funded by:

Smart Analytics and Innovative

Technologies for Improving Efficiency of

Water Distribution Networks

Pavlos Pavlou (M.Sc)

Research Engineer

KIOS Research and Innovation Center of Excellence

18/1/2019, Cyprus Water Association

www.kios.ucy.ac.cy

Introduction

Projects:

SmartWater2020:

The project SmartWater2020 is coordinated by the KIOS CoE at the University of Cyprus. The project’s partners are

the Water Boards of Limassol and Larnaca, the Water Development Department, the Foundation for Research and

Technology (FORTH) and the Municipal Water Supply and Sewerage Company Malevizi (Greece).

This project is funded by the INTERREG V-A “Greece-Cyprus 2014-2020” Cooperation Program.

WaterAnalytics:

The project’s consortium consists of KIOS CoE and PHOEBE Research and Innovation Ltd.

This project is funded is funded by the Cyprus Research Promotion Foundation.

www.kios.ucy.ac.cy

Introduction

Current State:

▪ 15-20% - Non Revenue Water.

▪ High financial costs.

▪ Water scarcity and high desalination costs.

Objectives:

▪ Early detection of leakage and water quality problems through the installation of sensors, valves and meters in Water Distribution Networks (WDNs).

▪ Real-time monitoring of water quality and pressure in the network.

▪ Leakage reduction through pressure control with the use of Pressure Reduction Valves (PRVs).

▪ Reduction of water system’s operational expenses and energy usage.

www.kios.ucy.ac.cy

Tasks

Model design (EPANET, QGIS)

▪ Model reduction/Skeletonization plugin

Innovative Technologies for minimizing water losses

▪ Pressure control with PRV settings algorithm

▪ Leakage control algorithm

▪ Pipe risk estimation algorithm

▪ AMR Leakage Detection algorithm

www.kios.ucy.ac.cy

Model design

www.kios.ucy.ac.cy

Model reduction

Model design

▪ Through geographic information systems (GIS)

▪ Complex and detailed models

A skeletonization process is required to managesuch large-scale WDN and to reduce thesimulation time.

▪ Reduce the complexity of the original waternetwork model without affecting its hydraulicand water quality performance.

www.kios.ucy.ac.cy

Model reduction

▪ Junction removal (junctions with two connections)

▪ Pipe series removal/merge

➢ Basic Criteria:

▪ Set of criteria 1: Distribution of demands to the end nodes of the new pipes based on four strategies:

▪ proportional to the distance from the removed junction

▪ proportional to the already assigned demand of the end nodes

▪ proportional to the elevation of the end nodes

▪ equally distributed

▪ Set of criteria 2: Same pipe characteristics (diameter, loss coefficient, roughness, initial status, type, controls).

▪ Set of criteria 3: Junction characteristics: Critical elevation (elevation difference between nodes of the merged pipes) and demand pattern.

www.kios.ucy.ac.cy

Model reduction

➢ Secondary criteria – Customize skeletonization

▪ Set of criteria 4: Junctions parameters: Elevation, Demand, Demand Pattern, Type (pipe junction, sensor, consumption node, fire hydrant, air valve) and Emitter Coefficient.

▪ Set of criteria 5: Pipe parameters: Description, Status, Length, Diameter, Roughness, Minorloss, Bulk Coefficient, Wall Coefficient

▪ Set of criteria 6: Distance criteria - Placement of junctions after the skeletonization process in specific distances.

www.kios.ucy.ac.cy

Model reduction

Function: Model_Reduction_Network(model)

Inputs:

@model: Water network model (.inp file) including base demands.

Basic conditions:

@SetOfCriteria2: PipeCharacteristics: Diameter, Roughness, Minor Loss Coefficient, Length,

Initial Status,Controls

@SetOfCriteria1&3: JunctionCharacteristics: Demand, Demand Pattern, Elevation

Extra conditions for customize skeletonization:

@SetOfCriteria4: JunctionCharacteristics

@SetOfCriteria5: PipeCharacteristics:

@ SetOfCriteria6: Distance Criteria

Output:

@SkeletonizedModel: A simplified network

www.kios.ucy.ac.cy

Model reduction: Case study

DMA 131 – Limassol

1043 Junctions

1093 Pipes

www.kios.ucy.ac.cy

Model reduction: TypesScenario 2:• Network skeletonization

Scenario 1:• Pipeline skeletonization

154 Junctions

204 Pipes

1020 Junctions

1070 Pipes

www.kios.ucy.ac.cy

Model Reduction: Types

434 Junctions

484 Pipes

173 Junctions

223 Pipes

Scenario 3:• Junctions every 150 m• Junctions representing sensors

Scenario 4:• Junctions with base demands

www.kios.ucy.ac.cy

Model Reduction: Hydraulic performance

Head Loss equation

▪ Hazen - Williams formula (for water and turbulent flow)

hL =K × L × Q1.852

C1.852 × D4.871

[k: conversion factor for the unit system (k = 1.318 for US customary units, k = 0.849 for SI units)]

[C: roughness coefficient]

▪ Darcy - Weisbach formula (all liquids and all flow regimes)

hL = f8 × L × Q2

g × π2 × D5

[f: friction factor]

www.kios.ucy.ac.cy

Model Reduction: Hydraulic performance

▪ 𝛥𝑝𝑖 𝑘 = 𝑝𝑖,𝑜𝑟𝑖 𝑘 − 𝑝𝑖,𝑠𝑘𝑒𝑙 𝑘 𝑖𝜖 𝑛𝑜𝑑𝑒𝑠 , 1 ≤ 𝑘 ≤ 𝑡𝑠𝑡𝑒𝑝𝑠

▪ 𝛥𝑓 𝑘 = 𝑓𝑜𝑟𝑖 𝑘 − 𝑓𝑠𝑘𝑒𝑙 𝑘 1 ≤ 𝑘 ≤ 𝑡𝑠𝑡𝑒𝑝𝑠

𝑝𝑜𝑟𝑖: 𝑛𝑜𝑑𝑒 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 (𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 𝑚𝑜𝑑𝑒𝑙)𝑝𝑠𝑘𝑒𝑙: 𝑛𝑜𝑑𝑒 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 (𝑠𝑘𝑒𝑙𝑒𝑡𝑜𝑛𝑖𝑧𝑒𝑑 𝑚𝑜𝑑𝑒𝑙)𝑓𝑜𝑟𝑖: 𝑖𝑛𝑓𝑙𝑜𝑤 (𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 𝑚𝑜𝑑𝑒𝑙)𝑓𝑠𝑘𝑒𝑙: 𝑖𝑛𝑓𝑙𝑜𝑤 (𝑠𝑘𝑒𝑙𝑒𝑡𝑜𝑛𝑖𝑧𝑒𝑑 𝑚𝑜𝑑𝑒𝑙)

Method for demand reallocation:

▪ proportional to the distance from the removed junction

▪ equally distributed

www.kios.ucy.ac.cy

Model Reduction: Hydraulic performance Proportional to the distance from the removed junction

Equally distributed

www.kios.ucy.ac.cy

Model Reduction: Water Quality

▪ Chlorine fate - Initial dose of 10 mg/L at Reservoir

▪ Advective transport (Rossman & Boulos, 1996): 𝜕𝐶𝑖

𝜕𝑡= −𝑉𝑖

𝜕𝐶𝑖

𝜕𝑥+ 𝑟(𝐶𝑖)

▪ Ci the concentration (mass/volume) as a function of the distance x and time t,

▪ r the reaction rate (mass/volume/time)

▪ V the fluid velocity (length/time)

www.kios.ucy.ac.cy

Model Reduction

Reallocation of demands (Proportional to the distance from the removed junction)

▪ High accuracy with the initial model’s hydraulic performance.

▪ Reduces the efficiency of the skeletonized model to simulate water quality and the fate ofcontaminants since the water velocities are affected.

▪ Improve water quality performance : Keep more nodes - No removal of consumption nodes.

▪ Drawback: Not enough skeletonization of the network.

www.kios.ucy.ac.cy

Innovative Technologies for minimizing water losses

www.kios.ucy.ac.cy

Pressure and leakage control

1. Pressure control with Pressure Reduction Valves (PRV) setting optimization

▪ Calculates the optimal PRV setting at each time step based on daily consumption data.

▪ Achieve the minimum acceptable pressure in the water network/DMA.

2. Reduce Leakages with Pressure control

▪ Estimate leakage flow based on SCADA data.

▪ Create “leakage models” – Emitter coefficients (simulate leakage flow – depends on pressure)

▪ 𝑞 = 𝐶 𝑝𝛾 q = leakage flow, C= discharge/emitter coefficient, p=pressure and γ= pressure exponent

▪ Apply PRV setting algorithm.

▪ Calculate new leakage flow.

www.kios.ucy.ac.cy

Pressure control

Function: PRV_setting(model,hmin,hmax,Pmin,Pmax)

Inputs:

@model: Water network model (.inp file), weekly consumption patterns, time duration and time step

@Pmin: Minimum acceptable pressure

@Pmax: Maximum acceptable pressure

@hmin,hmax: Range of proposed PRV settings

Output:

@OptimalHead: PRV setting for each time step

@AverageHead: Average PRV setting

@HeadTimePattern: Assign new head pattern

www.kios.ucy.ac.cy

Leakage control

Function_2: PRV_setting_leakage(Leakage_model, hmin, hmax, Pmin, Pmax)

Inputs:

@Leakage_model: Water network model (.inp file), weekly consumption

patterns, time duration and time step

@Pmin: Minimum acceptable pressure

@Pmax: Maximum acceptable pressure

@hmin,hmax: Range of proposed PRV settings

Output:

@TotalFlow: Flow fluctuation (m3/h) per PRV setting

@TotalLeakage: Leakage fluctuation (m3/h) per PRV setting

@TotalVolumeLeakage: Leakage volume (m3) per PRV setting

@Optimized_model: Model with new PRV settings

Function_1: Set_EmitterCoef(Model, ActualConsumption)

Inputs:

@model: Water network model (.inp file), weekly consumption

patterns, time duration and time step

@RealConsumption: Actual Consumption (m3/h)

Output:

@Leakage_model: Flow fluctuation (m3/h) per PRV setting

www.kios.ucy.ac.cy

Pressure and leakage control

▪ Case study: DMA 131

▪ Acceptable pressure: 20-40 m

▪ Range of PRV settings: 10-60 m

www.kios.ucy.ac.cy

Pressure and leakage control

www.kios.ucy.ac.cy

Pressure and leakage control

www.kios.ucy.ac.cy

Pipe Risk Estimation

3. Pipe-risk estimation

▪ Identification of high-risk pipes based on pressure variation, material, age, historical data -NOPB (number of previous breaks), length and diameter.

▪ Pressure variation parameter (3 cases):

▪ Pressure change per time step per pipe and the frequency exceeding the stated pressure change.

𝛥𝑝𝑖 𝑘 = 𝑝𝑖 𝑘 − 𝑝𝑖 𝑘 − 1 𝑖𝜖 𝑝𝑖𝑝𝑒𝑠 , 1 ≤ 𝑘 ≤ 𝑡𝑠𝑡𝑒𝑝𝑠

▪ Maximum pressure change of each pipe for the total simulation time.

𝛥𝑝𝑖 = max(𝑃𝑖)𝑘=1𝑡𝑠𝑡𝑒𝑝𝑠

−min(𝑃𝑖)𝑘=1𝑡𝑠𝑡𝑒𝑝𝑠

𝑖𝜖{𝑝𝑖𝑝𝑒𝑠}

▪ Average pressure of each pipe.

𝐴𝑣𝑒𝑃𝑟𝑒 𝑖 =1

𝑡𝑠𝑡𝑒𝑝𝑠

𝑘=1

𝑡𝑠𝑡𝑒𝑝𝑠

𝑝𝑖 𝑘

www.kios.ucy.ac.cy

Pipe Risk Estimation

Function_1: PipeRisk_PressureVariation (model, Threshold1, Threshold2)

Inputs:

@model: Water network model (.inp file), weekly consumption patterns, time duration and time step

@ Threshold1: Threshold for case 1 – pressure variation

@ Threshold2: Threshold for case 2 – pressure variation

Output:

@RiskFactor1: Risk of failure based on pressure change per time step per pipe

@RiskFactor2: Risk of failure based on the maximum pressure change of each pipe for the total

simulation time

@RiskFactor3: Risk of failure based on the average pressure

@plot: A graph of the network with the high-risk pipes highlighted

www.kios.ucy.ac.cy

Pipe Risk Estimation: Case study

www.kios.ucy.ac.cy

Pipe Risk Estimation

www.kios.ucy.ac.cy

Pipe Risk Estimation

www.kios.ucy.ac.cy

AMR – Consumption Abnormal Behavior

Minimum Flow (L/h) per day

4. AMR – Leakage Detection• Sends alerts when abnormal behavior is detected

www.kios.ucy.ac.cy

AMR – Consumption Abnormal Behavior

Minimum Flow (L/h) per day

www.kios.ucy.ac.cy

KIOS CoE Water Team

Dr. Demetrios

Eliades

Dr. Maria

AnastasiadouMarios

Kyriakou

Dr. Agathoklis

Agahokleous

Stelios

Vrachimis

Marilena

Chrysanthou

Pavlos

Pavlou

Constantinos

Heracleous

Magdy

Abdullah Eissa

Prof. Marios

Polycarpou

Prof. Christos

Panayiotou

www.kios.ucy.ac.cy

Thank You