Planning for the unknown: Why water networks are like home renovations

Author: Prof Dragan A Savić, University of Exeter, UK

Water industry professionals must balance cost, resilience and uncertainty to avoid costly mistakes. Professor Dragan Savic at the University of Exeter explores these challenges, drawing on examples from his recent book, Water Supply and Distribution Systems.

Imagine you're planning a major renovation for your home. But there’s a catch:

• You don’t know how many kids you’ll have in 20 years
• You’re unsure if you’ll need a home office or a pottery studio
• Your future income? A complete mystery

Sounds absurd, right? Yet this is exactly the kind of uncertainty engineers face when designing and planning operations of the underground pipe networks that deliver water to our cities. Get it wrong, and we either waste millions on oversized systems or scramble to fix shortages when demand outpaces supply. 

This blog draws on themes from our book, Water Supply and Distribution Systems, covering everything from basic hydraulics, water resources and treatment, to asset management and water finance, to explore how water network planning is changing with technological advances.

For example, water availability is the foundation of any supply system. Chapter 3 of the book explores the hydrological cycle, surface and groundwater sources, and the challenges of seasonal variability and climate change. Understanding these dynamics is essential for long-term planning and sustainable resource management. Before water reaches consumers, it must be treated to meet health and safety standards. Chapter 4 covers the principles of water treatment, including coagulation, filtration, and disinfection. It also discusses emerging technologies and the trade-offs between treatment efficiency, cost, and environmental impact.

The ageing maze beneath our feet

Water pipes aren’t just buried metal and plastics, but part of a living system. Over time, they rust, crack, and leak. Cities grow, people move, and weather patterns shift. As discussed in Chapter 10 on asset deterioration and asset management, water pipes are subject to corrosion, fatigue, and external loads. All this means we must constantly upgrade our water networks. But doing so is expensive. Think billions, not thousands! Furthermore, upgrading these systems involves not just engineering but also financial planning, covered in Chapter 11 on water economics, finance and project appraisal.

And here’s the real challenge: the future is foggy. We are notoriously bad at predicting the future and can’t know exactly where people will live, how climate change will affect rainfall, or how water-saving habits will evolve. Planning becomes a guessing game. 

The cost of guessing wrong

Traditionally, engineers have had to make educated guesses:

• Guess too high? We build oversized systems that cost a fortune and may never be fully used.
• Guess too low? We risk water shortages, emergency repairs, and expensive retrofits.

It’s a lose-lose situation.

Smarter ways to build: Lessons from everyday life

To tackle this, engineers have tried different strategies, each with its own pros and cons:

1. Build in phases. Like adding rooms to your house only when you need them, phased building spreads costs over time and allows for adjustments. But it still relies on decent predictions.
2. Build for the worst. This is the ‘just in case’ approach, like building your house to survive a meteor strike. It’s safe, but wildly expensive and often unnecessary.
3. Build to adapt. This is the sweet spot: create a system that’s easy to upgrade later. Think modular wiring in your walls, which is ready for new sockets without tearing everything apart.

The problem? Finding these flexible plans used to be painfully complex. Engineers would have to map out every possible future in giant flowcharts, which is impossible. It would also be slow, clunky, and hard to scale.

Meet the AI urban water planner

Enter AI multi-objective optimisation, which is a type of AI that uses heuristics derived from nature. Chapters 7 and 8 provide illustrative examples of the AI tools used to help with network modelling and design. For example, Genetic Algorithms (GAs) are introduced as AI tools commonly used for design and operations optimisation. Think of them as evolution applied to engineering. Instead of evolving better swimmers or stronger giraffes, GAs evolve better water network designs, i.e., pipes, pumps, and pressure constraints included.

• Start with a population: Begin with a variety of random design solutions, some promising, others wildly inefficient.
• Fitness evaluation: Each design is scored based on cost and hydraulic performance. Designs that fail to meet pressure requirements are penalised.
• Selection & crossover: The best designs are combined to produce new ones, mixing traits like cost-efficiency and layout effectiveness.
• Mutation: Occasionally, random tweaks are introduced. Some improve the design; others don’t, but they keep the search creative.
• Iterate: Over many generations, the designs improve. It’s not guaranteed perfection, but it’s far better than guessing.

GAs are a powerful tool for optimising complex systems like water distribution networks, especially when traditional methods struggle with scale or uncertainty.

The bottom line

Our water pipes are ageing. Fixing them is urgent. But planning for an uncertain future is hard. This book provides basic principles and tools, including AI methods, which can help us design systems that are both cost-effective and resilient.

It is important to emphasise that all these tools are not about replacing engineers. They provide them with tools to build cities that are more sustainable, no matter what the future holds.

And that’s not just clever, but it’s essential.

“We are notoriously bad at predicting the future and can’t know exactly where people will live, how climate change will affect rainfall, or how water-saving habits will evolve” 

Author

• Prof Dragan A Savić is Professor of Hydroinformatics at the University of Exeter, UK. He is also Global Advisor on Digital Sciences and CEO Emeritus at KWR Water, Netherlands, and Visiting Professor at the University of Belgrade, Serbia.