufacturing Processes Changed

Flight ControlSoftware Changed

Gross Takeoff Weight

Increased

Center of Gravity Shifted


F/A-18 Lessons LearnedChanges increased cost per aircraft by O(~$10M)

encountered “surprises” along the way

Changing a system after its initial design isoften required to accommodate new requirementsexpensive, and time-consuming if change was not anticipated in the original design

Change propagationsome changes are local and remain localother changes start local, but propagate through the system in complex, unanticipated ways Switching costs include: engineering redesign cost, change in materials, manufacturing changes, change in operational costs, others …


Key Research Questions

What causes lock-in and how can it be avoided?

What are important exogenous uncertainties?

how can they be classified, modeled, considered during design, mitigated (=risks), or taken advantage of (=opportunities)?

What are components of switching costs?

How can switching costs be lowered?


Meta-Controls View

UncertaintyModeling

DecisionModeling

System ChangeModeling


Uncertainty Modeling

Domains:Operational Environment

- terrain, weather,…Market (Demand)

- customers, competition

Economic- interest rates, prices

Technology- disruptive

Regulatory- e.g. CAFÉ,

Political/Policy- Funding

TechniquesDiffusion Models

- GBMLattice Models

- binomial, trinomial…Scenario PlanningDelphi [RAND 1959]Times Series Forecasting

0 5 10 150.40.60.81

1.21.41.6

x 105

Time [years]

Dem

and

[Nus

ers]

Geometric Brownian Motion Model

GBM model, Δt = 1 month, Do = 50,000, μ = 8% p.a., σ = 40% p.a. – 3 scenarios are shown


Decision Modeling

When to switch to a new system/configuration?

Real Options “in” Projects [de Neufville, others…]MAUA (Multi-Attribute Utility Analysis)Staged Development and DeploymentSpiral Approaches/Rapid PrototypingOptimization: Stochastic Programming, Dynamic Programming, Multiobjective, …Controls-inspired Approaches


System Change ModelingTraditional Systems Engineering Methods (QFD, DSM,…)

Change Propagation Analysis [Clarkson, Eckert 2004]

System Architecting Principles, “Illities”Modularity, Flexibility, Scalability, Reconfigurability,…

Product/System Platforms [Meyer 1996, Simpson 2001 ..]

… all these attempt to address part of the problem, when do these methods apply, is there a unifying framework …?


Is there a unifying framework?

Maybe

Non-Homogenous Markov Chains

Time-Expanded Decision Networks


Markov Modeling - I

π n( )= π 0( )Φ 0,n( )= π 0( )Pn

In the most general case, for k operational states thereare k(k-1) reconfiguration states

π i =1

1− pi− i

p j− jij=1

k

∑ π j , j ≠ i

For n→∞:

π jj=1

k

∑ +p j− ji

1− p ji− jij=1

k

∑ π ji=1

k

∑ =1, i ≠ j

In the Time Homogeneous Case:

Operational State Reconfiguration State

P n( )= P ∀n

Ref: Afreen Siddiqi, PhD Thesis, 2006


Markov Modeling - II

( ) ( ) ( ) ( ), 1 1 , 1m n P m P m P n m nΦ = + − ≤ −L

For many reconfigurable systems, the state transition probabilities can be assumed to be conditioned on some external time varying process:

In the Non - Homogeneous Case:

pi− j n( )= f u n( ),i, j( )i* = argmaxJ u,S( )pm−mj > pm−mk, J j > Jk > Jm

State selection can be based on how sharply the performances differ

Path of PSV across varying soil conditions

Drawbar pull as PSV travels over varying terrainOpportunity stuck on Mars (NASA/JPL)


Application of NHMC – Meta-Controls

Emerging Technologies supporting Reconfigurability

‘Skate board’ chassisDrive-by-wireReconfigurable wheelsSMART carsFPGAs

Siddiqi A., de Weck O., Iagnemma K., “Reconfigurability in Planetary Surface Vehicles: Modeling Approaches and Case Study”, Journal of the British Interplanetary Society (JBIS), 59, 2006


Time-Expanded Decision NetworksWe developed concept of time expanded decision networks, to formalize effect of lock-in and identify opportunities for flexibility

3.) Create a Time Expanded Network based on the Static Network

5) Modify System Configurations to exploit easier Switching

4.) Create Operational Scenarios, Evaluate Optimal Paths

1.) Design Set of feasible System Configurations

2.) Quantify Switching Costs and Create a Static Network

Silver M.R., de Weck O.R., “Time Expanded Decision Networks: A Framework for Designing Evolvable Complex Systems , Systems Engineering, SE-061003 (under review)

exit with optimal initialconfiguration and switching embedded

( , )SWC C B

A

B

C ( , )SWC A C

( , )SWC C A

( , )SWC B A

( , )SWC A B

( , )SWC B C

0=SWC 0=SWC

0=SWCStep 1Step 2


Time-expanded Decision Networks (TDN)Expand static network in timeModel operations arcs and switching arcsChance Nodes and Decision NodesCost Elements identified

( )DC A

1( )F vC C D+

T1 T2 T3

1 2

7 8

13 14

3 4

9 10

15 16

5 6

11 12

17 18

S-19 Z-20

A

B

C

( , )SWC A B

( , )SWC A C

2( )F vC C D+ 3( )F vC C D+

( , )SWC A B

( , )SWC A C

Beginningof Lifecycle

Endof Lifecycle

T time periods

( )DC A

1( )F vC C D+

T1 T2 T3

1 2

7 8

13 14

3 4

9 10

15 16

5 6

11 12

17 18

S-19 Z-20

A

B

C

( , )SWC A B

( , )SWC A C

2( )F vC C D+ 3( )F vC C D+

( , )SWC A B

( , )SWC A C

Beginningof Lifecycle

Endof Lifecycle

T time periods

( )1

( )( , , )

1

T Fi Vi jLCi Di j

j

C C DC D T r C

r=

+= +

+∑

Design and Ops Costs

Switching Costs( ) [ ]( , , ) ( , ) ( , )SW O D RC A B N c N C B A C A B= × Δ + Δ

Step 3


Scenarios and Path Optimization

Time Period Scenario T1 T2 T3 T4 Scenario Description

1 1 0 0 1 0 0 0 0

Low Moon; No Mars

5 5 0 0 2 0 0 0 0

High Moon; No Mars

1 1 0 0 3 0 0 1 1

Low Moon; Low Mars

3 3 0 0 4 0 0 1 1

High Moon; Low Mars

2 1 0 0 5 0 0 1 2

High-Low Moon; Low-High Mars

5 5 5 0 6 0 0 5 5

High Moon; Hi Mars

3 2 1 7 0 1 2 3

Fade from Moon to Mars

1 1 0 0 8 0 0 3 3

Low Moon; High Mars

0 0 0 0 9 0 0 1 1

No Moon; Low Mars

0 0 0 0 10 0 0 5 5

No Moon; High Mars

Step 4

Major advantage over traditional decision trees !

Step 5Evaluate optimal paths through TDN:

-acyclic network-topological ordering-reaching algorithm (find shortest path)

Method Scaling

2 2N C T= × × +# of nodes

# of paths

optimal path solvable in

# Tpaths C=

2( 1) ( 2)M T C C T= − × + × +

# of arcs


Application of TDN to Launch VehiclesHow to choose launch vehicle configurations for space exploration when future traffic model is uncertainDemand at the campaign level vehicle demandEffort started under NASA funded Draper/MIT CER effort 2004-2005, continued development after that

4 initial configurations consideredPlot

Shortest Path; Discounted LCCOptimization

Discounted TEN

Time Expanded Network

Yearly Variable Cost/VehicleLV Cost

LV properties: Name, Capacity,

FRC, Shroud SizeLV Define

Schedule (IMLEO/yr)Traffic Model

Switching Cost Matrix; DDT&E

Cost; Fixed Cost, Rec. Cost;

Time Horizon Moon/Mars IMLEO Mission Schedule

Manual Input (Excel)

Step 1

Step 2

Step 3

Step 4

Step 5Plot

Shortest Path; Discounted LCCOptimization

Discounted TEN

Time Expanded Network

Yearly Variable Cost/VehicleLV Cost

LV properties: Name, Capacity,

FRC, Shroud SizeLV Define

Schedule (IMLEO/yr)Traffic Model

Switching Cost Matrix; DDT&E

Cost; Fixed Cost, Rec. Cost;

Time Horizon Moon/Mars IMLEO Mission Schedule

Manual Input (Excel)

Step 1

Step 2

Step 3

Step 4

Step 5

Model Development


TDN Launch Case Study - ResultsInitially: only two launch vehicles are ever used

1:SDV-A(80mt) = large, 2:EELV+(62 mt)=smallNo switching during lifecycle observed (= “lock-in”)Idea: gradually reduce switching cost to:

see if/when switching will occurwhat switches are selected most often?how valuable is it?

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T-34

FR+VR(D )SC(A1,A2)

DEV

(A1)

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T-34

FR+VR(D )SC(A1,A2)

DEV

(A1)

0460088486718from 4

8900088486718from 3

8900460000from 2

8900460022850from 1

to 4to 3to 2to 1Switching

Costs

Switching cost matrix

SDV-A (80 mt)

EELV+ (62 mt)


TDN Launch Case Study – Results (cont.)T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S- 33 T-34

)

1

2

3

4

SwitchingCost

CSW(3,1)

$4B

$1B

$2B

$3B

Scenario 8 Scenario 10 Scenario 7 Scenario 5T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T- 34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S-33 T -34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S- 33 T-34

)

1

2

3

4

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S- 33 T-34

T1 T2 T3 T4

1 2

9 10

17 18

25 26

3 4

11 12

19 20

27 28

5 6

13 14

21 22

29 30

7 8

15 16

23 24

31 32

S- 33 T-34

)

1

2

3

4

SwitchingCost

CSW(3,1)

$4B

$1B

$2B

$3B

Scenario 8 Scenario 10 Scenario 7 Scenario 5

As switching cost is reduced switching becomes more frequentScenario 8 most interestingMax savings $0.6B/$2.8BGuideline for embedding flexibility in 3:EELV+ and how much it is worth

Life Cycle Cost - Average and Sc. 08

15 16 17 18 19 20 21 22

SC_1000 SC_2000 SC_3000 SC_4000 SC_5000 SC_Normal

Switching Cost EELV(3) – SDV-A (1)

LCC

($B

illio

n)

Average

Scenario 08

redu

ce s

witc

hing

cos

t


Real Options and TDN(Real) Options are “simply” switching cost reducersCan model calls or puts as separate configurationsNeed to find optimal exercise rules, since TDN finds shortest paths assuming perfect knowledge of futureOptimal balance between upfront cost, and later switch cost

cash in

buy stockhold stock

buy call hold call

buy otherinvestment hold

invest-ment

cash outexercise

callsell stock

Expiration

don’texercise

sell otherinvestment

European Call


Degree of OutcomeUncertainty

RelativeSwitchingCosts

wirelesssensor

networks

highwayinfrastructure

(some) consumerproducts

communicationsatellites

automotiveplatforms

System Landscape

water supplysystems

commercialaircraft

ΔC/LCC

σNPV

E[NPV]

volatileflexible

volatileinflexible

stableinflexible

stableflexible

“Lock-in” Quadrant

embedflexibility

facilitateswitching

designrobustly

avoidswitching


Emerging PrinciplesFlexibility is a relative, not absolute system property.

There are two types of flexibility: operational (inner loop), configurational (outer loop).

Flexibility only makes sense in the context of specified uncertainty, which might lead to specific classes of changes in the future. General flexibility does not exist.

Fully reconfigurable systems are those systems that exhibit the highest degree of flexibility where the switching costs between a finite set of configurations (states) has been reduced to nearly zero.

Given an uncertain future operating environment, there exists anoptimal degree of flexibility that will balance reduction in switching costs with upfront system design/build complexity and ops cost.

Real options are switching cost reducers.


Some Recent Relevant PublicationsSurvey Papers

de Neufville R., de Weck O., Frey D., Hastings D., Larson R., Simchi-Levi D., Oye K., Weigel A., Welsch A., et al, “Uncertainty Management for Engineering Systems Planning and Design”, Monograph, 1st Engineering Systems Symposium, M.I.T., March 29-31, 2004.

Haberfellner R., de Weck O.L., “Agile SYSTEMS ENGINEERING versus engineering AGILE SYSTEMS”, INCOSE 2005 - Systems Engineering Symposium, Rochester, NY, July 10-15, 2005

Research Papersde Weck, O.L., de Neufville R. and Chaize M., “Staged Deployment of Communications Satellite Constellations in Low Earth Orbit”, Journal of Aerospace Computing, Information, and Communication, 1 (3), 119-136, March 2004

de Weck O.L., Suh E.S., “Flexible Product Platforms: Framework and Case Study”, DETC2006-99163, Proceedings of IDETC/CIE 2006 ASME 2006 International Design Engineering Technical Conferences, September 10-13, 2006, Philadelphia, Pennsylvania USA, submitted to Research in Engineering Design (in revision)

Siddiqi A., de Weck O., Iagnemma K., “Reconfigurability in Planetary Surface Vehicles: Modeling Approaches and Case Study”, Journal of the British Interplanetary Society (JBIS), 59, 2006

Hauser D., de Weck O.L., “Flexible Parts Manufacturing Systems: Framework and Case Study”, Journal of Intelligent Manufacturing, 2006 (accepted, to appear mid-2007)

Silver M., and de Weck O., “Time-Expanded Decision Network Methodology for Designing Evolvable Systems”, AIAA-2006-6964, 11th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Portsmouth, Virginia, 6 - 8 Sep 2006, submitted to Systems Engineering


Thesis ResearchMonica Giffin (SDM student, Raytheon), supported by Gergana Bounova (AA PhD candidate, networks)

Analyze Change Propagation Processes on an Actual ProgramComplex technical system at RaytheonChange and Configuration Management: regulate changes in systemWhat happens when one change unexpectedly causes another?

Database of 41,000 Requested Changes:~500 individuals named in requests~9 years project lifeDetailed design → implementation, test & sell-offChanges requested involved hardware, software, documentation

ApproachBuild underlying system mapMine database in an automated fashionCreate change propagation networks and compare to system mapFind change propagation patterns, multipliers, opportunities for flexibility, better architecture, better change management processes…?

Ongoing Work: Change Propagation


Change Requests Written per Month

0

300

600

900

1200

1500

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65 69 73 77 81 85 89 93

Month

Num

ber W

ritte

n

Change Request Generation

[Eckert, Clarkson 2004]

Discovered new changepattern: “inverted ripple”

component design

subsystem design

systemintegrationand test

bug fixes

major milestonesor managementchanges


Network plot of largest change network in the dataset, with 2579 associated change requests.Created by Gergana Bounova

Data from Monica Giffin (SDM)

Change Propagation Network


Future Work

Uncertainty Characterization (jump-diffusion = continuous change + discrete events superimposed)

Demonstrate benefits and challenges of reconfigurable systems [NSF Proposal ARES-CI, teaming with NU, Penn State, CMU]

Gather more data on real world large scale systems [e.g. MIT-Portugal program]

Change propagation in complex systems (Raytheon, others…)

Empirical/Field work on system evolution over time (a la Whitney Subway study)

More information on web: http://strategic.mit.edu

strategic engineering nov 2006 v2strategic.mit.edu/docs/designunderuncertaintyv2.pdfwhat is...

Documents