You are on page 1of 76

REAL OPTIONS ANALYSIS IN HYDRAULIC ENGINEERING

NGUYEN TAN THAI HUNG

DEPARTMENT OF CIVIL ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE

2008/2009
REAL OPTIONS ANALYSIS IN HYRAULIC ENGINEERING

NGUYEN TAN THAI HUNG

A THESIS SUBMITTED

FOR THE DEGREE OF BACHELOR OF ENGINEERING

DEPARTMENT OF CIVIL ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE


ACKNOWLEDGEMENTS

First and foremost, the author would like to express his deepest gratitude to his supervisor,

Associate Professor Vladan Babovic, for his invaluable guidance and encouragement. It

would have been much harder to complete this project without the advice and motivation

from Professor Babovic.

Secondly, the author feels grateful to Mr. Zhang Xu, Stephen of the Singapore-Delft

Water Alliance for his consistent and helpful guidance during the last two semesters, as

well as his valuable comments and suggestions on the programming and debugging of the

MATLAB model.

Appreciation is also extended to Ms. Sally Teh, Mr. Albert Goedbloed and the Singapore-

Delft Water Alliance for the privileges that the author was provided during the period of

the project.

Acknowledgements are also attributed to those who have contributed to this research in

one way or another, and also to the authors of various papers and materiel referred to in

this thesis.

Last but not least, the author would like to thank his family for their support throughout

his years of education and to his many friends who had helped him in school works and

who had made his years in university memorable ones.

i
TABLE OF CONTENTS

ACKNOWLEDGEMENTS .................................................................................................. i

TABLE OF CONTENTS ..................................................................................................... ii

SUMMARY ......................................................................................................................... v

NOMENCLATURE............................................................................................................ vi

LIST OF TABLES ............................................................................................................ viii

LIST OF FIGURES.............................................................................................................. x

1. INTRODUCTION ........................................................................................................ 1

1.1 Background .......................................................................................................... 1

1.2 Objectives and scope of project ........................................................................... 1

2. LITERATURE REVIEW ............................................................................................. 3

2.1 Financial options .................................................................................................. 3

2.1.1 Definitions........................................................................................................ 3

2.1.2 Types of options ............................................................................................... 3

2.1.3 Examples .......................................................................................................... 4

2.2 Real options.......................................................................................................... 5

2.3 Monte Carlo simulation........................................................................................ 6

3. CASE STUDY – PARKING GARAGE....................................................................... 7

3.1 Is average a good approach? ................................................................................ 7

3.2 Dealing with uncertainties.................................................................................... 8

3.3 Case description ................................................................................................... 9

3.4 Planning with traditional Net-Present Value (NPV) model: ................................ 9

3.4.1 The model......................................................................................................... 9

3.4.2 Sensitivity analysis - What if demand changes? ............................................ 10

ii
3.5 Planning with randomized demand .................................................................... 13

3.5.1 Changing demand projection factors.............................................................. 13

3.5.2 Sensitivity analysis ......................................................................................... 15

3.6 Considering expansion option............................................................................ 17

3.7 Conclusion for the case study............................................................................. 18

4. SINGAPORE WATER – A PICTURE ...................................................................... 20

4.1 Demand .............................................................................................................. 20

4.1.1 Domestic demand ........................................................................................... 20

4.1.2 Industrial demand ........................................................................................... 24

4.2 Supply................................................................................................................. 24

4.2.1 The national four-tap model........................................................................... 25

4.2.2 The potential "fifth tap".................................................................................. 29

5. MONTE CARLO SIMULATION MODEL............................................................... 31

5.1 Demand model ................................................................................................... 31

5.1.1 Domestic consumption model........................................................................ 31

5.1.2 Total consumption model............................................................................... 34

5.2 Supply model...................................................................................................... 34

5.2.1 Configuration and risk premium .................................................................... 34

5.2.2 Imported water modelling .............................................................................. 35

5.2.3 Cost and payment modelling.......................................................................... 36

5.2.4 "Unit Tariff per Output" modelling................................................................ 40

5.2.5 "Two-part tariff" modelling............................................................................ 44

5.3 Risk modelling ................................................................................................... 44

6. RESULTS AND DISCUSSIONS............................................................................... 46

6.1 “Unit tariff” vs. “two part tariff” ........................................................................ 46

iii
6.2 Effects of the “fifth tap” ..................................................................................... 46

6.2.1 Mean NPV...................................................................................................... 46

6.2.2 Water security risks and water scarcity risks ................................................. 47

6.3 Effect of risk premium ....................................................................................... 48

6.4 Effect of β and m................................................................................................ 50

6.5 Recommendation for future researches.............................................................. 50

REFERENCES................................................................................................................... 51

APPENDIX A – CASE STUDY RESULTS ..................................................................... 53

APPENDIX B – POPULATION PROJECTION RESULTS ............................................ 56

APPENDIX C– SIMULATION RESULTS ...................................................................... 59

iv
SUMMARY

This dissertation aims to provide a framework for the planning of Singapore’s water

supply system in 40 years, using real options analysis with Monte Carlo simulation.

A hypothetical case study on a parking garage is conducted to examine the advantages of

real options analysis over the traditional Net-Present-Value model. The case study

remarked the value of flexibility, the ability to wait for more information before making

decision, and the sensitivity analysis is a guide to which kind of information worth

attending to. The case study also emphasized the strength of simulation in capturing

reality.

The review of Singapore’s water issues pointed out that there are a lot of uncertainties, a

lot of potential risks and opportunities that require a flexible management.

The Monte Carlo simulation model involving Singapore’s four national taps and a

potential fifth tap suggested that LNG-desalination is a valuable option with high chance

of exercising. The simulation model also studied the effect of different management

approaches and noted that flexibility is not always in favour. The model also examined the

effect of risk premium on imported water.

Keywords: real options, Monte Carlo simulation, MATLAB, water supply system, four-

tap model, expansion option, flexibility, risk premium

v
NOMENCLATURE

β Proportion of break-even point volume to total capacity

DBOO Design-Build-Own-Operate

Df Final demand

DPM Deputy Prime Minister

D(t) Demand at time t

E Excessive amount of demand to supply

FC Fixed Cost

I Investment cost

kF Fixed Cost coefficient

kI Investment cost coefficient

LNG Liquid natural gas

m Economies of scale factors

O&M Operation & Maintenance

NPV Net Present Value

P The averaged unit cost of producing a unit volume of water

PM Prime Minister

PUB Public Utilities Board

pti Technical cost at year i

vi
pfi Energy cost at year i

Q Total capacity

q Break-even point volume

qe Expansion amount for capacity

qr Reduction amount for capacity

RO Reverse-Osmosis

RSP Required selling price

RV Revenue

S Saving

s Individual water consumption

TC Total Cost

TFR Total Fertility Rate

U Tap usage

u Unit production cost

x number of days Sinagpore can survive without imported

water

y year

vii
LIST OF TABLES

Table 3.1 – Summary of parking garage problem................................................................ 9

Table 3.2 - Sensitivity results for demand projection factors ............................................ 12

Table 3.3 - Mean NPV of randomized demands ................................................................ 14

Table 3.4- Sensitivity results for initial demand’s volatility.............................................. 15

Table 3.5 - Sensitivity with respect to first year's volatility and with expansion option ... 17

Table 3.6- Case study summary ......................................................................................... 19

Table 4.1 - Domestic and individual water consumption................................................... 24

Table 4.2 -Break-down of Singapore's water consumption ............................................... 24

Table 4.3 - Major reservoirs............................................................................................... 26

Table 5.1 - Scenarios for 2011 Agreement ........................................................................ 35

Table 5.2 - Scenarios for 2061 Agreement ........................................................................ 35

Table 5.3 - Cost components.............................................................................................. 36

Table A.1 - Sensitivity results, demand by year 10 ........................................................... 54

Table A.2 - Sensitivity results, demand after year 10 ........................................................ 54

Table A.3 - Sensitivity results, volatility in year 10 demand ............................................. 55

Table A.4 - Sensitivity results, volatility in demand after year 10..................................... 55

Table B.1 - Resident population projection ....................................................................... 56

Table B.2 - Non-resident population projection................................................................. 56

Table B.3 - Total population projection ............................................................................. 57

Table B.4 - Age-gender-specific population projection..................................................... 58

Table C.1 - Comparison between Unit tariff and two-part tariff, m = 1, β = 0.5............... 60

Table C.2 - m = 0.5, β = 0.5, two-part tariff....................................................................... 61

Table C.3 - Comparison between β = 0.5 and β = 0.75 (Unit tariff).................................. 62

viii
Table C.4 – Comparison between big and small LNG capacity ........................................ 63

ix
LIST OF FIGURES

Figure 3.1 - The flaw of averages (from Savage, 2000) ...................................................... 7

Figure 3.2 - Hypothetical demand projection..................................................................... 10

Figure 3.3 - Different demand curves associated with different estimation of initial
demand ............................................................................................................................... 11

Figure 3.4 - Sensitivity results for demand projection factors ........................................... 12

Figure 3.5 - A typical demand scenario ............................................................................. 14

Figure 3.6 - Sensitivity results for different volatility........................................................ 16

Figure 4.1 - Average monthly water bill, inclusive of tax (in S$) 1980 - 2005 ................. 22

Figure 4.2 - Domestic Sector Breakdown, Singapore........................................................ 23

Figure 4.3 - The blue map of Singapore............................................................................. 25

Figure 4.4 - Contribution of four national taps with maximum capacity for local taps..... 29

Figure 5.1 - Individual water consumption model ............................................................. 33

Figure 6.1 – Effect of enlarging LNG capacity.................................................................. 47

Figure 6.2 - Mean NPV result ............................................................................................ 48

Figure 6.3 - Partial NPV curve from $0.00 to $0.50.......................................................... 50

Figure A.1 - Best, base and worst cases, demand by year 10 ............................................ 53

Figure A.2 - Base and worst cases, demand by year 10..................................................... 53

Figure A.3 - Base and best cases, demand by year 10 ....................................................... 54

Figure A.4 - Effect of volatility on net present value......................................................... 55

Figure C.1 - Unit tariff, m = 0.5, β = 0.5............................................................................ 59

Figure C.2 - Two part tariff, m = 1, β = 0.5 ....................................................................... 59

Figure C.3 - Unit tariff, m = 1, β = 0.5............................................................................... 60

Figure C.4 - Demand and supply spectra (100 simulation runs)........................................ 62

x
Figure C.5 - A typical random walk for desalination variable costs.................................. 63

xi
1. INTRODUCTION

1.1 Background

In project evaluation, the traditional discounted cash flow model, for which decision

making is based on net present value (NPV), has been widely practiced. However, this

model is a rigid approach and is incapable in risky projects that involve a high level of

uncertainty such as water supply system planning.

A better approach

1.2 Objectives and scope of project

This dissertation aims to provide a framework for the planning of Singapore’s water

supply system in 40 years, using real options analysis with Monte Carlo simulation.

The thesis consists of:

• a review on the theory of real options;

• a case study to demonstrate the advantages of the real options analysis compared

to traditional NPV;

• a summary of the current picture of the Singapore water supply and the

uncertainties involved;

• and an application of real options analysis to the planning of Singapore water

supply system.

1
As will be shown in the thesis, the planning for Singapore’s water supply system involves

not only many risks and uncertainties but also a lot of potential. As a framework, the

thesis will demonstrate how real options analysis can solve this problem, how it is better

compared to traditional method. To achieve that, a decision making model, involving

demand, supply and some expansion and reduction rules will be built and tested. This

thesis will be a start for subsequent researches towards a reliable decision making solution

for the planning works.

The time horizon for this project is 40 years from 2008 to 2047.

The programming language used in the case study is Microsoft Excel, and that for the

main model is Mathworks MATLAB.

2
2. LITERATURE REVIEW

2.1 Financial options

Since real options employs financial options theory but applies for real assets (hence the

name "real options"), it is beneficial to start our options understanding by going through

financial options theory.

2.1.1 Definitions

According the Australian Stock Exchange (2004), an option is a right, but not the

obligation, of the option holder to buy or sell a security with a specific price, called the

exercise or strike price, on or before a specific date in the future, called the expiry date.

Technically speaking, an option is a contract between two parties: the buyer of the right

(called the taker) and the seller (called the writer). The price for having this right is called

the option premium. Note that in options trading, the buyer and seller are not buying or

selling the underlying security directly. They are trading the right to buy or sell the

security. Clearly, there are two types of options: options to buy (call options) and options

to sell (put options).

2.1.2 Types of options

A call option gives the taker the right to buy the underlying security at an agreed price.

When the taker chooses to exercise the right, which means he decides to buy the security,

the writer has to sell the security at the agreed price, even if it is higher than the market

price. It is important to note that the taker does not have to exercise this right. Obviously,

the writer will choose to do so when the market price is higher than the agreed price.

3
Otherwise, if the market price is lower than the agreed price, the writer would rather buy

from the market. If the two prices are the same, the writer is indifferent.

Similarly, a put option gives the taker the right to sell the underlying security at an agree

price, and the taker has to buy the security once the right is exercised. Obviously, the taker

will sell the security to the writer when the agreed price is higher than the market price,

otherwise he would prefer to sell to the market. If the two prices are the same, the taker is

indifferent.

2.1.3 Examples

Example of call option: suppose share price of a particular firm is $5.00 each. If you

expect the price to rise, you take a $100 option to buy 1000 shares at $5.20 each in 3

months.

If 3 months later, share price rises to $5.50, and you buy at $5.20, you have made a profit

of ($5.50 - $5.20) x 1000 = $300, less the option premium $300 - $100 = $200. So, you

have achieved a capital gain of $200.

If, unfavourably, share price falls to $4.80, you will not buy the shares from the writer and

lose the option premium of $100. Should you have bought the share directly 3 months

earlier, you would have lost ($5.00 - $4.80) x 1000 = $200. So, the option limits your loss

to only $100.

Example of put option: suppose share price of a particular firm is $5.00 each. If you

expect the price to fall, you take a $100 option to sell 1000 shares at $4.80 each in 3

months.

4
If 3 months later, share price falls to $4.50, and you sell at $4.80, you have made a profit

of ($4.80 - $4.50) x 1000 = $300, less the option premium $300 - $100 = $200. So, you

have gained a profit of $200.

If, unfavourably, share price rises to $5.20, you will not sell the shares to the writer at

$4.80, and lose the option premium of $100. Should you have sold the share directly 3

months earlier, you would have suffered a capital lost ($5.20 - $5.00) x 1000 = $200. So,

the option limits your loss to only $100.

The examples show that options are a type of risk mitigation that limits loss over a

security to the option premium.

2.2 Real options

According to Wang (2003), Myers (1987) was one of the first authors to point out the

limitations of the traditional discount cash flow model. Then, Dixit and Pindyck (1994)

noted the irreversibility of investment decision making and the uncertainties involved.

Trigeorgis (1996) was attributed to have collected scattered knowledge about real options.

A few other authors wrote some introductory books and finally, the theory develops

rapidly at the beginning of this century. So, the theory of real options has just had a firm

ground in about 10 years.

Similar to financial option, a real option is the right (and not the obligation) to execute a

project. Unlike financial options which involve intangible assets (shares, securities), real

options often involve tangible assets: a factory, a patent etc – hence the name real options.

For example, a factory manager may have the right to expand his factory to meet new

productivity level. This right may be gained through a land acquisition - purchasing the

5
parcel of land next to his factory. With this parcel of land in possess, the factory manager

is technically taking a call option. He paid the option premium - the land acquisition cost.

Once market for his product turns good, he will exercise the option, expand the factory

and produce more products. If the market does not turn into his favour, he is not obliged to

expand the factory still.

Hence, with real option, the decision maker can wait until more information is available,

and then choose to execute if the situation is in favour, or abandon if not. The price for

this right to wait is the option premium, which is much smaller than the project execution

cost. Should the decision maker opt to abandon the project, the most loss suffered would

only be this premium, not the entire execution cost.

There are many ways to evaluate an option, the most common being Black-Scholes

model, decision tree analysis, and Monte Carlo simulation.

2.3 Monte Carlo simulation

Monte Carlo simulation is a process in which a large amount of random inputs (with an

assumed statistical distribution) are generated, and through a known model, converted to

outputs. It has some advantages over an analytical (mathematical) model. Usually an

analytical model requires simplification of the problem while a simulation can describe

the problem closely. As inputs are generated, it helps to save data acquisition time.

However, a simulation for a large problem may require significant computer resources

(Mun, 2006).

6
3. CASE STUDY – PARKING GARAGE

3.1 Is average a good approach?

In daily life circumstances, we tend to take an average of a quantity. For example, when

different clocks show different times and a standardized clock is unavailable, we take the

average of them.

However, an average is not always a good approach, especially when the quantities are

quite scattered, or even when there are some adverse situation where something might

happen far from the average. This situation is illustrated the case of a scientist who

drowned in a river that is of “average” 3 feet deep (see Figure 3.1) (Savage, 2000).

Figure 3.1 - The flaw of averages (from Savage, 2000)


Cartoon by Jeff Danziger

7
A more technical example is the case of a car parking garage at a shopping centre. The

investor of the garage predicts that the average demand for parking in a particular area in

the next few years will be 750 spaces per year. He builds a facility to cater for 750 cars. In

the bad years when car parking demand is less than 750 spaces, he is not able to utilize

full garage capacity and may make a loss. In the good years when parking demand is more

than 750 spaces, he does not have sufficient facility to meet the entire demand. So he

cannot make extra profit to cover the bad years. Thus, even though demand averages up to

his correct estimate of 750 spaces per year, profit cannot average up. An average approach

of input does not lead to the average output. This is an expression of the so-called “The

Flaw of Average” (Savage, 2000).

3.2 Dealing with uncertainties

The moral of the story is that we need to pay attention to uncertainty. For the case of the

parking garage, demand is the main uncertainty. For various reasons, it is virtually

impossible to predict exactly how many parking slots will be demanded in a future year.

The Figure depends on how many people visiting the shopping centre, which in turn is

affected by economic conditions, consumers’ behaviour and other uncertainties. In

additions, it also changes due to infrastructure conditions such as public transport

availability, the overall development of the town in which the shopping centre is located

etc. In short, demand is an uncertainty input for the planning and using the average is not a

good strategy, as shown above. Therefore, a better strategy for decision making must be

adopted. In the following case study, we will examine the matter in greater details, with

numerical results.

8
3.3 Case description

The following case is adopted from de Neufville and Scholtes (2004).

Table 3.1 – Summary of parking garage problem

Inputs
Demand in year 1 750 spaces
Additional demand by year 10 750 spaces
Additional demand after year 10 250 spaces
Average annual revenue $5,000 per space used
Average operating costs $1,000 per space available
Land lease and other fixed costs $1,800,000 p.a.
Capacity cost $8,000 per space
10% growth per level for every level above 2
Capacity limit 200 cars per level
Time horizon 15 years
Discount rate 12%

An investor plans to construct a parking garage next to a commercial building in a

developing area. He forecasts that the immediate demand right after the car park is

constructed is 750 spaces and there will be a subsequent 750 spaces needed over the next

10 years. The revenue and cost associated with running the car park is summarized in

Table 3.1.

The site is large enough to accommodate 200 cars per level.

3.4 Planning with traditional Net-Present Value (NPV) model:

3.4.1 The model

In this model, demand is predicted as an exponential curve as shown in equation 3.1

D (t ) = D f − α e − β t (3.1).

9
The hypothetical demand curve is expressed graphically in Figure 3.2.

Figure 3.2 - Hypothetical demand projection

Based on this demand projection, the revenue and cost for each year is then calculated.

After that, the net present value of the whole investment is computed. The process is

repeated for each different capacity of the garage in order to find out the best investment.

For this particular problem, building 6 levels is the best investment strategy, with an NPV

of $3,119,208.

3.4.2 Sensitivity analysis - What if demand changes?

As mentioned above, it is very unlikely that demand will be the same as prediction due to

its high level of uncertainty.

The investment’s outcome depends heavily on demand, which is an interpolation based on

three factors, the value of which are the investor’s estimation.

- Initial demand at year 1

- Demand at year 10

10
- Additional demand after year 10

Now, the investor expects that each of the above factors can be 50% higher or lower than

his prediction. The former is called the “best case”, the latter the “worst case” and his

prediction is called the “base case”. The investor carried out a sensitivity analysis in which

each of the factors is set as their best and worst cases in order to compare with their base

case. The different demand curves associated with different estimation of initial demand is

shown in Figure 3.3. For the complete set of Figures for all different values of the factors,

see Appendix A – Case study results.

2500

2000
Demand (spaces)

1500
Base case

1000 Worst case


Best case
500

0
0 5 10 15 20 25
Year

Figure 3.3 - Different demand curves associated with different estimation of initial
demand

The different estimations and results are summarized in Table 3.2.

11
Table 3.2 - Sensitivity results for demand projection factors

Additional by Year Additional demand


Year 1
10 after Year 10
Spaces NPV Spaces NPV Spaces NPV
Worst case 375 -$5,575,322 375 -$649,900 125 $3,528,689
Base case 750 $3,118,834 750 $3,118,834 250 $3,118,834
Best case 1125 $6,929,612 1125 $4,349,167 375 $2,865,465

The change in NPV with respect to different values of demand factors is plotted in Figure

3.4. It can be seen that the NPV changes for year 1 demand is significantly large, for year

10 is moderate and for demand after year 10, the change in NPV is very small.

$8,000,000

$6,000,000

$4,000,000

$2,000,000 Year 1
NPV

$0 Year 10
Worst case Base case Best case After year 10
-$2,000,000

-$4,000,000

-$6,000,000

-$8,000,000

Figure 3.4 - Sensitivity results for demand projection factors

Each factor affects demand in a different way. It is seen that the most sensitive factor is

the initial demand. This is reasonable because the demand curve is very steep at first and

12
getting flatter towards the future. So any change in the early proportion of the curve will

affect the entire curve more significantly than the latter part.

3.5 Planning with randomized demand

3.5.1 Changing demand projection factors

We have seen different possible results for the best cases and worst cases. So which

Figure should the investor use for his project? Should he take an aggressive one and hope

for the best, or should he be defensive and prepare for the worst? Can he take somewhere

in between? If so, which one?

In real life, demand can be much different from the predicted figures, which leads to

various possible outcomes, some of which are disastrous. In other words, the predicted

demand is only an average of the many possible situations. Therefore, the reliability of the

analysis based on this average demand is questionable.

To capture a more realistic picture, a randomized demand model is created.

Firstly, the investor expects demand to rise or fall 50% away from his prediction, with 5%

volatility in demand growth. Figure 3.5 shows a typical demand scenario, among the

various possible.

13
Demand

2400
2200
2000
1800
Demand (spaces)

1600
1400
Demand pro jectio n
1200
Demand scenario
1000
800
600
400
200
0
1 3 5 7 9 11 13 15 17 19

Tim e (years)

Figure 3.5 - A typical demand scenario


A Monte Carlo Simulation is programmed in Excel and performed with 100,000 runs. The

mean NPV for different capacities are shown in Table 3.3.

Table 3.3 - Mean NPV of randomized demands

Capacity Mean NPV Static NPV


4 -$38,200 $615,996
5 $1,681,600 $2,615,732
6 $1,996,800 $3,119,208
7 $620,100 $2,022,130

Although the optimum capacity is still 6 levels, the expected net present value is much

smaller than that in the static model ($2.0 million versus $3.2 million). This represents the

flaw of the average: “The result of average of input is not necessarily the same as the

average results of input” (Savage, 2000). In this case, when demand is higher than the car

14
park’s capacity, the profit is not realized. Therefore, the mean NPV is lower than that from

the static model, which did not capture this characteristic of demand.

3.5.2 Sensitivity analysis

In the above simulation, demand is expected to fall 50% on either side of the projection.

This percentage indicates how reliability the estimation is. What if there is more

information which leads to a more convincing projection, so that demand will only fall

within 10% away from its expected value? Or what if development in the area becomes

volatile and demand can differ as much as 75% away from the projection? The sensitivity

of the expected NPV with respected to different volatility levels is examined through

simulations with various values for the demand estimation parameters. The results with

respect to initial demand at year 1 are shown numerically in Table 3.4 and graphically in

Figure 3.6. For entire results, see appendix A.

Table 3.4- Sensitivity results for initial demand’s volatility

Capacity 10% 25% 50% 75%


4 $594,741 $288,809 -$38,200 -$711,709
5 $2,438,317 $2,418,779 $1,681,600 $710,959
6 $2,720,151 $2,434,126 $1,996,800 $1,085,276
7 $1,344,570 $1,216,553 $620,100 -$173,881

15
Sensitivity - Demand at year 1
$4,000,000

$3,000,000 4 levels

$2,000,000 5 levels
6 levels
$1,000,000
7 levels
$0
0% 20% 40% 60% 80%
-$1,000,000

Figure 3.6 - Sensitivity results for different volatility

As we can see in the results, the higher the volatility, the less the expected profit. In

addition, the slopes of the NPV curves are steeper towards the larger volatility, which

means that NPV is more sensitive with volatility as it grows larger. This is because the

higher the volatility, the smaller demand can be compared to the projection, hence the

larger unused capacity there is. On the other hand, there is also higher chance that demand

will be much higher than expected, but there are not enough spaces to be utilized then.

Therefore, “the flaw of averages” is more prominent in markets with high volatility.

The results show that with different volatility, year 1 demand is still the most sensitive

factor, followed by year 10 demand and that after year 10. In addition, expected NPV also

decreases with increasing volatility. This is in alignment with the analysis we have

mentioned above.

This information is very valuable since it guides the investor as where should he put effort

to find out more market information.

16
We also examine the sensitivity of NPV with respect to the volatility in demand growth

while keeping those of demand projection constant as 50%.

3.6 Considering expansion option

It is not necessary that the investor must have a big garage right from the start to cater for

the demand in 20 years. In fact, he may choose to build a small garage in the early years

when demand is small, and expand one or a few more levels in the subsequent years when

demand increases.

A simulation with this expansion option is performed with 100,000 runs. In this model,

the investor will build additional floors when demand has exceeded capacity in the

previous two years. Sensitivity analysis is also incorporated into the model. The mean

NPV corresponding to 50% volatility of all demand factors is $5,309,424, which is

significantly higher than both static NPV and randomized demand models. The common

decision rule is to build 4 levels first and expand 1 each time. The sensitivity results in

regards of first year demand’s volatility is shown in Table 3.5.

Table 3.5 - Sensitivity with respect to first year's volatility and with expansion option

Volatility 10% 25% 50% 75%


Mean NPV $5,937,170 $5,832,101 $5,309,424 $4,521,155
Initial levels 4 4 4 4
Additional levels
1 1 1 1
each expansion

As expected, the common strategy is to start with a small capacity and build up additional

capacity later rather than going for a big facility right at the start.

17
The results are significantly higher than those obtained without the expansion option. This

is because with the option, the investor is able to capture the whole demand curve. When

it was low, he has a small facility so he does not suffer lost from unused spaces and when

it grows, he is able to expand his capacity to cater for that demand.

We can also see that even though the NPV is still sensitive to the volatility in the same

way as in the case without expansion option (i.e. most sensitive to demand at year 0 and

least to demand after year 10), the magnitude of sensitivity is less.

High volatility still decreases the profit. As volatility is high, the good period is short and

hence cannot totally make up for the bad period.

Without the expansion option, the investor has to make decision before the project is

implemented. Thus, he does not have sufficient information about future demand. When

he has the option to expand, he can choose to wait until more information is available and

his decision is more accurate. He is faced with less risk and hence the volatility has less

effect.

3.7 Conclusion for the case study

From the results of the three different models (static NPV, randomized demand without

expansion option, randomized demand with expansion option), it can be seen that the

model with expansion option is the best approach, being the most realistic model and

yielding the highest NPV (see Table 3.6).

18
Table 3.6- Case study summary

Model NPV Decision rule


Static NPV $3,119,208 Build 6 levels (and maintain)
Randomized
$1,996,800 Build 6 levels (and maintain)
demand
Build 4 levels and expand 1 level when demand is
With expansion
$5,309,424 higher than supply for 2 consecutive years (can end
option
up with 4, 5, 6 or 7 levels)

The different NPVs resulted from different models reflect the value of information

(demand in this case), thus they reflect the value of the option to wait and decide when

there is more information. Therefore, it is the amount the investor is willing to pay for the

right to wait for more information.

In this case the expansion option offers the investor more flexibility in making his

decision. Rather than making a single decision, he can, from time to time, make

continuous decisions when the situation changes and more information is available. This

is very important when there is a lot of uncertainty and the market is volatile. In such

circumstances, the investor is able to response correctly. If the demand is not as good as

expected, he can choose not to exercise the right to expand. In this case what he has lost is

just the option premium, rather than the construction cost of the excessive facility.

The case study also showed that the option’s value depends on the volatility of the market.

The higher the volatility is, the higher the option’s value is. However, it has different

sensitive level with respect to different factors, which guides the decision maker on where

to put efforts to find more information.

19
4. SINGAPORE WATER – A PICTURE

4.1 Demand

Currently, Singapore’s demand for water is 1.3 million m3/day. Its total land area is 697

km2 with a population of 4.5 million and annual rainfall of 2400mm (Tan, 2007).

Singapore's water demand comprises of domestic and non-domestic sectors (Table 4-2).

Since domestic demand is the larger proportion and involves more uncertainties, it will be

viewed as the product of population and individual water consumption, and will be

examined down to that level.

4.1.1 Domestic demand

4.1.1.1 Population

The more people the more the water consumption. It is important to study the future

population in order to plan for water supply.

As of June 2008, the population of Singapore is 4,839,400, of which 3,642,700 are

residents (Department of Statistics).

The population growth rate is estimated as 1.135% (CIA World Fact Book, 2008) but

there are many uncertain factors that obstruct population prediction.

The high proportion of non-resident population (1,196,700 people) which contributes

24.7% of the total population indicates a fluctuation to some extent of the total number of

people living on the island. There are a large flow of transient foreign workers.

20
On average, there are 52,500 PRs and 12,800 new citizens granted every year from 2004.

In 2007 there were 63,600 PRs and 17,300 new citizens (DPM, 2008).

A general picture of the population is sketched out by PM Lee in his National Day Rally

Speech. There are significant decreases in the population trend. Total Fertility Rate (TFR),

which is the average number of children per woman over her life time, is only 1.29, far

lower than the replacement level of 2.1. When times are hard, people tend to have fewer

babies and there were more babies born in the years of the dragon (Lee, 2008). The recent

financial crisis has great impact on Singapore and this can strongly affect the population in

the near future.

In order to increase the fertility rate, the government is encouraging couples to get married

and have children. They have created special agencies for this purpose. In addition, a

series of policies such as child care leave, extended maternity leaves, baby bonuses are

and will be implemented (Lee, 2008).

The effectiveness of these policies will affect the population and thus the water demand.

4.1.1.2 Individual water consumption

As a result of appropriate demand management, individual water consumption has kept

decreasing from 176 lit/day in 1994 to 158 lit/day in 2006 (Tan, 2007).

Tortajada (2006) remarked that various demand management schemes from PUB such as

water tariff and water conservation tax has helped to convey and reinforce the water

conservation message to the households. Average monthly bill, in general, has been

increasing in the period 1980 - 2005 (Figure 4.1). Consequently, there are disincentives

for households to consume more water.

21
Figure 4.1 - Average monthly water bill, inclusive of tax (in S$) 1980 - 2005
(from Tortajada, 2006)

Beside increasing tax, a better tariff structure involving a targeted subsidy for the poor and

eliminating the case that commercial and industrial users subsidize domestic users, and a

penalty for households that use more than 40m3/month, helped to address the problem

without creating pressure on those who cannot afford the high water bill to meet their

basic needs. Campaigns to raise awareness such as the "10 litre a day" programme also

contributed to the decrease in consumption per capita. The fact that the average monthly

bill slightly decreased in the period 2000-2005 (Figure 4.1), but the consumption per

capita also decreased in that period has proved the success of the demand management

scheme.

Individual water consumption can be further reduced by targeting its major components

such as shower and flushing system (Figure 4.2).

22
Figure 4.2 - Domestic Sector Breakdown, Singapore

In Australia, water-efficient showerhead and smart flushing system (including no-flush

system) have been used in order to target the two large segments of water demand. It is

possible that these measures will be adopted in Singapore. In fact, the no-flush system has

been adopted in ViVo City, one of the largest shopping centers on the island.

4.1.1.3 The results

As individual consumption declines, the effect of increasing population on domestic water

consumption is less prominent (see Table 4.1).

23
Table 4.1 - Domestic and individual water consumption
Year 2002 2003 2004 2005 2006 2007
Population 4,176,000 4,114,800 4,166,700 4,265,800 4,401,400 4,588,600
Domestic water
consumption 687 690 686 694 702 725
3
(1000 m /day)
Individual 165 168 165 163 159 158

4.1.2 Industrial demand

Industrial water demand has remained about 43% of the total water demand.

Table 4.2 -Break-down of Singapore's water consumption

2002 2003 2004 2005 2006 2007


Domestic 687 690 686 694 702 725
Industrial 572 534 517 512 528 521
Total 1259 1224 1203 1206 1230 1246
% Industrial 45.43% 43.63% 42.98% 42.45% 42.93% 41.81%

4.2 Supply

Singapore’s water supply system is solely managed by the Public Utilities Board (PUB).

The Board has envisioned Singapore to be the ASEAN hydro hub, and has conFigured the

nation’s water supply system with four “national taps”: local catchment, imported water

from the State of Johor, Malaysia, NEWater, and desalinated water (PUB).

24
4.2.1 The national four-tap model

4.2.1.1 Water catchment

With the completion of the Marina Barrage in 1 November 2008, two-third of the island’s

area is now catchment area (PUB) (See Figure 3-2. In 2005, around 680,000 m3 of water

was drawn from these catchment areas (Lee, 2005).

Figure 4.3 - The blue map of Singapore


(from Young, 2008)

Some of the major reservoirs and their capacities are listed in Table 4.3.

25
Table 4.3 - Major reservoirs

Name of reservoir Year completed Storage capacity (million m3)


MacRitchie 1867 (*1894) 4.2
Lower Pierce 1912 2.8
Selatar 1935 (*1969) 24.1
Upper Pierce 1974 27.8
Kranji/Pandan 1975 22.5
Western Catchment 1981 31.4
Bedok/Sungei Selatar 1986 23.2
Total 142.0
* Year when the reservoir is enlarged

There are totally 15 reservoirs in commission, and 2 more in progress (PUB). Therefore, it

can be predicted that more water will be drawn from the reservoirs in the future.

4.2.1.2 Imported water

About 40% of Singapore’s water supply is imported from Malaysia, following the 1961

and 1962 Water Agreements that end in 2011 and 2061 respectively. The first agreement

allows Singapore to draw 391,000 m3 of raw water per day from Malaysia. The second

one in allows Singapore to draw 1,136,500 m3 of raw water per day from Johor, but it also

requires Singapore to sell 168,200 m3 of treated water per day back to Johor (Lee, 2003),

hence the maximum capacity is 1,359,300 m3/day.

Singapore is no longer depending extremely on imported water, and in fact, can become

self-sufficiency (Lee, 2003). However, the Singapore government has “expressed the view

that it would like to continue to purchase water, under fair terms, from Malaysia or any

other country willing to be its long-term supplier” (Lee, 2003).

26
The price revision and the future of these agreements when they come to lapse have been

brought to the political front of the two countries for many years. Lee (2003) described in

details the political and historical background of this matter. His paper showed that these

conflicts pose great uncertainties to the planning of future water supply system for

Singapore.

4.2.1.3 NEWater

According to the PUB website,

“NEWater is high-graded re is high-grade reclaimed water produced from treated

used water that is purified further using advanced membrane technologies, making

the water ultra-clean and safe to drink.”

NEWater has passed thousands of scientific tests and well surpasses WHO and USEPA

standards. Water reclamation was first mentioned in 1970s but was hindered due to

technological limitations as well as high cost at that time. With the latest membrane

technology and decreasing membrane prices, NEWater proved to be more feasible. A

board of expert initiated NEWater study in 1998 and NEWater was born in 2003.

NEWater is pioneered and directed by the PUB to become a “key pillar” in Singapore’s

water supply system.

At the moment, there are 4 NEWater plants: Bedok (7 mgd), Kranji (12 mgd), Seletar (5

mgd) and Ulu Pandan (32 mgd). They contribute totally 56 mgd, about 20% of

Singapore’s water demand. The promising future of NEWater has triggered early

construction of the fifth NEWater factory at Changi, with capacity 50mgd (Lee, 2005).

NEWater’s contribution to total water supply was set to increase to 30% by 2012. There is

27
potential to expand existing plants (e.g. the Ulu Pandan plant was expanded from 25 mgd

to 32 mgd) and build more plants.

4.2.1.4 Desalinated water

Singapore has commissioned its first desalination plant in Tuas with capacity 30 mgd

(136,380 m3/day) in 2005.

Recently, desalinated water has seen much technological advancement. Multi Stage Flash

(MSF) and Multi-Effect Distillation (MED) can be combined with Reverse-Osmosis (RO)

to reduce cost (Methnani, 2006). Advanced technology has made membranes cheaper,

modern technology in pre-treatment and design have led to “better performance, lower

capital and operating costs” ("Cover story: Singapore shines at desalination conference,"

2005).

However, there are considerable uncertainties involved in the process of desalination.

Firstly, the reverse-osmosis technology that desalination employed requires a fairly large

amount of energy, ranging from 30% to 50% of the total cost (Methnani, 2006). Since fuel

price fluctuates significantly from year to year, desalination cost is expected to change

accordingly. Secondly, technology, even though advancing rapidly now, is a noTable

unknown, and “making it [desalination] economically viable is something of an art”

("Cover story: Singapore shines at desalination conference," 2005).

4.2.1.5 Summary on the four-tap model

The three local taps (catchment, NEWater, desalination) can contribute up to 236 mgd

(1,072,856 m3/day or 82.5% of the total demand). Only 50 mgd (227,300 m3/day or 17.5%

of the total demand) is dependent on the imported tap (see Figure 4.4). However,

28
desalination is still an expensive source, with the Tuas plant’s tender price of $0.78/m3,

more than three times as expensive as imported water ($0.25/m3) (Lee, 2005), and

NEWater is also 20% more expensive than imported water ($0.3/m3 for the latest Ulu

Pandan and Changi plants). Therefore, it is not economically feasible to rely only on the

local sources. In order to become self-sufficient, given the above potential and

uncertainties, water supply system planning needs special attention. New researches, both

on planning and on technology, are of critical importance.

17.48% Foreign reliance

NEWater

52.45% 19.58%
Desalinated water
Catchment water
10.49%

Figure 4.4 - Contribution of four national taps with maximum capacity for local taps

4.2.2 The potential "fifth tap"

Shaik, Ooi and Pehkonen (2006) outlined a promising future for desalination using liquid

natural gas (LNG) as a heat sink instead of the traditional RO method. The authors

remarked that 70% of Singapore's electricity originated from natural gas. Currently,

natural gas is imported into Singapore as its gaseous form through a system of pipelines.

However, it is more economical to import natural gas in liquid form, as it only occupies

29
1/600th of its gaseous volume, can be shipped easily by tankers and is easier to store. As

natural gas demand is expected to exceed the current level of supply, the Singapore

Energy Market Authority is investigating the potential to build an LNG terminal.

The drawback of LNG, however, is that it has to be regasified before use, and this process

consumes a huge amount of energy.

It has been known in literature since the 1960s that LNG regasification process, since

taking in large amount of energy, can be made use of as a heat sink for another process,

such as refrigeration (Shaik et all, 2006). Researchers such as Shwartz and Probstein

(1969), Cravalho et al (1977) analyzed the use LNG regasification to freeze seawater in

order to separate salt from it. Shaik et al (2006) applied mass-enthalpy balance and

economic considerations within Singapore context to examine the potential of a linked

LNG-desalination terminal with capacity 34,000m3/day (2.6% of Singapore's total water

demand). Their calculation showed that the required selling price (RSP) for the

desalinated water as a result of this process is $0.53/m3, which is favourably comparable

to RO desalinated water ($0.78/m3) and NEWater ($0.3/m3). Although the authors noted

the key constraint of limited LNG terminal capacity (6mtpa), they recommended that LNG

regasification-desalination should be further studied so that its potential can be explored,

the process can be improved and the capacity can be increased (Shaik et al, 2006).

This thesis will consider this new LNG-desalination technology as an option and examine

its value.

30
5. MONTE CARLO SIMULATION MODEL

5.1 Demand model

5.1.1 Domestic consumption model

Domestic consumption is the product of total population and individual water

consumption. These two factors will be examined in details.

5.1.1.1 Population model

Resident population: As suggested by Hashmi and Hui (2002), previous studies on

Singapore's population such as Saw (1987), Shantakumar (1996) are outdated because

they did not include migration in the projection, and their base years were only as latest as

1990. Hashmi and Hui pointed out that immigration and total fertility rate (TFR) will be

the key parameters contributing to Singapore's population in the future. They provided a

new population projection which allowed for migration and used 1999 as the base year.

However, as their assumptions for migration are outdated at the moment, and the base

year 1999 is far ago, an improvement to Hashmi and Hui's projection is performed with

aids of the PDE Projection software (PDE), using 2007 as the base year. The data are

provided in the Population Year Book 2008 by the Department of Statistics.

The projection uses the component method. The total population is divided into 2 genders,

male and female. Each gender group is then divided into 5-year age groups: 0-4, 5-9, 10-

14 and so on. The oldest age group is 85 and above (85+). Age-specific data such as

migration rate, fertility rate and mortality rate are entered into PDE for the base year. The

40-year analysis horizon is also divided into 5-year periods. The population of age group

31
(x to x + 4) at period (t to t + 5) is calculated based on the migration rate and mortality

rate of age group (x - 5 to x - 1) at period (t-5 to t). The population of age group (0 to 4) is

calculated based on the age specific fertility rate of all the female from 14 to 44 years of

age in the previous period (Saw, 1983 and PDE manual). All calculations were performed

by PDE.

In this projection, a total fertility rate is assumed to increase 10% each year, from 1.29 in

2007 to 1.39 in 2047. Migration rate is also assumed to increase 10% each year, starting

from 36,500 males and 44,483 females.

To take uncertainty into account, each year's population is randomized as a normal

distribution with the mean being the PDE's estimated value and standard deviation 4,200.

The standard deviation will ensure that the randomized population is within 50,000 away

from the mean, according to Chebyshev’s theorem.

Non-resident population: the base projection is calculated as a linear regression line to

the historical data. Then it is randomized with mean is the base projection and standard

deviation 10% of mean.

Total population: is the sum of resident and non-resident population.

5.1.1.2 Individual water consumption model

According to the analysis in section 4.1.1.2, it is reasonable to assume that water

consumption per capita will continue to decrease, but with a decreasing rate, and will

remain constant at a steady state.

One way to describe this nature is to use an exponential curve with negative power, and

the following equation is used:

32
s = ae − by + c (5.1)

where s is consumption per capita (litre/day), y is the year (from 0 to 40) and a, b, c are

constants.

Consumption is 158 litre/day in year 0 (2007).

Since it took 4 years (from 2003 to 2007) to reduce consumption by 10 litre/capita/day

(Table 4-1), it can be assumed that another 10 litre reduction will take 8 years into the

future. That means consumption per capita will be 148 litre/day by year 8 (2015).

It is also reasonable to let the consumption per capita be 140 litre/day at its steady state at

the end of the analysis period, year 40 (2047).

Substituting the above values to equation 5.1, and perform Goal Seek in Excel, we obtain

s = 18.359e −0.098 y + 139.641 (5.2)

The resulted consumption per capita each year is plotted in Figure 5.11.

160
158
156
154
152
150
148
146
144
142
140
138
2007 2012 2017 2022 2027 2032 2037 2042 2047

Figure 5.1 - Individual water consumption model

33
To consider uncertainty, each year's Figure will be randomized within 0.5 litres away from

the projected value, according to a uniform distribution.

5.1.2 Total consumption model

From Table 4-2, it can be calculated that industrial demand takes up an average of 43.21%

the total demand, or in other words, domestic demand takes up 56.79% of the total. The

standard deviation of this percentage is calculated as 0.012466. Therefore, total water

consumption will be calculated as follow:

• Generate a random number x following a normal distribution with mean 0.5679

and standard distribution 0.012466.

• Total water consumption = Domestic water consumption / x

5.2 Supply model

5.2.1 Configuration and risk premium

Each year, facing with a demand and different water prices from the four taps, PUB has to

configure the contribution of the four taps to minimize the cost of supplying water. The

key strategy is to choose the tap with cheapest unit cost, use as much as possible, and then

move on to use the next cheapest tap.

It is obvious that currently, the import tap is the second cheapest and should be utilized as

much as possible. However, if PUB takes the cost of imported water as its face value, it is

omitting the potential risks that this tap is cut off. Therefore, a risk premium should be

incorporated into the cost of imported water when compared with other taps. How much

this risk premium should be is an unknown. This study will assign values from $0.00 to

34
$3.50 with an increment of $0.10 to the risk premium to see the effect of different

considerations.

5.2.2 Imported water modelling

For the Agreement ending 2011, the Singapore government has expressed the intention to

let it lapse (Lee, 2003). However, there are possibilities that a revision be made in 2011.

The following scenarios are adopted for the purpose of this study (Table 5.1).

Table 5.1 - Scenarios for 2011 Agreement

Description Probability
Lapse 0.7
Renew at $0.25 (same price) 0.1
Renew at $1.00 (unfavourable) 0.1
Renew at $3.45 (Very unfavourable) 0.1

For the Agreement ending 2061, its expiry date is out of our study horizon; therefore we

need not consider the revision. However, as the capacity of water in this Agreement is

very high, the small likelihood of it being disrupted will lead to severe consequences.

Therefore, we have to take into account the risk of it being broken out. The following

scenarios are to be examined:

Table 5.2 - Scenarios for 2061 Agreement

Description New unit price ($/m3) Probability


Unchanged Same as the value of 2011 revision 0.97
Cut off 0.01
Quantity reduced by half Same as the value of 2011 revision 0.01
Price doubled Twice the value of 2011 revision 0.01

35
5.2.3 Cost and payment modelling

5.2.3.1 Cost components

There are 3 components of cost in the simulation: investment cost, fixed cost (operation

and maintenance), variable costs (production cost). The initial proportion of each

component for desalination, NEWater and LNG desalination are shown in Table 5.3. The

proportion for desalination is based on Methnani (2006) and Zhang (2006). The Figures

for NEWater is an adjustment from desalination and based on Zhang (2006). The Figures

for LNG desalination is based on Shaik et al (2006)

Table 5.3 - Cost components

Desalination NEWater LNG-Desalination


Cost ($
Components % % Components
million)
Energy (variable) 46.90% 23.45% Capital charge 1.65
Investment 29.50% 42.53% Investment 5.67
O & M (fixed) 17.90% 25.80% O & M (fixed) 2.93
Technical (variable) 5.70% 8.22% Variable costs 1.41

For desalination and NEWater, since they employ reverse-osmosis technology which

consumes a large amount of energy, the variable cost is broken down to energy cost and

technical cost, each of which follows a different random walk:

Technical cost: as efforts are made to increase efficiency, the technical cost at year i is

pt i = pt i −1 × r (5.4)

where r is a random number following a normal distribution of 0.926 and standard

deviation 0.0926. This means that generally, technical cost is declining over time. But in

36
some certain simulation run, it can be either declining or increasing from one year to

another.

Energy cost: due to the high uncertainty involved in the crude oil prices, energy cost at

year i is

pei = pei −1 × r (5.5)

where r is a random number following a normal distribution of mean 1 and standard

deviation 0.5.

Technical cost and energy cost add up to unit (variable) production cost, which does not

depend on capacity.

O&M cost and investment cost depends on capacity and they remain unchanged if there is

no change in capacity. When there is an expansion, both investment and O&M will

increase. When there is a reduction, O&M will decrease while investment remains

unchanged because it is considered sunk cost.

5.2.3.2 Economies of scale

The new investment and O&M cost after expansion or reduction is assumed the following

cost curve:

C = kQm (5.6)

where C is the cost at capacity Q, k is a constant and m is the economies of scale factor.

• When m > 1, we have economies of scale. An increase in capacity will lead to

smaller averaged unit cost and expansion is favourable.

37
• When m < 1, we have diseconomies of scale. An increase in capacity will lead to

larger averaged unit cost and expansion is not favourable.

• When m = 1, the cost curve is linear. An increase in capacity will lead to a

proportional increase in fixed costs. Therefore, whether expansion is favourable

depends on other factors.

It is difficult to estimate the correct m value. In practice, m also depends on the operators.

For instance, if all the desalination plants are operated by one company, they can take

advantage of the common resources and there will be economies of scale.

For the purpose of this study, m will be assigned values of 0.5, 0.8 and 1.

5.2.3.3 Ownership and payment scheme simulation

The three latest water treatment plants in Singapore, namely Tuas Desalination Plant, Ulu

Pandan NEWater Plant, and Changi NEWater Plant, have undertaken a PPP (Public-

Private-Partnership) scheme, in which the project procurement followed the DBOO

(Design-Build-Own-Operate) model (Sanmuganathan, 2008). Accordingly, private

companies will bid to be PUB's water supplier. The tender will include a first year's

selling price and an adjustment formula with respect to inflation, energy cost, taxes and so

on. Once a company is awarded the contract - since then called the "concession company"

(Sanmuganathan, 2008) - it will employ subcontractors to design, build and operate the

plant, and it is the concession company that owns the project, not PUB.

The reasons behind this is to "offer a win-win solution" for both the public and private

sectors as well as the services users - members of the public. This strategy also helps to

utilize the strength of both the public and private sectors (Sanmuganathan, 2008). Besides,

38
by involving the private sector, PUB will be offered a more competitive price, and

competition in the private sector will also help to increase efficiency, optimize technology

configuration and keep cost affordable (Sivaraman, 2006).

Under DBOO scheme, there are three possible payment structures that PUB can adopt

(Sanmuganathan, 2008):

• "Unit Tariff per Output" structure: PUB is not obliged to purchase any minimum

amount of water; therefore this structure offers the highest flexibility. However, in

terms of financing for the project, the concession company cannot get a bank loan

because there is no security against the loan; hence it is harder to implement the

project of this type.

• "Take or Pay" pay structure: PUB agrees to pay at least a minimum of money to

cover the capital charge for the concession company, even though it does not need

to buy the equivalent amount of treated water. This minimum purchase amount can

be used as a security for a bank loan, thus procurement for the project of this type

is easier. However, it does not offer as much flexibility.

• "Two-part tariff" structure is the hybrid of the previous two and is the one adopted

in all three DBOO plants. Under this structure, PUB pays the fixed costs to cover

the project capital, thus partially reduce the risks for the concession company. On

the other hands, variable payments are dependent on production and the

concession company still faces market risks.

39
As the "two-part tariff" structure is being used and the "unit tariff per output" structure can

offer high flexibility, they will be simulated in the model. The "take or pay" structure will

not be addressed.

Whilst all the current DBOO plants are following the third structure, it is possible that

some future plants will follow the first scheme. It is also important to note that the first

three NEWater plants are self-operated by PUB. However, it is very complicated to

simulate this composition, as it requires the model to track the process to the plant level,

while the purpose of this research is to examine the impact only at the "tap" level.

Therefore, in the Monte Carlo simulation model, the two payment structures will be

modelled separately.

As we are not interested in modelling to the plant level, where each plant can be operated

by a different concession company, we will consider that there are two sides in the

transaction: one single concession company who operates the plants and sell water to

PUB, and PUB as the buyer, and we are interested in the cost born by PUB.

5.2.4 "Unit Tariff per Output" modelling

This model employs the "break even point" concept in economics.

The treatment of water will require the concession company to pay some fixed cost

(investment, fixed operation and maintenance cost, staff salary etc) and some variable

cost, which depends on the unit cost per unit volume of water treated.

After treatment, the company will then sell the treated water to PUB at a required selling

price (RSP). This is the price that PUB will buy water from the concession company.

40
5.2.4.1 Selling price

Let q (m3) be the volume of water treated.

The cost of treating q m3 of water is

TC = u × q + FC + I (5.7)

where TC is the Total Cost ($), u is the unit variable cost ($/m3), FC is the Fixed Cost ($)

and I is the Investment cost or Capital charged ($).

The revenue (RV, in $) of selling q m3 of water is

RV = RSP × q (5.8)

At break-even point where revenue equals cost and the company does not gaining any

profit or suffering any loss

RV = TC (5.9)

u × q + FC + I = RSP × q (5.10)

FC I
RSP = u + + (5.11)
q q

Let Q be the total capacity that can be supplied by the concession company, so

q = βQ (5.12)

where β is the break-even point coefficient and 0 < β ≤ 1. This means β is the proportion

of the break-even quantity to the total capacity.

Substituting (5.12) to (5.11)

FC I
RSP = u + + (5.13)
βQ βQ

41
Applying economies of scale; let

I = kIQm (5.14)
where
I ($) is the capital needed to invest in having the total capacity Q
kI is a capital cost coefficient
m is the economies of scale factor.

Similarly, let

FC = k F Q m (5.15)
where
FC ($) is the fixed cost needed to operate the total capacity Q
kF is a fixed cost coefficient
m is the economies of scale factor.

Note that for simplicity, we have assumed that I and FC have the same economies of scale

factor.

Substituting (5.14) and (5.15) into (5.13)

Q m −1
RSP = u + (k F + k I ) (5.16)
β

In this simulation model, two values β = 0.5 and β = 0.75 will be examined.

5.2.4.2 Expansion

When a tap is in full capacity for 5 consecutive years and its price is the cheapest of all the

taps that are expandable, PUB will consider expanding its capacity by inviting the

concession company to build a new plant or expanding existing plant.

In that case, the concession company will have to put in more investment cost and fixed

cost to operate the larger facility.

42
If the capacity is expanded an amount qe, the new RSP will then be

(Q + q e ) m−1
RSPnew = u + (k F + k I ) (5.17)
β

Supposed the tap will be in full capacity after expanding, hence, PUB can save money Q

m3 at the new price and the additional qe m3 is used to replace the amount previously

bought through the more expensive taps. Let S be the saving, then

S = Q ( RSP − RSPnew ) + ∑ q e ( RSPnew − RSPj ) (5.18)


j ≠i

where
i is the tap of interest
j is any tap that is more expensive than i

If in a year there are more than one expandable taps, then the one with the highest saving

will be expanded.

Note that as the unit variable cost of each tap changes over time, in calculating the RSP

values in (5.18), we use u as the average of the previous 5 years.

5.2.4.3 Reduction

If a tap has been running below its break-even point for 5 consecutive years, then the

concession company may decide to shut down some part of the plants to save operating

and maintenance cost. The amount of reduction would be (1 - β) Q and the capacity after

reduction is βQ.

However, it has to resume its original capacity quickly since this is the contractual

obligation that it has to provide the full capacity upon request.

43
In reducing capacity, the concession company can save the fixed cost but cannot recover

the investment costs, as they are considered sunk cost. The new cost after reduction would

be:

k F (βQ) m k I Q m
RSP = u + + (5.19)
β ( βQ) βQ

If the new RSP is too high, the product will become less competitive, and then PUB

and/or the concession company may choose to shut down the source completely and wait

for favourable demand. In this study the threshold is set at $2.00.

5.2.5 "Two-part tariff" modelling

This model is similar to the "Unit tariff per output" model, but the Investment cost term is

not presented. On the other hand, the investment cost is considered in the tap ranking

when PUB decides how much of each tap is used, and the savings when expansion takes

place must make up for the additional investment expense.

5.3 Risk modelling

Apart from economic consideration when calculating net present value, it is inadequate to

consider economic values only. In this model, we propose two types of risk:

• Water security risk: is when the averaged unit cost of supplying a unit volume of

water for a particular year is greater than $1.50. The averaged unit cost of

supplying water is calculated as:

∑ RSP × U
i
i i
P= (5.20)
TWC

where

44
i denotes the tap

Ui is the usage of tap i in that year

TWC is total water consumption in that year.

• Water scarcity risk: how many days Singapore can survive without the import tap.

If for a particular year, the total available capacity is greater than total water consumption,

then clearly there is no risk.

Otherwise, let E be the excessive amount per day that consumption has more than supply.

This amount will have to be met by the water stored in the reservoir. But we have already

taken out catchment water to consider in the total supply capacity, so, let x be the number

of days we can survive,

Reservoir storage capacity


x= (5.21).
E + Catchment Capacity

If x is greater than 365 then there is no risk. Otherwise risk is calculated as the difference

between x and 365 days.

These two risks will be calculated in the simulation.

45
6. RESULTS AND DISCUSSIONS

6.1 “Unit tariff” vs. “two part tariff”


On average, the “unit tariff” approach helps to reduce NPV by 7.29% for the case without

LNG and 7.35% for the case with LNG (Table C.2), because under the “unit tariff”

structure, project cost is part of the variable purchase costs. Hence, PUB does not always

have to bear this cost, but only when it purchases some water. On the other hand, for the

“two-part tariff” structure, this payment is fixed so PUB always has to pay. However, the

difference between the two structures is small because the proportion of investment cost is

moderate. In addition, PUB has the option to expand when needed in stead of having to

pay capital upfront. Therefore, it does not have excessive capacity and wasted capital.

The result suggests that the savings from the “unit tariff’ does not justify the difficulties in

financing the projects of its type. Therefore, in the future, PUB should continue on the

“two-part tariff” structure. In this case the dispatch flexibility cannot balance the

procurement costs.

6.2 Effects of the “fifth tap”

6.2.1 Mean NPV

Table () showed that by keeping the prospective LNG-desalination terminal capacity at

$34,000 m3/day as the current limit, LNG as the “fifth tap” can only help to save about

$3.50 - $3.60 million SGD in terms of NPV for the lower range of political risk premium

($0.00 to $1.50) and about $6.50 million for the higher range. However, if we can make

the capacity three times bigger (up to 102,000 m3/day) the saving would increase

46
significantly, to $29.3 million for the lower range and $47.0 million for the higher range,

which means 6 to 8 times the original savings.

Figure 6.1 – Effect of enlarging LNG capacity

This suggests that Singapore is in the situation of holding a call option – and Singapore

should invest in a research on LNG-desalination and realize its potential economic value.

The research’s success provide Singapore the right to implement LNG-desalination in the

future. Table () suggests that the funding for the research should be around $2.0 million,

and it is also the option premium. Once Singaopre has the technology in hand, it has

another option – to enlarge the capacity by three times. This option to expand is called

“option on option”.

6.2.2 Water security risks and water scarcity risks

47
The fifth tap, due to its small capacity, does not contribute significantly to reducing the

water security and scarcity risks.

6.3 Effect of risk premium


Figure 6.2 shows the mean NPV versus political risk premium, with m = 0.8 for

desalination and NEWater, LNG's maximum capacity set as 34,000 m3/day.

Figure 6.2 - Mean NPV result

The results showed that the mean NPV of the total water supply cost over the analysis

period is very sensitive with risk premiums valued from $0.00 to $1.50. Beyond $1.50, the

NPV curve flattens.

It is interesting to note that, generally, the mean NPV increases as the risk premium

increases. However, between $0.00 and $0.10, NPV actually decreases. A finer simulation

for the range from $0.00 to $0.50 with $0.05 increment proved this (Figure 6.3).

48
Figure 6.3 - Partial NPV curve from $0.00 to $0.50

This is because the risk premium of $0.10 makes imported water generally more

expensive than NEWater. Therefore, the model chooses to expand NEWater and use less

imported water. As this is the case where economies of scale exist (m = 0.8), the

expansion of NEWater is beneficial and helps to save cost. The same result is obtained

when m = 0.5 (Figure C.1). However, political cost is still not too high to make

desalination more favourable than imported water. Therefore, the cost of supplying is still

low. As the risk premium goes higher, less imported water is used, more desalinated water

is produced and this increases the cost of supplying water.

49
6.4 Effect of β and m
From the various tables in appendix C, we can see that the model is not quite sensitive to β

and m. In fact, from Table C.2, changing from β = 0.75 to β = 0.5 only decreases NPV by

less than 2%.

6.5 Recommendation for future researches

This study has not examined the water supply system to the plant level. In further

research, the problem should be addressed to that level of details to be realistic.

In addition, in this study, the time step is one year each. Therefore, we take averaged daily

consumption as the constant consumption for every day in a year. To have a more realistic

simulation, further researches should also take a daily time step so that the reaction to

disruption can be captured more accurately.

50
REFERENCES

• DPM’S SPEECH ON POPULATION AT COMMITTEE OF SUPPLY (2008).

http://www.nps.gov.sg/files/news/DPM%27s%20speech%20on%20population%20at

%20COS%2027%20Feb%202008.pdf. D. P. s. Office.

• Department of Statistics (2008). Yearbook of Statistics 2008.

• Deparment of Statistics (2008). Population if Brief 2008.

• Department of Statistics (2008). Population Trends 2008.

• de Neufville, R., Scholtes, S (2004 ). Designing a parking garage.

• Australian Stock Exchange (2004). Options - Understanding Options trading.

• Hashmi, A. R., Hui, W. T. (2002). Population and Labour Force Projections for

Singapore (1999 - 2049). IUSSP Conference on Southeast Asia's Population in a

Changing Asian Context. Bangkok, Thailand.

• Lee, P. O. (2003). "The Water Issue between Singapore and Malaysia: No Solution in

Sight?" Economics and Finance 1: 1-29.

• Lee, P. O. (2005). Water Management Issue in Singapore. Water In Mainland

Southeast Asia. Siem Reap, Cambodia.

• Methnani, M. (2007). "Influence of fuel costs on seawater desalination options."

Desalination 205(1-3): 332-339.

• Mun, J. (2006). Real Options Analysis: Tools and Techniques for Valueing Strategic

Investment and Decisions. Hoboken, New Jersey: John Wiley & Sons Inc.

51
• Sanmuganathan, D. (2008). PUB Singapore's Experience in Public-Private-Partnership

(PPP) Projects. CAPAM Biennial Conference.

• Saw, S.-H. (1983). Population Projections for Singapore 1980 - 2070.

• Seah, H. (2008). "PUB's NewWater Programme."

• Shaik, S. O. T., Ho; Pehkonen Simo, O (2006). Potential for Seawater Desalination in

Singapore using LNG Regasification as a Heat Sink. UltraPure Water Asia.

• Sivaraman, A. (2006). Desalination and Ulu Pandan NEWater DBOO Project, PUB's

Experience. Capacity Building Workshop. Bangkok.

• Tan, N. S. (2007). Integrated Urban Water Management in Singapore. Urban

Sustainability Conference.

• Tortajada, C. (2006). "Water Management in Singapore." Water Resource

Development 22(2): 227-240.

• Young, J. C. (2008). "Meeting the challenges of sustainable water supply - Integrated

water resource management."

52
APPENDIX A – CASE STUDY RESULTS

Results without expansion option

2500

2000

1500

1000

Base case
500 Year 10 - Worst
Year 10 - Best

0
0 5 10 15 20 25

Figure A. 1 – Best, base and worst cases, demand by year 10

1800

1600

1400

1200

1000
Base case
800 > yr 10 - Worst
600

400

200

0
0 5 10 15 20 25

Figure A. 2 – Base and worst cases, demand after year 10

53
Figure A. 3 – Base and best cases, demand after year 10

Table A. 1- Sensitivity results, demand by year 10


Capacity 10% 25% 50% 75%
4 $31,263 $5,966 -$38,200 -$172,296
5 $1,910,019 $1,815,273 $1,681,600 $1,329,153
6 $2,246,297 $2,226,523 $1,996,800 $1,455,020
7 $1,094,273 $1,057,458 $620,100 $167,421

Table A. 2 – Sensitivity results, demand after year 10

Capacity 10% 25% 50% 75%


4 -$84,534 -$62,990 -$38,200 -$46,314
5 $1,644,533 $1,685,499 $1,681,600 $1,732,331
6 $1,887,795 $1,973,699 $1,996,800 $2,000,341
7 $708,070 $706,963 $620,100 $700,847

54
Results with expansion option

Table A. 3 - Sensitivity result, volatility in year 10 demand

10% 25% 50% 75%


Max Profit $5,529,409 $5,420,879 $5,321,847 $5,600,358
Initial levels 4 4 5 4
Additional levels 1 1 1 2

Table A.4 – Sensitivity result, volatility in demand after year 10

10% 25% 50% 75%


Max Profit $5,481,395 $5,443,167 $5,294,556 $5,031,332
Initial levels 4 4 4 4
Additional levels 1 1 1 1

Figure A.4 – Effect of volatility on net present value

55
APPENDIX B – POPULATION PROJECTION RESULTS

Table B.1. Resident population projection


Year Population Year Population Year Population Year Population
2007 3,583,200 2017 3,887,131 2027 4,151,326 2037 4,337,086
2008 3,615,064 2018 3,914,503 2028 4,172,564 2038 4,348,906
2009 3,646,929 2019 3,941,876 2029 4,193,803 2039 4,360,726
2010 3,678,793 2020 3,969,248 2030 4,215,041 2040 4,372,547
2011 3,710,658 2021 3,996,621 2031 4,236,280 2041 4,384,367
2012 3,742,522 2022 4,023,993 2032 4,257,518 2042 4,396,187
2013 3,771,444 2023 4,049,460 2033 4,273,432 2043 4,407,492
2014 3,800,366 2024 4,074,926 2034 4,289,345 2044 4,418,796
2015 3,829,287 2025 4,100,393 2035 4,305,259 2045 4,430,101
2016 3,858,209 2026 4,125,859 2036 4,321,172 2046 4,441,405
2047 4,452,710

Table B.2. Non-resident population projection

Year Population Year Population Year Population Year Population


2007 941,322 2017 1,259,458 2027 1,577,593 2037 1,895,729
2008 973,135 2018 1,291,271 2028 1,609,407 2038 1,927,543
2009 1,004,949 2019 1,323,085 2029 1,641,221 2039 1,959,357
2010 1,036,762 2020 1,354,898 2030 1,673,034 2040 1,991,170
2011 1,068,576 2021 1,386,712 2031 1,704,848 2041 2,022,984
2012 1,100,390 2022 1,418,526 2032 1,736,661 2042 2,054,797
2013 1,132,203 2023 1,450,339 2033 1,768,475 2043 2,086,611
2014 1,164,017 2024 1,482,153 2034 1,800,289 2044 2,118,425
2015 1,195,830 2025 1,513,966 2035 1,832,102 2045 2,150,238
2016 1,227,644 2026 1,545,780 2036 1,863,916 2046 2,182,052
2047 2,213,865

56
Table B.3. Total population projection

Year Population Year Population Year Population Year Population


2007 4,524,522 2017 5,146,589 2027 5,728,919 2037 6,232,815
2008 4,588,199 2018 5,205,774 2028 5,781,971 2038 6,276,449
2009 4,651,878 2019 5,264,961 2029 5,835,024 2039 6,320,083
2010 4,715,555 2020 5,324,146 2030 5,888,075 2040 6,363,717
2011 4,779,234 2021 5,383,333 2031 5,941,128 2041 6,407,351
2012 4,842,912 2022 5,442,519 2032 5,994,179 2042 6,450,984
2013 4,903,647 2023 5,499,799 2033 6,041,907 2043 6,494,103
2014 4,964,383 2024 5,557,079 2034 6,089,634 2044 6,537,221
2015 5,025,117 2025 5,614,359 2035 6,137,361 2045 6,580,339
2016 5,085,853 2026 5,671,639 2036 6,185,088 2046 6,623,457
2047 6,666,575

57
Table B.4. Age-gender-specific population projection

58
APPENDIX C– SIMULATION RESULTS

Figure C.1 – Unit tariff, m = 0.5, β = 0.5

Figure C.2. – Two part tariff, m = 1, β = 0.5

59
Figure C.3 – Unit tariff, m = 1, β = 0.5

Table C.1 – Comparison between Unit tariff and Two-part tariff. m = 1, β = 0.5

Unit tariff Two part tariff Percentage different


Risk Without Without Without With
premium LNG With LNG LNG With LNG LNG LNG
$0.00 $3,015,533 $3,008,627 $3,272,130 $3,270,173 7.84% 8.00%
$0.10 $3,027,891 $3,023,089 $3,301,439 $3,299,574 8.29% 8.38%
$0.20 $3,049,612 $3,044,994 $3,312,934 $3,311,012 7.95% 8.03%
$0.30 $3,060,626 $3,055,942 $3,334,274 $3,331,475 8.21% 8.27%
$0.40 $3,078,751 $3,073,727 $3,386,621 $3,382,838 9.09% 9.14%
$0.50 $3,115,894 $3,110,825 $3,417,319 $3,413,273 8.82% 8.86%
$0.60 $3,185,308 $3,180,380 $3,432,941 $3,429,744 7.21% 7.27%
$0.70 $3,212,544 $3,207,580 $3,441,390 $3,438,934 6.65% 6.73%
$0.80 $3,225,455 $3,220,251 $3,448,328 $3,446,043 6.46% 6.55%
$0.90 $3,232,320 $3,226,898 $3,455,083 $3,452,267 6.45% 6.53%
$1.00 $3,236,426 $3,230,846 $3,461,502 $3,457,759 6.50% 6.56%
$1.10 $3,238,967 $3,233,239 $3,466,929 $3,462,645 6.58% 6.63%
$1.20 $3,240,808 $3,234,855 $3,472,728 $3,467,742 6.68% 6.72%
$1.30 $3,242,245 $3,236,005 $3,477,780 $3,472,148 6.77% 6.80%
$1.40 $3,243,341 $3,236,889 $3,481,756 $3,475,793 6.85% 6.87%
$1.50 $3,244,118 $3,237,510 $3,485,512 $3,479,392 6.93% 6.95%
$2.00 $3,246,284 $3,239,049 $3,494,710 $3,488,046 7.11% 7.14%
$2.50 $3,247,224 $3,239,718 $3,496,567 $3,489,672 7.13% 7.16%
$3.00 $3,247,625 $3,240,002 $3,496,907 $3,489,836 7.13% 7.16%
$3.50 $3,247,878 $3,240,169 $3,497,080 $3,489,769 7.13% 7.15%
Average 7.29% 7.35%

60
Table C.2 m = 0.5, β = 0.5, Two-part tariff
mean water mean water scarcity
mean NPV
security risk risk (days)
Risk Without Without With Without With
With LNG Savings
premium LNG LNG LNG LNG LNG
$0.00 $2,894,257 $2,891,995 $2,262 0.0098 0.0094 796.54 796.00
$0.10 $2,882,516 $2,880,622 $1,895 0.0014 0.0014 128.68 128.73
$0.20 $2,890,272 $2,888,341 $1,931 0.0008 0.0008 53.64 53.42
$0.30 $2,909,638 $2,907,571 $2,067 0.0008 0.0008 25.46 21.45
$0.40 $2,961,849 $2,959,125 $2,724 0.0006 0.0006 11.81 9.81
$0.50 $2,992,541 $2,989,739 $2,802 0.0006 0.0006 4.67 3.39
$0.60 $3,010,187 $3,007,705 $2,482 0.0006 0.0006 3.00 2.07
$0.70 $3,020,616 $3,018,120 $2,496 0.0006 0.0006 2.38 1.93
$0.80 $3,027,899 $3,025,130 $2,768 0.0006 0.0006 2.08 1.72
$0.90 $3,034,004 $3,030,722 $3,281 0.0006 0.0006 1.94 1.70
$1.00 $3,039,130 $3,035,221 $3,909 0.0006 0.0006 1.88 1.61
$1.10 $3,043,679 $3,039,002 $4,678 0.0006 0.0006 1.80 1.59
$1.20 $3,047,564 $3,042,206 $5,358 0.0006 0.0006 1.72 1.52
$1.30 $3,050,324 $3,044,595 $5,729 0.0006 0.0006 1.66 1.54
$1.40 $3,052,697 $3,046,717 $5,980 0.0006 0.0006 1.61 1.52
$1.50 $3,054,178 $3,048,019 $6,158 0.0006 0.0006 1.55 1.46
$2.00 $3,055,381 $3,049,511 $5,870 0.0006 0.0006 1.36 1.36
$2.50 $3,055,578 $3,049,544 $6,033 0.0006 0.0006 1.36 1.36
$3.00 $3,055,461 $3,049,264 $6,197 0.0006 0.0006 1.36 1.36
$3.50 $3,055,253 $3,048,873 $6,380 0.0006 0.0006 1.36 1.36

61
Table C.3 Comparison between β = 0.5 and β = 0.75 (Unit tariff)

m = 0.8 β = 0.75 m = 0.8 β = 0.5 Difference


Without Without
Without LNG With LNG LNG With LNG LNG With LNG
$0.00 $3,009,991.0 $3,004,592.4 $2,984,145 $2,967,281 0.86% 1.24%
$0.10 $2,986,255.7 $2,981,542.5 $2,954,483 $2,943,342 1.06% 1.28%
$0.20 $3,002,009.8 $2,998,249.8 $2,969,547 $2,958,694 1.08% 1.32%
$0.30 $3,011,314.2 $3,006,763.6 $2,979,408 $2,968,560 1.06% 1.27%
$0.40 $3,028,231.6 $3,023,122.4 $2,994,851 $2,983,551 1.10% 1.31%
$0.50 $3,064,618.2 $3,058,980.3 $3,028,315 $3,016,630 1.18% 1.38%
$0.60 $3,133,300.5 $3,127,272.5 $3,094,548 $3,084,250 1.24% 1.38%
$0.70 $3,160,034.6 $3,153,487.1 $3,117,049 $3,105,888 1.36% 1.51%
$0.80 $3,172,819.7 $3,165,646.2 $3,127,051 $3,115,160 1.44% 1.59%
$0.90 $3,180,402.0 $3,172,414.9 $3,132,172 $3,119,624 1.52% 1.66%
$1.00 $3,185,154.8 $3,176,289.9 $3,135,240 $3,122,245 1.57% 1.70%
$1.10 $3,188,414.9 $3,178,833.2 $3,137,622 $3,124,271 1.59% 1.72%
$1.20 $3,191,146.9 $3,180,772.8 $3,139,878 $3,125,900 1.61% 1.73%
$1.30 $3,193,047.1 $3,182,078.7 $3,142,035 $3,127,427 1.60% 1.72%
$1.40 $3,194,413.7 $3,182,924.0 $3,144,017 $3,128,836 1.58% 1.70%
$1.50 $3,195,723.7 $3,183,783.2 $3,146,178 $3,130,258 1.55% 1.68%
$2.00 $3,200,357.6 $3,186,833.7 $3,152,335 $3,134,884 1.50% 1.63%
$2.50 $3,202,043.6 $3,187,878.0 $3,154,202 $3,136,537 1.49% 1.61%
$3.00 $3,202,774.0 $3,188,303.5 $3,154,954 $3,137,278 1.49% 1.60%
$3.50 $3,203,275.9 $3,188,567.0 $3,155,348 $3,137,663 1.50% 1.60%

Figure C.4 – Demand and supply spectra (100 simulation runs)

62
Table C.4 – Comparison between big and small LNG capacity

LNG = 34,000 LNG = 102,000


Without Without
LNG With LNG Savings LNG With LNG Savings
$0.00 $2,894,257 $2,891,995 $2,262 $3,276,045 $3,272,484 $3,561
$0.10 $2,882,516 $2,880,622 $1,895 $3,302,558 $3,296,086 $6,472
$0.20 $2,890,272 $2,888,341 $1,931 $3,313,832 $3,300,735 $13,097
$0.30 $2,909,638 $2,907,571 $2,067 $3,335,355 $3,317,620 $17,734
$0.40 $2,961,849 $2,959,125 $2,724 $3,388,161 $3,366,512 $21,649
$0.50 $2,992,541 $2,989,739 $2,802 $3,417,727 $3,394,917 $22,810
$0.60 $3,010,187 $3,007,705 $2,482 $3,433,078 $3,409,945 $23,133
$0.70 $3,020,616 $3,018,120 $2,496 $3,442,436 $3,417,591 $24,845
$0.80 $3,027,899 $3,025,130 $2,768 $3,449,889 $3,422,157 $27,733
$0.90 $3,034,004 $3,030,722 $3,281 $3,456,252 $3,424,860 $31,392
$1.00 $3,039,130 $3,035,221 $3,909 $3,463,418 $3,426,714 $36,705
$1.10 $3,043,679 $3,039,002 $4,678 $3,469,649 $3,428,201 $41,448
$1.20 $3,047,564 $3,042,206 $5,358 $3,474,729 $3,429,535 $45,194
$1.30 $3,050,324 $3,044,595 $5,729 $3,479,512 $3,430,722 $48,790
$1.40 $3,052,697 $3,046,717 $5,980 $3,483,255 $3,431,801 $51,454
$1.50 $3,054,178 $3,048,019 $6,158 $3,486,500 $3,432,998 $53,503
$2.00 $3,055,381 $3,049,511 $5,870 $3,496,031 $3,440,315 $55,717
$2.50 $3,055,578 $3,049,544 $6,033 $3,498,750 $3,448,423 $50,328
$3.00 $3,055,461 $3,049,264 $6,197 $3,499,259 $3,455,689 $43,570
$3.50 $3,055,253 $3,048,873 $6,380 $3,499,352 $3,460,885 $38,467

Figure C.5 – A typical random walk for desalination variable cost

63

You might also like