Amarasinghe, Pradeep Barnes, Paul H. Egodawatta, Prasanna Mcgree, James Goonetilleke, Ashantha
Amarasinghe, Pradeep Barnes, Paul H. Egodawatta, Prasanna Mcgree, James Goonetilleke, Ashantha
Amarasinghe, Pradeep Barnes, Paul H. Egodawatta, Prasanna Mcgree, James Goonetilleke, Ashantha
ABSTRACT
This paper presents an approach for identifying the limit states of resilience in a water supply system
when influenced by different types of pressure (disturbing) forces. Understanding of systemic resilience
facilitates identification of the trigger points for early managerial action to avoid further loss of ability to
provide satisfactory service availability when the ability to supply water is under pressure. The approach
proposed here is to illustrate the usefulness of a surrogate measure of resilience depicted in a three
dimensional space encompassing independent pressure factors. That enables visualisation of the transition
of the system-state (resilience) between high to low resilience regions and acts as an early warning trigger
for decision-making. The necessity of a surrogate measure arises as a means of linking resilience to the
identified pressures as resilience cannot be measured directly. The basis for identifying the resilience
surrogate and exploring the interconnected relationships within the complete system, is derived from a
meta-system model consisting of three nested sub-systems representing the water catchment and
reservoir; treatment plant; and the distribution system and end-users. This approach can be used as a
framework for assessing levels of resilience in different infrastructure systems by identifying a surrogate
measure and its relationship to relevant pressures acting on the system.
Key words: Limit states, systemic resilience, resilience surrogate, water supply system, trigger
points
1. INTRODUCTION
The term “infrastructure systems” is used in its broadest sense, encompassing both built infrastructure
(buildings, roads, bridges, pipe networks, treatment facilities) and infrastructure services that rely on
integrated built and natural systems to provide fundamental needs of society. It is natural that these
infrastructure systems are subject to different type of disturbances. “Resilience” (as used here) refers
to the ability of such infrastructure systems (including their interconnected ecosystems and social
systems) to absorb the effect of disturbances and still retain their basic function and structural
capacity.
When a system is subject to effects of disturbances, the ability of the system to absorb the ‘pressure’
determines the magnitude of the effect that it can withstand, without transforming into a different state
(i.e: with reduced functionality). In a water supply system this transformation may be from
‘satisfactory state’ to ‘unsatisfactory state’. Knowledge of the resilience of the system to identified
pressures gives an insight to the potential operational range of the system within which [it] can sustain
outputs within set acceptable parameters. An ability to note when the system is transitioning from
acceptable to un-acceptable state is an important capability for both owners/operators and regulators
of such essential service. However, in order to undertake such assessments, it is necessary to precisely
define the system boundaries and the thresholds. Our primary objective for identifying resilient
1
characteristics of a water supply system is to use the knowledge of the dynamics of the system as an
early warning, allowing for proactive precautionary measures to be enacted. Thus, this paper outlines
an approach to identify the limit states of resilience of a water supply system relative to a defined
failure criterion under the influence of two categories of pressure: variation of rainfall and consumer
demand.
The ‘ability to withstand’ pressure (some degree of disturbance) while maintaining functionality is the
principle definition of resilience used in this study. However, other interpretations of resilience can be
found in research literature such as, amount of disturbance that a system can absorb without changing
state (Gunderson 2000)1, measure of the persistence of systems (Holling 1973)2, measure of the speed
of a systems return to equilibrium (Brock et al 2002)3, potential of a system to remain in a particular
configuration (Walker et al. 2002)4, and buffer capacity or the ability of a system to absorb
perturbations, or the magnitude of disturbance that can be absorbed (Adger 2000)5. The fundamental
premise of these definitions is similar, but addresses different aspects of the resilience concept.
In order to assess the selected resilience characteristic/s of a water supply system, it is necessary to
define the ‘system’ properly. Therefore, the initial steps of the process are based on modelling a
complete water supply system as a meta-system and identifying a surrogate measure of resilience
applicable to the system.
A water supply system consists of a water catchment and the reservoir; treatment plant; and the
distribution system and end-users. These components range from socio-ecological to technical
domains making a complex combination of different sub-systems. The concept of a ‘meta-system’
presents a holistic approach that avoids the limitation of considering each sub-system separately in
relation to the same pressure. A meta-system in the case of a water supply system is formed by
considering the integration of the three sub-systems as shown in the Figure 1 below.
Urban users
(Socio-technical subsystem)
In typical operation, all the components may operate below the maximum capacity as it needs only to
satisfy the demand. However, each sub-system illustrated in Figure 1, has a maximum capacity for the
appropriate operation. For example, the reservoir has a maximum storage capacity and the treatment
plant has a maximum treatment capacity. The final output of the system is determined by sub-system
which has the lowest maximum capacity as that will limit the potential of the system. Therefore,
although another sub-system has an excess capacity, it will serve only as an excess capacity that may
not be useful under normal operational conditions. However, this excess storage capacity could act as
a redundant capacity which can be applied in a crisis situation to enhance the resilience of the system.
The challenge is to recognise the response of the system to pressure variations. For this purpose, a
dynamic parameter applicable to the complete system that responds to pressure variations needs to be
identified. Resilience characteristics of the system can then be described by analysing response
variations of this parameter to varying pressure levels.
2
3. CASE STUDY – SEQ WATER GRID
The South East Queensland water supply system (SEQ Water Grid) was selected as the case study for
assessing systemic resilience. SEQ Water Grid is a diverse system of water storages and treatment
facilities with interconnecting bulk water pipelines. The main infrastructure of SEQ Water Grid
includes 12 connected dams, 10 connected drinking water treatment plants, 3 advanced water
treatment plants producing purified recycled water, 1 desalination plant, 28 water reservoirs, 22 bulk
water pump stations and 535 km of potable bulk water mains (SEQ Water 2012)7. The service area of
SEQ Water grid includes retail distributing areas of Sunshine Coast, Moreton Bay, Logan, Gold
Coast, Redlands, Brisbane, Somerset, Lockyer Ipswich and Scenic Rim local government areas as
shown in Figure 2. The system yield is about 350,000 ML/a compared to current demand of about
290,000 ML/a (Spiller et al. 2011)8. Water Grid transfers an average of 600ML/d, where water is
needed most (Link Water 2012)9.
Historical records help to understand climate change effects on rainfall variations. However, annual
and seasonal average rainfall is also influenced by local factors (such as topography, vegetation) and
broader scale weather patterns (El Nino-Southern Oscillation –ENSO- events). The annual rainfall
averaged over the last decade for SEQ region has decreased by 18% compared to the 1961-1990
average (BOM 2008)11. High resilience of the system to low rainfall conditions play an important role
under a reduced rainfall regime to maintain undisturbed service.
3
Figure 3 indicates the Stock and Flow diagram for a single catchment and reservoir. For development
of the complete model all the catchments of the Water Grid need to be combined.
Precipitation
Ev aporation
Fig. 3- Stock and Flow diagram for a single catchment and reservoir
The following section discuss the relatioships and the equations used to develop model inputs
parameters. The quantity of rainfall excess (runoff) in to the reservoir was obtained using the
following basic equation by Wanielista et al. (1997)12.
R = CP
However, the volumetirc runoff coefficicient of each catchment had to be derived. The volumetric
coefficient was calculated for relevant catchments by using regression analysis for historical rainfall
and stream flow data. Closest possible rainfall data were obtained and using a scatter plot, the best fit
line was drawn to derive the volumetric runoff coefficient (C) value. For verification of the C values,
the actual storage data of each reservoir for each month (during the period 2008 to 2010) were
obtained and then compared with simulated storage volumes using the derived C values. For very low
monthly rainall values, the runoff generaton was observed as negligible. Therefore, the volumetric
runoff coefficient (C) was considerd as ‘0’ for monthly rainfall less than 50mm.
The main stock at the catchment level is the water storage in the reservoir. Generally, stock
accumulates the difference between its inflow and outflow. Thus the accumulation of stock with
inflow and outflow can be formulated as;
tn
Stock(t) = ∫
t0
( inflow (t) – outflow (t) ) dt + Stock (t 0 )
4
Accordingly, the water balance of each reservoir was considered as;
tn
The following relationship was used to compute the evaporation from reservoirs based on CSIRO
(2008)13 and Kohler et al. (1955)14 recommendations. For estimation of evaporation from reservoirs,
the pan coefficient was recommended between 0.6-0.8 in the above references and hence 0.75 was
considered as the pan coefficient in this study.
E r = PE x C p x SA r
The surface area does not remain constant as the storage level varies. Therefore the actual water
surface area was obtained when the reservoir is full and considered that the water surface area is zero
when the reservoir is empty. The data for water surface area were obtained by interpolating between
the maximum and minimum value.
Water drawn to the relevant treatment plant was allowed only if the storage level of the reservoir was
greater than 20% of the full capacity and limited to the maximum capacity level of the treatment
plant. When water storage was greater than the reservoir capacity, the excess water was released. The
in-built mathematical functions were used to allow these conditional directives in the model.
The potential output of the system was obtained by simulating under variable pressure conditions.
Rainfall was reduced by 10% at a time and demand was increased by 10% increment at a time. 50
simulations were run for a five year period for combination of pressure conditions. Inability to supply
up to failure threshold in any month in the five year period, was denoted as a failure. Number of
failure simulations out of 50 was considered as the probability of failure.
In the process of exploring the selected resilience characteristics of the system, two mandatory
requirements were considered. They are the existence of a pressure (a disturbing force/s) and a
method to assess the ability of the system to withstand the pressure and not fail. The ‘ability’ of the
system is assessed by performance variations. In order to assess performance variations, the dynamic
process of output production has to be identified.
The dynamic process of a water supply system is the ‘water flow’ and the physical measure of
performance is the potential supply ‘volume’ of potable quality water. Therefore the ‘ability’ under
pressure can be assessed by the likelihood (probability) of failure. For example, considering a
particular level of pressure, lower failure probability (say 1%) indicates higher ability while higher
failure probability (say 80%) indicates low ability to withstand pressure. Therefore, the surrogate
measure of resilience of the water supply system considered here is the probability of failure.
However, the failure criterion should be defined to evaluate probability of failure.
The failure criterion of the system is defined here as follows. The system is expected to supply water
for the population of a particular region. Based on an average (pre determined) per capita
consumption, the expected demand of the system for that population can be determined. The failure
state of the system is defined here such that the maximum supply level of the system fails to satisfy
5
50% of the demand. It is assumed that regulations such as water restrictions can reduce typical
demand to 50% of the average demand in a significant pressure situation. It is also important to
specify the conditions under which the resilience is assessed.
The probability of failure can be determined under given conditions. However, due to the diversity of
the resilience concept, a mathematical relationship between failure probability and resilience is
difficult to derive. Therefore, at this stage, resilience is conceptualised as an inversely proportional
vector to failure probability. The relationship of surrogate measure (probability of failure) and the
pressures is developed based on the following relationship as resilience is inversely proportional to
probability of failure:
1
𝑅𝑒𝑠𝑖𝑙𝑖𝑒𝑛𝑐𝑒 =>
𝑃𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑜𝑓 𝑓𝑎𝑖𝑙𝑢𝑟𝑒
Probability of failure is disaggregated further in order to relate it to the adverse pressure conditions.
Probability of failure is expected to be higher with increase in adverse pressure conditions. Therefore,
the adverse pressure is one factor that influences failure. As the failure threshold is defined here as a
function of demand, demand is also another factor that influence failure. Hence, the relationship can
be expressed as given below;
Depends on
Probability of failure (adverse pressure, demand based failure threshold)
In order to determine probability of failure using the above relationship, the adverse pressure
conditions that create stresses on the system should be identified. The following section discusses the
main stress generating factors.
As the main purpose of a water supply system is to supply the designed quantity of potable water, any
force that creates an adverse effect to disturb output generation is a pressure and this pressure creates
stress on the system. The stresses are the ‘less water availability’ and ‘low water quality’. Climate
change impacts to reduce inflow to the reservoir due to low rainfall.
Apart from the reduction of inflow, climate change can also contribute to degrade water quality
(Delpla et al. 200915, Park et al. 201016, Ducharme 200817). For example, increased temperature
associated with high nutrient loads can lead to eutrophication and algae growth in the reservoir. When
the raw water quality is degraded, the treatment plant may not be able to produce its maximum
capacity, lowering the rate of system output (in terms of volume). Therefore, climate change is a
major pressure generator on a water supply system which is dependent on surface water for storage.
Rainfall and temperature are the two main parameters of climate change that influence the processes
in the natural water cycle. The temperature change is a gradual phenomenon which takes place over a
period of time. However, the variations in rainfall parameters are more significant and difficult to
predict accurately even within a shorter period. Therefore, rainfall variation has been selected as a
major pressure for developing the analytical framework in this study.
Increases in population generate stress on the system due to the compounding increase in water
demand. A concomitant consequence of population increase is rapid urbanisation which will also
contribute to degrading water quality due to the creation of new pollutant sources and increase in
pollutant loads (Goonetilleke and Thomas 2003)18. The responses of the system to these pressures
indicate the level of resilience and the response can be evaluated in terms of performance (output
volume).
6
6. ASSESSING RESILIENCE
Attempts have been made by different researchers to quantify resilience of water resource systems
(Kjeldsen and Rosdjerg 200419, Moy et al. 198620, Wang and Blackmore 200921). However, Haimes
(2009)22 has pointed out that resilience of a system cannot be characterised with a single numerical
descriptor and resilience must be understood and evaluated in the context of a probabilistic and
dynamic set of input threat scenarios to the system and in terms of a complex set of associated
consequences attached to any such threat. Furthermore, the resilience of a system could be measured
in terms of a myriad of sub-states that characterise the system for a specific time period and threats.
Hence, measuring the efficacy of a system’s resilience might be achieved through the unique
functionality of that particular system and its responses (outputs) to specific inputs.
The probability of failure is proposed here as the way of measuring the response of the system to the
input threat scenarios. Considering the probability of failure as the dependent variable and the other
two threatening forces (pressures) as independent variables, a failure surface is developed by plotting
the three variables in a three dimensional space. Then by identifying high and low resilience regions
in the three dimensional space and observing proximity of the surface to the respective regions, status
of resilience of the system can be interpreted.
In the Figure 4 below, region close to ‘G’ represents the region of lowest pressure as the pressure 1
increases towards ‘H’ and pressure 2 increases towards ‘E’. The region close to ‘C’ indicates high
failure probability even under low pressure conditions. Therefore, the region close to ‘C’ represents
low resilience region. Similarly the region ‘F’ represents low probability of failure under high
pressure conditions indicating high resilience region.
B
A
Low resilience D
C
Probability of failure
E High resilience
F
Pressure component 2 H
G
Pressure component 1
Fig. 4 3D Representation of high & low resilience regions with respect to two pressure components.
Managing a water supply system requires an understanding of the behaviour characteristics of the
system. The three stages (disaster prevention, damage propagation and recovery) of a system
performance response curve (Ouyang and Osorio 2011)23 given in Figure 5 help to understand the
different stages of decision making.
7
Fig. 5 - System performance response curve (Ouyang and Osorio 2011)
Under normal operations, the system performance level is at 100%. However, when the initial failure
occurs at point A, the damage propagates and the performance level drops to a certain value I.
Therefore, it is important to identify the trigger points for early actions. The failure surface introduced
here can help to identity the trigger points. The Figure 6 shows the failure surface obtained from
simulation data for the SEQ Water Grid for the 100% full combined storage conditions.
120
100
80
Probabiity of failure
60
40 0
20
40
20 60
80
0 100
120
80
60 80
Inc 70
rea 40 60
se 50 (%)
in D 20 40
infall
em
30 in ra
and 0 10
20
re ase
(% 0 D ec
)
Fig.6 Modelled surface of failure for 100% combined initial storage of SEQ Water Grid
The surface is far from the low resilience region when the demand increase is zero percent and
decrease in rainfall is zero percent indicating high resilience. There is no considerable difference until
the rainfall decrease is approximately 40% even for a demand increase. It indicates that the system is
highly capable of providing required level of service without failure until the rainfall reduces
approximately 40% below the reference level. This knowledge allows management to formulate pre-
disaster strategies for a predicted long term low rainfall period. Therefore, that is the disaster
8
prevention period. A common approach by water supply authorities is the introduction of water
restrictions for maintaining storage levels for a longer period as a precaution in the event of a
predicted drought. The failure surface helps to identity the trigger points for early actions such as
introduction of water restrictions, as the shift of the surface indicates loss of resilience. Managerial
decisions should be taken at the trigger points. For example decrease in rainfall below 40% combined
with increasing demand shows a major shift of surface towards low resilience region indicating high
rate of resilience loss. It is notable that the rate of resilience loss is greater. Therefore, timely water
saving measures such as the introduction of suitable water restriction levels for the predicted level of
low rainfall conditions can be properly formulated by observing the changing points of the surface.
This surface has been developed considering 100% full initial combine storage levels. However, if the
initial storage levels are lower, loss of resilience is expected for a lower decrease in rainfall.
Therefore, storage level is also an important consideration for determining the trigger point.
8. SUMMARY
This paper proposes a three dimensional surface as a tool for analysing systemic resilience of a water
supply system. The approach is illustrated here by considering SEQ Water Grid as a case study. The
surface provides a structure for visualising the conditions for loss of resilience of the system and also
highlights the importance of urgent proactive precautionary measures under predicted low rainfall
conditions. The failure surface presented here does not provide an explanation of resilience, but rather
a robust tool to systematically examine systemic resilience. Analytical methods for systematically
assessing resilience, such as the one proposed here, may be critical for defining current and future
management needs and policy decisions.
REFERENCE
1. Gunderson, L. (2000). Ecological Resilience - In Theory and Application. Annual Review of Ecology and
Systematics, Vol. 31:pp 425-439.
2. Holing, C.S. (1973). Resilience and Stability of Ecological Systems. Annual Review of Ecology and Systematics,
Vol. 4: pp 1-23.
3. Brock, William A., Karl-Goran Maler and Charles Perrings. (2002). Resilience and Sustainability: The Economic
Analysis of Nonlinear Dynamic Systems. In Panarchy: Understanding Transformations in Human and Natural
Systems, edited by L. Gunderson and C. S. Holling. Washington DC: Island Press.
4. Walker., Brian., Carpenter ,S., Anderies, J., Abel, N., Cumming, G., Janssen, M.A., Lebel, L., Norberg, J.,
Peterson, G.D., Rusty Pritchard. (2002). Resilience Management in Social-Ecological Systems: A Working
Hypothesis for a Participatory Approach. Conservation Ecology ,Vol.6 (1):pp14.
5. Adger, W. Neil. (2000). Social and Ecological Resilience: Are They Related? Progress in Human Geography, Vol
24 (3): pp 347-364.
6. Barnes, P.H., Goonetilleke, A., Amarasinghe, P., Egodawatta, P. (2011). Assessing the resilience of potable water
supplies in South East Queensland, Australia, Proceedings of the International Conference on Building Resilience:
Interdependency approaches to disaster risk reduction and the development of sustainable communities and cities.
Sri Lanka: 19th - 21st July 2011. Can be accessed at http://eprints.qut.edu.au/43602/.
7. SEQ Water (2012). http://www.seqwater.com.au. Sighted on November 3, 2012.
8. Spiller, D., Owens, C., Horton, G,. Banks,S.(2011). Resilience of WEQ Water Grid -Water –Journal of the
Austrian Water Association, Vol, 38(7): pp 88-92.
9. Link water (2012). www.linkwater.com.au/media/docs/31394_AdultsFactSheetVF_WEB.pdf -
Sighted on December 5, 2012.
10. Bureau of Statistics. (2011). http://www.abs.gov.au. Sighted on February 10, 2012.
11. Bureau of Meteorology. (2008). Climate of Australia, BPA Print Group Pvt. Ltd.
12. Wanielista, M., Kersten, R., Eaglin, R. (1997). Hydrology, Water Quantity and Quality control (second edition)
John Wiley & Sons, Inc.
13. CSIRO (2008). Estimating open water evaporation for Murray Darling Basin. Can be accessed at www.csiro.au
14. Kohler, M.A., Nordenson, T.J., Fox, W.E. (1955). Evaporation from pans and lakes. Research paper No.38, U. S.
Department of Commerce, Washington.
15. Delpla, I., Jung, A. V., Baures, E., Clement, M., Thomas, O. (2009). Impacts of climate change on surface water
quality in relation to drinking water production. Environment International, Vol,35: pp.1225–1233.
16. Park, J .H., Duan, L., Kim, B., Mitchell, M. J., Shibata, H. (2010). Potential effects of climate change and
variability on watershed biogeochemical processes and water quality in Northeast Asia. Environment
International, Vol,36: pp. 212–225.
9
17. Ducharne, A. (2008). Hydrology and Earth System Sciences Importance of stream temperature to climate change
impact on water quality. Hydrol. Earth Syst. Science. Vol,12: pp. 797–810.
18. Goonetilleke, A., Thomas, E. C (2003). Water quality impacts of urbanisation. Technical report. Centre for Built
Environment and Engineering. Can be accessed at http://eprints.qut.edu.au/4209/.
19. Kjeldsen, T. R., Rosbjerg, D. (2004). Choice of reliability, resilience and vulnerability estimate for risk
assessments of water resources systems. Hydrological Science Journal, Vol,49(5): pp. 755-767.
20. Moy, W. S., Cohon, J.L., ReVelle, C.S., (1986). A programming model for analysis of the reliability, resilience
and vulnerability of a water supply reservoir. Water Resources Research,Vol, 22(4): pp 489-498
21. Wang, C., Blackmore, J. M. (2009). Resilience Concepts for water resources systems. Journal of Water Resources
Planning and Management, Vol,:135(6), pp. 528-563.
22. Haimes, Y.Y. (2009). Perspective on the definition of resilience of systems. Risk Analyse, Vol, 29(4): pp. 498-501.
23. Ouyang, M and Osorio, L. D. (2011). Resilience Modelling and simulation of Smart Grids. Structures Congress
pp. 1996- 2009.
10