Author: Rick Mullin / Source: acs.org
New information technology partnerships announced by major chemical companies signal that the industry is poised to fundamentally change inhow it deploys digital technology. Companies are starting to bring techniques such as machine learning and quantum computing to bear on vast troves of research data. The objectives are increased innovation in the lab and a significant reduction in the time and cost of commercializing new products.
It is not unusual for a consulting firm to coin a term to generate some buzz around a business practice on which its experts advise.
It’s sometimes as simple as turning an adjective into a noun, as McKinsey & Co. did with the word “digital” in a recent report on information technology in the chemical industry. It’s somewhat bold, however, when the word seems to describe a practice at which targeted clients have excelled for decades.
“There is a lot of excitement about the potential of digital,” McKinsey proclaims. It goes on to define digital as an “umbrella term for all the digital-related topics, including new or enhanced ways of operating businesses by using data, analytics, and new digitally enabled technologies, communications, and platforms.”
But hasn’t the chemical industry long been a digital leader, installing advanced financial management software and spending billions of dollars preparing for the shock of Y2K? Aren’t plants and laboratories today highly automated, filled with the latest computers? There wouldn’t appear to be much chance for IT excitement in chemicals.
Yet enthusiasm for digital as the “next big thing” is borne out in recent announcements by BASF, Dow Chemical, Evonik Industries, and others that describe partnerships with computer giants such as IBM and Hewlett Packard Enterprise (HPE). These new deals are touted as taking the computing-intensive components of science and business into a new realm that Evonik, for one, calls “digitalization.”
The announcements highlight investments in a new generation of high-powered supercomputers such as the HPE Apollo 6000 that BASF is installing as a digital processing center for its worldwide operations.
Cost conscious
Because advanced materials take many years to develop with little certainty of market success, firms are looking to harness digitalization.
| R&D time (in years) |
R&D cost (millions) |
Commercialization cost (millions) |
Technology uncertainty |
Market uncertainty |
|
|---|---|---|---|---|---|
| Advanced materials |
5 to 15 |
$2 to $20 |
$50 to $500 |
High ↑ | High ↑ |
| Biotechnology |
10 to 15 |
5 to 10 |
300 to 900 |
Very High ↑ | Medium ↑ |
| Software |
0 to 2 |
0 to 3 |
1 to 10 |
Low ↑ | Medium ↑ |
Source: Elicia Maine, associate professor, Beedie School of Business, Simon Fraser University, and Purnesh Seegopaul, Pangaea Ventures, Nat. Mater. 2016, DOI: 10.1038/nmat4625
But the news also has to do with applications for that computing power. Firms are adopting artificial-intelligence- and machine-learning-enabled methods of computing that accelerate the speed and broaden the scope of experimentation by making use of the vast amounts of data now available in the lab and manufacturing plant.
With cloud computing making data storage and access an inexpensive commodity, the focus in IT development has shifted to crafting software that supports methods of extracting meaning from data generated in materials science R&D.
BASF says the Apollo 6000, to be installed at its headquarters in Ludwigshafen, Germany, will be one of the world’s largest supercomputers for industrial research.
“The new supercomputer will promote the application and development of complex modeling and simulation approaches, opening up completely new avenues for our research at BASF,” Martin Brudermüller, the chemical company’s chief technology officer, said at the time of the announcement.
The supercomputer will include Intel’s Xeon processors and other components to boost bandwidth. It will be deployed to answer complex questions and reduce the time required to obtain results from several months to days. At 1.75 petaflops, a unit of computing speed equal to 1,000 million million (1015) operations per second, the Apollo 6000 has 10 times the computing power currently dedicated to scientific computing at BASF and will rank as the 65th largest supercomputer in the world, according to the company.
Meanwhile, Evonik will work with IBM as part of its newly declared digitalization program, for which the German company says it is committing $120 million. The partnership is casting a wider net than BASF’s and is looking at bringing an elevated level of computing firepower to a broad range of business functions. It will incorporate cloud-based technologies, management of distributed device networks via the so-called internet of things, and blockchain, a digital ledger that records transactions between two parties through a peer-to-peer computer network.
Evonik will also tap into IBM’s new capabilities in quantum computing for R&D modeling. In a break from standard binary computing systems, quantum computing performs calculations using bits, or “qubits,” that can exist in a state of superpositioned zeros and ones. It is a methodology nearly catered to materials research in that it employs principles of physics developed in the study of molecular composition. Other companies, including Dow, are developing applications for quantum computing, prompted, according to some industry watchers, by the imminent debut of quantum computers from conventional hardware companies and Google.
But Dow’s new partnership is not with a computer company. Instead, the chemical maker has signed with a quantum computing software firm, 1QBit. Through the collaboration, Dow hopes to glean insight on how quantum computing methodologies can advance research efforts. 1QBit, meanwhile, seeks to use Dow’s experiment design expertise in developing software for materials science.
Similarly, Solvay is hoping to gain an inside track on next-generation computing by investing $2 million in MultiMechanics, a virtual testing software firm focused on modeling and failure prediction in materials research. Solvay says the software will advance its high-performance polymer and composites pipeline for auto and aerospace markets.
“We generally see digitalization bringing huge opportunities in value creation,” says Stephan Schenk, team leader for high-performance computing databases at BASF. “All the other companies are pushing for opportunities, too, which is why you are seeing a lot of stories.” But every industry—indeed every company—has its own take on what digitalization means, he says. “To us, it means the use of digital tools to boost our development process.”
Although BASF has decades of experience with high-powered computing, the partnership with HPE will significantly boost its activity. “It’s not new, really,” Schenk says, “but it’s the first time we’ve had this amount of computing capabilities.”
The partners will focus not only on increasing speed in R&D and commercialization but also on identifying areas in which advanced modeling and testing methodologies can break new ground in the lab.
“There are a lot of questions in our product development cycle that can be answered or augmented with information from simulation,” Schenk says. “You can narrow down the number of experiments you need to do to create…
Click here to read more
