Reading Group Meeting
PhD thesis:
Modelling the Performance of an Integrated
Urban Wastewater System under Future
Conditions
Maryam Astaraie-Imani
29 August 2013
 BACKGROUND
Aim
 INTEGRATED URBAN WASTEWATER SYSTEM (IUWS)
 IMPACT ANALYSIS
 Sensitivity Analysis
 OPTIMISATION OF THE IUWS PERFORMANCE
 Climate Change and Urbanisation Scenarios
 Operational Control Optimisation Model
 Design Optimisation Model
 Risk-based Optimisation Model
 Summary of findings
2
BACKGROUND
 BEng/BSc in Civil Engineering (1996-2001)
 MEng/MSc in Water & Hydraulic Engineering (2004-2006)
 Thesis Title: Risk-based Floodplain Management
 PhD in Water Engineering (2008-2012)
 Thesis Title: Modelling the performance of an Integrate Urban Wastewater System
under future conditions
 Associate Research Fellow in Safe & SuRe project (2013-2015)
3
Improving an Integrated Urban Wastewater System (IUWS) performance
under future climate change and urbanisation
aiming to maintain the quality of water in water recipients
4
SIMBA library
 Matlab/simulink based
 User friendly
 Capable of integrated modelling
of urban wastewater system
 Sewer system
 Wastewater treatment plant (WWTP)
 River
5
Case Study
 Semi-real
 Norwich wastewater treatment plant
Sewer System
Wastewater Treatment Plant
CSO
flow
(Tank)
Inflow
SC4
Pump 1
(Tank)
SC7
(Tank)
SC5
Return Flow
SC3
Primary
Clarifier
Reactor
Secondary
Clarifier
Return Sludge
Pump 2
Waste
Sludge
SC6
(Tank)
CSO discharge
Effluent
SC1
SC2
Dispose
Storm Tank
Discharge
Reach 7
River
Reach 10
6
Impact analysis of climate change and urbanisation
on the IUWS performance
 IUWS model input parameters
 Climate change parameters
 Urbanisation parameters
 Operational control parameters
 IUWS model output parameters
 Dissolved Oxygen concentration (DO)
 Ammonium concentration (AMM)
 Local sensitivity analysis
 One-at-a-time method (Tornado Graph)
 Global sensitivity analysis
 Regional sensitivity analysis (RSA) Method
7
IUWS model input parameters
 Climate change parameters
 Rainfall depth increase (RD)
 Rainfall intensity increase (RI)
 Urbanisation parameters




Per capita water consumption (PCW)
Population increase (POP)
Imperviousness increase (IMP)
Ammonium concentration in DWF (NH4+)
 Operational control parameters





Maximum outflow rate from the sewer system (i.e. last storage tank) (Qmaxout)
Maximum inflow to the wastewater treatment plant (Qmaxin)
Threshold at which the storm tank is triggered to be emptied (Qtrigst)
Emptying flow rate of storm tank (Qempst)
Return activated sludge is taken from the secondary clarifiers (QRAS)
8
Parameter
Unit
Nominal value
Value/Range
RD
%
0
[10, 20, 30]
RI
%
0
[10, 20, 30]
POP
%
0
[4.5, 15]
IMP
%
0
[5, 15]
PCW
litre/person/day
180
[80, 260]
NH4+
mg/l
27.7
[20, 30]
Qmaxout
m3/d
5× DWF*
[3×DWF*, 8×DWF*]
Qmaxin
m3/d
3× DWF*
[2×DWF*, 5×DWF*]
Qtrigst
m3/d
24192
[16416, 31104]
Qempst
m3/d
12096
[6912, 24192]
QRAS
m3/d
14688
[6912, 24192]
9
Sensitivity Analysis
One at a time method
 Select one IUWS model input and change its value from default to upper or
lower value in the considered range. Keep the other input parameter values
at their nominal values.
 Run the IUWS model and evaluate the relevant IUWS model outputs.
 Calculate the relative difference (percent change) for the analysed IUWS
model outputs relative to the BC.
 Rank the obtained relative differences in a descending order and identify the
most sensitive IUWS model input parameters.
10
Regional Sensitivity Analysis
 Identify the most important parameters from LSA
 Generate samples by using Latin Hypercube Sampling (LHS)
 Run the IUWS model
 Determine the behavioural (B) & non-behavioural (NB) groups of samples
 Provide the CDF of B & NB samples
 Kolmogorov-Smirnov (KS) test
11
LSA Results
Qmaxin RD
RD
RI
Qmaxout
PCW
RDPCW
Qmaxin
POP
PCW
IMP
POP
NH4 RI
IMP NH4
QRAS POP
RI IMP
Qmaxout
Qmaxin
Qtrigst
QtrigstQRAS
QempstQtrigst
Qempst
Qmaxout
-10 -100
0
-80
10
-60
20
-40 30
-20 40
0 50
2060
70
40
Relative
variation
of AMM
concentration
to to
thethe
BCBC
forfor
maximum
Relative
variation
of DO
concentration
minimum
values
of the
IUWS
model
input
parameters
(%)(%)
values
of the
IUWS
model
input
parameters
12
GSA Results for AMM concentration
1
1
B
NB
0.8
0.6
0.8
0.6
0.4
0.4
0.2
0.2
0
1
3
4
5
6
7
8
0
3
4
Qmaxout(*27500,m3/d)
5
6
7
8
Qmaxout(*27500,m3/d)
1
0.5
0.5
0
80 100 120 140 160 180 200 220 240 260
0
80 100 120 140 160 180 200 220 240 260
PCW (lit/person/day)
PCW (lit/person/day)
13
Optimisation of the IUWS performance
14
Climate change and urbanisation scenarios
Climate Change
Parameters
Urbanisation Parameters
Scenarios
POP
(%)
IMP
(%)
PCW
(lit/person/day)
Base Rainfall
0
0
180
0
0
180
Climate
Change
Scenarios
RD
Scenario B (SCB)
Scenario C (SCC)
RI
0
0
180
Combined
Climate Change
with
Urbanisation
Scenarios
Scenario A (SCA)
RD, RI
Scenario K1 (SCK1)
RD
1.045
1.05
80
Scenario K2 (SCK2)
RI
1.045
1.05
80
Scenario L1 (SCL1)
RD
1.15
1.15
260
Scenario L2 (SCL2)
RI
1.15
1.15
260
15
Operational control optimisation model
 Objectives
 Maximise the minimum DO concentration in the river
 Minimise the maximum AMM concentration in the river
 Decision variables
 Qmaxout , (m3/d)
 Qmaxin , (m3/d)
 Qtrigst, (m3/d)
 Optimisation algorithm
 Modified MOGA-ANN algorithm (CCWI, 2011)
16
17
Ng=3000, Nd=50
NSGA-II
4.5
4
3.5
3
2.8
3
3.2
3.4
3.6
3.8
4
4.2
4.4
DO Concentration (mg/l)
69
Ratio of computational time
reduction (%)
AMM Concentration (mg/l)
Modified MOGA-ANN performance
5
Training set size 1000
Training set size 2000
Training set size 3000
68
67
66
65
64
63
50
200
Size of new data set
500
18
Optimal Pareto fronts under climate
change scenarios
19
20
Design optimisation model
Increasing the storage capacity of whole the catchment
 Objectives
 Maximise the minimum DO concentration in the river
 Minimise the maximum AMM concentration in the river
 Optimisation algorithm
 Modified MOGA-ANN algorithm
21
IUWS redesign
parameters
IUWS operational
control parameters
Design optimisation model decision variables
Decision variables
Decision variables description
QST2 (m3/d)
QST4 (m3/d)
The maximum outflow rate of ST2
QST6 (m3/d)
The maximum outflow rate of ST6
Qmaxout (QST7) (m3/d)
The maximum outflow rate of sewer system (ST7)
Qmaxin (m3/d)
The maximum inflow rate to the WWTP
Qtrigst (m3/d)
The threshold triggering emptying the storm tank
a2 (%)
Contribution-coefficient of ST2
a4 (%)
Contribution-coefficient of ST4
a6 (%)
Contribution-coefficient of ST6
a7 (%)
Contribution-coefficient of ST7
The maximum outflow rate of ST4
22
Minimum storage capacity increase- coefficient
Scenario
Minimum increasecoefficient (%), (c)
Increased storage capacity
(m3)
Cost
(Million $), (C)
SCB
100 %
13,200
494,340
SCL1
675 %
89,100
1,219,800
SCL2
500 %
66,000
1,058,400
23
24
Operational
parameters
Design control
parameters
in SCL1in SCL1
Storage Tank's contribution-coefficient (%)
Operational control parameters
100
12
10
80
608
406
4
20
2
Qmaxout
0
ST7
Qmaxin
ST2
Qtrigst
QST2
ST4
QST4
QST6
ST6
25
Summary of the results from the design and operational control
optimisation models
 Operational control optimisation has the potential to improve the quality of water
under the considered climate change scenarios.
 Operational control optimisation under the combined climate change with
urbanisation scenarios can improve the water quality indicators to some extent.
 RD has more potential than RI in worsening the quality of water under future




climate change.
The values of the urbanisation parameters (specifically PCW) are very decisive as
water quality indicators.
Combination of urbanisation with climate change (in some extent) have the
potential to intensify water quality deterioration.
Improving the system performance only by optimising the operational control is not
adequate enough, to meet both economic and water quality criteria, under the
examined climate change and urbanisation scenarios.
Considering the combined impacts of climate change and urbanisation for the
system performance improvement, increases costs over just climate change impacts.
26
Risk-based improvement of the IUWS
 Risk-based IUWS optimisation model objectives
 Minimising the risk of DO concentration failure
 Minimising the risk of un-ionised Ammonia concentration failure
Risk= Consequence × Probability of water quality failure
 Risk-based IUWS optimisation model decision variables
 Operational control decision variables (similar as above)
 Design decision variables (similar as above)
 Modified MOGA-ANN algorithm
 Uncertainty in urbanisation parameters
27
Consequence of Water Quality Failure
28
1
0.9
0.8
Consequence
0.7
Empirical CDF of freshwater long
term data for DO concentration
(mg/l)
0.6
0.5
0.4
0.3
0.2
0.1
1
2
3
4
5
6
DO concentration (mg/l)
29
Probability of Water Quality Failure
Risk of Water Quality Failure
30
31
32
Decision
variables
the operational
Design
decision
variables
in the design
Operational
control of
decision
variables
control
optimisation
model
optimisation
model
in the
design
optimisation
model
14
1490
Decision
Decision variable
variable value
value
Decision variable value
80
12
12
70
1060
10
50
8
8
40
6630
20
4
4
10
22 0(*27500)
Qmaxout
ST7
Qmaxin (*27500)
Qmaxout (*27500)
Qmaxin (*27500)
ST2 (*2400)
QST2 (*DWF)ST4 QST4 (*DWF)
Qtrigst
Storage Tank
33
(*2400)
Qtrigst ST6
QST6 (*DWF)
Summary of the risk-based optimisation model results
 Uncertainty of the urbanisation parameters under RD brings about greater risk to the
IUWS than RI.
 The risk of failures under the considered climate change and urbanisation
parameters results in greater stress for DO than un-ionised Ammonia.
 The duration and frequency of water quality failures are determining factors of the
tolerable risk level for the health of aquatic life.
 Improving the considered operational control of the IUWS in isolation did not show
enough potential to reduce the risk of water quality failures to meet the tolerable risk
levels.
 Improving the design of the IUWS (in addition to the operational control) was
required in this study to mitigate the risk of water quality failures.
 Decisions about the tolerable level of risk are vital to determine the required strategy
(ies) for the system improvement(s) in the future. Therefore, having comprehensive
knowledge about the ecosystem under study is important for the planners to reduce
the future unavoidable risks in their decisions.
34
35