麻豆传媒

As a child growing up in Kolkata, Suchi Sanyal was used to power cuts.

Doing homework with her brother by a light powered by a portable generator gave her a special appreciation of the importance of affordable, reliable energy.

At 22, Suchi began studying for a postgraduate degree in Materials Science and Engineering at the Indian Institute of Science in Bangalore. Since then, she has been fascinated by new materials and the way they shape our world.

鈥淚f you look at the history of human civilisation, from the Stone Age to the Iron Age and what we now call the 鈥楽ilicon Age鈥�, we have classified civilisation in terms of materials as they have evolved,鈥� she says.

New materials are essential in the world鈥檚 transition to a cleaner energy system, because many of the raw materials we rely on for energy today are in finite supply or produce greenhouse gas emissions when they are mined or burned.

Today, as General Manager of Computational Science at the Shell Technology Centre Bangalore, Suchi and her team are using digital technologies including artificial intelligence (AI) to accelerate the discovery of new materials that go into batteries for storing renewable power, solvents for carbon capture, and low-carbon fuel blends.

Working in Bangalore, the 鈥淪ilicon Valley of India鈥�, allows the team to tap into a vast pool of technological expertise. The city boasts prestigious academic institutions, technology business parks and a thriving community of start-ups and entrepreneurs.

Computational technologies including AI enable researchers to rapidly digest and summarise decades of scientific research, predict the properties of new materials, and simulate their behaviour under various conditions.

For Shell, such process improvements help accelerate the discovery of materials for the energy transition, while also saving money by reducing the amount of lab work and field testing required.

But AI is only as effective as the data fed into it, and this is where scientists continue to play a vital role.

鈥淎I can give you fast results, but it can also give you meaningless results if you don鈥檛 train it with the right information,鈥� says Suchi. 鈥淭he chemistry intuition comes from our computational and experimental chemists.鈥�

Lithium-ion batteries power today鈥檚 smartphones, laptops and electric vehicles, and are also used to store wind and solar power. But the rapid growth in demand for raw materials has put pressure on supply chains and driven the search for alternatives.

Suchi鈥檚 team is working on a new type of rechargeable battery, known as a flow battery, which uses organic molecules 鈥� those that contain carbon 鈥� to provide long-term energy storage at a low cost.

By training AI models to identify molecules with the required properties, the team was able to screen 112 million molecules in just a month and identify 67 promising options, far exceeding what people would be able to achieve in that time. 

鈥淭raditionally, chemists have relied on experimentation 鈥� on trial and error 鈥� to discover new materials, and this can take a long time,鈥� Suchi explains. 鈥淐omputational models essentially narrow down your research space to focus and guide that experimentation.鈥�

Sodium, an abundant material in nature, offers another potential alternative. But exchanging lithium for sodium in the electrodes requires changes to several components of the battery, including the electrolyte.

Beena Rai, chief scientist at IT services company Tata Consultancy Services (TCS), was one of around 300 attendees of a major conference on AI for sciences in Bangalore in October 2023, convened by Shell. She says that AI significantly reduces the time it takes to filter through thousands of potential electrolytes 鈥� and can sometimes generate options that scientists hadn't considered.

鈥淭he most powerful thing about AI is that it has that generative ability,鈥� says Beena. 鈥淵ou can discover materials which probably we would not have conceived of.鈥�

The success of the modelling process for flow batteries has inspired the use of AI in another Shell operation 鈥� the molecular modelling of solvents for capturing carbon dioxide (CO鈧�), the most common greenhouse gas.

厂丑别濒濒鈥檚 CANSOLV process uses molecules known as amines to capture CO鈧傗�痜rom waste gas generated by industrial facilities. But amines naturally degrade over time, so there is a need to find longer-lasting solutions.

The structure of amine molecules affects their rate of degradation and, by feeding an AI model with chemical insights, the computational science team was able to quickly filter through the molecular database and identify amines which are slower to degrade.

They are now testing these amines in the lab with the aim of creating a more durable version of CANSOLV.

Mathias Steiner, Sustainable Materials Lead at IBM Research, who is also researching materials for CO鈧� capture, says that AI allows scientists to design new materials, while also simulating and verifying their properties.

鈥淲e are trying to move more of the research effort from the lab into the computational environment because there it can be automated, it removes a lot of work from the hands of the researchers and frees up their time and resources so they can focus on higher-value activities,鈥� he says.

The use of computer simulations and AI in material design is increasing, with further applications in lower-carbon fuel production.

Software company Dassault Syst猫mes is using machine learning and AI to develop cheaper but more efficient electrodes for generating hydrogen using electrolysers.

鈥淯sing scientific AI to simulate the performance of different electrodes allows us to design a commercially viable device without the need for thousands of costly lab experiments,鈥� says Lalitha Subramanian, Science Fellow and Senior Research Director at Dassault Syst猫mes.

As part of its innovation partnership with Scuderia Ferrari, Shell has developed a race fuel containing 10% advanced bioethanol, using computational modelling to predict the combustion behaviour and performance of each fuel blend, significantly reducing the development time.

The team is now working on a 100% sustainable race fuel, which could be made from advanced biofuels and synthetic fuel, to meet requirements for the 2026 Formula 1 season. This research is expected to inspire the development of future lower-carbon road fuels for ordinary drivers.

Suchi believes the scientific work she is doing with digital technologies and AI is vital to the energy transition. But she also knows that no one company has all the answers, or can make all the breakthroughs.

鈥淓nergy transition requires a movement that goes beyond one organisation,鈥� says Suchi. 鈥淒igital technologies can accelerate innovation, but we also need to collaborate more to build a cleaner energy system for all of us.鈥�