Why AI Matters in the Water Industry

AI/ML are invaluable in demand forecasting. By analysing historical data for water consumption patterns, weather conditions, population growth, and other relevant factors, utilities can develop accurate models to predict future operational needs, thereby optimising production, distribution, and capacity management.

For clean water, ensuring ample water is available at the right time and place, is a clear goal, considering increasing droughts, population growth and ageing assets. For wastewater, it covers some gaps from long- term underinvestment and focus. The drivers are similar, in that climate change and increasing populations impact networks not designed to withstand these challenges.

This has been Metasphere and Meniscus’ focus for years. Highly accurate hyperlocal weather forecasts are changing how councils control localised flooding, utilities manage remote raw water sources and how pollution events are reduced. Metasphere’s mantra is ‘No Spills’ – we believe shifting from a reactive ‘monitor and alarm’ to proactive ‘Predict and Prevent’ operational mentality, is the answer to achieve Zero Spills and with technology playing a critical role.

AI/ML technology isn’t new. Energy consumption and optimisation are proven use cases for AI/ML within water utilities, having been widely adopted in recent years. With energy being one of the highest costs for utilities, the driver is clear. Water treatment and distribution require significant energy inputs. Analysing energy consumption and system performance data, offer utilities improvement opportunities. Machine-learning algorithms can optimize pumping schedules, adjust treatment processes based on real-time conditions, and predict energy demand. This means operating at the most efficient rate whilst optimising pumps against network demands and performance; thereby reducing cost and environmental impact.

Utilities have historic strength in Asset Management, together with supply chain knowledge and capacity. However, part of the challenge is moving away from management of isolated assets and transition to automated and integrated networks. This is a big shift but viewing networks holistically (drinking water being more advanced than waste) and taking a catchment-based approach, is the way forward.

This approach includes overlaying weather data against catchment and/or drainage areas and utility network data (level, flow and increasingly water quality measurements) to predict impact on natural watercourses and coastlines. For years, modelling of tidal patterns has been used together with sewer discharge data to signal the quality of bathing water and shellfish sites.

Predicting water quality, including soil moisture data and information on Agricultural use, such as fertiliser by crop type and proximity to water courses, is an obvious and valuable albeit not simple addition to catchment-based management. The ability to handle multiple data sources, including traditional, new IoT solutions and even data delivered through citizen science, present challenges to historic functional silos (Operational and Data) developed over time. Integrating these disparate data sources (including nomenclature, varying quality, and time sampling rates) is still a critical role for people who are subject matter experts, and who understands that assets and networks are more important than ever.

In conclusion, AI/ML has a critical role to play in water utility operations. Leveraging supply chain expertise with AI technologies, coupled with their own operational specialists; utilities can improve efficiency, reduce costs, optimise their networks, and protect the environment; all whilst delivering sustainable water services to the communities they serve.

More News and Articles