The value in battery optimisation: Insights by Liam Brannigan

How does battery optimisation deliver value? 

In an increasingly dynamic energy landscape, the ability to anticipate, not just respond to demand is what separates truly intelligent systems from the rest.

At the heart of Joulen’s PARIS platform and battery optimisation approach is a forward-looking capability: forecasting. By predicting how energy will be consumed and generated at a residential, commercial or industrial site before it happens, we enable smarter, more efficient decisions that maximise value. This shift from reactive control to predictive planning enables energy systems to operate with greater precision, resilience, and economic efficiency. 

The key reason that Joulen’s PARIS battery optimiser is able to deliver value is that it looks ahead to understand what the electricity requirements of a site are likely to be and then it plans accordingly.

Whether the site is residential or a commercial/industrial site PARIS forecasts two things: the energy consumed and the energy generated. We make forecasts for each of these for every half-hour of the day. 

We use advanced, deep learning models that analyse the patterns of energy use and generation at each site and produce forecasts. We feed these models with the historic energy flows from our sites along with other inputs such as the weather. The machine learning model learns the patterns of site energy usage and generation and uses it to forecast the coming days. 

We produce forecasts for every individual house or commercial property. This allows the algorithm to learn typical patterns that hold across many sites but also the individual quirks of each site.  For example, our models capture which houses typically charge their electric vehicles at night and which commercial properties are closed on Sundays. But our models also detect finer patterns such as how energy use at a retail property may gradually ramp up before opening. 

Our energy generation models capture how the solar panels at a site generate different amounts of energy at different times of the year. Our models also capture patterns, such as whether the solar panels on a house get shaded by trees in the afternoon. 

As well as capturing the regular patterns of energy flows at a site, our machine learning models also adapt to changes in these patterns. For example, a household may have a different pattern of energy use on a day when residents are working from home compared to days when they are working in an office. We run our machine learning models continually through the day and night to detect these changes and update our forecasts accordingly.  

The real value of battery optimisation lies in this combination of foresight and adaptability.

Ultimately, the future of energy optimisation lies in systems that continuously learn, adapt, and anticipate. By combining granular, site-specific insights with advanced machine learning, we move beyond one-size-fits-all approaches and towards truly intelligent energy management. As patterns evolve, whether driven by human behaviour, weather, or technology adoption, our models evolve with them, ensuring that every decision is grounded in the most current and relevant data. This is not just about improving battery performance; it’s about redefining how energy is managed in a world where foresight is the most valuable resource of all.