Published

Why is AI so vital for foundries? And what is AI anyway? Here’s a quick guide.

Norican leads the foundry industry with its Monitizer® suite of digital tools, including the cutting-edge Monitizer | PRESCRIBE which harnesses Artificial Intelligence (AI) from DataProphet to optimise entire foundry processes.

Here, we catch up with Norican’s top digital expert Dr Dennis Janitza, Senior Vice President at ‎Norican Digital Labs, to find out why AI is such a game-changing technology for foundries of all shapes and sizes.

Why should foundries investigate AI?

It’s simple. Scrap increases production costs. AI helps minimise scrap and cuts those costs to a minimum. Think of AI as a virtual engineer that sits alongside the foundrymen and guides their actions. AI’s huge value lies in knowing which parameters are the most influential and when and how they have to be adjusted.

What is AI and how does it work in foundries?

AI is intelligence demonstrated by a machine, which includes functions similar to (but not the same as) human behaviours like learning and problem solving. Combined with cloud computing power, an AI can analyse far more complex data sets than any human could possibly manage, with hundreds or even thousands of variables in the mix.

A foundry AI is “trained” on historic data to work out the relationship between process parameters like metal temperature, pouring time or injection speed and the output of interest. In most cases, this will be the amount of scrap. The AI “learns” which parameters affect scrap and all the possible links between them, building a unique AI model of the entire process.

It uses that AI model to predict which machine settings and other parameters will deliver stable, high quality production. Working in real time, Monitizer | PRESCRIBE predicts the next best action to take before any scrap is produced. So it doesn’t just predict scrap, it prevents it.

What is the difference between AI and Industry 4.0?

Industry 4.0 technologies provide the smart factory infrastructure necessary for AI deployment, for example, the connectivity and networking that support data collection. AI is a computer application.

As well as reducing scrap, what else can AI do for foundries?

AI can help reduce energy use which, along with lower scrap, cuts both costs and CO2 emissions significantly. It can also predict which parts have hidden defects that only become apparent once shipped. Identifying these parts and not shipping them reduces the number of customer returns. Because it increases quality, an AI-supervised process will become a competitive advantage. Customers will see that parts made in a process supervised by an AI service are inherently better made.

Does AI support new business models for foundries?

Because AI software can handle very complex processes, foundries can bid for more challenging castings which normally have higher margins. That helps them move into new parts of their existing market and be competitive. AI improves process efficiency, letting foundries produce at lower cost – and so compete with competitors in lower- cost countries.

Is AI only for green sand foundries or Norican equipment?

Not at all. The AI used in Monitizer | PRESCRIBE can be applied across many different types of production systems and foundries. Norican is already collecting and applying process data from non-Norican equipment. The AI doesn’t care whose name is on the machine.

How do you get started?

To be accurate and effective, an AI needs data – lots of it. To optimize a complete process, you need to collect data from melting all the way to final quality control.

But to start your digital journey, just collect data from one system or sub-process. Technologies like Norican’s NoriGate make data collection cost-effective and easy to implement – just plug the yellow box into the machine and your network.

Would I need to buy a lot of new IT hardware or employ data scientists?

Norican delivers AI as a service so there’s little or no new hardware and no new staff. Infrastructure, system operation, generating the AI models and tailoring them to your process are all part of the service. You can start with a small investment and so a very low business risk, connect machines and gather data, then get the first results in a very short time.

Norican’s typical AI project installation time is around three months. Clients see results as early as one month after installation. Almost all the work can be done remotely, which was extremely useful during the pandemic.

Details