Optimisation of Optimisation of Water ManagementWater Management
Prof. Graeme DandyProf. Graeme DandySchool of Civil, Environmental and School of Civil, Environmental and
Mining EngineeringMining EngineeringUniversity of AdelaideUniversity of Adelaide
Acknowledgement of Co-ResearchersAcknowledgement of Co-Researchers
Prof Holger MaierProf Holger Maier Postdocs:Postdocs:
Matt GibbsMatt Gibbs Postgrads:Postgrads:
Abby GoodmanAbby Goodman Dan PartingtonDan Partington
Honours students:Honours students: Fiona PatonFiona Paton John BaulisJohn Baulis Ben StanifordBen Staniford Lisa LloydLisa Lloyd Rebecca TennantRebecca Tennant Jason NicolsonJason Nicolson Liam HarnettLiam Harnett
OutlineOutline
Background: Optimisation ModelsBackground: Optimisation Models Case Studies:Case Studies:
Optimum planning of urban water systems Optimum planning of urban water systems (regional scale)(regional scale)
Resource optimisation framework for the Resource optimisation framework for the Upper South East Region of SAUpper South East Region of SA
Optimum design of greywater reuse systems Optimum design of greywater reuse systems (cluster scale)(cluster scale)
ConclusionsConclusions
Types of ModelsTypes of Models
DescriptiveDescriptive How does the system behave? How does the system behave? What will be the consequences of certain What will be the consequences of certain
actions?actions?Simulation ModelsSimulation Models
PrescriptivePrescriptive What are the best actions to achieve a What are the best actions to achieve a
particular objective or set of objectives?particular objective or set of objectives?Optimisation modelsOptimisation models
MethodologyMethodology
Systems approachSystems approach
Optimisation Module
Selection of Alternative
Simulation of Alternative
Evaluation of Alternative
Selection of Objectives
Results
Selection of Objectives
Optimisation Module
Form of an Optimisation ModelForm of an Optimisation Model
Choose values for a set of decision Choose values for a set of decision variables so as to maximise (or minimise) variables so as to maximise (or minimise) a particular objectivea particular objective
Subject to a set of constraintsSubject to a set of constraints
Genetic Algorithm Genetic Algorithm OptimisationOptimisation
What Are Genetic AlgorithmsWhat Are Genetic Algorithms ??
Guided search procedures that work by Guided search procedures that work by analogy to natural selectionanalogy to natural selection
Include embedded computer simulationInclude embedded computer simulation Each solution is represented by a string of Each solution is represented by a string of
numbersnumbers Work with a population of solutionsWork with a population of solutions Algorithm can run for any length of timeAlgorithm can run for any length of time Can’t prove that you have reached the Can’t prove that you have reached the
optimum solutionoptimum solution
Typical GA stringTypical GA string
Distribution Network Pipe
MaterialDistribution
Network Pipe Diameters
Pump Size
Collection Network Pipe
Material
So
luti
on
Co
st
($ m
illi
on
)
30
40
50
60
70
80
90
100
0 50,000 100,000 150,000 200,000
Number of Solution Evaluations
The GA conducts adirected search foroptimal solutions
Repeat Towards Convergence
Multi-Objective Multi-Objective OptimisationOptimisation
Multi-Objective OptimisationMulti-Objective Optimisation
Pareto Optimal Front
Optimum Planning of Optimum Planning of UrbanUrban
Water Systems Water Systems (Regional Scale) (Regional Scale)
TEMPORAL SCENARIOS
SUPPLY TYPE ALTERNATIVES
Run simulation
model
Check constraints
Output results
Water Simulation Model (WaterCress)
Calculate: - Reliability - Resilience - Vulnerability
2020 – 72ML/day 2060 – 225ML/day 2100 – 300ML/day
0KL < 30GL/yr
Risk Based Performance Assessment
0
10
20
30
40
50
60
70
80
90
2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060Date
Wat
er S
up
ply
(G
L/y
r)
MypongaReservoir
Happy ValleyReservoir
River Murray
DesalinationPlant
MaximumRiver MurrayFlow
ReliabilityResilience
(years-1)Vulnerability
(GL)
225ML/day 0KL 85% 0.63 26.0
Risk Based Performance Assessment
OptimisationOptimisation
Objectives:Objectives: Minimise present value of total system costMinimise present value of total system cost Minimise greenhouse gas emissionsMinimise greenhouse gas emissions
Constraint:Constraint: Availability of water from the Murray (30 Availability of water from the Murray (30
GL/year)GL/year)
OptimisationOptimisation
Decision Variables:Decision Variables: Capacity of desalination plant Capacity of desalination plant
(ML/day)(ML/day) Size of rainwater tanks for all Size of rainwater tanks for all
households (kL)households (kL) Operating rules for the systemOperating rules for the system
0
1
2
3
4
5
6
7
0 1 2 3 4Cost ($/kL)
GH
G e
mis
sio
ns
(kg
CO
2-e
/kL
)
Desalination PlantRainwater TankRiver MurrayHappy ValleyMyponga
Approximate trade-offs between supply types
The 2060 Pareto Front
21.76
21.78
21.80
21.82
21.84
6.990 7.000 7.010 7.020
Cost ($2007 billion)
GH
G e
mis
sio
ns
(Meg
ato
nn
es o
f C
O2-e
)
Optimal Tradeoffs – Southern SystemOptimal Tradeoffs – Southern System
The 2060 Pareto Front
21.76
21.78
21.80
21.82
21.84
6.990 7.000 7.010 7.020
Cost ($2007 billion)
GH
G e
mis
sio
ns
(Meg
ato
nn
es o
f C
O2-e
)
Breakpoint (250ML/day, 2KL)
(251ML/day, 1.8KL)
(248ML/day, 2.6KL)
Optimal Tradeoffs – Southern SystemOptimal Tradeoffs – Southern System
The 2060 Pareto Front
21.76
21.78
21.80
21.82
21.84
6.990 7.000 7.010 7.020
Cost ($2007 billion)
GH
G e
mis
sio
ns
(Meg
ato
nn
es o
f C
O2-e
)
$45/tonne
$1000/tonne
Breakpoint (250ML/day, 2KL)
(251ML/day, 1.8KL)
(248ML/day, 2.6KL)
Optimal Tradeoffs – Southern SystemOptimal Tradeoffs – Southern System
Range depends on optimal rainwater tank size, which depends on average yearly water supply per tank:
Optimisation Process
Average yearly water supply
per tank
1KL 24KL $3.08/KL 0.96kgCO2-e/KL
20KL 48KL $3.27/KL 3.34kgCO2-e/KL
0 200 400 600
Lifetime of the Desalination Plant
Lifetime of Rainwater Tanks
Climate Change Impacts
River Murray Supply Constraint
Demand
Social Discount Rate
Economic Discount Rate
247.1
241.9
247.8
206.7
168.0
245.5
247.7
251.1
251.1
287.0
296.0
512.0
251.0
251.6
Desalination Plant Size (ML/day)
0 2 4 6
Lifetime of the Desalination Plant
Lifetime of Rainwater Tanks
Climate Change Impacts
River Murray Supply Constraint
Demand
Social Discount Rate
Economic Discount Rate
1.8
1.8
0.9
1.1
1.8
1.7
1.7
3.1
5.3
2.9
3.2
2.9
3.6
2.8
Tank Size (KL)
Sensitivity Analysis of the Optimisation
Process
Planned Extensions to this ResearchPlanned Extensions to this Research
•Include more objectives (reliability, social factors)•Add more alternatives (e.g. stormwater reuse)
ConclusionsConclusions
Future expansion of Adelaide’s water Future expansion of Adelaide’s water supply will use a combination of non-supply will use a combination of non-traditional sources including desalination, traditional sources including desalination, rainwater tanks and stormwater and rainwater tanks and stormwater and wastewater reusewastewater reuse
Tradeoffs exist between the costs and Tradeoffs exist between the costs and environmental impacts of these sourcesenvironmental impacts of these sources
Multi-objective Optimisation can be used Multi-objective Optimisation can be used to quantify some of these tradeoffsto quantify some of these tradeoffs
Resource Optimisation Resource Optimisation Framework for the Upper Framework for the Upper South East Region of SASouth East Region of SA
110 km
Dune and flat topologyDune and flat topology Very flat, slope of 1:6000Very flat, slope of 1:6000 Prone to floodingProne to flooding
Flats cleared for agriculture, Flats cleared for agriculture, dunes contain wetlands of high dunes contain wetlands of high conservation valueconservation value
Area of 1 Million HaArea of 1 Million Ha 40% affected by dryland 40% affected by dryland
salinitysalinity Over 640 km of groundwater Over 640 km of groundwater
drains installeddrains installed 100 regulators throughout the 100 regulators throughout the
regionregion
Case Study – Upper South EastCase Study – Upper South East
Management DecisionsManagement Decisions
Management decisions involve movement of available waterManagement decisions involve movement of available water Regulators in the drainage network allow water to be Regulators in the drainage network allow water to be
directed around the landscapedirected around the landscape Decisions are based on a number of considerations:Decisions are based on a number of considerations:
• Water quantityWater quantity• Water qualityWater quality• Wetland prioritiesWetland priorities
Conflicting objectives:Conflicting objectives: Manage dryland salinityManage dryland salinity Maintain wetland biodiversityMaintain wetland biodiversity Mitigate floodingMitigate flooding
Proposed Decision Support SystemProposed Decision Support System
A multidisciplinary approach is proposed to A multidisciplinary approach is proposed to produce a dryland salinity decision support tool:produce a dryland salinity decision support tool: Groundwater modelling Groundwater modelling Rainfall-runoff modelling Rainfall-runoff modelling Salt-transport modelling Salt-transport modelling Ecological modellingEcological modelling
Models combined to produce an integrated Models combined to produce an integrated modeling framework to assist management modeling framework to assist management decisionsdecisions
Integrated Modelling FrameworkIntegrated Modelling Framework
groundwaterresponse
RegulatorDecision
Point
shut openflow flow
runoff,salinity
rainfallevap.
runoff,salinity
runoff,salinity
runoff,salinity
rainfallevap.
rainfallevap.
rainfallevap.
wetlandresponse
environmentalconditions
Water Quantity ModellingWater Quantity Modelling
Hydrologic model requiredHydrologic model required Good GIS information available:Good GIS information available:
Drain, wetland and regulator locationsDrain, wetland and regulator locations Catchment wide LiDAR elevation data are currently Catchment wide LiDAR elevation data are currently
being processedbeing processed Very sparse flow data recordsVery sparse flow data records
Historically not much data recordedHistorically not much data recorded Drains installed since the late 1990sDrains installed since the late 1990s Very little to measure in the last few yearsVery little to measure in the last few years
Hydrologic ModellingHydrologic Modelling
HEC-HMS adopted for modellingHEC-HMS adopted for modelling Different models can be selected for each component of Different models can be selected for each component of
rainfall-runoff modelrainfall-runoff model Loss (Infiltration)Loss (Infiltration) Transformation (Rainfall-Runoff)Transformation (Rainfall-Runoff) BaseflowBaseflow
Hydrologic ModellingHydrologic Modelling
Water Quality ModellingWater Quality Modelling
Salinity is very important variable to decision making Salinity is very important variable to decision making processprocess
No water quality models in HEC-HMSNo water quality models in HEC-HMS CATSALT (Tuteja et al., 2003) to determine salt loadCATSALT (Tuteja et al., 2003) to determine salt load
Considers flow from groundwater and unsaturated Considers flow from groundwater and unsaturated zone separatelyzone separately
Other considerations, such as evaporation in storages Other considerations, such as evaporation in storages QSW
QGW
Unsaturated
Saturated
QUW
Integrated Modelling FrameworkIntegrated Modelling Framework
groundwaterresponse
RegulatorDecision
Point
shut openflow flow
runoff,salinity
rainfallevap.
runoff,salinity
runoff,salinity
runoff,salinity
rainfallevap.
rainfallevap.
rainfallevap.
wetlandresponse
environmentalconditions
Groundwater ModellingGroundwater Modelling
Regional groundwater flow is relatively well Regional groundwater flow is relatively well understoodunderstood
Local effects of drains on groundwater table largely Local effects of drains on groundwater table largely unknown, and highly controversialunknown, and highly controversial
Groundwater ModellingGroundwater Modelling
Groundwater modelling to answer important Groundwater modelling to answer important questions, such as:questions, such as: What is the zone of influence of the drain?What is the zone of influence of the drain? Once a regulator is changed, how long will it take Once a regulator is changed, how long will it take
for the groundwater table to adjust?for the groundwater table to adjust? What is the expected effect on the soil salinity?What is the expected effect on the soil salinity?
Integrated Modelling FrameworkIntegrated Modelling Framework
groundwaterresponse
RegulatorDecision
Point
shut openflow flow
runoff,salinity
rainfallevap.
runoff,salinity
runoff,salinity
runoff,salinity
rainfallevap.
rainfallevap.
rainfallevap.
wetlandresponse
environmentalconditions
Ecological Response to SalinityEcological Response to Salinity
A further criterion on the management problem is to sustain A further criterion on the management problem is to sustain the wetlands in the regionthe wetlands in the region
Project aims to answer questions such as:Project aims to answer questions such as: What are the impacts of elevated salinities on the health What are the impacts of elevated salinities on the health
and survival of aquatic species?and survival of aquatic species? How long can elevated salinities be tolerated?How long can elevated salinities be tolerated? How can we best use water from the drains to optimise How can we best use water from the drains to optimise
wetland health and function? wetland health and function? Field and laboratory studies to collect necessary dataField and laboratory studies to collect necessary data Modelling to allow expected effects to be predictedModelling to allow expected effects to be predicted Decision making process can then make use of modelling Decision making process can then make use of modelling
resultsresults
Bayesian Network ModellingBayesian Network Modelling
Interaction Between ModelsInteraction Between Models
Rainfall,Evaporation
CurrentConditions
Groundwater Models
Wetland Models
Rainfall-Runoff Models
Salt TransportModels
Regulator Settings
Catchment Routing
Environmental Response
Dryland Salinity
Flooding
Evaluate OptionSimulation/Optimization
Summary Summary
Currently, the information required to Currently, the information required to tackle the problem of dryland salinity is tackle the problem of dryland salinity is incompleteincomplete
A multidisciplinary approach is proposed to A multidisciplinary approach is proposed to adequately address the problemadequately address the problem Water quality and quantityWater quality and quantity Groundwater Groundwater EcologyEcology
Project OutcomesProject Outcomes
Integrated simulation model of the systemIntegrated simulation model of the system Considering all aspects that affect Considering all aspects that affect
regulator operationregulator operation Optimisation component to determine Optimisation component to determine
optimal operating schemeoptimal operating scheme Multi-Objective evolutionary algorithmsMulti-Objective evolutionary algorithms
Outcomes (2) Outcomes (2)
Novel aspects include:Novel aspects include: Ungauged catchment model calibrationUngauged catchment model calibration Groundwater modellingGroundwater modelling Ecological modellingEcological modelling Integrated catchment modellingIntegrated catchment modelling Optimisation and reliability aspectsOptimisation and reliability aspects
SummarySummary Whereas simulation models can be used to Whereas simulation models can be used to
assess the likely effects of various actions on a assess the likely effects of various actions on a system, optimisation models are useful for system, optimisation models are useful for providing guidance in identifying the best set of providing guidance in identifying the best set of actionsactions
Optimisation models require a clear definition of Optimisation models require a clear definition of objectives objectives
Multi-objective optimisation models can be used Multi-objective optimisation models can be used to assist in managing scarce water resourcesto assist in managing scarce water resources
Wastewater Treatment
Wetland
ASR
Industry
House or Cluster
Mains water
Stormwater
Total Urban Water ManagementTotal Urban Water Management
Optimum Design of GreywaterOptimum Design of Greywater
Reuse Systems (Cluster Scale) Reuse Systems (Cluster Scale)
Research ObjectivesResearch Objectives
1.1. Develop a new Develop a new methodology for the methodology for the planning of greywater planning of greywater reuse schemes in urban reuse schemes in urban areas that considers areas that considers their sustainabilitytheir sustainability
2.2. Apply methodology to Apply methodology to development in Streaky development in Streaky BayBay
MethodologyMethodology
Systems approachSystems approach
Optimisation Module
Selection of Alternative
Simulation of Alternative
Evaluation of Alternative
Selection of Objectives
Results
Selection of Objectives
Optimisation Module
Case Case StudyStudyScale 019 housesScale 0219 housesScale 0347 housesScale 0468 housesScale 05117 housesScale 06147 houses
System ComponentsSystem Components
Individual house reuse systemIndividual house reuse system Treatment systemTreatment system PumpPump Storage tankStorage tank
System ComponentsSystem Components
Cluster scale reuse schemesCluster scale reuse schemes Greywater collection networkGreywater collection network Treatment systemTreatment system Storage tankStorage tank PumpPump Treated greywater distributionTreated greywater distribution
networknetwork
LayoutLayout
Sustainability ObjectivesSustainability Objectives
Sustainability:Sustainability: Environmental: Total energy consumption Environmental: Total energy consumption
(GJ)(GJ) Economic: Present value of life cycle cost ($)Economic: Present value of life cycle cost ($) SocialSocial TechnicalTechnical
Simulation of AlternativesSimulation of Alternatives
Many design variablesMany design variables Collection network material and pipe diameterCollection network material and pipe diameter Distribution network material and Distribution network material and pipe pipe
diametersdiameters Greywater Treatment SystemGreywater Treatment System Pump SizePump Size
However, we need to simulate each However, we need to simulate each componentcomponent
21
23
25
27
29
10,000 12,000 14,000 16,000NPV per Household ($)
Tota
l Ene
rgy
per H
ouse
hold
(GJ)
1
2
3
4
5
6
21
23
25
27
29
10,000 12,000 14,000 16,000NPV per Household ($)
Tota
l Ene
rgy
per H
ouse
hold
(GJ)
1
2
3
4
5
6
ResultsResults
Comparison of ResultsComparison of Results
Case StudyCase Study Between $3 and $6 per kLBetween $3 and $6 per kL
Rouse Hill (Sydney)Rouse Hill (Sydney) Between $3 and $4 per kLBetween $3 and $4 per kL
Sensitivity AnalysisSensitivity Analysis
Doubling the population densityDoubling the population density
Extending the available pipe Extending the available pipe
materials and diametersmaterials and diameters
17
19
21
23
25
27
8,000 10,000 12,000 14,000NPV per Household ($)
Tota
l Ene
rgy
per H
ouse
hold
(GJ)
Sensitivity AnalysisSensitivity Analysis
ConclusionsConclusions
Cluster scale is more sustainable than Cluster scale is more sustainable than individual householdindividual household
Reuse schemes are more sustainable Reuse schemes are more sustainable with:with: Increased population densityIncreased population density Network design standards Network design standards
that allow different pipethat allow different pipematerials and diametersmaterials and diameters
Further WorkFurther Work
Include other objectivesInclude other objectives Ecological impactsEcological impacts ReliabilityReliability
Include other optionsInclude other options Rainwater tanksRainwater tanks Stormwater reuseStormwater reuse Blackwater reuseBlackwater reuse Aquifer storage and recoveryAquifer storage and recovery
Apply to larger scalesApply to larger scales
SummarySummary Whereas simulation models can be used to Whereas simulation models can be used to
assess the likely effects of various actions on a assess the likely effects of various actions on a system, optimisation models are useful for system, optimisation models are useful for providing guidance in identifying the best set of providing guidance in identifying the best set of actionsactions
Optimisation models require a clear definition of Optimisation models require a clear definition of objectives objectives
Multi-objective optimisation models can be used Multi-objective optimisation models can be used to assist in managing scarce water resourcesto assist in managing scarce water resources