With a strong focus on growing its international business, machine learning solutions provider DataProphet has signed a partnership with leading metallic parts enhancement company Norican Group to accelerate the adoption of artificial intelligence (AI) in the foundry industry. This came after a successful partnership with local automotive castings provider Atlantis Foundries.
“Atlantis Foundries faced challenges with complex interrelationships between process variables, as well as a high cost of scrap. This makes traditional process control methodologies uncompetitive,” says DataProphet chief technology officer Dr Michael Grant.
Grant says that DataProphet started working with Atlantis in 2017. DataProphet’s PRESCRIBE solution, now known as Monitizer | PRESCRIBE in the foundry market, was used to improve Atlantis’ processes.
Monitizer | PRESCRIBE is a deep learning solution that provides operational parameter set-point prescriptions across the manufacturing process to reduce product scrap that is produced.
“Monitizer | PRESCRIBE halved the scrap rate in the first month of deployment at Atlantis Foundries and achieves a 0.1% external scrap rate on an ongoing basis,” confirms Grant.
He notes that as Atlantis Foundries continued to grow and benefit from this partnership, “DataProphet started to appear on the radar of many international foundries and other manufacturers”.
This included green sand foundry technology supplier DISA, part of the Norican Group.
Norican visited Atlantis’ operations in 2018 and was impressed with DataProphet’s solutions and impact on the partnership. This started the partnership discussions with DataProphet and Norican Group, and led to a number of joint innovation projects over the last 12 months with DISA.
This led to the industry-exclusive partnership with Norican, announced in January this year.
Grant asserts that these projects are being used to develop commercial AI application products and the first AI application developed under the collaboration will go live in a customer’s manufacturing operation in a few months.
Grant emphasises that Monitizer | PRESCRIBE is crucial to these partnerships, and describes it as an expert execution system. The solution coordinates manufacturing systems’ responses with production and quality plans, with costly process management operations also streamlined.
Owing to the importance of the Monitizer | PRESCRIBE solution, the broad support for Norican’s equipment offering has been achieved through integration with its NoriGate technology, which forms the foundation for the Monitizer solution.
This means that DataProphet | PRESCRIBE has been integrated across all the Norican Group companies.
“This application can prescribe optimum plant control parameters that frequently reduce the cost of nonquality by more than 50% through a customised single model approach while taking higher order effects into consideration.”
The new AI applications developed from this partnership are available on solutions from other Norican brands, as well as non-Norican equipment, through the company’s NoriGate, an Industrial Internet of Things gateway.
“DataProphet has a full product development roadmap. Exciting upcoming innovations in the foundry space will be released through Norican and its distribution network,” he notes.
Meanwhile, Grant argues that a significant challenge of implementing AI systems in developing countries, such as South Africa, is the level of maturity of the processing systems active in operations, such as manufacturing operations.
While the use of AI is slowly being integrated into the local industry, he comments that it is still underused, as the aim is to improve manufacturing processes using AI, rather than building processes around AI.
“With this in mind, we take time to understand the unique environments of each client before deployment. During this process, we can see where there are blindspots from a data perspective and can advise on the relevant improvements that need to be made to be AI-ready.”
Grant explains that having a clear Fourth Industrial Revolution (4IR) roadmap is critical, as it provides direction on what needs to be done and when. “We have managed to play a significant role in some of our clients’ 4IR journeys, which has seen great stakeholder acceptance and substantial value delivery.”
He suggests that the complexity of the foundry process – with many variables that have intricate relationships with each other – resulting in various quality measurements, makes it ideal for 4IR technologies.
“Traditional approaches to analysing data and optimising for quality falls short because they are, mostly, reactive. These approaches were also not built to handle the volume and level of complexity of the data that is now being collected,” he concludes.