How AI is transforming the chemicals and materials industry

While the impact of AI in virtually every aspect of our lives is being discussed everywhere these days, the chemicals and materials industry has perhaps been quieter about it than some industries. The fact is, though, that AI is being thoroughly explored by industry players who see many possible applications for AI.

In a recent IBM survey opens in new tab/window of 400 chemicals executives done, 80% said AI will be important to the success of their business in the next three years, and 74% of their chemical companies have begun implementing AI in R&D already.

A recent McKinsey & Company article opens in new tab/window contends that there is still much untapped potential for AI in the chemicals sector and that it stems from “the industry’s reliance on scientific data for innovation, the availability of (often fragmented) customer data, and the industry’s nuanced and complicated manufacturing processes.” The authors argue that generative AI adds intelligence and completeness to the data, and they see significant growth potential, estimating that the application of GenAI in energy and materials could create up to $140 billion in value.

Data and informatics

Harnessing all that data to make it ready for deployment with AI is a significant part of the challenge, and one that Elsevier is at the forefront of — often working with organizations to help them organize data according to the FAIR principles opens in new tab/window of findability, accessibility, interoperability and reusability. Elsevier’s SciBite has helped Johnson Matthey, a global leader in sustainable technologies, tackle their “unstructured data problem” (as detailed here), and the JM team in charge of the project is eager “to see generative AI put on top to make it even more accessible.”

Materials informatics is where data and AI unite, leveraging machine learning for discovery, design and optimization. A critical goal of informatics is to connect the work of materials scientists to that of data scientists. As IDTechEx’s Sam Dale observes in this article opens in new tab/window: “If integrated correctly, materials informatics will become a set of enabling technologies accelerating scientists’ R&D processes while capitalizing on their specialized knowledge.”

Move fast and create things

AI and machine learning have the power to accelerate the often-cumbersome molecule and materials discovery process, marking a big leap for chemical companies who will enjoy major time and cost savings by speeding up the process. While GenAI is not without its challenges — ranging from the need for vigorous oversight to maintain its accuracy and integrity to negative environmental impact from its use of energy — it is already being utilized with impressive results by some businesses.

McKinsey offers an example opens in new tab/window of a North American chemical company that wanted to formulate a new coating to address challenges faced in one of its end-market applications:

Using GenAI, the company mined the universe of materials using external data as well as its own proprietary R&D data and ultimately identified the molecules that could offer the desired functionality. This allowed the company to move from a slow and expensive user-customization cycle to a rapid customization cycle at a fraction of the cost.

Another notable success story of AI in chemicals has been a collaboration opens in new tab/window between Dow Polyurethane and Microsoft’s Azure Machine Learning that significantly accelerates Dow’s work on new polyurethane formulations: “All the team’s prior knowledge, expertise, and records about previous flexible foam formulations are fed into machine learning AI models,” the Microsoft article explains. “The models are then able to take that knowledge and predict where there are gaps that could suggest possible novel flexible foams. Rather than taking four to six months of work by multiple experts, the algorithms can sort through millions of possible combinations and suggest promising areas for experimentation — in seconds.”

Enhancing safety in chemicals and materials manufacturing

AI can also improve safety in chemical and materials processes and manufacturing. On the manufacturing facility floor, it can be used to analyze data generated by sensors and equipment to better determine when failures might happen. With this predictive information in hand, an organization can get well ahead of potential trouble by scheduling maintenance before problems arise.

For instance, Surveily opens in new tab/window is a company that uses AI for real-time hazard detection and predictive analytics to reduce accidents. Surveily’s software identifies pathway obstructions, monitors worker headcounts, detects worker safety compliance and monitors other possible safety hazards, providing alerts when these issues are detected so that the issue can be addressed proactively, not reactively.

In search of more sustainable materials

AI’s ability to make processes move faster and more safely is reason enough for players in the chemicals and materials industry to be thinking seriously about how to implement it into their R&D. Even more important, however, is the potential for AI to drive sustainability through the development of greener materials and more ecologically friendly manufacturing processes.

Of course, the health of the planet is reason enough to prioritize green manufacturing, but there is always the risk that concerns about the bottom line will get in the way of pursuing sustainability. By accelerating the search for sustainable options, AI can increase the discovery and adoption of chemicals, materials and manufacturing processes that are cleaner, greener and healthier for our environment.

The data foundation

While many chemicals and materials organizations plan to utilize AI and some already are, an industry-wide transformation will really depend on laying a good data foundation. Organizing internal data and being able to integrate it with valuable external data is key to unlocking insights and solving problems.

Elsevier has been at the forefront of this work, curating high-quality datasets in a process that involves:

• Data scaffolding — structuring data and filling in missing gaps using internal and external sources

• Data enrichment — adding ontological details based on internal expert knowledge, finding hidden links using existing external ontologies, etc.

• Knowledge graphs — creating the linkages between documents or terms to find insights in how they connect

“We have a long history in building ontologies and using them to create successful products,” said Dr Chris Cogswell opens in new tab/window, Customer and Engineering Global Consultant at Elsevier, in a recent webinar opens in new tab/window on AI and knowledge management. “And those are the kind of services we can provide to your organization, either as a data service vendor or as a part of your push towards creating AI data retrieval tools.”