Civil Engineering and Environmental Systems

ISSN: 1028-6608 (Print) 1029-0249 (Online) Journal homepage: http://www.tandfonline.com/loi/gcee20

Developing efficient management strategies for

a water supply system using system dynamics
modelling
Suwan Park, Dae-Hoon Jeon & So-Yeon Jung
To cite this article: Suwan Park, Dae-Hoon Jeon & So-Yeon Jung (2014) Developing efficient
management strategies for a water supply system using system dynamics modelling, Civil
Engineering and Environmental Systems, 31:3, 189-208, DOI: 10.1080/10286608.2013.820281
To link to this article: http://dx.doi.org/10.1080/10286608.2013.820281

Published online: 06 Aug 2013.

Submit your article to this journal

Article views: 374

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=gcee20
Download by: [Universiti Teknologi Malaysia]

Date: 22 September 2015, At: 08:04

Civil Engineering and Environmental Systems, 2014

Vol. 31, No. 3, 189208, http://dx.doi.org/10.1080/10286608.2013.820281

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Developing efficient management strategies for a water supply

system using system dynamics modelling
Suwan Park*, Dae-Hoon Jeon and So-Yeon Jung
Department of Civil and Environmental Engineering, Pusan National University, Busan 609-735,
Republic of Korea
(Received 3 July 2013)
In this paper, the causal feedback relationships among the components that make up the working mechanism
of water supply systems management, including key factors and their relationship to the management of
water pipes, were identified based on the conceptual framework established for water supply systems
management. Subsequently, a system dynamics computer simulation model, which can be used to aid
efficient management of water supply systems, was developed. The computer model consisted of water
supply, pipe maintenance, and water supply business finance sub-models. The model was verified using
historical data from a water supply service case study. Using the verified model, long-term managerial
and operating conditions of the case study system were predicted under optimistic, basic, and pessimistic
management condition scenarios. In addition, sensitivity analyses on major indicators of the case study
system have been presented to show that the developed model can facilitate identification of the best policy
for achieving a specific management objective of a water supply system.
Keywords: causal feedback; computer model; management; simulation; system dynamics; water supply system

1.

Introduction

Water supply systems are one of the most significant engineering accomplishments in history
and have greatly contributed to the wellz-being of mankind. However, due to waters status as a
public service, many water supply systems around the globe are confronted with problems related
to their overall efficiency. South Koreas water supply system has also faced these problems and
suffered from inefficient operations and poor monetary management. As a result, the Korean water
supply service incurred debts equivalent to 28% of their total income in 2006 (Korea Ministry of
Environment 2007).
Given the advancement of international standardisation of water and wastewater services
(ISO/TC224), and the addition of environmental services to the agenda of the World Trade
Organisations multilateral trade negotiations, developed countries, such as the EU member states,
continuously demand the participation of their companies in the Korean water supply services
market. To cope with current internal problems and external pressures, the water supply services in
Korea seek to improve the efficiency of their systems. They hope to accomplish this by employing
*Corresponding author. Email: swanpark@pusan.ac.kr
2013 Taylor & Francis

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

190

S. Park et al.

various strategies such as reducing the gap between consumer price and production cost, benchmarking operation and management strategies, and evaluating system performance based on the
level of water supply services and financial management.
Determining sound and sustainable policies is of the utmost importance for the efficient operation and management of a water supply system. The application of poor policies which are not
based on an understanding of the underlying relations among various components that make up
the working mechanism of a system may only lead to a short-lived, partial, or superficial solution
to the problem. Moreover, determining an appropriate policy based only on experience may be
problematical. This is because human experience alone may not be able to identify the inherent mechanisms connecting the input (or policy) to the output of a system. Consequently, any
connections between a policy and indicators that represent the state of a system (e.g. physical
condition of system components and financial performance and service improvement indicators)
established during the evaluation process of a policy may not be ascertained.
Therefore, an understanding of the working components of a system, as well as the correlations
among them, is essential for appropriate analysis of problems associated with water supply systems and establishment of policies that address problems of interest. A very useful and efficient
methodology suited for modelling such multiple component systems, where these components
influence each other, is system dynamics (SD).
Grigg and Bryson (1975) and Lee, Park, and Park (2007) modelled water supply service management using the SD methodology. However, these manuscripts did not explicitly consider
modelling the effects of pipe maintenance on other aspects of system operations and management, such as mechanisms for modifying the condition of pipes and the corresponding effects in
water supply systems management.
Given that the majority of investments made by water utilities are for pipe rehabilitation,
extension, and replacement, one cannot neglect the significance of pipe maintenance on water
supply services. In Korea, the main policy orientation until the end of the 1990s was to expand
water supply service with low investments in pipe rehabilitation and replacement. As a result of
this policy, the amount of water lost from deteriorated water pipes between 1998 and 2007 was
estimated at 8.4 billion tons (Korea Ministry of Environment 2009). Due to the enormous water
losses caused by leakage, and the heightened public awareness of tap water quality, the water
supply authorities in Korea have spent large amounts of money on pipe renewal in recent years.
In this context, an SD computer simulation model has been developed to aid the efficient
management of water supply systems. To develop the SD computer model, the conceptual framework for the working mechanisms of water supply systems was established, and then the causal
feedback loop relationships among the components of the systems management, including the
management of water pipes, were identified.
Data from a water supply service in South Korea were used as a case study to validate the models
structure. Some important indicators of the system under the study were analysed via management
scenario simulations, with sensitivity analyses conducted on some major indicators of the case
study system. This study was conducted in order to illustrate the developed SD models ability to
facilitate the identification of the best policy for achieving a specific management objective for a
water supply system (i.e. identify decisions which provide policy leverage).

2. The SD methodology
The SD Methodology is a simulation methodology based on systems theory. It deals with the
interpretation of the dynamic nature of systems in which information and material feedback are
present. The characteristics of systemic approaches adopted in the systems theory were well
presented by Beard (1999) in which 14 systemic ideas were provided, with each idea explained in

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Civil Engineering and Environmental Systems

191

terms of the associated philosophical concepts. The methodology can facilitate understanding of a
system by extracting structures essential to its working mechanisms, and, based on an analysis of
feedback structures inherent to the system, lead to development of efficient management strategies.
Computer simulation models that are developed based on an SD methodology are composed
of four basic components: stocks, flows, converters, and interrelations among them, which are
graphically represented as arrows and mathematically modelled as the finite difference equations.
The value of each component is calculated at each delta time for a specified simulation time
period defined in a model, starting at the initial values of the stocks and based on the functional
relations among components. Computer simulation experiments using an SD methodology are
realised using object-oriented modelling software such as Vensim, Powersim Studio, AnyLogic,
STELLA, etc. and generally take the following steps (Richardson and Pugh 1981).
First, relevant problems are identified and defined, in order to provide assurance that the
established model functions consistently with the purpose of the model. During the second step,
feedback phenomena inherent in a systems working mechanisms are conceptualised and a causal
loop diagram representing the cause and effect relations among the systems components is drawn.
A computer simulation model is created in the third step, based on the conceptualised working
mechanisms of a system. In the fourth step, the temporal behaviour of a system is analysed via
the computer simulation model and compared to the recorded behaviour of the system. In the
fifth step, the established model is calibrated and validated. Finally, the validated model is used
to analyse the behaviour of the system through various management scenarios. After this process
one can develop management strategies that meet the long-term operation goals of the system.
Figure 1 provides an example of an SD computer model that shows a causal feedback loop diagram of a reservoir system with outflows by gravity and the corresponding computer simulation
model using STELLA.
The plus (+) sign in Figure 1 implies that an effect is changed in the same direction as the
cause. In other words, the plus sign indicates a situation in which an effect is increased when
its cause is increased and vice versa. The minus sign () represents the opposite, a situation in
which the difference of the effects in the current and the previous time step is decreased when the
cause is increased and vice versa. The finite difference equations generated by STELLA for the
example model are given in Table 1.
Hannon and Ruth (2001) provide a general introduction to modelling dynamic systems and
many real-world examples involving chemistry, genetics, ecology, economics, and engineering.
There are also many textbooks that introduce the methodology for more specialised types of
systems such as economic (Ruth and Hannon 1997), business (Sterman 2000, McGarvey and

Figure 1.

Causal diagram and corresponding computer simulation model using STELLA.

192

S. Park et al.

Table 1.

Finite difference equations for the example model.

Stocks
WaterVolume(t) = WaterVolume(t dt) + (Out Flow) dt
INITIAL VALUE: WaterVolume = 100 [m3 ]

Outflows
Out Flow = WaterVolume Flow Rate
Flow Rate = 0.2 [m3 /hr/m3 ]

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Hannon 2004), biological (Hannon and Ruth 1997), and environmental (Ford 1999, Deaton and
Winebrake 2000).

3.

Previous studies on the applications of SD for water resources and water supply
services management

After Forrester (1958) introduced the concept of SD to model systems with complex feedback
structures, the applications of SD methodology have focused on modelling of management strategies and, personnel management and inventories in industrial fields. More recently, SD has been
developed into a general methodology that has been used to address management problems not
only in industrial areas, but also in other fields, such as public policy evaluation, engineering
problem solving, and social issues.
Drew et al. (1985) applied the SD methodology to the planning of a hypothetical civil and
environmental system that included water resources, sewage treatment, land use, and urban transportation. Water resources planning and management problems have especially benefitted from
the use of this methodology. Examples in this field include: Crawford and Linsleys (1966) comprehensive watershed simulation model, which included physical processes of water flows and
socio-economic factors of water management in a watershed, drought analysis by Keyes and
Palmer (1993) using SD, Shawwash and Russells (1997) SD model dealing with water management problems in Jordan, Palmers (1998) application of SD methodology to watershed planning,
application of the SD methodology as a decision support system for management of surface water
by Fletcher (1998), and studies by Simonovic, Fahmy, and Elshorbagy (1997) and Simonovic and
Fahmy (1999) on long-term planning and policy analyses for the use of water resources in the
Nile watershed, Egypt.
Application of SD methodology to the field of water resources management has continued
into the twenty-first century, including Ahmad and Simonovics (2000) work on applying SD
to reservoir operations for flood prevention, Teegavarapu and Simonovics (2000) SD model for
multiple reservoir operations for hydroelectric power generation, Staves (2000) application of
SD to policy determination for water resources problems in Las Vegas, Ahmad and Simonovics
(2001) study on SD as a decision support tool for evaluating the effects of flood protection policies,
the SD model developed by Li and Simonovic (2002) for prediction of floods due to snowmelt
in the prairies in of North America, Simonovics (2002) world water model, and a study by
Simonovic and Li (2003) that analysed the effects of climate change on the performance of flood
protection systems. Simonovic (2009) also presented the SD simulation along with optimisation
and multi-objective analysis as an integrated modelling tool for contemporary water resources
management problems. A comprehensive review of the applications of SD methodology for water
resource management problems can be found in Winz, Brierley, and Trowsdale (2009).
Applications of SD methodology to modelling of urban water supply service management
was initiated by Grigg and Bryson (1975), where a computer programme was used to develop
an SD model of the water supply system in Fort Collins, Colorado. The model consisted of
water balance, water use, population, and financial accounting sectors; it focused on controlling
the price of water while meeting the water demand of the growing population. Although Grigg

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Civil Engineering and Environmental Systems

193

and Brysons (1975) model captured several factors inherent to urban water management, some
important causal relations, like the one between condition of capital and investments, were not
considered. In addition to this, the model was not verified via calibration processes, even though
it was based on actual data.
Recently, Lee, Kim, and Park (2006) presented an SD model, used to assist with the management
of water supply systems; it modelled the dynamic nature of the water revenue ratio, supplied
population and financial status of water supply businesses, and mainly analysed the effects of
transparency of operational and managerial status of a case water supply system. Lee, Park, and
Park (2007) modified the model of Lee, Kim, and Park (2006) to more comprehensively analyse
managerial conditions of a case water supply system. However, the modifications to the model
made in Lee, Park, and Park (2007) were not of any significance. Even though Lee, Kim, and
Park (2006) and Lee, Park, and Park (2007) modelled the causal feedback mechanisms of pipe
extension and rehabilitation, they did not explicitly model the dynamic nature of pipe condition
due to deterioration, rehabilitation, and extension, or the causal feedback mechanisms among pipe
condition and other working components in water supply system management.

4.

Developing causal relationships of working mechanisms in water supply systems

management

4.1. Conceptual Framework for the Mechanism

The goal of establishing the underlying causal relationships in water supply systems management
was to identify key drivers of managerial action and the corresponding consequences of action.
As a result, the fundamental working mechanism of a water supply system was established
conceptually, in relation to the causal relationship between each component of the mechanism
(Figure 2).
In Figure 2, the state of a water supply system is represented using a Managerial Indicator.
A system may also have a Target for the indicator of interest. The gap between the Target and
current value of the Managerial Indicator plays a key role in driving the mechanism, as this
difference defines the Required Amount of Action. If the required amount cannot be fulfilled

Figure 2.

Conceptual framework for the working mechanism of water supply systems management.

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

194

S. Park et al.

in a given year, the Yearly Amount of Action is calculated based on the specified Planning
Time Period. During this period the indicator is to be incrementally improved, so that that the
Yearly Amount of Action will enable the indicator to reach target levels after the specified time
period. The Yearly Amount of Action put on a system will produce corresponding Consequences of Action and, as a result, the current state of the system represented by the Managerial
Indicator will be changed. After examining the polarity of the relationships between the causes
and effects, it was determined that the fundamental causal relationship of water supply systems
management has the form of a negative feedback loop with variables that exhibit stabilising
behaviour.
Typical Managerial Indicators of a water supply system are water supply ratio [dimensionless or %], revenue water ratio [dimensionless or %], and water price [the US dollar/m3 or
the Korean currency 1000 Won/m3 ]. Action to improve managerial conditions of water supply system may include capital investments or water price increases. Consequences of Action
are dependent on the nature of the Action and may include changes in the state of the systems pipelines. Through additional causal relationships between Consequences of Action and
Managerial Indicator, the Consequences of Action eventually contribute to changes in the
Managerial Indicator.
4.2. Realisation of a conceptual framework as a causal feedback loop diagram
A causal feedback loop diagram for water supply systems was developed based on a previously
established conceptual framework that represents the working mechanism of water supply systems management. Figure 3 shows the essential causal feedback loop relationships among the
components of the management mechanism of water supply systems identified in this study.

Figure 3.

Causal feedback loop relationships among the components of water supply systems management.

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Civil Engineering and Environmental Systems

195

The three main Managerial Indicators used to establish causal relationships were (water)
supply ratio [dimensionless], Revenue water ratio [dimensionless], and water price
[1000 Won/m3 ]. (Water) Supply ratio [dimensionless] represents the portion of the population that has access to a water supply system and has a value between 0 and 1. Revenue water
ratio represents the ratio of the amount of billed (or consumed) water to the total amount of
water produced and also has a value between 0 and 1. The differences between supply ratio
and target supply ratio, revenue water ratio and target revenue water ratio, and water price
and target price increase trigger the working mechanisms of water supply systems and play key
roles in driving the management mechanism of water supply systems. Therefore, the dynamics of
water supply was assumed to be driven by the differences between the targets and actual values
of the three main Managerial Indicators. The actual water demand was considered indirectly by
considering the causal effects among yearly consumed water, serviced population, and daily
consumed water per person.
As shown in Figure 3, the causal feedback relationship between supply ratio and (yearly)
investment costs for pipe extension [1000 Won/yr] was determined to be a negative feedback
loop in which the variables exhibit stabilising behaviour. This feedback loop was established
by examining the following causal relationship between investment costs for pipe extension
and supply ratio. Increases in investment costs for pipe extension will result in an increased
supply ratio. These increases in the supply ratio will in turn reduce the difference between
the supply ratio and target supply ratio which was initially assumed to be 1. Therefore, the
increase in required costs for supply ratio increase [1000 Won] in the current simulation time
step will be reduced compared to the previously calculated value for a specified one simulation time step earlier. Subsequently, increases in the investment costs for pipe extension in the
current simulation time step will also be reduced compared to the previously calculated values
completing the stabilising negative feedback loop that stems from investment costs for pipe
extension.
The causal feedback relationship between the revenue water ratio and (yearly) investment
costs for pipe rehabilitation [1000 Won/yr] was also determined to be a negative feedback loop.
In this stabilising feedback mechanism, increases in investment costs for pipe rehabilitation will
result in increased rehabilitated pipe lengths[km], reduced (yearly) leakage (amount)[m3 /yr],
increased revenue water ratio, decreased required pipe rehabilitation length[km], and reduced
increases in or stabilised investment costs for pipe rehabilitation. The difference between the
revenue water ratio and target revenue water ratio [dimensionless] plays a key role in this
stabilising feedback mechanism.
The causal feedback relationship between the water price and investment costs for pipe rehabilitation was modelled based on the following causal feedback mechanisms. First, an increase
in investment costs for pipe extension will yield an increase in supply ratio and revenue
water ratio but a reduction of increases in, or stabilisation of, prime cost (of water production)
[1000 Won/m3 ]. This will also stabilise (water) price increase (per year) [1000 Won/yr] due
to a decreased difference between the prime cost and water price. This in turn will stabilise
the variables water price, recognition of profitability [dimensionless], and investment costs
for pipe extension. However, the actual behaviours of the variables in the causal relationship
between prime cost and income (per year) [1000 Won/yr] may be dominated by the consumer price indices (CPI)-trend [dimensionless], which can cause the values of these variables
to continuously increase.
In addition, there are two other counteracting causal feedback loops affecting cost of investment: an increase in investment due to increased income and a decrease in investment due
to increased production costs (per year) [1000 Won/yr]. These two counteracting feedback
mechanisms are represented by the bifurcating routes from prime cost to total balance ratio
[dimensionless] via production costs and income. Total balance ratio was calculated using

196

S. Park et al.

Equation (1).
total balance ratio =

income
investment costs for pipe rehabilitation + investment costs for pipe extension

(1)

[1000 Won/yr] was used for the units of the variables in Equation (1) that are income,
investment costs for pipe rehabilitation, and investments costs for pipe extension.

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

5.

Description of the developed SD computer simulation model

An SD computer simulation model was developed based on the causal feedback loops shown in
Figure 3. The model was composed of three sub-models: a water supply sub-model, a pipe maintenance sub-model, and a water supply business finance sub-model. Figure 4 shows the essential
stock and flow diagram of the computer model constructed using STELLA. The ghost components, which have italicised names, shown in Figure 4 can be tracked to find inter-connections
between the sub-models.
The water supply sub-model examined changes in supply ratio due to population changes and
pipeline extension. It also examined long-term changes in total (volume of) water produced (per
year) [m3 /yr], resulting from changes in leakage due to pipe deterioration. In this sub-model,
supply ratio was modelled as a stock variable, whereas yearly supply ratio change [1/yr] was
modelled as yearly extended length of pipe [km/yr] divided by supply ratio improvement ratio
[km]. Supply ratio improvement ratio represents improvements made in the supply ratio, with
respect to changes in yearly extended length of pipe. The variable yearly (volume of) consumed

Figure 4.

Stock and flow diagram of the SD computer simulation model.

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Civil Engineering and Environmental Systems

197

water [m3 /yr] was calculated as the product of serviced population [persons] and (average)
daily consumed water (volume) per person [m3 /(dayperson)]. In this sub-model, the variable
yearly consumed water was also designed to influence the variable income in the water supply
business finance sub-model. Serviced population was modelled as the product of supply ratio
and total population which can be provided by the modeller as predicted values. Therefore, the
SD model developed in this study simplified modelling of water demand by considering only the
simple relationship among yearly consumed water, daily consumed water per person, serviced
population, total population, and supply ratio. A more sophisticated modelling is needed to
fully incorporate the effects of changes in water demand in the SD modelling of water supply
systems management.
In the pipe maintenance sub-model, condition of the pipes was examined, and pipes were
assigned one of three possible groups: deteriorated pipes [km], non-deteriorated pipes [km], or
disposed-off pipes [km]. Deteriorated pipes represented those pipes that were assumed to have
leaks caused by deterioration, while non-deteriorated pipesrepresented pipes not in a deteriorated
condition. It should be noted that the group non-deteriorated pipesincluded pipes ranging in condition from newly installed pipes to pipes in the last stage of non-deteriorated condition. In order
to account for change of condition over time non-deteriorated pipes was modelled to become
deteriorated pipes at a rate of deterioration rate of (non-deteriorated) pipe [dimensionless].
Disposed-off pipes represented pipes that were discarded during rehabilitation or replacement
activities. Based on consultation with management officials regarding pipe maintenance history in
the case study Water Supply Authority, pipes in the group non-deteriorated pipes were also modelled to become the group disposed-off pipesconsidering the historical non-optimal management
practice of the Authority that eliminated non-deteriorated but unnecessary pipes.
The three types of pipes were affected by the following indicators: disposal rate for deteriorated
pipe [dimensionless], disposal rate for non-deteriorated pipe [dimensionless], deterioration rate
of pipe, yearly deteriorated pipe length [km/yr], and yearly rehabilitated pipe length [km/yr].
The yearly deteriorated pipe length and yearly rehabilitated pipe length were themselves
affected by investment costs for pipe rehabilitation and investment costs for pipe extension in
the water supply business finance sub-model. The unit cost of pipe rehabilitation [1000 Won/km]
and unit cost of pipe extension [1000 Won/km] in the pipe maintenance sub-model were used, in
turn, to calculate investment costs for pipe rehabilitationand investment costs for pipe extension
in the water supply business finance sub-model.
The transition from non-deteriorated pipesto deteriorated pipesand from deteriorated pipes
to non-deteriorated pipes were modelled using yearly deteriorated pipe length and yearly
rehabilitated pipe length, respectively. The yearly rehabilitated pipe length was modelled as the
product of 1/(unit cost of pipe rehabilitation) and investment costs for pipe rehabilitation.
In the water supply business finance sub-model, the indicators able to represent the financial
status of a water supply system were modelled; they included income, production costs, investment costs for pipe rehabilitation, and investment costs for pipe extension. In this sub-model,
total balance ratio was modelled to have a direct relationship with income and an inverse relationship with investment costs for pipe rehabilitation and investment costs for pipe extension.
Target price increase [1000 Won/m3 ] was estimated as the difference between prime cost and
water price. Yearly price increases [1000 Won/(m3 yr)] were determined by dividing target
price increase by TAG3 [yr]. TAG3 represented the period of time over which target water
price increase is realised, or alternatively the period of time over which water price was to be
incrementally increased. The resulting model was structured such that yearly price increases
will enable the water price to reach target price increase after TAG3.
Investment costs for pipe rehabilitationwere determined based on required pipe rehabilitation
length [km], which was calculated by utilising revenue water improvement ratio [km] (which
represents the improvements in revenue water ratio with respect to changes in rehabilitated pipe

198

S. Park et al.

length). Investment costs for pipe extension were modelled using target supply ratio, supply
ratio, and supply ratio improvement ratio.

6.

Model verification and scenario analyses

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

6.1. Description of model input data

The developed SD computer model was verified using data from the water supply service of Busan,
South Korea. Using the necessary input data and results of the model simulation, predicted values
of the systems operational and managerial indicators were compared with historical data reported
by the Statistics on Busan Water Supply Services in 2008 (Busan Metropolitan City Water Supply
Authority, 2009), abbreviated to Busan (2009) in the following texts. Specifically, data in Busan
(2009) from 1995 to 2007 were used to run and verify the model.
Necessary input data for the model included initial stock variable values and appropriate values
or trends as functions of time for the exogenous variables. Exogenous variables may be interpreted
as inputs supplied beyond the assumed system boundary, independent of the working mechanism
of the system. Data for 1995 in Busan (2009) were used as the initial stock variable values, and
data from 1995 to 2007 were analysed to estimate the values or trends of the exogenous variables.
Since some of the necessary input data were not available, these were estimated using trial-anderror during the model verification processes. Determination of these values consisted of testing
assumed values for unavailable input data until the simulated results of the model closely followed
reported data for the case study system. Tables 24 show the stock and exogenous variables used
in the model as well as their values or trends. The time scale used in the computer simulation
model was designed so that time 1 in the model represented 1995. This definition of time was
the same across all variables, and has been named time in the following tables.

Table 2.

Stock and exogenous variables of the water supply sub-model.

Variable
type
Stock
variable

Exogenous
variable

Variable
name

Initial value/
function of time

Supply ratio

0.974

Accumulated leakage
Accumulated water
meter under
registration and other
non-revenue water
Accumulated nonrevenue water
Accumulated total water
produced
Total population

Unit

Source of initial value/

basis for estimation

95,378,573
42,444,546

Not applicable
(N.A.)
m3
3
m

Busan (2009)
Busan (2009)

172,049,104

m3

Busan (2009)

540,694,859

m3

Busan (2009)

Figure 5

Persons

Busan (2009)

Daily consumed water

per person

270

/(personday)

Supply ratio
improvement
ratio
Yearly leakage per unit
deteriorated pipe
length

6 109
loge (time) + 6
106
38,225e(0.087time)

1/km

Estimated during
the calibration
(model verification)
processes
Analysis of Busan
(2009)

(m3 /yr)/km

Busan (2009)

Analysis of Busan
(2009)

Civil Engineering and Environmental Systems

Table 3.

Stock and exogenous variables of the pipe maintenance sub-model.

Variable
type

Variable
name

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Stock variable

Exogenous
variable

km

Deteriorated pipes

3221

km

Disposed-off pipes

92

km

Unit cost of pipe

rehabilitation

If time 8 then
7957.7e0.1749time
else 150,000
0.06

1000 Won/km

Income per unit

disposed-off pipe

58,856loge (time) +
18784
If time < 7 then 0.08
else if time < 10
then 0.095 else if
time < 13 then 0.03
else 0.005
If time 13 then
0.015 else 0.004

200

Source of initial value/

Basis for estimation

Unit

5982

Disposal rate of
deteriorated pipe

Estimated as 65% of the

initial value of total
pipe length during the
calibration processes
Estimated as 35% of the
initial value of total
pipe length during the
calibration processes
Assumed as 1% of the
initial value of total
pipe length
Analysis of Busan (2009)

1/yr

Estimated during the

calibration processes
Analysis of Busan (2009)

1000 Won/km
1/yr

Estimated during the

calibration processes
utilising expert opinion
from personnel in the
Busan water authority
Estimated during the
calibration processes
utilising expert opinion
from personnel in the
Busan water authority
Consultation with the
personnel in the Busan
water authority

1/yr

1000 Won/km

Stock and exogenous variables of the water supply business finance sub-model.

Variable
type
Stock
variable

Initial value/
function of time

Non-deteriorated
pipes

Deterioration rate of
pipe
Unit cost of pipe
extension
Disposal rate of
non-deteriorated
pipe

Table 4.

199

Variable
name
Capital

Water price
Recognition of profitability
Exogenous TAG1 (time duration during
variable
which required costs for
pipe rehabilitation is spent)
CPI-trend
Delay time
TAG2 (time duration during
which required costs for
supply ratio increase is
spent)
Target revenue water ratio
Revenue water improvement
ratio
TAG3 (time duration during
which target water price
increase is realised)
Target service ratio

Initial value/
Function of time
141,577,480

Unit
1000 Won

Source of initial value/

basis for estimation
Busan (2009)

0.320
1000 Won/m3
0.4
N.A.
If time 7.5 then 2.5
yr
else if time 10
then 9 else 6
2.6797 time +
yr
70.161
1
yr
If time 7 then 5 else
yr
2.5

Busan (2009)
Assumed
Estimated during the calibration
processes

Maximum probable revenue

water ratio
Analysis of Busan (2009)

N.A.

Statistics Korea (2009)

Assumed
Estimated during the calibration
processes

5902.75

km

4.5

yr

Estimated during the calibration

processes

N.A.

Maximum probable service ratio

200

S. Park et al.

Table 5. Values or trends of exogenous variables for the scenario simulations.

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Variable name
Deterioration rate of
pipe (1/yr)
Disposal rate of
non-deteriorated
pipe (1/yr)
Disposal rate of
deteriorated pipe
(1/yr)
Total population
(persons)
CPI-trend

TAG1 (yr)
TAG2 (yr)

TAG3 (yr)

Optimistic scenario

Basic scenario

Pessimistic scenario

Fixed as 0.03 during the

scenario simulation period
Fixed as 0.006 during the
scenario simulation period

Fixed as 0.06 during the

scenario simulation period
Fixed as 0.005 during the
scenario simulation period

Fixed as 0.08 during the

scenario simulation period
Fixed as 0.003 during the
scenario simulation period

Fixed as 0.007 during the

scenario simulation period

Fixed as 0.004 during the

scenario simulation period

Fixed as 0.003 during the

scenario simulation period

Figure 5

Figure 5

Figure 5

Stabilised from 105 in year

2007 to 138 in year 2030
(52.299 time0.2699 )

Increased linearly from

Increased exponentially
105 in year 2007 to
from 105 in year 2007
167 in year 2030
to 199 in year 2030
(2.6797 time + 70.161)
(73.362 e0.0277time )
Fixed as 6 during the scenario Increased linearly from 6 in
simulation period
year 2007 to 10 in year 2030
(0.1739 time + 3.7391)
Fixed as 2.5 during the
Increased linearly from 2.5
scenario simulation period
in year 2007 to 4.5 in
year 2030 (0.087 time +
1.3696)
Fixed as 4.5 during the
Increased linearly from 2.5
scenario simulation period
in year 2007 to 6.5 in
year 2030 (0.087 time +
3.3696)

Decreased linearly from 6 in

year 2007 to 4 in year 2030
(0.087 time + 7.1304)
Decreased linearly from 2.5
in year 2007 to 1.5 in year
2030 (0.0435 time +
3.0652)
Decreased linearly from 4.5
in year 2007 to 2.5 in year
2030 (0.087 time +
5.6304)

The value for supply ratio improvement ratio was obtained by analysing data on length of
new pipe installed yearly and corresponding improvements in supply ratio. Similarly, the value
of revenue water improvement ratio was estimated based on available data on length of pipe
rehabilitated yearly and corresponding improvements made in the revenue water ratio. Values for
unit cost of pipe rehabilitation and unit cost of pipe extension were also based on actual data
for yearly rehabilitated/extended length of pipes and corresponding investments.

6.2. Management scenario analysis

Management scenarios for the case study system were created for the period between 2008 and
2030 using different values for the exogenous variables to predict the status of the system under
optimistic, basic, and pessimistic managerial conditions. The optimistic scenario simulated a managerial condition that would improve the systems financial status compared to the basic scenario,
while the pessimistic scenario was designed to simulate the opposite managerial condition.
Table 5 shows values or trends of the exogenous variables used to create the scenarios. Exogenous variables not given in Table 5 were assumed to retain the same values or functions given in
Table 4 during the scenario simulation period from 2008 to 2030. Figure 5 shows the value of
total population of Busan in each of three scenarios; since the citys population has continuously
declined for more than 20 years, population for the optimistic scenario was assumed to be the
same as for the basic scenario, with the exception of the period after 2020, in which population
was assumed to increase.
Figures 610 show the results of the scenario simulations for supply ratio, revenue water
ratio, total balance ratio, cost recovery ratio, total pipe length, and deteriorated pipes in
each of the three scenarios. Figures 610 also show the simulation results before 2008, which
were obtained using the data in Table 4. Since the input data used for the period before 2008 were

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Civil Engineering and Environmental Systems

Figure 5.

Scenarios for total population.

Figure 6.

Scenario simulation results for supply ratio.

201

the same as those used for the basic scenario, the label for the simulation results before 2008 was
also given as basic in Figures 610.
Equations (2) and (3) show equations used to calculate revenue water ratio and cost recovery
ratio, respectively, which were, together with total balance ratio, selected as important indicators
in the representation of managerial conditions of the case study system.
yearly consumed water
,
total water produced
water price
cost recovery ratio =
.
prime cost

revenue water ratio =

(2)
(3)

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

202

S. Park et al.

Figure 7.

Scenario simulation results for revenue water ratio.

Figure 8.

Scenario simulation results for total balance ratio.

Units of the variables in Equations (2) and (3) are as follows: [m3 /yr] for yearly consumed water
and total water produced; and [1000 Won] for water price and prime cost, respectively.
By comparing historical data and corresponding simulation results for the model variables, as
shown in Figures 610, it was determined that the developed SD model succeeded in reasonably
simulating historical trends of the case study system. Figure 6 revealed that, in the basic scenario,
the systems supply ratio should essentially reach a state of dynamic equilibrium around 2015;
results of the optimistic and pessimistic scenarios were almost identical. Under the basic scenario,
cost recovery ratiois also expected to reach a state of dynamic equilibrium around 2015, as shown
in Figure 9. However, as shown in Figures 7 and 8, revenue water ratio and total balance ratio
are predicted to continuously improve under any assumed scenario.

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Civil Engineering and Environmental Systems

Figure 9.

Figure 10.

203

Scenario simulation results for cost recovery ratio.

Scenario simulation results for pipe length changes.

Figure 10 shows the scenario simulation results for total pipe length and deteriorated pipes.
Since there are no historical records for deteriorated pipes in the system, only the simulation
results of deteriorated pipes are shown in Figure 10. From Figure 10, total pipe length was
observed to remain within the range of 75008000 km during the planning horizon, but the general
condition of pipes in the system will deteriorate in any of the assumed scenarios.

7. Analysis of exogenous variable sensitivity and identification of policy leverage

The policy leverage for a system may be defined as the policy that most efficiently directs a
given system to its desired state; this policy can be identified using experiments with various

204

S. Park et al.

Table 6.

Results of the sensitivity analyses of the exogenous variables.

Index
Revenue water ratio
Cost recovery ratio

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Total balance ratio

Exogenous
variable

Change rate of
corresponding index

Maximum change
occurrence time (yr)

TAG1: 6 yr 1.5 yr
Deterioration rate of pipe: 0.06/yr 0.015/yr
TAG3: 4.5 yr 1.125 yr
TAG1: 6 yr 1.5 yr
Deterioration rate of pipe: 0.06/yr 0.015/yr
TAG3: 4.5 yr 1.125 yr
TAG2: 2.5 yr 10 yr
Deterioration rate of pipe: 0.06/yr 0.015yr

2% (91% 93%)
2% (92% 94%)
7% (90% 97%)
1% (90% 91%)
1% (90% 91%)
7% (83% 90%)
3% (70% 73%)
2% (80% 82%)

2014
2017
2013
2010
2011
2020
2008
2015

management alternatives on a constructed SD model. Exogenous variables may be regarded as

inputs to a system, and work as operational or managerial decisions. Additionally, exogenous
variables in an SD model trigger changes in the systems behaviour by providing a means for
policy implementation, and can, therefore, be regarded as candidates for policy leverage.
The policy leverage that will enable a system to reach goals most efficiently can be determined
by conducting analyses of the sensitivity of the exogenous variables on major operational or
managerial indicators within a system. These sensitivity analyses record the behaviour of major
indicators within the system according to changes in the exogenous variables values. The indicators
analysed to obtain policy leverage of the case study system were revenue water ratio, total
balance ratio, and cost recovery ratio. The exogenous variables selected to analyse the behaviour
of these indicators were TAG1, TAG2, TAG3, deterioration rate of pipe, disposal rate of
non-deteriorated pipe, and disposal rate of deteriorated pipe.
The changes in the selected system indicators were observed after calibrated values of the
exogenous variables used during the model verification (given in Tables 24) were multiplied by
1/4, 1/2, 1, 2, and 4. Only one of the selected exogenous variables was allowed to change in a
given model run; this method allowed modellers to find the exogenous variable with the most
significant effect on an index.
Table 6 shows results of the sensitivity analyses of the exogenous variables on revenue water
ratio, total balance ratio, and cost recovery ratio. For example, a 2% maximum increase in
revenue water ratio (i.e. from 91% to 93%) was observed in 2014 after reducing TAG1 from
6 to 1.5 years (6 being the previously calibrated value of TAG1). Decreasing deterioration
rate of pipe from 0.06 to 0.015/yr had a similar impact on revenue water ratio, resulting in
a 2% maximum increase in 2017, i.e. from 92% to 94%. Improvements made by changes in
other exogenous variables on revenue water ratio were not as significant as those in TAG1 and
deterioration rate of pipe.
Although TAG1 and deterioration rate of pipe had similar effects on revenue water ratio,
adjusting TAG1 may be a more practical policy than changing deterioration rate of pipe,
given the obvious difficulty in controlling deterioration rate of pipe. Since TAG1 determines
investment costs for pipe rehabilitation, a policy to adjust TAG1 may be stated in more general
terms as an adjustment in yearly allocation of funds for pipe rehabilitation. It was concluded that
adjusting TAG1 or the yearly allocation of funds for pipe rehabilitation is the policy leverage
that would most effectively improve revenue water ratio. The important role of TAG1 in
determination of revenue water ratio can also be found through the causal relations of the model
shown in Figure 3. In this model TAG1 determines investment costs for pipe rehabilitation,
rehabilitated pipe length, yearly deteriorated pipe length, leakage, and revenue water ratio.
Results of the sensitivity analyses on cost recovery ratio and total balance ratio are also
summarised in Table 6. Results for cost recovery ratio reveal that a reduction of TAG3 or

Civil Engineering and Environmental Systems

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Table 7.

205

Effects of deterioration rate of pipe (1.5% and 24%).

Calculation
criteria

Deteriorated
pipes (km)

Leakage
(m3 /yr)

Total balance
ratio (%)

Revenue water
ratio (%)

Absolute maximum differences

Absolute maximum change ratios
with regard to base condition (%)
Absolute max occurrence times (yr)

2215; 2337
37; 48

7,611,940; 16,318,230
26; 47

3; 5
3; 6

2; 4
2; 4

2030; 2016

2015; 2012

2015; 2012

2017; 2012

an increase in price of water (achieved by yearly price increases in the model) was the policy
leverage that most effectively improved cost recovery ratio.
Results of TAG2on total balance ratioin Table 6 show that total balance ratiowas increased
by 3% in 2008 when a TAG2 of 10 years was used. However, differences in the values of total
balance ratio between TAG2 values of 2.5 and 10 years gradually diminished over time, and
after 2014 values of total balance ratio for a TAG2 of 10 years became lower than that for a
TAG2 of 2.5 years. Therefore, even though total balance ratio was improved when TAG2 was
increased, i.e. when investment costs for pipe extension was decreased, the improvements made
in total balance ratio by increasing TAG2 were only temporary.
Observations of the long-term behaviour of total balance ratioas a result of changes in TAG2,
combined with improvements made in total balance ratio as a result of changes in TAG3, lead
to the conclusion that an increase in income that comes from raising water price more efficiently
increases total balance ratio than a decrease in investment spending. Therefore, TAG2 may
not be considered an effective policy to improve total balance ratio of the system, as TAG3 is
considered to be the policy leverage for improving total balance ratio.
A sensitivity analyses of TAG1versus total balance ratioshowed that the effects of decreasing
TAG1 were a decrease in total balance ratio and an increase in revenue water ratio. However,
further analyses revealed that the decrease in total balance ratio caused by a decrease in TAG1
gradually recovered during the planning period. Therefore, it was concluded that reducing TAG1,
or increasing the yearly allocation of funds for pipe rehabilitation, is the policy leverage to improve
revenue water ratio as well as an efficient policy for overall improvement of the managerial
conditions of the system.
Additional sensitivity analyses were conducted to analyse the effects of the exogenous variables related to pipe management. Calibrated values for deterioration rate of deteriorated pipe,
disposal rate of deteriorated pipe, and disposal rate of non-deteriorated pipe, which are given
in Table 3, were multiplied by 1/4, 1/2, 1, 2, and 4, and changes in the selected system indicators were observed after each model run, each incorporating variation in only one of the given
exogenous variables.
The second rows of Tables 79 represent the absolute maximum differences between system
indicator simulation results under the base condition (given in Table 4) and those under the altered
condition (shown in the titles of the corresponding tables). The third rows of Tables 79 represent
the absolute maximum ratios of change in the corresponding indicators, while the fourth rows in
Tables 79 represent the times at which the absolute maximum differences were observed.
To illustrate the results given in Tables 79, results for deteriorated pipes in Table 7 are
explicated below; the other results in Tables 79 follow a similar convention. In Table 7, 2215 km
of deteriorated pipes were predicted for 2030, which was the maximum among the absolute
differences for deteriorated pipes; the absolute difference was calculated for every year in the
planning horizon. These absolute differences are the differences between the base condition, with
deterioration rate of pipe value of 6% and the case with a deterioration rate of pipe value of
1.5%. The corresponding ratio of maximum difference for the deteriorated pipes in the case of
a deterioration rate of pipe of 1.5% compared to the base condition was calculated to be 37%.

206

S. Park et al.

Table 8.

Effects of disposal rate of deteriorated pipe (0.1% and 1.6%).

Calculation
criteria

Accumulated disposed-off
pipes (km)

Deteriorated
pipes (km)

Revenue water
ratio (%)

Total balance
ratio (%)

342; 1182
6; 20

327; 1127
5; 20

0.2; 0.7
0.2; 0.7

0.2; 0.8
0.3; 1

2030; 2030

2030; 2030

2023; 2022

2022; 2019

Absolute maximum differences

Absolute change ratios with regard
to base condition (%)
Absolute max occurrence times (yr)

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Table 9.

Effects of disposal rate of non-deteriorated pipe (0.125% and 2%).

Calculation
criteria
Absolute maximum differences
Absolute change ratios with
regard to base condition (%)
Absolute max occurrence
times (yr)

Accumulated
disposed-off
pipes (km)

Nondeteriorated
pipes (km)

Deteriorated
pipes (km)

Revenue
water
ratio (%)

Total
balance
ratio (%)

252; 863
4; 15

130; 447
7; 20

123; 434
2; 7

0.06; 0.2
0.07; 0.2

0.08; 0.3
0.1; 0.4

2030; 2030

2030; 2025

2030; 2030

2026; 2025

2023; 2023

The values 2337 km, 48%, and 2016 in Table 7 corresponded to results associated with the case
in which a deterioration rate of pipe value of 24% was used.
As given in Table 7, noticeable changes in deteriorated pipes and leakage were observed due
to changes in deterioration rate of pipe. However, other important indicators, such as revenue
water ratio and total balance ratio, were not significantly changed. Also, analyses on the effects
of disposal rate of deteriorated pipe, the results of which are given in Table 8, revealed that
noticeable changes only occurred in accumulated disposed-off pipes and deteriorated pipes,
as no significant changes occurred in other important indicators, such as revenue water ratio and
total balance ratio.
Similar to deterioration rate of pipe, changes in disposal rate of non-deteriorated pipe
only had noticeable effects on accumulated disposed-off pipes, deteriorated pipes, and nondeteriorated pipes (Table 9). Based on these observations, deterioration rate of pipe, disposal
rate of non-deteriorated pipe, and disposal rate of deteriorated pipe were not considered as
the policy leverage for the major system indicators, such as revenue water ratio and total
balance ratio.

8.

Summary and conclusions

In this paper, a conceptual framework for management of water supply systems was established,
leading to identification of the working mechanisms inherent in water supply systems management based on causal feedback relationships. An SD computer simulation model was subsequently
presented based on these causal feedback relationships. In particular, the key factors and interrelationships associated with management of water pipes were identified and modelled based on the
SD modelling principles. An example of a modelled causal feedback relation is the causal feedback loop, which includes income, investment costs for pipe rehabilitation, rehabilitated pipe
length, yearly deteriorated pipe length, leakage, non-revenue water, total water produced,
revenue water ratio, prime cost, and water price.
The aim of model calibration in this study was to simulate the historical data reported in Busan
(2009), which were used to calibrate the model as closely as possible. Among the required input
values for the model, those that could not be obtained from the case study data, such as the initial

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

Civil Engineering and Environmental Systems

207

values of the stock variables, and values or trends of the exogenous variables, were assumed
through the model calibration and verification processes. The calibration process also took into
account expert opinions of managerial personnel of the case study system. During calibration, a
comparison of the simulated results and historical data of the model variables showed that the
constructed model reasonably simulated the historical trends of the case study system.
The analyses conducted on model simulation results were based on assumed management
scenarios, and showed that supply ratio and cost recovery ratio of the system were predicted to
reach a state of dynamic equilibrium under the basic scenario in approximately 2015. However,
revenue water ratio and total balance ratio were predicted to continuously improve in all of the
assumed scenarios. Meanwhile, pipes in the system were found to be in a deteriorating stage, even
in the most optimistic scenario. Therefore, major investments for pipe rehabilitation, even greater
than those assumed in the optimistic scenario, may need to be implemented to reduce amounts of
pipe deterioration in the system.
To identify the policy leverage for the case study, system changes for revenue water ratio, total
balance ratio, and cost recovery ratio, with respect to changes in exogenous variables, such as
TAG1, TAG2, TAG3, deterioration rate of pipe, disposal rate of non-deteriorated pipe,
and disposal rate of deteriorated pipe were analysed. These analyses showed that a reduction in
TAG1 or an increase in yearly allocation of funds for pipe rehabilitation was the policy leverage
for improvement in revenue water ratio. A reduction in TAG3 or increased water price was
the policy leverage for improvement in cost recovery ratio and total balance ratio.
Further sensitivity analyses of the exogenous variables showed that deterioration rate of pipe,
disposal rate of non-deteriorated pipe, and disposal rate of deteriorated pipehad no considerable
impact on some of the major indicators, such as revenue water ratio and supply ratio. It was
concluded that the limited effect of deterioration rate of pipe may have resulted from the models
own structure, since the direct costs of pipe repair and indirect costs related to pipe failure, such
as traffic congestion and flooding of nearby building basements, were not considered. Further
study will be required to model the working mechanism of indirect costs in water supply services
management. The model may need to be further expanded to include both direct and indirect costs
related to pipe failure.
Through the sensitivity and scenario analyses illustrated in this paper, the SD model developed
for water supply systems was shown to be sufficient in identification of policy leverage, leading
to efficient water supply system management; the model could also be utilised to determine longterm effects of policy change on the status of a water supply system. The principles associated
with establishing the causal relationships used in the SD computer modelling and the sensitivity
analysis methods for exogenous variables used for identifying policy leverages are also expected
to work as prototypical methods for modelling and solving the management problems of other
water supply systems.

Acknowledgements
This subject is supported by Korea Ministry of Environment as Projects for Developing Eco-Innovation Technologies
(GT-11-G-02-001-1).

References
Ahmad, S., and S. P. Simonovic. 2000. System Dynamics Modeling of Reservoir Operations for Flood Management.
Journal of Computing in Civil Engineering 14 (3): 190198.
Ahmad, S., and S. P. Simonovic. 2001. A Decision-Support Tool for Evaluation of Impacts of Flood Management
Policies. Hydrological Science and Technology 17 (1): 1122.
Beard, A. N. 1999. Some Ideas on a Systemic Approach. Civil Engineering and Environmental Systems 16 (3): 197209.

Downloaded by [Universiti Teknologi Malaysia] at 08:04 22 September 2015

208

S. Park et al.

Busan Metropolitan City Water Supply Authority. 2009. 2008 Statistics on Water Supply Services. Busan: Republic
of Korea.
Crawford, N. H., and R. K. Linsley. 1966. Digital simulation in hydrology: Stanford Watershed Model IV. Technical
Report 39. Stanford, CA: Civil Engineering Department, Stanford University.
Deaton, M. L., and J. I. Winebrake. 2000. Dynamic Modeling of Environmental Systems. New York: Springer.
Drew, D. R., L. C. Wadhwa, I. B. Santoso, and S. M. Moussavi. 1985. Scenariopolis: The Civil Engineering Systems
Policy Research Laboratory. Civil Engineering Systems 2 (2): 6676.
Fletcher, E. J. 1998. The Use of System Dynamics as a Decision Support Tool for the Management of Scarce Water
Resources. International conference on new information technologies for decision making in civil engineering,
Montreal, Canada, Vol. 1 (2), 909920.
Ford, A. 1999. Modeling the Environment. Washington, DC: Island Press.
Forrester, J. W. 1958. Industrial Dynamics A Major Breakthrough for Decision Makers. Harvard Business Review 36
(4): 3766.
Grigg, N. S., and M. C. Bryson. 1975. Interactive Simulation for Water System Dynamics. Journal of Urban Planning
Division 101 (1): 7792.
Hannon, B., and M. Ruth. 1997. Modeling Dynamic Biological Systems. New York: Springer.
Hannon, B., and M. Ruth. 2001. Dynamic Modeling (Modeling Dynamic Systems). New York: Springer.
Keyes,A. M., and R. Palmer. 1993. The Role of Object-Oriented Simulation Models in the Drought Preparedness Studies.
20th annual national conference of Water Resources Planning and Management Division, New York, 479482.
Korea Ministry of Environment. 2007. 2006 Statistics on Water Supply Services. Seoul: Republic of Korea.
Korea Ministry of Environment. 2009. Press Release: 2nd International Symposium for Integrated Water Network System.
Ministry of Environment Web. Accessed June 25, 2012. http://www.me.go.kr/web/288/me/common/board/detail.
do?idx=170537&decorator=me&searchOption=all&boardId=notice_02&categoryId=02&searchWord=%BB%
F3%BC%F6%B0%FC%B8%C1. (in Korean)
Lee, S., H. Kim, and H. Park. 2006. Analyzing Effects of Transparency on a Water Business Case Using System
Dynamics. Journal of the Korean Society of Water and Wastewater 20 (4): 605616. (in Korean)
Lee, S., H. Park, and H. Park. 2007. Application of System Dynamics Methodology for Comprehensive Analysis of the
Water Business System. Journal of the Korean Society of Water and Wastewater 21 (1): 6573. (in Korean)
Li, L., and S. P. Simonovic. 2002. A System Dynamics Model for Predicting Floods from Snowmelt in North American
Prairie Watersheds. Hydrological Processes 16 (13): 26452666.
McGarvey, B., and B. Hannon. 2004. Dynamic Modeling for Business Management. New York: Springer.
Palmer, R. 1998. A History of Shared Vision Modeling in the ACT-ACF Comprehensive Study: A Modelers Perspective.
In Special session of ASCEs 25th Annual Conference on Water Resources Planning and Management, edited by
W. Whipple Jr., 221226. Reston, VA: ASCE.
Richardson, G. P., and A. L. Pugh. 1981. Introduction to System Dynamics Modeling with DYNAMO. Cambridge, MA:
The MIT Press.
Ruth, M., and B. Hannon. 1997. Modeling Dynamic Economic Systems. New York: Springer.
Shawwash, Z. K., and S. O. Russell. 1997. Use of Interactive Simulation Environments for Managing Water in Jordan.
The 1996 international system dynamics conference, Cambridge, MA, July. CSCE, 156169.
Simonovic, S. P. 2002. World Water Dynamics : Global Modeling of Water Resources. Journal of Environment
Management 66 (3): 249267.
Simonovic, S. P. 2009. Managing Water Resources: Methods and Tools for a Systems Approach. London: Earthscan.
Simonovic, S. P., and H. Fahmy. 1999. A New Modeling Approach for Water Resources Policy Analysis. Water Resources
Research 35 (1): 295304.
Simonovic, S. P., H. Fahmy, and A. Elshorbagy. 1997. The Use of Object-Oriented Modeling for Water Resources
Planning in Egypt. Water Resources Management 11 (4): 243261.
Simonovic, S. P., and L. Li. 2003. Methodology for Assessment of Climate Change Impacts On Large-Scale Flood
Protection System. Journal of Water Resources Planning and Management 129 (5): 361372.
Statistics Korea. 2009. 2008 Consumer Price Index Survey. Seoul: Republic of Korea.
Stave, K. A. 2000. Using System Dynamics Models to Facilitate Public Participation in Water Resources Management: A
Pilot Study Using the Las Vegas, NV Water System. 18th international conference of the System Dynamics Society,
Bergen, Norway, August 610.
Sterman, J. D. 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World. New York: Irwin/
McGraw-Hill.
Teegavarapu, R., and S. P. Simonovic. 2000. System Dynamics Simulation Model for Operation of Multiple Reservoirs.
10th World Water Congress: Water, the Worlds Most Important Resource. Melbourne, Australia. International Water
Resources Association, 10091016.
Winz, I., G. Brierley, and S. Trowsdale. 2009. The Use of System Dynamics Simulation in Water Resources
Management. Water Resources Management 23 (7): 13011323.