system in water distribution network M ASHREKI I SLAM S AMI AGENDA - - PowerPoint PPT Presentation

MASHREKI ISLAM SAMI

Effective placement of sensors for efficient early warning system in water distribution network

MASHREKI ISLAM SAMI Chalmers 2

AGENDA

Introduction and Problem Statement Methodology Experimentation and Analysis Limitations Key Findings Conclusion

MASHREKI ISLAM SAMI Chalmers 3

Introduction and Problem Statement

▪ Water distribution network is important part of infrastrucutural development for any town/city/country. ▪ Water safety and quality are fundamental to human development and well-being, WHO ▪ UN SDG 6 also prioritizes accessibility and availability of safe driking water

MASHREKI ISLAM SAMI Chalmers 4

Problem Statement ▪ Contamination threats between WTP and consumer ▪ Chances of affecting mass population ▪ Uncertainties (Intrusion, micorbial growth, leakage)

❖ In 2007, an outbreak in Nokia, Finland affected 8453 people with waterborne gastroenteritis for pipeline cross-connection. ❖ In 2007, poisoning of water supply caused 71 people poisoned in China.

Introduction and Problem Statement

MASHREKI ISLAM SAMI Chalmers 5

Problem Statement ▪ Sensors are expensive ▪ Deploying sensors at every node is impractical ▪ Optimization requires vast computional resources

A large WDN with 10,000 nodes each contaminated and sensor sampling time is 10 min. 72h simulation will give 4.32 million scenarios of

• contaminations. Storage capacity required is 173 GB and 200 days to

end simulation.

Introduction and Problem Statement

MASHREKI ISLAM SAMI Chalmers 6

Objective of Study

Introduction and Problem Statement

• Event Detection: Using sensor resposne

to generate signal for water quality change

• Sampling and Identify: Collecting water

sample and testing to identify type of contamination

• Biomarks: Tracing source of

contamination and characterization (Leakage, Roads, fecal sources etc.)

2. Sampling and Identify 3. Biomarkers and Origin Analysis

• 1. Event

Detection

MASHREKI ISLAM SAMI Chalmers 7

Objective of Study

Introduction and Problem Statement

• Event Detection: Using sensor resposne

to generate signal for water quality change

• Sampling and Identify: Collecting water

sample and testing to identify type of contamination

• Biomarks: Tracing source of

contamination and characterization (Leakage, Roads, fecal sources etc.)

2. Sampling and Identify 3. Biomarkers and Origin Analysis

• 1. Event

Detection

MASHREKI ISLAM SAMI Chalmers 8

Criteria for Event Detection

Introduction and Problem Statement

• Detection time: How quick sensors can response and

generate a signal

• Detection likelihood: How often and efficiently sensors can

detect in corresponding to their placement.

• Population size: Number of consumers that may be affected

MASHREKI ISLAM SAMI Chalmers 9

Mathematical Algorithms Graph Theory

Linear algebra Numerical analysis Complex methods Genetic algorithm Greedy algorithm Parallel computing Heuristic approach Fast optimization Complex Network Theory Centrality Matrices Stochastic approach Critical region optimization

Advancement

Research on Sensor Placement

Introduction and Problem Statement

MASHREKI ISLAM SAMI Chalmers 10

Methodology

▪ Contamination event simulation using EPANET ▪ Constructing a pilot scaled WDN in laboratory ▪ Analyze real-time data and simulated data ▪ Possible analysis for source back-tracking

MASHREKI ISLAM SAMI Chalmers 11

EPANET Simulation ▪ The WDN was modelled in EPANET ▪ Chemical injection at each node ▪ Chemical concentration at each node ▪ Optimization for sensor placement and intrusion point

Methodology

MASHREKI ISLAM SAMI Chalmers 12

Design of water distribution network

Methodology

MASHREKI ISLAM SAMI Chalmers 13

EPANET Simulation

Methodology

Intrusion

MASHREKI ISLAM SAMI Chalmers 14

Methodology

Intrusion

Intrusion N1 N2 N3 N4 N5 N6 N7 N8 N9 N10 N11 N12 N13 N14 N15 N16 N17 N18 N19 N20 N21 N22 N23 N24 N25 N26 N27 Node ID Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. Conc. N1 5.03 5.04 4.95 3.51 3.51 3.02 3.51 3.51 1.48 0.21 0.21 4.83 4.83 0.21 1.47 N2 5.03 4.95 3.51 3.51 3.02 3.51 3.51 1.48 0.21 0.21 4.83 4.83 0.21 1.47 N3 5.03 3.66 3.66 3.15 3.66 3.66 1.36 5.04 5.04 1.35 N4 5.01 5.02 4.31 5.01 5.01 0.71 N5 5.01 4.31 5.01 5.01 0.71 N6 5.01 5.01 5.01 N7 5.01 5.01 N8 5 N9 5.01 5.01 5.02 5.02 N10 0.55 0.55 0.48 4.25 4.25 4.27 5.02 0.76 0.76 4.26 N11 0.55 0.55 0.48 4.25 4.25 4.27 5.03 5.02 0.76 0.76 5.03 4.26 N12 0.55 0.55 0.48 4.25 4.25 4.27 5.03 5.04 5.03 0.76 0.76 5.03 4.26 N13 1.29 1.29 1.26 1.16 1.16 1 2.04 2.03 2.04 2.41 2.31 2.31 5.02 2.46 1.6 1.6 2.6 2.41 0.18 0.18 0.18 0.28 0.28 2.04 N14 2.66 2.66 2.59 1.85 1.85 1.59 2.01 2.01 2.02 2.38 2.38 2.39 2.46 5.06 2.55 2.55 2.38 2.02 N15 2.66 2.66 2.59 1.85 1.85 1.59 2.01 2.01 2.02 2.38 2.38 2.39 5.04 5.04 2.55 2.55 2.38 2.02 N16 3.66 3.66 3.15 3.66 3.66 1.36 5.02 5.04 1.35 N17 3.66 3.66 3.15 3.66 3.66 1.36 5.02 1.35 N18 0.51 0.51 0.44 3.96 3.96 3.97 4.68 4.49 4.49 0.71 0.71 5.03 4.68 0.35 0.35 0.35 0.55 0.55 3.97 N19 0.55 0.55 0.48 4.25 4.25 4.27 5.03 0.76 0.76 5.02 4.26 N20 0.17 0.17 0.15 1.31 2.74 1.32 1.55 0.23 0.23 1.55 5.02 5.04 2.75 2.75 3.46 4.32 4.32 1.32 N21 0.17 0.17 0.15 1.31 2.74 1.32 1.55 0.23 0.23 1.55 5.02 2.75 2.75 3.46 4.32 4.32 1.32 N22 5.01 5.01 5.02 5 N23 5.01 5.01 5 N24 5.01 N25 0.2 0.2 0.17 1.53 3.2 1.54 1.81 0.27 0.27 1.81 3.21 3.21 3.2 5.02 5.04 1.54 N26 0.2 0.2 0.17 1.53 3.2 1.54 1.81 0.27 0.27 1.81 3.21 3.21 3.2 5.02 1.54 N27out1 4.8 4.8 4.7 3.35 3.35 2.88 3.44 3.44 1.41 0.2 0.2 0.11 0.24 4.6 4.6 0.2 1.4 N28out2 0.71 0.71 0.69 0.72 0.72 0.62 1.71 1.71 1.72 2.02 1.27 1.27 2.76 0.24 1.35 0.98 0.98 1.43 2.02 2.27 2.27 1.34 1.34 1.67 2.1 2.1 1.71 N29out3 5.01 1.35 5.02 5.02 5 N31 5.01 5.01 5.01 Total Nodes contamin- ated 6 7 8 20 21 22 25 30 20 16 10 7 2 4 5 18 19 3 15 2 3 9 10 10 6 7 23

MASHREKI ISLAM SAMI Chalmers 15

EPANET Simulation

Methodology

▪ Optimization

▪ Graph Theory ▪ Simple sorting matrix ▪ Strategic Decisions

Node Number Number of times Contaminated Node Number Number of times Contaminated Node Number Number of times Contaminated Node Number Number of times Contaminated N1 2 N1 N1 5 N1 5 12 N2 2 N2 N2 5 N2 4 11 N3 2 N3 N3 5 N3 3 10 N4 N4 N4 N4 5 5 N5 N5 N5 N5 4 4 N6 N6 N6 N6 3 3 N7 N7 N7 N7 2 2 N8 N8 N8 N8 1 1 N9 N9 N9 N9 4 4 N10 N10 N10 N10 5 5 N11 N11 N11 N11 7 7 N12 N12 N12 N12 8 8 N13 7 N13 10 N13 N13 1 18 N14 3 N14 14 N14 N14 1 18 N15 3 N15 13 N15 N15 2 18 N16 2 N16 N16 5 N16 2 9 N17 2 N17 N17 5 N17 1 8 N18 N18 N18 4 N18 5 9 N19 N19 N19 N19 6 6 N20 5 N20 3 N20 1 N20 4 13 N21 5 N21 3 N21 1 N21 3 12 N22 N22 N22 N22 4 4 N23 N23 N23 N23 3 3 N24 N24 N24 N24 1 1 N25 5 N25 N25 4 N25 2 11 N26 5 N26 N26 4 N26 1 10 N31 2 N31 1 N31 4 N31 3 10 Total Number of times Contaminated 1< Cx <2 mg/l 2< Cx <3 mg/l 3< Cx <4 mg/l Cx >4 mg/l

MASHREKI ISLAM SAMI Chalmers 16

EPANET Simulation

• Intrusion Points

Upstream

• Sensors Location

Downstream

Methodology

MASHREKI ISLAM SAMI Chalmers 17

Scaled WDN Model

Methodology

50 mm diameter PVC pipes supported on wooden frame and attached with clips. Flexible PVC pipes used for inflow and outflow.

MASHREKI ISLAM SAMI Chalmers 18

Materials and Intruments

Methodology

Electrode L8281 HD Electrode L7781 HD

pH electrodes

MASHREKI ISLAM SAMI Chalmers 19

Chemical Injections Three different solutions were used for the experiment of sensor resposne.

• Acetic Acid 24% - Very Good
• Sodium Chloride - Slow & inconsistent
• Commerical Chlorine soln. - Not Good

Methodology

Electrode L8281 HD Electrode L7781 HD

MASHREKI ISLAM SAMI Chalmers 20

Experimentation & Analysis

Electrode L8281 HD Electrode L7781 HD

Experimentation and Analysis

• 27 scenarios
• Inflow: 22 l/m
• Repeat 2-3 times

MASHREKI ISLAM SAMI Chalmers 21

Sensor Activities

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Raw data graphs merged into one

MASHREKI ISLAM SAMI Chalmers 22

Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N4 Outflow at node N27 & N28

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 23

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N4 Outflow at node N27 & N29

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 24

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Real-time Epanet

Intrusion at Node N4 Outflow at node N28 & N29

MASHREKI ISLAM SAMI Chalmers 25

Analysis of Intrusion at N4

• Intrusion chemical travel in laminar flow.
• Dispersion/diffusion rate is low.
• Advective flow pattern observed.

Electrode L7781 HD

Experimentation & Analysis

Electrode L8281 HD

MASHREKI ISLAM SAMI Chalmers 26

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N23 Outflow at node N27 & N28

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 27

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N23 Outflow at node N27 & N29

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 28

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N23 Outflow at node N28 & N29

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 29

Analysis of Intrusion at N23

• Intrusion chemical travel in laminar flow.
• Dispersion/diffusion rate is low.
• Advective flow pattern observed.
• Quickly out of the system at N29

Electrode L7781 HD

Experimentation & Analysis

Electrode L8281 HD

MASHREKI ISLAM SAMI Chalmers 30

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N31 Outflow at node N27 & N28

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 31

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N31 Outflow at node N27 & N29

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 32

Electrode L8281 HD Electrode L7781 HD

Experimentation & Analysis

Intrusion at Node N31 Outflow at node N28 & N29

Real-time Epanet

MASHREKI ISLAM SAMI Chalmers 33

Analysis of Intrusion at N31

• Intrusion chemical travel in laminar flow.
• Dispersion/diffusion rate is high.
• Advective flow behaviour is less.

Electrode L7781 HD

Experimentation & Analysis

Electrode L8281 HD

MASHREKI ISLAM SAMI Chalmers 34

Source Back-tracking

• Stochastic approach
• Contour mapping
• Critical region
• ptimization

Experimentation & Analysis

Electrode L8281 HD Electrode L7781 HD

MASHREKI ISLAM SAMI Chalmers 35

Source Back-tracking

Electrode L7781 HD

Experimentation & Analysis

Electrode L8281 HD Electrode L8281 HD Electrode L8281 HD Intrusion N4 Intrusion N23 Intrusion N31

Sensor N1 Sensor N13 Sensor N26 1 1 1 1 1 1 1

MASHREKI ISLAM SAMI Chalmers 36

Source Back-tracking

Experimentation & Analysis N1 N13 N26 N4

Intrusion Node N4 Intrusion Node N23 Intrusion Node N31 Sensor N1 1 1 Sensor N13 1 1 1 Sensors N26 1 1

MASHREKI ISLAM SAMI Chalmers 37

Limitations

▪ pH sensor data insufficient for for source back-tracking. ▪ More dynamic and hydraulic data are necessary for source back-tracking. ▪ EPANET time outputs changes with user defined parameters. ▪ Flow charateristics and pipe properties were not analyzed.

MASHREKI ISLAM SAMI Chalmers 38

Key Findings

▪ Simple sorting matrix, Graph theory, Complex network theory

• etc. are easy to use and efficient optimizing techniques.

▪ Chemical tends to flow in laminar in the pipe system ▪ EPANET is convenient to simulate WDN for hydraulic analysis but time-variant chemical flow is not available. ▪ Simple & cheap pH sensors are efficient in detecting sudden changes in pH of water.

MASHREKI ISLAM SAMI Chalmers 39

Conclusions

▪ The study highlights the possibilities of sensor localization in WDN using simpler methods ▪ Intrusion at downstream nodes for the designed WDN can be detected by sensors upstream enabling early warning signal ▪ Source tracking was not possible due to limitations and uncertainties associated with EPANET and simple sensors ▪ Computer program specifically developed for source tracking can contribute greatly to future water distribution network.

MASHREKI ISLAM SAMI Chalmers 40

Further Works ▪ Probabilistic Analysis ▪ Mike Urban by DHI ▪ WaterGEM by Bentley ▪ WaterCad by Bentley ▪ Machine Learning

chalmers.se/sv/centrum/dricks/Publikationer/Sidor/default.aspx

Thank You