While we’ve looked at some of the overarching themes of how AI is changing the cleantech industry in our miniseries, a number of specific breakthroughs and innovations emerging in 2019 demonstrate just how important AI is to the future of renewable energy. As the pressure to scale up cleaner and more sustainable solutions to global issues around power provision mounts, AI is quickly establishing itself as essential to many different parts of this process – a skeleton key capable of unlocking much more of clean energy’s potential.
5 ways AI is making renewable energy greener
1: From start… (manufacturing)
Current manufacturing techniques used in the creation of solar panels require the use of rare earth elements and temperatures of 2,000˚C, which is so high that it requires fossil fuel-generated power to achieve. So despite the clean and renewable energy it provides during its lifecycle, each solar panel is far from being emissions-free.
AI is set to change this, however, by boosting research into the development and usage of new materials suitable for solar panels. Experimentation into such questions is generally done through a rather torturous trail-and-error approach which may require thousands of individual tests before paying off with a big breakthrough. With AI in the picture, many of the mundane and even more complex trails and tasks can be automatically undertaken and analysed, speeding up this vital process and forging the next generation of solar panels so they can be manufactured without incurring a huge ecological burden.
ADA, (Autonomous Discovery Accelerator): the world’s first self-driving AI laboratory is the perfect example of how AI can take clean energy manufacturing to the next level. Ada’s current aim is to make solar panels more resilient and to turn carbon dioxide into useful fuels.
2: …to finish (recycling)
Looking to the end of any clean energy assets’ lifecycle is as important as its beginning. If we are going to achieve a truly circular economy where sustainability is embedded into every major industry, then the recycling of solar panels, wind turbines and other clean energy hardware cannot be overlooked.
Sadly, in the case of solar panels, recycling remains a serious problem with no clear answers having yet emerged. The US National Renewable Energy Laboratory recently revealed that the country now installs around 7 million pounds of solar panels per day, and given that these panels are largely designed to last around 30 years, this is a massive recycling issue in the making.
This is another reason why leveraging AI for the purposes of testing for suitable materials to be used in next-gen solar panels is essential. By alleviating the economic and ecological costs of their manufacture as well as their repurposing at the end of their life cycle, AI will help make solar the truly green energy source it needs to be.
3: Smarter utilisation
India’s power provision crisis is well documented and isn’t going away. Solar is rapidly posing a viable solution, but given the temperamental energy landscape of the country, it needs to be delivered in a much more flexible and adaptable setup.
Enter ThingsCloud, a startup dedicated to using AI and solar power together to deliver clean, connected, and intelligent home devices. Their solar inverter solutions for residential energy systems allow users to accurately gauge and monitor their energy consumption, conserve 20-30% excess energy, and store backup power to use whenever there’s a blackout, something that happens rather frequently across much of India.
Through this AI-powered solution, residential energy system operators can not only enjoy 20-30% greater energy efficiency through this flexible storage feature, initial case studies suggest that overall operational costs can also be reduced by up to 50%.
4: Predicting output for a more strategically sound energy grid
Even more than solar, generating power through wind can end up being an unpredictable business given the generally unpredictable nature of the wind itself. Once again, the sheer computational power of AI is at hand to help make even the seemingly random nature of the wind more predictable.
Google subsidiary DeepMind claims that its machine-learning algorithms can help improve the accuracy of wind farm output predictions, which makes said wind farm much more valuable to the grid as it can reliably deliver power at set times that the grid can count on. This has a number of positive implications in terms of improving the grid’s overall ability to receive, store and then deliver power as efficiently as possible.
5: Taking clean energy ‘to the edge’
According to Ali Farhadi, co-founder of Xnor: “Power will become the biggest bottleneck to scaling AI.” Xnor’s tiny application-specific integrated circuit (ASIC) technology is designed to equip devices in the field with AI-powered deep learning abilities but, critically, they can run for years without even needing a battery, requiring only a solar cell instead. This is achieved by moving AI algorithm processing away from the cloud (thus incurring a heavy power burden as millions of interconnected devices rely on cloud computing to draw energy from data centres) and towards ‘edge processing’.
Greener and cleaner, AI is paving the way for solar and wind’s expansion
While both renewable energy sectors are already enjoying political support and burgeoning private investment across the world, what is needed more than ever is scalability. In order to successfully make the case that renewable energy sources are able to literally take over the reins of power from fossil fuels, it will be necessary to show that they can be quickly ramped up to reliably satisfy demand. In a growing number of scenarios and use cases, AI is making that case seems increasingly viable.