## 6A Systems modelling

## Please note the time shown on this page is automatically adjusted by the server according to the time zone set in your computer.

## Blue

## Dragan Savic

## Chair:

##

##

## to

##

## A flexible and interactive toolbox to support reservoir operation decisions under multiple objectives and hydrological uncertainty

Andres Penuela

Presenter:

Authors:

Andres Penuela and Francesca Pianosi

Mathematical models enable the simulation of water fluxes in water resource systems in response to natural forcing inputs and human actions. Such mathematical models are typically implemented in software packages, but this implementation is generally different depending on the developer background and goals. On the one hand, ‘research’ software is generally developed to study a specific scientific question, it is written as an open-source code easily adaptable without a high level of software development expertise but without the thought for a more generalizable application. On the other hand, ‘professional’ software is generally produced by software engineering companies, it is written in a closed source code that cannot be read, explored or adapted to specific needs or priorities but can be easily run by users with no coding skills. The objective of this study is to develop a Python toolbox that can serve both the research and professional community in water systems operation and hence contribute to bridge the gap between them and promote the collaboration and integration of research advances in reservoir operation into the real-world reservoir system management. For this purpose, the toolbox uses Jupyter Notebooks to be accessible to users with different levels of expertise in mathematical modelling and numerical coding, be transparent, flexible and customizable by more experienced users, be fast without compromising its accessibility and flexibility and show data and results in a visual and interactive way.

# to

#

#

## Improved NSGA-II Multiobjective Genetic Algorithm for Optimization of Water Distribution Network Design

Naveen Naidu Maddukuri

Presenter:

Authors:

Naveen Naidu Maddukuri, Sriman Pankaj Boindala, Vasan Arunachalam and Murari Raja Raja Varma

Research into the design optimization of water distribution networks has been going on over the past few decades due to two reasons: (i) the nonlinear relationships between pipe discharges and head losses introduce complex nonlinear constraints, and (ii) the discrete pipe diameters lead to a combinatorial optimization problem. Traditionally, the design of water distribution networks has been focused on as only a least-cost optimization problem. Optimization based on cost minimization has serious limitations due to the uncertainty in future demands. The challenge for optimization modellers is to move away from the cost minimization and work towards capturing the true multiobjective nature of the problem. The focus of this research is to work towards enhancing the Pareto-optimal front of the multiobjective model of water distribution using Random Multi-Point Crossover operator in NSGA-II. This improvisation has been used in NSGA-II to obtain better Pareto optimal solutions for two benchmark problems, 1) one small network i.e., two loop network and 2) one medium network i.e., Hanoi water distribution network. Two objectives considered in this paper are namely minimization of cost and maximization of resilience for both the benchmark problems. For Hanoi water distribution network, a significant number of additional Pareto front solutions have been obtained. These Pareto optimal solutions give support for decision making to decision makers

# to

#

#

## Calculation of chlorine concentration bounding estimates in Water Distribution Networks using real-time learning

Stelios Vrachimis

Presenter:

Authors:

Stelios Vrachimis, Demetrios Eliades and Marios Polycarpou

On-line chlorine sensors can enhance contamination event detection in water distribution networks due to the fact that certain contaminants will affect chlorine residuals in a specific way. A key challenge to this approach is the ability to estimate chlorine residuals, which are affected by various dynamics and uncertainties, such as flow uncertainty and chlorine reaction rate uncertainty. To detect abnormalities in water quality, reliable thresholds are needed which quantify estimation errors and are tight enough to produce useful insight. This work proposes a comprehensive methodology which calculates bounded chlorine concentration estimates at sensor locations, suitable to be used for sensor fault and contamination detection purposes. The methodology combines an interval hydraulic-state estimator and a chlorine concentration bounding estimator available in the literature, and modifies them accordingly to be used in conjunction with a real-time learning algorithm. The learning algorithm is able to learn bulk reaction coefficients of water originating from different sources. The methodology is demonstrated on a real transport network of a large city in Cyprus using real quality and hydraulic data from sensors installed in the network. Artificial faults are added to the data, i.e. the effect of a sensor fault or contamination event, to demonstrate how the proposed methodology can be used for fault detection.

# to

#

#

## Statistical Comparison of 1-Scenario Estimators of Reliability in Water Distribution Networks

Yves Filion

Presenter:

Authors:

Diego Paez and Yves Filion

Estimating the reliability of water distribution networks (WDN) has proven to be a challenging problem, as it involves considering the stochastic nature of different perturbations that may affect the systems. One common approach are reliability surrogate measures that do not consider explicitly the stochastic nature of the perturbations, but instead they estimate the reliability of the network using only one design/base scenario. One problem with all these proposed 1-scenario reliability estimators is that they might be redundant in terms of the quantification of the reliability of a WDN, especially considering the many power-based indexes (RI, NRI, MRI, API). This paper uses five real WDNs, to perform a statistical comparison between the previously described 1-scenario reliability estimators in order to evaluate their similarity and redundancy. To compare the estimators, different optimization problems were solved for each of the five WDNs. A correlation analysis between all the computed reliability values was performed using the Spearman’s rank coefficient. Additionally, a Multidimensional Scaling was applied to the complement of the correlation matrices for the estimators in order to show graphically the similarities and differences between the 1-Scenario Estimators. Results show that power-based indexes are indeed highly similar and representative of the hydraulic reliability, while the flow entropy represents a different kind of reliability which also does not correlate with the mechanical reliability estimator.

# to

#

#

## A Study on the Selection of Water Pressure Monitoring Points in Water Distribution System using Entropy Theory and Genetic Algorithm

Jinseok Hyung

Presenter:

Authors:

Jinseok Hyung, Jeewon Seo, Kibum Kim, Taehyeon Kim and Jayong Koo

In Korea, about 682.5 million tons of leakage annually occurs because of deterioration of water supply pipes and inadequate maintenance. When the average cost of production in 2016 is applied, an economic loss is about 685 million dollars. Therefore, advanced operation and maintenance for leakage reduction is urgently needed. Water pressure monitoring to detect hydraulic problems such as leaks in water distribution systems can be one of the advanced operating methods for rapidly maintenance. It is necessary to monitor the water pressure at appropriate locations for effective management of water distribution networks. Decisions on the selection of appropriate water pressure monitoring points affect waterworks plans and budgets. So this should be done on a quantitative basis. This study proposes a methodology for selection of optimal water pressure monitoring points using genetic algorithm by applying entropy theory to quantify hydraulic information. In this study, water pressure monitoring points optimized for pressure changes in the water distribution system were selected using entropy theory and genetic algorithm. It is possible to select the optimal location and number of water pressure monitoring points, which can be used efficiently in water supply planning decision making.

# to

#

#

## Towards an Open, Low-Cost and Enhanced Standards-Based IoT Architecture for Autonomous and Smart Water Quality Control and Monitoring

Aitor Corchero Rodriguez

Presenter:

Authors:

Edgar Rubión Soler, Aitor Corchero Rodriguez, Xavier Domingo Albin, Lluis Echeverria Rovira and Gabriel Anzaldi Varas

Currently, water quality monitoring is a very important undertaking research area that would ensure safe water to be delivered to the users and reusable water run off to the environment. Nevertheless, despite this need, the existing Wireless Sensor Networks (WSN) are applied to large areas and based on expensive hardware and proprietary software hindering the wide deployment to water quality monitoring. We propose a novel, smart, and low-cost, ICT architecture to monitor and control an unattended an innovative ecological on-site sanitation system by using open hardware and software solutions. The main outcome consists of an IoT full stack solution, enhanced with local intelligence and edge-computing, to perform humanlike actions and enhance the measurements accuracy by using complex correlations. It thus provide a major contribution to: (i) reduce water monitoring deployment costs by using virtual sensors; (ii) minimize the human intervention thanks to edged-analytics; (iii) integrate heterogeneous sensors and actuators by using IoT standards; (iv) real-time transformation and correction of RAW data by using embedded-knowledge; (v) improve the quality metering by using machine learning models; and (vi) encourage the water quality monitoring fostering behavioral change. This article depicts the advancements in water quality monitoring proposed by INNOQUA project.