Waternomics is an ongoing EU-funded research project that addresses key challenges regarding the efficient management of, and the need for decision support tools in the water supply sector. A novel aspect of the Waternomics project is to apply Automated Fault Detection Diagnostics (AFDD) to building water networks and systems in order to identify faults (leaks, malfunctioning equipment, inefficient operation etc.) and to enhance understanding of a building's water usage. AFDD is a measurement science which has traditionally been used in other similar applications, most notably Heating Ventilation and Air-Conditioning (HVAC) systems to identify and rectify faults resulting in energy savings of between 10 and 30%. To date these AFDD tools have not been applied systematically to water supply infrastructure. This study aims to develop and implement, in a number of large scale pilots, AFDD tools that can be applied to water networks. The ability to detect faults will bring added value to water management systems by increasing awareness of the role of system faults in (i) increasing costs, (ii) causing meter problems at the consumer level and (iii) contributing to higher water consumption rates. Detecting faults at the earliest possible stage can lead to reduced maintenance costs, higher customer satisfaction and increase the efficiency of water supply systems. This study leverages a number of unique pilots that target a range of end-users, from a large commercial building to public buildings and residences in Italy, Ireland and Greece.
The application of AFDD tools within water networks in buildings requires the development of rules and models relating to the water network's operation. These can generally be divided into rule-based or model-based AFDD tools.
1. Rule based AFDD should utilise simple logic applied to the water network to decide whether any given system is operating as designed or not e.g. a fault could be identified if there are two identical pumps servicing the same load in a building, yet consuming different amounts of energy. Traditional flow metering may not detect such issues.
2. Model based AFDD can be segregated into Law-driven and Data-driven models. Law-driven or forward models apply physical laws to the system to forecast its operation under a given set of conditions. Data-driven or inverse models require the actual water usage trends to be quantified and characterised using sensors, placed at critical locations in the water service of a building. Comparison of the building's real-time water usage to the Law and Data-driven models, in conjunction with model tolerances, can identify faults, their severity and enable rapid diagnosis.
The AFDD tools developed within Waternomics form part of four high impact experimental pilots that target various different end users/stakeholders (i) Domestic users in Thermi, Greece (ii) Corporate operator in Levante airport, Rome, Italy and (iii) a Municipal water based demonstration in NUI Galway's Engineering building (a university building) and a school in Galway City (both in Ireland). In each pilot the installation of targeted sensor and meter equipment (e.g. water and energy meters) will inform the development of AFDD tools.
Results and Discussion
To commence the investigation, the water flows need to be baselined in each pilot. Where possible, the associated energy users e.g. pumping equipment will be investigated as well. The initial work has been concentrated on the NUI Galway building. This is a newly constructed engineering building which is equipped with a range of sensors monitoring building structural performance, mechanical operation etc. including 11 individual water meters. These measure various parts of the water network including the entire mains water use (32,000 L/day) down to an individual drinking fountain (15 L/day). The meters measure cumulative water flow totals physically and are connected to a Building Management System (BMS) which reports and stores the water usage from each meter remotely. The first step in developing AFDD Data-driven models was to observe the existing physical meters and to compare them with what the BMS recorded. The building's water infrastructure and systems have been investigated also, so as to develop the AFDD Law-driven models and rules.
Observing the meters gave vital information relating to the building's main water users. However, it also revealed that there is potential to characterise the building's water usage in more detail with additional meters and with the generation of virtual meters, which will create data relating to the water network inferred from the existing meters. This will add to the quality of the models produced for the Data-drive AFDD. Comparing the physical readings to what the BMS recorded highlighted 4 faulty remote sensors. A strategic plan to implement additional meters and virtual meters which will characterise the buildings water use further has been devised and a solution to record the meter readings securely are being worked on.
The work to-date has shown a significant potential to inform increased water network efficiency, reduced maintenance costs, higher customer satisfaction and increased savings in buildings with the use of AFDD tools. To attain these savings, rules and models must be developed relating to specific water systems and networks. Investigation of the NUI Galway engineering building has revealed that calculated implementation of metring and in-depth study into the network infrastructure are required to implement AFDD to the building. It is clear that AFDD tools must ensure that the data potential from such metering is mined, analysed and provides accessible information to relevant stakeholders that informs the design of new systems, maintenance and on-going decision making. Each pilot will involve its own unique elements and intricacies, the AFDD tools in this project will evolve and adapt to this however, to create an all-encompassing fault detection platform which will access untapped water and energy savings in various buildings.