Peter Chalk Centre

University of Exeter

Stocker Road

Exeter

EX4 4QD

Tel: +44 (0)1392 263637

E-mail: CCWI2019@exeter.ac.uk 

17th International Computing & Control for the Water Industry Conference

1st - 4th September 2019
University of Exeter, UK
  • Facebook Social Icon
  • Twitter Social Icon
  • Instagram

2D Big data management

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

2.1-2.2

Juan Saldarriaga

Chair:

to

Investigating drinking water behaviour treated by different disinfection with the use of a machine learning technique on water quality datasets

Grigorios Kyritsakas

Presenter:

Authors:

Grigorios Kyritsakas, Vanessa Speight, Claire Thom and Joby Boxall

Investigating the impact of switching from one disinfection type to another is very important for water companies in order to guarantee that the quality of the drinking water that they serve to their customers is of a high quality. In this paper, a research on the switching from chlorination to chloramination with the use of a clustering machine learning technique on water quality samples taken from the Assets and the customer taps of a Water Company in the North of the UK is presented. Findings indicate that factors such as low disinfectant residual and total organic carbon could increase the bacteriological regrowth risk in the water distribution network, an iformation water companies require in order to make decisions for interventions

to

Enabling Equitable Water Supply in a Mega-city using a Big Data Analytics Platform

K R Sheetal Kumar

Presenter:

Authors:

Prithvi Alva, K R Sheetal Kumar, Yogesh Simmhan and M S Mohan Kumar

Bangalore is one of the fastest growing mega-cities in India with a population of 12 million. But its water infrastructure suffers from inequitable distribution and a significant Unaccounted for Water (UFW) losses. The availability of billing and observational data from the water distribution network using Internet of Things (IoT) technologies gives the opportunity for data-driven decision making to address these limitations. We propose a Big Data platform to acquire, analyze and visualize the behavior of the water network, and effectively exploit hydro-informatics for water managers. Our study area covers 15% of the Bangalore city with over 700 sensors and 1.5 million residents. Our platform architecture consists of a data ingest pipeline, a data storage layer, a query and analytics engine, and a visualization portal. It acquires observational data periodically, and uses Apache Spark SQL to perform data quality checks and transformations to generate sanitized and standardized data. This is stored in HDFS using a compressed columnar format to reduce disk access costs. An analytics service offers ad hoc and pre-defined queries related to inequity and UFW. This is accessed from a visualization portal for water managers and analysts, which has a geo-spatial interface to conduct independent analysis on individual DMAs. Also, a plotting interface allows more complex studies across different regions, variables, time, and indices like Theil and Gini. Currently, our platform has ingested 18 months of data, that translates to 10.4 millions rows of flow, pressure and billing data from 185,316 bulk and residential meters, which are being analyzed. Our platform reduces the time to science and decision making, ensures repeatability, and is designed to scale across a Mega-city. It also includes novel indices for continual evaluation of inequity and UFW, which can translate to policy planning and interventions.

to

Burst detection in district metering areas based on deep neural network

Xiaoting Wang

Presenter:

Authors:

Xiaoting Wang, Shuming Liu and Xue Wu

Early detection of hydraulic accidents or abnormal situations has always been an essential but difficult task in urban water distribution system management. This paper proposes a sensitive deep neural network for burst detection in district metering areas (DMAs). A bidirectional long short-term memory network model is developed to predict time series flow meter data. The relationship between flow patterns and predicted residual performance is analyzed and the redundant residual patterns are excluded from the identified outliers. The capability of the proposed method to detect outliers caused by bursts is tested and the reason for reducing false positive rate has been discussed in detail. The results from a case study for a real-life DMA in South southern China show that the method is relatively high successful at triggering burst alarms with a 99.64% detection accuracy, a 98% true positive rate and a 0.36% false positive rate. Moreover, the proposed method outperforms the previous distance-based method and shows good potential for water losses control in water distribution systems.

to

A nexus ontology to support the generation of cross-domain policies

Aitor Corchero

Presenter:

Authors:

Aitor Corchero, Lluis Echeverria, Eugene Westerhof, Sara Masia, Gabriel Anzaldi, Xavier Domingo, Edgar Rubion, Janez Susnik, Roberto Garc�a, Chrysi Laspidou, Lydia Vamvakeridou-Lyroudia and Floor Brouwer

A current trend in water-nexus governance is to build cohesive policies to guarantee long-term sustainability of the resources. With the advent of the IoT revolution, nexus governance has moved on to involve intelligent sensing, systems integration, advanced modelling, data-driven intelligence and continuous integration of different stakeholders. This kind of technology has permitted the policy-makers to have fine-grained information about the different domains (water, energy, climate, land-use, etc) involved in decision making. Hence, the elaboration of different scales of policy models has permitted a more efficient use of resources. Despite these benefits, a lack of common understanding of nexus variables and their potential links has immersed water governance in a bottleneck that hinders reaching higher efficiency levels. With that aim, significant advancements have been made in terms of variables harmonization at the domain-specific level. However, the subsequent complexity in the interrelation between the variables at cross-domain has fragmented the holistic decision-making process, requiring expert judgement and analysis in order to understand it. In this situation, the present research line proposes a water-nexus ontology and a software platform which uses it to provide a common representation of the information involved in the nexus. This helps the involved stakeholders create and design newer policies and decision-making tools. Both insights are part of the EC H2020 project SIM4NEXUS (GA 689150), which focuses on elaborating intelligent tools, in the context of a serious game environment, that facilitate the design and impact assessment of nexus policies.

to

The SIM4NEXUS design challenge: making a User Interface for a Serious Game based on a large size System Dynamic Model.

Mehdi Khoury

Presenter:

Authors:

Mehdi Khoury, Lluis Echeverria, Niels Rigel, Luqman Perviz, Barry Evans, Nikos Mellios, Chrysi Laspidou, Lydia Vamvakeridou-Lyroudia and Dragan Savic

The ongoing H2020 research project SIM4NEXUS [1,2,3] focuses on exploring the interactions between the five Nexus sectors (water, energy, land, food, climate). The project’s emphasis is on the development and application of policy-relevant Serious Games for 12 case studies ranging from sub-national to global scale. In the context of one of those, the Greek case study, a scientifically-robust System Dynamics Model (SDM) was first developed to quantify Nexus relationships and interactions. This paper focuses on the next task undertaken while creating the Serious Game: the creation of a User Interface (UI). The main objective of the UI is to help users to explore the SDM model in all its complexity (more than 10 thousand variables and 14 interconnected regions) without causing confusion for players. Being clear about the Nexus system is important as the players range from members of the public to policy stakeholders and experts. Although this challenge is more a design problem than a technical issue, it is not trivial, and, to our knowledge, no other Serious Games has attempted to create behavioural understanding for a model of this size. The developments of the UI were undertaken through a trial and error procedure and a continuous iterative co-creation process with stakeholders to ensure that the visual information available was pertinent to situational awareness and strategic policy making. A beta version of the Serious Game is currently operational and will be available to play and experiment at a special session of CCWI 2019.

to