StormHarvester hindsight is turning into wastewater resilience

From rain to gain – how weathered wisdom from the UK and hard-won lessons in Australia could reshape wastewater resilience.

Queensland’s 2012 floods made headlines for their devastation – but for one engineer on the ground, they also sparked innovation. As the swollen creeks receded and the cleanup began, Brian Moloney was asking a more profound question: Why had the system behaved so differently from how it was modelled?

This disconnect between simulation and reality eventually gave rise to StormHarvester. More than a decade later, Moloney is bringing the company’s AI-powered solutions back to Australia, armed with real-world learnings from the United Kingdom’s utility frontlines.

When the 2012 deluge struck, Moloney, now Chief Executive Officer of StormHarvester, was living inland from Brisbane. Working with South Burnett Regional Council in the aftermath, he saw a stark gap between expected and actual network behaviour.

“Everything was kind of simulated in digital twins,” he said, “but those digital twins didn’t have a great relationship to what was actually happening in the sewers.”

Flooding appeared in places the models hadn’t predicted, and the infrastructure responded far less effectively than anticipated. That experience exposed a fundamental issue: drainage systems were reactive. The infrastructure remained passive despite humans watching weather forecasts and bracing for storms.

“We were all watching the weather forecast, but our infrastructure wasn’t. It didn’t optimise, it didn’t prepare,” Moloney said. “That was where the idea came from – to build a drainage system that wasn’t blind to what was coming. We started tinkering with prototypes, linking them to weather data and trying to drain tanks ahead of the rain. That was the first time it felt like we could shift from reaction to prevention.”

Engineering insight meets AI

StormHarvester’s early work focused on retrofitting intelligence into existing infrastructure—linking weather forecasts to real-time control of storm tanks and sewer systems. The aim was to optimise network performance and free up capacity before major rainfall. Once seen as ambitious, this idea of proactive control is now rapidly gaining traction.

Moloney, who holds a background in engineering and wastewater network operations, co-founded StormHarvester with the goal of fusing domain expertise with machine learning. The company has since become a specialist in AI-driven sewer monitoring, alarm rationalisation, and predictive network control.

With a team of over 60, including more than 40 software engineers focused on AI and data analytics, StormHarvester’s value lies in its dual focus: technical capability and operational context. One of its early innovations was applying rainfall radar data to drainage modelling—but not just in raw form.

“We’re very much domain-specific in business,” Moloney said. “It’s using the rainfall, but not strictly speaking your regular radar files. It’s a catchment-based approach, layering your catchment over your radar grid to give you the most accurate rainfall model possible.”

That precision feeds into the machine learning models, which grow smarter with every confirmed outcome.

The UK case study

While StormHarvester may have been born in Queensland, its real-world testing ground was the UK. There, increasing regulatory scrutiny of environmental performance and pollution events led water companies to deploy sensor networks at an unprecedented scale. One utility alone has installed close to 40,000 monitors.

StormHarvester now hosts data from more than 230,000 sensors within its platform, making it one of the most extensive AI-backed datasets in global wastewater management.

“Machine learning thrives on big data,” Moloney said. “But it’s not just about having the data, it’s about turning it into operational action. We get feedback from operators who confirm when blockages were found and cleared. That feedback loop helps the system learn. We now have 8000 confirmed blockages in our dataset.”

This confirmation loop is essential. Rather than training algorithms in a vacuum, StormHarvester refines its models using validated field reports. Over time, this has enabled the platform to deliver greater accuracy in detecting infiltration, identifying asset deterioration, and prompting timely interventions.

The result? Nine of the UK’s 12 major utilities now use the system as part of standard operations. Moloney said these tools have often allowed them to reduce costly site investigations and prioritise upgrades with far greater precision. He quoted one utility representative who reported spending £500 million on upgrades. It makes sense to spend £15 million on sensors to understand what the utility was upgrading.

Beyond dashboards: Delivering operational action

One of StormHarvester’s central messages is that sensor networks alone are not enough. Many utilities are collecting more data than ever but insights are often lost in the noise. Utility staff can struggle to turn that information into action without an interface between the data and decision-makers.

“One of the key learnings is that you need something between the data and the operator,” Moloney said. “You might install thousands of sensors, but unless a platform makes sense of it, you’ll be overwhelmed. The goal is to turn that raw data into a plan – get the right crew to the right place at the right time. That’s what prevents tomorrow’s pollution event. And that’s what we’re hoping to help deliver in Australia.”

StormHarvester’s platform addresses this through AI-powered alerting and alarm rationalisation. It learns each network’s baseline behaviour and flags anomalies – whether due to blockages, infiltration, or asset failure – before they escalate. Operators receive a prioritised list of actions based on real-time performance and weather data.

Avoiding déjà vu in Australia

Australia is no stranger to digital water programs. Smart meter rollouts, pressure monitoring, and leak detection have featured in both urban and regional utilities. But Moloney said the challenge ahead is different: interpreting complex, decentralised data in ways that lead to faster, smarter decisions in the field.

“There’s a huge amount of learning that exists,” he said. “We don’t really, in Australia, need to make the same mistakes that we made in the UK early days around sensor deployment, what works, what doesn’t, what density of censoring you need, what sensor locations are critical, what outcomes are possible.”

Rather than treat each deployment as a blank slate, StormHarvester is positioning itself as a conduit between utilities across hemispheres. The company has already connected several Australian utilities directly with their UK counterparts, enabling direct peer learning between operations teams that manage similarly scaled programs.

Moloney doesn’t view this as a market expansion exercise. Instead, he describes the company’s role as one of collaboration, experience-sharing, and long-term partnership.

“We’re not in this for a quick buck,” he said. “It’s about protecting the environment and communities and ensuring the network works as efficiently as possible.”

He will visit several Australian utilities before his Ozwater appearance in May. The message is simple: the tools are ready, and the UK’s experience is available–now it’s a matter of partnership.

“We don’t need to relearn what the UK already went through,” Moloney said. “We’ve seen what works, fails, gets traction, and gets ignored. This isn’t about selling software but building long-term relationships that improve environmental outcomes. That’s what makes it exciting to come back home and share what we’ve learned.”

For more information, visit stormharvester.com