Architecting flexible projects and systems using portfolios of real options a decision support framework and two case studies

The resulting flexible solutions consist of an initial design and a portfolio of real options with their exercising conditions to adapt according to the manners the future unfolds - in c

DECISION SUPPORT FOR ARCHITECTING FLEXIBLE PROJECTS AND SYSTEMS: AN EVOLUTIONARY

FRAMEWORK AND TWO CASE STUDIES

by Stephen Xu ZHANG

(B Eng., Nanyang Technological University)

A THESIS SUBMITTED

TO THE DIVISION OF ENGINEERING AND TECHNOLOGY

MANAGEMENT FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

AT THE NATIONAL UNIVERSITY OF SINGAPORE

2010

ACKNOWLEDGEMENT

This dissertation as well as the research is not possible without the many people who

have helped me for years

First, the gratitude belongs to my thesis adviser, prof Vladan Babovic, who has provided inspiration, direction, time, effort, feedback, catalyst, and support continuously throughout the last four years He knows what I need better than I do and provides more guidance than

I could expect

Thanks also to prof Hang for setting up the division of Engineering and Technology

Management and for his version and consistent effort, which helps and steers me through the PhD

Thanks are also due to the people and organizations that have provided funding support for the research The project from Defense Science and Technology Agency (DSTA) in

Singapore provided me with an opportunity to make explicit and to develop at some length

a real options approach that has been central to much of my research Another project,

done in Singapore Delft Water Alliance (SDWA), permitted me to amend and expand the real options approach through applying it to additional field in a water supply system Not only

do I need to thank them for the generous provision of funding, but also the excellent

community for carrying out research projects

In the course of carrying out the MDP case study, thanks to Maarten Keijzer for the

application of the multiple-objective evolutionary algorithms into the real options model He worked patiently with me to set up and make sense of the model immediately after long

flights from the Netherland It went way beyond the call of duty Thanks also to my other

partner in the project, Joost Buurman It would be hard to imagine a more professional

partner

Thanks to Nguyen Tan Thai Hung at the National University of Singapore for the

development of the MATLAB codes for the water supply system case His patience and

perseverance in working with me was admirable, and the case study would have taken much longer, without working together with him

Thanks are also extended to Joost, Rao Raghuraj, Sarah von Helfenstein, Onion, Yabin, James, S K Ooi, Jianghang, Mark Fielding, and Mark Womersley for helping me in writing as well as many valuable conversations Without their help, this thesis would not have had any chance of being completed by the same time frame at the same quality

Finally, my gratitude goes out to my family Without their love and support, especially

my mom and dad, I would have never made it to a Ph.D It is unfortunate the thesis is

written in a language they know little about, yet it simply confirms again their unconditional love and support on whatever I choose to do

TABLE OF CONTENTS

ACKNOWLEDGEMENT III TABLE OF CONTENTS V SUMMARY IX LIST OF TABLES XI LIST OF FIGURES XIII LIST OF ABBREVIATIONS XVII

1 INTRODUCTION 1

1.1 M OTIVATION 1

1.1.1 Uncertainty and flexibility 1

1.1.2 Flexibility in the design of systems and projects: a casino example 3

1.1.3 Multiple threads of flexibility –their benefits and associated complexity 8

1.2 R ESEARCH OPPORTUNITIES 9

1.2.1 Research gap 9

1.2.2 Research question 10

1.3 R ESEARCH APPROACH 11

1.4 O RGANIZATION OF THE THESIS 15

2 LITERATURE REVIEW 17

2.1 I NTRODUCTION 17

2.2 D ESIGN 18

2.3 U NCERTAINTY -A KEY ISSUE IN PROJECT AND SYSTEM DESIGN : 20

2.3.1 Means to reduce uncertainties 22

2.3.2 Means to cope with uncertainties 24

2.4 F LEXIBILITY 27

2.4.1 Modularity 27

2.4.2 Balancing costs and benefits of flexibility 29

2.5 O PTIONS AND REAL OPTIONS – FLEXIBILITY FORMALIZED 30

2.5.1 Options 30

2.5.2 Real options 32

2.6 R EAL O PTIONS V ALUATION 35

2.6.1 Options Valuation 35

2.6.2 Real options valuation 40

2.7 D ECISION ANALYSIS AND REAL OPTIONS FRAMEWORKS 48

2.7.1 Decisions 48

2.7.2 Decision analysis and real options 49

2.7.3 Real options frameworks 51

2.8 R ESEARCH GAP ANALYSIS 54

2.8.1 Practical issues of real options related to system and projects design 54

2.8.2 How the current real options frameworks deals with those practical issues 56

2.8.3 The techniques used to develop the proposed framework 57

2.8.4 Evolutionary algorithm 57

2.9 S UMMARY OF THE CHAPTER 62

3 PROPOSING A NEW REAL OPTIONS FRAMEWORK BY CONSOLIDATING AND EXTENDING THE PREVAILING REAL OPTIONS PRACTICES BY A STEP OF OPTIMIZATION 63

3.1 T HE PROPOSED REAL OPTIONS FRAMEWORK 63

3.2 S TEP 1: F RAMING 65

3.2.1 Define system and its objective(s) 65

3.2.2 Identify and quantify uncertainties 67

3.3 S TEP 2: D ESIGN 71

3.3.1 Generate alternative designs 71

3.3.2 Identify and generate options 71

3.3.3 The complexity associated with multiple interacting real options, their costs, benefits and exercising conditions 75

3.4 S TEP 3: C HOICES 76

3.4.1 Simulation-based valuation methods 77

3.4.2 Issues with using multiple real options in project design 81

3.4.3 How the prevailing real options frameworks deal with the issues of multiple interacting real options and varying exercising conditions 84

3.4.4 Issue of the representation of the value of real options from multiple perspectives 85

3.4.5 Specific methodological challenges to be further addressed 85

3.5 S TEP 3 E : T HE EXTENDED OPTIMIZATION STEP 86

3.5.1 Overview of the extended step 86

3.5.2 The mechanism of the proposed framework 87

3.5.3 The partitioning or determination of the exercising conditions of real options 95

3.5.4 Sensitivity Analysis 100

3.5.5 The focus on searching for better design solutions 100

3.6 T HE PROPOSED FRAMEWORK SUPPORTS DECISION MAKERS 101

3.7 T HEORETICAL EVALUATION OF THE FRAMEWORK 103

3.8 S UMMARY 108

4 CASE EXAMPLE: MARITIME DOMAIN PROTECTION (MDP) SYSTEM IN MALACCA STRAIT 111

4.1 I NTRODUCTION 111

4.2 B ACKGROUND OF MDP S YSTEM 114

4.2.1 MDP system description 114

4.3 M ODEL OF MDP SYSTEM 119

4.3.1 Uncertainties 120

4.3.2 Options of interest 122

4.3.3 Valuation of real options 123

4.3.4 Optimization by evolutionary algorithms 127

4.4 R ESULTS 131

4.4.1 Project value of a fixed design using deterministic & Monte Carlo simulations 135

4.4.2 Project value based on the hand-picked real options approach 135

4.4.3 Project value based on the “EA selecting real options without exercising rules” approach 136

4.4.4 Project value based on the “EA selecting real options together with exercising rules” approach 138

4.4.5 Real option values in detail using a Value-At-Risk graph 140

4.4.6 Sensitivity of the value of the real option to the level of the uncertainty 141

4.5 S UMMARY AND DISCUSSIONS 142

5 CASE EXAMPLE: SINGAPORE WATER SUPPLY MANAGEMENT SYSTEM 145

5.1 B ACKGROUND 147

5.1.1 Problem statement and approach 147

5.2 S YSTEM DESCRIPTION 148

5.3 A PPLICATION OF THE REAL OPTIONS APPROACH 152

5.3.1 An overview of the real options model of the water supply system: 152

5.3.2 Objective measures 153

5.3.3 Uncertainties 155

5.3.4 Options of interest 161

5.3.5 Modeling of the water supply system without and with the real options 164

5.3.6 The optimization and selection of option exercising conditions 169

5.4 R ESULTS 170

5.4.1 Sensitivity analysis 173

5.4.2 Determination of option exercising conditions 180

5.4.3 Discussions 184

5.5 S UMMARY 187

6 DISCUSSIONS AND CONCLUSIONS 189

6.1 A FORMAL METHODOLOGICAL EVALUATION OF THE PROPOSED FRAMEWORK 189

6.2 D ISCUSSIONS 193

6.3 R ESEARCH QUESTIONS REVISITED AND CONTRIBUTIONS 196

6.4 L IMITATIONS 199

6.4.1 Boundary of real options 199

6.4.2 The identification, nurturing and maintenance of real options 199

6.5 F UTURE W ORK 201

6.5.1 The application of the framework 201

6.5.2 The improvement of framework 202

6.6 S UMMARIZING STATEMENTS 204

REFERENCES 205

SUMMARY

The proper decision support to design flexible projects considering downstream

decisions under uncertainties is critical yet challenging While flexibility provides important leverage against uncertainty, decision making that involves the planning and exercising of multiple elements of flexibility under uncertainty is complex, and due to the complexity,

organizations often fail in practice to follow a well-structured, accountable and reproducible decision-making process for assessing and selecting flexible projects

This study inherits the prevailing real options practices in framing downstream decisions

as real options and establishing the cause-and-effect relationships between flexibility and project value under uncertainty The study extends that with an optimization step to

develop a framework that can handle portfolios of interacting real options in a

combinational design space more effectively than humans alone can with bounded

rationality The framework borrows and integrates simulation-based options valuation

methods, decision analysis techniques and evolutionary algorithms to search for flexible

solutions more effectively

The resulting flexible solutions consist of an initial design and a portfolio of real options with their exercising conditions to adapt according to the manners the future unfolds - in

contrast to the common real options practices which primarily aim to derive the option

value Evaluated and exemplified through two real-life cases, the evolutionary framework compares favorably with the traditional fixed design approach and delivers considerable

improvements over the prevailing real options practices In the MDP case study, it is found that designing multiple real options into the MDP system using the proposed framework can increase the system value by 13% beyond that from the prevailing real options practices In the other case study, it is found that the developments and incorporations of innovative

water technologies for the water supply system of Singapore are the dominant solutions

from multiple perspectives

The proposed decision support framework facilitates the exploration, analysis,

optimization, and selection of solutions effectively in a combinatorial solution space – the essence of a design process- and allows the decision process regarding multiple threads of interacting flexibility to be within the bounds of reasonable effort on the part of the decision maker Therefore, the framework can be used to search flexible designs with portfolios of real options, free up design space, and permit human experts to focus on the more creative process of generating design alternatives to harness more from flexibility

It changes fundamentally the design process by improving the otherwise obscure

understanding of flexibility, systemizing the design process regarding flexibility in project

design, and facilitating the leverages from flexibility and the ex ante planning of downstream decisions This framework is especially helpful for large-scale innovation projects, which are usually accompanied by huge technological and market uncertainty and complex

downstream decisions

LIST OF TABLES

TABLE 2-1DEFINITIONS AND CLASSIFICATIONS OF UNCERTAINTIES -ADAPTED FROM DELAURENTIS AND

MAVRIS (2000) 20

TABLE 2-2A CHRONICLE OF THE DEVELOPMENT OF REAL OPTIONS VALUATION METHODS 41

TABLE 3-1COMPARISONS OF SEVERAL DESIGN APPROACHES CONCERNING FLEXIBILITY BY DESIGN SPACE SETS AND DESIGN VALUES 104

TABLE 4-1FORMULATION OF THE MDP SYSTEM DESIGN PROBLEM BY DIFFERENT DESIGN APPROACHES 119

TABLE 4-2INPUT PARAMETERS FOR THE REAL OPTIONS MODEL IN THE MDP APPLICATION 133

TABLE 4-3.A SUMMARY OF DESIGN APPROACHES AND THEIR RESULTING DESIGN SPACES 134

TABLE 4-4.THE PROJECT VALUES BASED ON SEVERAL DIFFERENT DESIGN APPROACHES 134

TABLE 5-1LIST OF UNCERTAINTIES MODELLED BY GBM IN THE WATER SUPPLY SYSTEM OF SINGAPORE 159

TABLE 5-2LIST OF UNCERTAINTIES MODELLED BY SCENARIOS IN THE WATER SUPPLY SYSTEM OF SINGAPORE 160

TABLE 5-3BENEFITS OF REAL OPTIONS IN TERM OF ITS IMPROVEMENTS ON THE WATER SUPPLY SYSTEM OF SINGAPORE 172

TABLE 5-4THE EXPECTED OPTION VALUE OF DESALINATION-BY-REGASIFICATION WHEN IT IS EXERCISED SHOULD ITS COST BE LOWER THAN A PREDEFINED MONETARY THRESHOLD 181

TABLE 5-5THE EXPECTED OPTION VALUE OF DESALINATION-BY-REGASIFICATION WHEN IT IS EXERCISED SHOULD ITS COST BE LOWER THAN A FRACTION OF THE COST OF NEWATER OR DESALINATION BY RO 181

TABLE 5-6PARAMETERS FOUND BY GA IN A SET OF FUNCTIONS THAT DEFINE THE EXERCISING CONDITIONS OF DESALINATION-BY-REGASIFICATION OVER NEWATER 182

TABLE 5-7PARAMETERS FOUND BY GA IN A SET OF FUNCTIONS THAT DEFINE THE EXERCISING CONDITIONS OF DESALINATION-BY-REGASIFICATION OVER DESALINATION BY RO 182

LIST OF FIGURES

FIGURE 1-1A REAL OPTION TO EXPAND AN OFFICE BUILDING IN DOWNTOWN CHICAGO (GUMA 2008).THE INITIAL DESIGN (A) CONTAINS A REAL OPTION TO EXPAND VERTICALLY DEPENDING ON THE DEMAND TO THE BUILDING (B) 3

FIGURE 1-2A FLEXIBLE DESIGN WITH TWO OPTIONS TO EXPAND AND THEIR EXERCISING CONDITIONS 5

FIGURE 1-3THE TWO OPTIONS TO EXPAND PLUS THE OPTION TO CONTINUE AS-IS FORM THREE MUTUALLY

EXCLUSIVE EXERCISING REGIONS DEFINED BY THEIR EXERCISING CONDITIONS IN A DECISION NODE 7

FIGURE 1-4THE AREA IDENTIFIED FOR THIS RESEARCH IN TERM OF NUMBERS OF REAL OPTIONS AND THEIR

FIGURE 2-2THE GOAL OF DESIGN AND PEOPLE INVOLVED –FROM ROSS (2006) 19

FIGURE 2-3CHANGES IN COST, KNOWLEDGE, AND MANAGEMENT LEVERAGE OF A SYSTEM IN ITS LIFECYCLE –

ADAPTED FROM FABRYCKY AND BLANCHARD (1991) 26

FIGURE 2-4COSTS AS A FUNCTION OF THE DEGREE OF FLEXIBILITY –ADAPTED FROM FABRYCKY AND

BLANCHARD (1991) AND SCHULZ ET AL.(2000) 30

FIGURE 2-5THE OPTION VALUE OF A CALL OPTION 31

FIGURE 2-6 THE 8-STEP REAL OPTIONS FRAMEWORK PROPOSED IN MUN (2006), WITH THE PERMISSION

FROM THE AUTHOR 52

FIGURE 2-7A SIMPLE FLOWCHART OF AN EVOLUTIONARY ALGORITHM 59

FIGURE 3-1THE PROPOSED FRAMEWORK - CONSOLIDATING THE REAL OPTIONS PRACTICES BY THE 3-STEP

PROCESS OF SIMON AND EXTENDING IT BY A STEP OF OPTIMIZATION 65

FIGURE 3-2OPTIMIZATION BASED ON MULTIPLE OBJECTIVES YIELDS A PARETO FRONT 66

FIGURE 3-3REPRESENTING REAL OPTIONS RELATED TO A WATER TECHNOLOGY IN AN INFLUENCE DIAGRAM

(EACH CIRCLE WITHIN A STATE REPRESENTS THE OPTION TO CONTINUE IN THAT STATE) 73

FIGURE 3-4FIXED DESIGNS VS. FLEXIBLE DESIGNS ON PARETO GRAPHS 74

FIGURE 3-5VAR GRAPH OF THE VALUE OF THE FIXED DESIGN AND THE FLEXIBLE DESIGN WITH REAL OPTIONS OPTIMIZED BY GA– FROM ZHANG AND BABOVIC (2010A) 80

FIGURE 3-6VALUE OF SYSTEMS UNDER UNCERTAINTIES.THE VALUE OF A FIXED SYSTEM IS DEPICTED IN GRAPH

(A) WHEN THE REALIZATION OF DEMAND IS GENERALLY UPWARD AND IN GRAPH (C) WHEN THE

REALIZATION OF DEMAND IS GENERALLY DOWNWARD.GRAPH (B) AND (D) DEPICT HOW FLEXIBLE

SYSTEMS WITH REAL OPTIONS COPE WITH THE CHANGES IN DEMAND IN STAGES 83

FIGURE 3-7THE FLOWCHART OF THE PROPOSED REAL OPTIONS FRAMEWORK INTEGRATING GENETIC

ALGORITHMS,MONTE CARLO SIMULATIONS, AND DECISION TREE TECHNIQUES 87

FIGURE 3-8AN EXAMPLE OF MULTIPLE OBJECTIVE OPTIMIZATIONS WHICH FORM A PARETO FRONT 93

FIGURE 3-9A DECISION TREE WITH OPTIONS TO EXPAND, CONTINUE, AND ABANDON -ADAPTED FROM

BRANDAO AND DYER (2005) 96

FIGURE 3-10(A)A SCHEMATIZATION OF THE EVOLUTION OF UNCERTAINTIES ALONG TIME.(B)A

SCHEMATIZATION ON PARTITIONING THE DECISION SPACE (SIMULTANEOUSLY IDENTIFY THE

CONFIGURATIONS TO USE AND THEIR EXERCISING CONDITIONS) 97

FIGURE 3-11TRANSFORMATIONS OF A CLASSIfiCATION TREE TO A DECISION TREE –FROM BRYDON AND

GEMINO (2008) 99

FIGURE 3-12RELATIONSHIPS OF SETS OF DESIGN SPACE FROM DIFFERENT DESIGN APPROACHES CONCERNING FLEXIBILITY (A,B,C,D, AND E) 106

FIGURE 4-1.PIRACY AND TERRORISM INCIDENTS.SOURCE:IMO(2007) 112

FIGURE 4-2TANKER “LIMBURG” HIT BY AN TERRORISM ACT NEAR YEMEN IN 2002 WHILE SHIPPING OIL TO

MALAYSIA FOR PETRONAS 115

FIGURE 4-3.SEDP FLOW DIAGRAM.SOURCE:NPS(2005) 116

FIGURE 4-4.MDPOVERARCHING MODELING PLAN.SOURCE:NPS(2005).THE MDPOVERARCHING

MODELING PLAN PICTORIALLY SHOWS AN INTEGRATED MODEL THAT TRANSFORMS INPUTS FROM

INDIVIDUAL SYSTEM PERFORMANCE AND COST MODELS INTO THE DESIRED PERFORMANCE MEASURE

OUTPUTS 118

FIGURE 4-5BRANCHING OF A TRINOMIAL DECISION TREE 123

FIGURE 4-6A HANDPICKED REAL OPTIONS TRINOMIAL SCENARIO TREE OF 3 STAGES (R 2007, R 2010, AND R2013 REPRESENT THE DEGREE OF TERRORISM RISK IN YEAR 2007,2010, AND 2013 RESPECTIVELY) 124

FIGURE 4-7A FLOWCHART OF THE VALUATION SUB-MODULE OF THE REAL OPTIONS MODEL IN THE MDP APPLICATION 126

FIGURE 4-8A FLOWCHART OF THE REAL OPTIONS MODEL AS APPLIED IN THE MDP APPLICATION 128

FIGURE 4-9(A)A SCHEMATIZATION OF THE EVOLUTION OF UNCERTAINTIES ALONG TIME,(B)A SCHEMATIZATION ON PARTITIONING THE DECISION SPACE (IDENTIFYING CONFIGURATIONS TO USE AND SETTING UP THE EXERCISING CONDITIONS) 130

FIGURE 4-10AN EXAMPLE OF REFINING VALUE BY A MONTE CARLO SIMULATION WITH A LARGER SAMPLE 131 FIGURE 4-11GA SELECTING REAL OPTIONS IN A TRINOMIAL SCENARIO TREE (THE PARTS IN RED DENOTE DECISION VARIABLES OPTIMIZED BY GA– IN THIS CASE ALL THE SYSTEM CONFIGURATIONS IN THE 13 DECISION NODES IN THE 3 STAGES) 137

FIGURE 4-12GA SELECTING BOTH THE REAL OPTIONS AND THEIR EXERCISING CONDITIONS (PARTS IN RED DENOTE DECISION VARIABLES OPTIMIZED BY GA– IN THIS CASE THE CONFIGURATIONS AND THE DEGREE OF RISK TO EXERCISE THE CONFIGURATIONS) 139

FIGURE 4-13AVAR GRAPH OF THE SYSTEM VALUE OF THE FIXED DESIGN AND THE FLEXIBLE DESIGN WITH REAL OPTIONS SELECTED BY GA 141

FIGURE 4-14THE AVERAGE VALUE OF REAL OPTIONS WITH RESPECT TO THE UNCERTAINTY REGARDING THE TERRORISM LEVEL 142

FIGURE 5-1 THE WATER RESOURCE OPERATING STRATEGY THE USUAL WAY (A) AND THE WATER RESOURCE OPERATING STRATEGY THE INTEGRATED WAY WITH FLEXIBILITY (B) 148

FIGURE 5-2SINGAPORE HAS THE 2ND HIGHEST POPULATION DENSITY IN THE WORLD BUT ALSO HAS 14 RESERVOIRS IN THE CITY-STATE (YOUNG 2008) 150

FIGURE 5-3THE CONTRIBUTIONS TO THE WATER SUPPLY SYSTEM BY THE FOUR TAPS AT THEIR MAXIMUM CAPACITIES IN YEAR 2006(HUNG 2009) 150

FIGURE 5-4A REAL OPTIONS MODEL OF THE WATER SUPPLY SYSTEM TAKING ACCOUNT OF UNCERTAINTIES AND FLEXIBILITIES 152

FIGURE 5-5REAL OPTIONS RELATED TO A WATER TECHNOLOGY REPRESENTED IN AN INFLUENCE DIAGRAM 162 FIGURE 5-6THE EVOLUTION OF THE WATER SUPPLY SYSTEM ACCORDING TO THE EVOLUTION OF UNCERTAINTIES 166

FIGURE 5-7THE FLOWCHART TO DETERMINE WHICH TAP TO ADJUST IN A 3-TAP SYSTEM 167

FIGURE 5-8.THE FLOWCHART TO DETERMINE WHICH TAP TO ADJUST IN A 4-TAP SYSTEM 168

FIGURE 5-9THE FLOWCHART TO DETERMINE WHICH TAP TO ADJUST IN A 4.5-TAP SYSTEM 169

FIGURE 5-10THE SENSITIVITY OF THE EXPECTED FINANCIAL VALUE OF NEWATER AND DESALINATION-BY -REGASIFICATION TOWARDS POLITICAL PARAMETERS.THE VALUE OF THE NEWATER IS MARKED ON THE LEFT VERTICAL AXES IN EACH OF THE 9 GRAPHS, WHILE THE VALUE OF THE DESALINATION-BY -REGASIFICATION IS MARKED ON THE RIGHT VERTICAL AXES IN EACH OF THE 9 GRAPHS.IN ROW 1, THE HORIZONTAL AXIS REPRESENTS THE PROBABILITY OF 1961 AGREEMENT TO LAPSE.IN ROW 2, THE HORIZONTAL AXIS REPRESENTS THE PROBABILITY OF 1962 AGREEMENT TO BE CANCELLED OR HALVED IN CAPACITY.IN ROW 3, THE HORIZONTAL AXIS PRESENTS THE PREMIUM OF IMPORTING WATER 175

FIGURE 5-11THE SENSITIVITY OF THE POLITICAL RISK OF NEWATER AND DESALINATION-BY-REGASIFICATION TOWARDS POLITICAL PARAMETERS.THE VALUE OF THE NEWATER IS MARKED ON THE LEFT VERTICAL AXES IN EACH OF THE 9 GRAPHS, WHILE THE VALUE OF THE DESALINATION-BY-REGASIFICATION IS

MARKED ON THE RIGHT VERTICAL AXES IN EACH OF THE 9 GRAPHS.IN ROW 1, THE HORIZONTAL AXIS

REPRESENTS THE PROBABILITY OF 1961 AGREEMENT TO LAPSE.IN ROW 2, THE HORIZONTAL AXIS

REPRESENTS THE PROBABILITY OF 1962 AGREEMENT TO BE CANCELLED OR HALVED IN CAPACITY.IN

ROW 3, THE HORIZONTAL AXIS PRESENTS THE PREMIUM OF IMPORTING WATER 177

FIGURE 5-12THE SENSITIVITY OF THE SOCIOECONOMIC RISK OF NEWATER AND DESALINATION-BY

-REGASIFICATION TOWARDS POLITICAL PARAMETERS.THE VALUE OF THE NEWATER IS MARKED ON THE LEFT VERTICAL AXES IN EACH OF THE 9 GRAPHS, WHILE THE VALUE OF THE DESALINATION-BY-

REGASIFICATION IS MARKED ON THE RIGHT VERTICAL AXES IN EACH OF THE 9 GRAPHS.IN ROW 1, THE HORIZONTAL AXIS REPRESENTS THE PROBABILITY OF 1961 AGREEMENT TO LAPSE.IN ROW 2, THE

HORIZONTAL AXIS REPRESENTS THE PROBABILITY OF 1962 AGREEMENT TO BE CANCELLED OR HALVED IN CAPACITY.IN ROW 3, THE HORIZONTAL AXIS PRESENTS THE PREMIUM OF IMPORTING WATER 179

FIGURE 5-13THE OPTIMAL EXERCISING COST OF DESALINATION-BY-REGASIFICATION, BELOW WHICH

DESALINATION-BY-REGASIFICATION IS PREFERRED OVER NEWATER IN CASE OF AN EXPANSION 183

FIGURE 5-14THE OPTIMAL EXERCISING COST OF DESALINATION-BY-REGASIFICATION, BELOW WHICH

DESALINATION-BY-REGASIFICATION IS PREFERRED OVER DESALINATION BY RO IN CASE OF AN EXPANSION

184

LIST OF ABBREVIATIONS

3-tap The three taps of supplying water by means of catchment, desalination, and

import before recycling was introduced 4-tap The four taps strategy of water supply in Singapore by means of catchment,

desalination, recycling, and import 4.5-tap The water supply system in Singapore if desalination by LNG regasification is

used in addition to catchment, desalination (by Reverse Osmosis), import, and recycling

C3I Command and control, communications and intelligence (C3I)

DA Decision Analysis

DCF Discounted Cash Flow

DEEP Desalination Economic Evaluation Program

DSTA Defense Science and Technology Agency

EA Evolutionary algorithm

GBM Geometric Brownian Motion

IWRM Integrated Water Resource Management (IWRM)

LNG Liquefied Natural Gas

MAD Market Asset Disclaimer

MDP Maritime Domain Protection

NEWater The name for the recycled water in Singapore

NPS Naval Postgraduate School

NPV Net Present Value

PSR Punggol/Serangoon Reservoir

PUB Public Utility Board

SDWA Singapore Delft Water Alliance

VaR Value at Risk

WMD Weapons of Mass Destruction

1 INTRODUCTION

“Randomness fills the world and our life with charms of the unknown and provides us with all

kinds of opportunities … By ‘seizing the opportunity,’ we actually mean to snatch the

randomness.”

- Deyi Li and Yi Du in “Artificial Intelligence with Uncertainty”

1.1 Motivation

1.1.1 Uncertainty and flexibility

Managers of projects and systems tend to adopt a single scenario for the future, come

up with a fixed design, and compute a single performance measure for a project Ex ante

design decisions are based on forecasts that are “always” wrong (Ascher 1978, de Neufville 2000) If reality departs in one manner or another from what was originally anticipated –

which is usually the case (Savage 2000), especially innovative projects, the fixed design ends

up in suboptimal configurations regardless of how sound it is designed technically

Seikan Tunnel between Honshū and Hokkaidō and Iridium satellite telephone system

have given us classical examples (Takashima 2001, de Weck et al 2004) of how superb innovation projects, failed under uncertainties While the failures arise for many

technically-reasons, they can be attributed to a large extent to the fact that the analysis either has not incorporated risks and uncertainty into the process, or has failed to create plans to address the issue of uncertainty in the design and management process

To address uncertainty in projects, the conventional notion that risk is undesirable and must therefore be minimized is inadequate (Trigeorgis 2005) What we can do when faced with uncertainty is more than risk reduction, we can also incorporate flexibility so that

downstream value-maximizing decisions can be made to change the flexible design, to

reduce losses in case of downward risks and exploit upward opportunities arisen from

uncertainty For instance, projects with flexibility properly incorporated can have their

subsystems or project modules changed to exploit later shifts in market or to make use of new technologies to be developed in future

Multiple sources of flexibility exist in the design and management of systems and

projects Such flexibilities related to systems and projects are specifically known as “real

options” by an analogy with financial options which are purely contractual in monetary

terms (Myers 1984, Trigeorgis 1996, Copeland and Antikarov 2001) Both financial options and real options are rights but not obligations to take certain actions at some point of time Real options are valuable, as they embody flexibilities to enable downstream

management decisions for increasing the values of the projects Having real options would always be advantageous − if they were free Yet, incorporating flexibility into a project

involves costs, because it may involve extra capabilities, because it may involve making

investments in smaller stages and missing out on economies of scale, or because it may

cause delays and causes losses of potential benefits The questions thus are: what is the

value of each of the different forms of flexibility that might be added to the project? And

which ones justify their costs? This is the essential task of real options analysis

Real options analysis has been widespread, covering various types of real options, such

as the option to temporarily shut-down (Brennan and Schwartz 1985), option to continue or discontinue a series of investments (Majd and Pindyck 1987), option to stage a plan (Alvarez

1999, Benaroch 2002), option to switch (Baldwin and Ruback 1986, Kulatilaka 1993) and

option to increase/decrease capacity (Chou et al 2007) Real life projects often involve

combinations of such options

Recently, emerging interest in applying real options analysis to evaluate and design

innovative projects and systems is observed (de Neufville 2003) Researchers at MIT studied the design and management of the communication satellite systems and pointed out that if

they were prepared to deal with uncertainty with real options to be deployed in stages

according to the demands, the bankruptcies of those systems, such as Iridium and

Global-star, could have been avoided (de Weck et al 2003) Zhao and Tseng (2003) studied the real option of sizing the foundation of a parking garage so that additional floors can be added at

a later date if a large demand materializes The value of the parking garage with the extra sizing therefore includes not only its present value, but also the value associated with the

option to expand the floor, which is found to be significant Guma (2008) did a similar study for a real option to expand an office building (Figure 1-1)

Figure 1-1 A real option to expand an office building in downtown Chicago (Guma 2008) The initial design (a) contains a real option to expand vertically depending on the demand to the building (b)

1.1.2 Flexibility in the design of systems and projects: a casino example

Let’s use a simple and general case of building an infrastructural system with

flexibility/options to expand as an example to illustrate the concepts and issues of using real options in real-world projects Suppose the problem at hand is to build a new casino A key problem in designing the casino is that the number of customers the casino will attract in

future (i.e the realized demand of the casino) is uncertain

On one hand, a capacious casino may not have its cost recovered if the realized demand

is smaller, and on the other hand, a smaller compact casino will have to forgo the

opportunity to gain customers and profits in case the realized demand is bigger One means

to deal with this dilemma is to develop a flexible design with real options The flexible design starts with a small building and strengthened footings and columns to enable possible future expansions In doing so, the owner of the casino acquires real options- rights but not

respect to the capacity in two segments: 1) the NPV is linearly proportional to the capacity before the design reaches it full capacity, and 2) the NPV is capped and unchanged once the design reaches its full capacity

Each of the two dotted lines represents the flexible design having a real option exercised (Figure 1-2) Practically, designers may conceive many real options, but at the moment, let’s suppose there are only two real options to expand to two different levels of capacity and

analyze them One real option is to expand the capacity from C0 to C1, and the other option

is to expand the capacity from C0 to C2 (C2 > C1) To further simplify the matter, no option is available to expand the capacity from C1 to C2

So in principle if the realization of capacity is high, option A might be exercised, and if

the realization is very high, option B may be exercised But exactly how high is considered high and how high is very high? In another words, what are the conditions to exercise the two real options?

Figure 1-2 A flexible design with two options to expand and their exercising conditions

Given that the decision makers have a right but not an obligation to expand the flexible design, they will only choose to do so if it increases the project value In this case, if the

decision makers only have option A, they will exercise it if the capacity is larger than E01

(where the blue line of initial design and the dotted line of option A intersect) in Figure 1-2 This changes the targeted capacity of the flexible design from C0 to C1 Similarly, if the

decision makers only have option B, option B should be exercised when capacity reaches E02 (where the blue line of initial design and the dotted line of option B intersect) to expand the capacity from C0 to C2

However, if the designers have both option A and option B, it is better to exercise option

B if the realized demand is above E012, and it is it is better to use option A if the realized

demand falls between E01 to E012 (Figure 1-2) The exercising conditions when two options are available are different from those when either of them is available individually This

clearly shows that, where multiple real options are present, the exercising conditions of real options cannot be set in isolation

Even though the casino example is simplistic, it illustrates a few fundamental issues

about real options

Cost of options

Option A to expand to C1

Option B to expand to C2

Cost of options

Option A to expand to C1

Option B to expand to C2

First, unlike a financial options, whose exercising conditions are straightforward (a

financial option is exercised if the price of underlying asset is more than the sum of its strike price and option cost written on a contract), real options are exercised to increase the value

of the underlying asset –a real-life project, and their exercising conditions are not

contractual, but need to be defined so as to best increase the value of the underlying project

In this case, even though there are only two real options -both being expansion options, and many simplifying assumptions are made, it can be seen that there is optimality in setting the exercising conditions of the real options The exercising conditions should be set so that the project value will be maximized This is easy sometimes (as in this case where linear

convex functions are assumed), but more likely to be complicated (as in most real world

design cases which exhibit non-linearity and non-convexity) and deserves more careful

studies

Second, the exercising conditions when multiple real options are present simultaneously are different from the exercising conditions when each of them is present separately One reason for this is that the exercising regions, defined by the exercising conditions, could be mutually exclusive -only one real option can be exercised in a scenario In the casino

expansion example, the decision makers have in fact three options: two options to expand and an option to continue as-is Only one option will be exercised at any single scenario in a decision mode (as illustrated in Figure 1-3) The exercising conditions of the three real

options partition all the scenarios (i.e the realized demand in the casino example) into three mutually exclusive exercising regions Additionally, the combination of all the exercising

regions should be exhaustive No scenarios should be left out without a decision (to remain as-is is considered as a conscious decision)

Figure 1-3 The two options to expand plus the option to continue as-is form three mutually exclusive exercising regions defined by their exercising conditions in a decision node

Mutual exclusivity and exhaustiveness are some of the most typical relationships formed

by the exercising conditions of real options However the relationships of the options

exercising conditions are context dependent and could be highly intricate and convoluted For instance, real options can be sequential; e.g the results of an R&D option may enable the options to produce related products

Real options and their exercising conditions stand or fall together Exercising a wrong

real option and exercising a right option in a wrong condition are equally undesirable by the same token A flexible solution comprises of both the options and their exercising strategies, and to maximize the design value, both parts of the solutions (the options and the exercising conditions) must be analyzed and selected simultaneously to take account of the

interactions

However, this poses a heavy load upon our planning and decision-making capability,

when the number of real options considered is increasingly large and the assumptions about the real options interactions are less simplified As real options models approximate the

reality of project design better, the web of real options and interactions becomes larger,

Continue

A Decision Node

Continue

A Decision Node

tighter and more intricate, and it will be practically too complex for human cognition alone

to design systems with multiple real options without a proper decision support framework

1.1.3 Multiple threads of flexibility –their benefits and associated complexity

Despite of the many issues related, embedding multiple flexibility/real options and

exercising them to make changes according to how the future unfolds is a very important

means to deal with uncertainties in the design and management of projects and systems

(Triantis and Borison 2001) Both the initial design and a sequence of intermediate decisions

to change the project must be determined ex ante under the consideration of uncertain

project environment to improve the design (Hazelrigg 1998)

The important benefits of using multiple real options as well as the difficulties, such as the non-additivity and complex interactions among real options, have long been

acknowledged (Trigeorgis 1996, Vassolo et al 2004) The difficulties are caused by the

complex structure of project pay-offs generated from the paths of uncertainty and the

interactions among the portfolio of real options where each option may alter the boundary conditions of other options (McGrath 1997) The value of a combination of real options is

not the combined value of each option in isolation (Trigeorgis 1993) To value a portfolio of real options, Anand (2007) analyzed the determinants to explain the portfolio effects, and Baldwin and Clark (2000) calculated the option value of modules in a modular architecture theoretically

However, it remains unclear how to quantitatively assess a portfolio of real options

when the number of interacting real options becomes large and the design space becomes non-linear and discontinuous (an area needs further research as illustrated in Figure 1-4)

Figure 1-4 The area identified for this research in term of numbers of real options and their exercising conditions

No conventional real options methodologies are able to systematically and holistically valuate and select multiple interdependent real options and their exercising conditions for project and system designs Decision making that involves the planning and exercising of

multiple options under uncertainty is complex, and due to the complexity, organizations

often fail in practice to follow a well-structured, accountable and reproducible

decision-making process for assessing and selecting a dynamic strategy for flexible project solutions (Hilhorst et al 2008)

1.2 Research opportunities

1.2.1 Research gap

The conceptual appeal of designing systems and projects with real options is strong,

however, the valuation and selection of real options to be incorporated in a complex system

is non-trivial (McManus and Hastings 2006) The bounded rationality of humans limits our capability to take account of multiple real options and anticipate all the possible

Areas identified for further research

Areas identified for further research

contingencies in an inherently combinational and computational design process (Simon

1995)

A framework that supports rational decision making within the bounds of reasonable

effort on the part of decision makers (Todd and Benbasat 1994) is crucially needed to give managers and designers discriminating tools with which to choose the most valuable

portfolio of flexibilities to be incorporated Meanwhile, methods for flexible or evolutionary designs for handling uncertainties are in their infancy (McManus and Hastings 2006), and

tools need to be developed to evaluate and select an often very large number of possible

design configurations under a wide range of uncertain future scenarios (Cardin and de

Neufville 2009)

The essence of the problem in designing projects and systems is to facilitate the

exploration, analysis, optimization, and selection of solutions effectively in a combinatorial design space This requires an appropriate design process (Olewnik et al 2004) Hence the focus of a decision support framework for designing flexible projects and systems is not to strive for the absolute value of a real option as in prevailing real options analysis (Kalligeros and De Weck 2004) but to establish a systematic design process that lead to better design solutions

1.2.2 Research question

This study centers on the following issue:

The proper decision support to account for and construct a portfolio of real options as well as their exercising strategies for projects and systems under uncertainty

To answer that question, this study develops a decision support framework to expose, evaluate, systematize, and partly routinize the use of a portfolio of real options for designing

systems and projects and tests the framework by two cases In doing so, a myriad of

intermediate questions arise:

• What are the deficiencies of the existing real options frameworks that prevent them from accounting for multiple real options and exercising conditions?

• Which parts of the existing real options frameworks can be reused?

• Can the proposed framework be built upon the existing real options frameworks to maintain a high degree of compatibility?

• Can the prevailing real options frameworks be consolidated by a more structured

process as found in the field of decision analysis?

• What new developments are needed to enable the proposed framework to take

account of multiplicities of real options (i.e multiple real options and multiple

exercising conditions over multiple decision nodes under multiple sources of

uncertainty, as well as their interactions)?

• How can the framework explore and search the large flexible design space

associated with the multiplicities?

• Is the framework useful and can it be applied effectively (able to search the large real options design spaces for better design solutions) in real-life design cases?

• Can the framework evaluate and select designs based on multiple objectives that commonly exist in system and project design?

• Is the framework able to change the design decision and formalize the design process?

1.3 Research approach

To be able to take account of a large quantity of real options and their exercising

conditions in a systematic manner to search for better flexible design solutions, this research consolidates the state-of-the-art real options framework based on the three step-wise

decision process developed by Simon (1977) and develops an extension necessary (Figure 1-5) Chapter 3 will elaborate the state-of-the-art real options process vis-a-vis the proposed real options decision support framework

Figure 1-5 Consolidating the prevailing real options framework and extending it to develop the proposed real options framework (to be elaborated in chapter 3)

Without the extended step of optimization, the prevailing real options practices contain

a very selective process so that the number of real options and consequently the design

space are both small (Wang 2005), whereas the proposed real options framework relaxes such stringent requirements and considers a larger pool of real options thanks to its

expanded capability to value and select them in a large design space (e.g the design space is

in the scale of 10 to the power of 58 in the MDP case study to be discussed in chapter 4)

In addition, while option exercising conditions are predetermined and not optimized in

an integrated manner in the prevailing real options practices, the exercising conditions and the real options are valued and selected in an integrated procedure to further improve the flexible design solutions

3.1 Simulations of the design under risks without real options

3.2 Simulations of the design with real options and compare the results and select the better design

Description

of processes

3e.1 Optimize options and their exercising conditions 3e.2 Optimize the representation

of states and options

1 Framing:

define system and its

objectives and risks

2 Design: formulate solutions and generate alternatives

3 Choice: simulate, valuate and select design

Estimated exercising condition

Interact with decision makers

E N

Optimized exercising condition

E N*

Steps of frameworks

Current real options best practice

Proposed real options framework with an optimization step

A fixed system based on deterministic forecast

3.1 Simulations of the design under risks without real options

3.2 Simulations of the design with real options and compare the results and select the better design

Description

of processes

3e.1 Optimize options and their exercising conditions 3e.2 Optimize the representation

of states and options

1 Framing:

define system and its

objectives and risks

2 Design: formulate solutions and generate alternatives

3 Choice: simulate, valuate and select design

Estimated exercising condition

Interact with decision makers

E N

Optimized exercising condition

E N*

Steps of frameworks

Current real options best practice

Proposed real options framework with an optimization step

A fixed system based on deterministic forecast

Those improvements are made possible by the extended step of optimization The

optimization step employs evolutionary algorithms to take account of many real options,

their exercising conditions, and options interactions altogether The optimization step is

integrated with other techniques such as decision trees and Monte Carlo simulations that are popular in prevailing real options analysis Such techniques have been applied separately

in various applications of real options but now woven together in a framework to practically design a portfolio of real options in an otherwise intractable design space

In order to examine and validate the proposed framework -specifically its effectiveness and efficiency in both finding suitable pieces of real options and formulating a dynamic

adaptive plan to increase the project value, the framework is applied to two real-life cases The proposed framework are compared with the conventional real options practices as well

as the classical generic system engineering approach in their abilities to yield better flexible solutions, free up (flexible) design space, and allow humans to focus on the more creative process of generating design alternatives to harness more from flexibility To permit a valid and consistent comparison, the objective hierarchy, architectural principles, and technical modules are always kept the same

In addition, the proposed framework not only can evaluate projects and associated

flexibilities from a single financial metrics commonly found in extant real options literature, but also evaluate real options from multiple objectives The values of flexibilities are

represented from multiple perspectives or utilities This better imitates the reality where

real options are used not just for their monetary values but also their other benefits -for

example, Zhang and Babovic (2010b)

This study applies the proposed decision support framework to two cases Such

applications are very different from most of the extant real options applications, which are very restrictive as discussed earlier The two cases both involve volatile uncertainties and

large design spaces, due to the large number of real options, possible exercising conditions and their interactions The two cases are:

1 Maritime Domain Protection (MDP) system in Malacca Strait

The requirements of the MDP system depend on the degree of terrorism, which is highly uncertain As the terrorism level fluctuates, the MDP system requirements change as well Therefore, it is desirable to design a flexible system that can be reconfigured according to how terrorism could possibly evolve over time This becomes a complex practical challenge

to model, because the numbers of pathways, whereby uncertainty may unfold and system can be reconfigured, is very large

2 Singapore water supply system

The various sources of water supply to the city-state face financial, socioeconomic, and political uncertainties Facing those uncertainties, Singapore has constructed a portfolio of 4 sources of water (4 taps) and is constantly looking for new means of making water, which is

a tremendous success (Tortajada 2006) Even so, there is a lack of integrated analysis to

assess the value that the four-tap strategy provides in its flexibilities to deal with

uncertainties

The case studies are carried out to verify the feasibility (proof of concept) of the

framework in real-world complex and interdisciplinary design problems They also serve to demonstrate and illustrate the practical issues found in the applications of the framework Moreover, despite of the conceptual appeal of real options for designing systems and projects, applications have not been widespread The two cases mark new areas for real

options applications where flexibility has not been studied formally before Therefore, the applications not only help to better understand the design and valuation of flexibility for

systems and projects in general but also yield insight about applying real options in those

two areas

1.4 Organization of the thesis

Chapter 2 starts with presenting the literature of uncertainty and flexibility related to

designing real world innovation project The research gap is identified and promising

techniques that may help to address the research gap are reviewed Chapter 3 proposes a decision support framework to fill the research gap The framework is constructed from

consolidating the prevailing real options practices and extending them by a step of

optimization Also in chapter 3, the design space and maximum possible design value

resulting from the framework is compared with those from other approaches Chapters 4

and 5 cover two case studies: Maritime Domain Protection system (MDP) and the Water

Supply System of Singapore These two case studies were chosen to test out the proposed framework –its concepts, techniques and processes Both chapters provide overviews of the systems, uncertainties, as well as potential options of interest for the two case studies

Quantitative analysis of the options is carried out based on data collected from public source materials Finally, chapter 6 provides final discussions, conclusions, limitations, and

suggestions for future work

2 LITERATURE REVIEW

2.1 Introduction

This chapter frames, organizes, and reviews the literature related to the problem of

designing flexible projects under uncertainty As the nature of the problem spans multiple disciplines, so does the body of literature review covered in this chapter Only areas that are closely related to the research questions are reviewed, and Figure 2-1 maps those areas by how they are related to the problem For such a multidisciplinary topic, an exhaustive

literature review is not possible and unlikely to be helpful, and this chapter aims to be

succinct and focuses on the latest issues

Figure 2-1 Mapping the multidisciplinary literature associated with the research question of this study

Another issue related to multidisciplinary research is that there is a high degree of

ambiguity in definitions across domains It is not the purpose of this work to synthesize them

Decision analysis field Financial trading

Decision process & decision analysis techniques Valuation (DCF)

Valuate managerial flexibility (real options)

Real option valuation methods

Simulation methods based on DCF

Incorporate flexibilities (real options)

Valuation of real options associated with system design

Valuation of real options associated with system design

Decision support for designing systems with multiple real options

Conceptual analogy

Conceptual analogy

lend tech- niques

provide value of real

Optimization field

Evolutionary Algorithm

Evolutionary Algorithm

Issues due to the complexity of real options with systems design

Decision analysis field Financial trading

Decision process & decision analysis techniques Valuation (DCF)

Valuate managerial flexibility (real options)

Real option valuation methods

Simulation methods based on DCF

Incorporate flexibilities (real options)

Valuation of real options associated with system design

Valuation of real options associated with system design

Decision support for designing systems with multiple real options

Decision support for designing systems with multiple real options

Conceptual analogy

Conceptual analogy

lend tech- niques

provide value of real

Optimization field

Evolutionary Algorithm

Evolutionary Algorithm

Issues due to the complexity of real options with systems design

into domain-neutral constructs or to develop new constructs; instead this chapter only

attempts to clarify the important terms used in this research

This chapter is organized so that the scope of the literature review is gradually narrowed down to the specific research gap of lacking of a framework to support decision regarding using multiple real options for designing projects and systems

First, section 2.2, 2.3, and 2.4 present a brief review of design, uncertainty, and flexibility This shows uncertainty is a key issue in design, and one fundamental means to deal with

uncertainty is flexibility, which can be analyzed as real options (section 2.5) Next, the

valuation of real options is examined (section 2.6) in a chronical order, from the classical

Black Scholes methods to the recent shift to developing other methods Section 2.7 presents the central topic of supporting decision making regarding flexible design by state-of-the-art real options frameworks Such frameworks borrow methods and techniques from decision analysis and present more structured and more powerful processes to support decision-

making under uncertainty Such frameworks are promising, but so far the frameworks

developed are not sophisticated enough to address the many practical and complex issues related to designing projects with multiple real options This represents a research gap

(section 2.8), and some techniques that can be used to develop a new framework to address the research gap are reviewed Section 2.9 concludes the chapter

2.2 Design

Design refers to the conception of objects, processes, and ideas for accomplishing goals, and showing how the objects, processes, or ideas can be realized (Simon 1995) In design, the goals and constraints are given or formulated, and the question is to find a design or

designs that satisfy those goals and constraints Figure 2-2 shows a graphical depiction of the goal and context of design The context surrounds the entire endeavor, including the

stakeholder, who influence over the definition and evaluation of the needs, the funders,

who influence over the allocation of the resources The decision makers take account of the needs and resources, define the system, create designs and select the best design(s) that

efficiently fulfill the requirement and maximize the payoff

Figure 2-2 The goal of design and people involved –from Ross (2006)

This study only discusses human designed1 projects or systems The design of projects and systems is often studied separately in their particular professional domains from which the relevant technical knowledge is borrowed Recently, people have started recognizing the commonality among different designs from different domains and attempted to find and

formulate design rules regardless of domains The recognition and attempt seek out insights into the design process itself or “the art and science of design” (Simon 1995)

For example, in system design, people have found that in the structure of hierarchical systems the relationship between the inner subsystems and their goals are maintained

largely independent of variations in the outer environment or other subsystems This

property has long been known by biologists under the label of homeostasis It provides a key system architectural advantage and is an important property of most good designs (Simon 1981) Other research suggests that the inherent organization of complex non-biological

1

This distinguishes itself from the systems designed by nature Nature builds complex systems, yet in quite a different manner How nature designs or how nature lets systems evolve is without a

performance or functionality goal in mind -nature doesn’t have a mind (Abbott 2007) Nature gets

designs from putting together existing pieces (or somewhat random variants of existing pieces) by

principle of survival of the fittest.

systems displays the same topologic scaling properties demonstrating striking similarities

and complying with the design principles to metabolic networks, despite of their significant variances in individual constituents and pathways (Barabasi and Albert 1999, Jeong et al

2000)

2.3 Uncertainty -A key issue in project and system design:

Uncertainty is a foundational issue in design Many changes will inevitably happen along the life time of systems, while decisions regarding how to design have to be made an ante in the planning phase before the changes take place

There is no consensus on the definition of uncertainty In the field of economics, Knight separated out risk and uncertainty, calling risk as situations that probabilities can be

assigned to events, while uncertainty as situations cannot be expressed in terms of a specific probability (Knight 1921) In the field of system design, uncertainty is predominantly

characterized in terms of probability and statistics (DeLaurentis and Mavris 2000) A general sampling of the definitions and classifications of uncertainties based on the contexts and

sources is given in Table 2.1

Table 2-1 Definitions and classifications of uncertainties -adapted from DeLaurentis and

phenomena or processes

• Natural (inherent to physical process)

• Model (inability to perfectly model nature)

• Input (stochastic inputs)

• Measurement/data transfer and manipulation

• Operational/environmental Zhao

(1995)

Controls Uncertainty refers to the

differences of errors between models and reality

• Unstructured, representing those that

is un-modeled or not possible to model

• Structured, representing those for which information on the likely behavior

is available Hazelrigg

• Insufficient knowledge of the laws of nature

• Inability to assess or measure a

element with non-zero probability in an experiment,

phenomenon, initial or boundary conditions

• Inherent randomness of a physical process

• Statistical (results from incompleteness

of statistical data, e.g too small sample size)

• Modeling (use of simplified analysis models)

• Human error Ober-

• Uncertainty- due to incomplete information

• Error- recognizable deficiency in modeling/simulation that is not due to lack of knowledge

• Variability- Inherent variation DeLauren-

• Incomplete knowledge in decision making

Synthesized from an array of definitions of uncertainty, DeLaurentis and Mavris (2000) provided a general and consistent definition that uncertainty represents the incompleteness

in knowledge (either in information or context) that causes model-based predictions to

differ from reality in a manner described by some distribution functions This study adopts this definition, and uses uncertainty to describe something that is known probabilistically

but not exactly In order to better understand and characterize uncertainty, this study also classifies uncertainty by its functional areas, such as:

• change in requirements

• institutional change

• technological discontinuities (Tushman and Anderson 1986, Funk 2008b)

• change in (trans-boundary) dependencies

• change in complexity (structural/functional)

• change in priorities, such as between short term gains and long-term sustainability issues (Thissen and Herder 2003)

• change in economic environments

Uncertainty, organizationally, often falls to the duty of the managers They set targets amid uncertainty and try to achieve the targets at the least possible cost Modern managers often rightfully allocate resources to carry out risk analysis, but risk analysis is often done ex post and in silo by different groups of professionals and not integrated to the design process This disconnection leads to technically optimized, yet inflexible designs, incapable of

adjusting to changes in the environments they are built for

For example, the tunnel between Honshū and Hokkaidō was built based on the need at the time of design It was designed with the 1971 traffic predictions, which turned out to be overestimates as air transport facilities improved and deregulation and competition stepped

up in Japanese domestic air travel As a result, the cost of rail using the tunnel becomes

more expensive than air travel and the usage of the tunnel that cost US$ 3.6 billion turned out to be low (Takashima 2001) A decade later, another tunnel project at British strait

channel again failed to account for uncertainty The actual construction costs of the tunnel doubled the forecasted (Finnerty 1996)

Many large-scale systems are important to our society and public users, and the

consequences of uncertainty are well documented in public media, such as in a commentary

in New York Times by Ada Huxtable: “the textbook scientistic-utopian planning of long-range policies based on statistical extrapolations and translated into massive rebuilding schemes has proved such a conspicuous failure in the last 25 years”

Fortunately, there are means to reduce and deal with uncertainty

2.3.1 Means to reduce uncertainties

There are two ways to reduce uncertainty,