Transcript
Sam: Doctor Moyne, thank you for your time and I’m truly excited to hear your insights and expertise.
James: Please call me James.
Sam: Thanks, James. Let’s begin by discussing the significance of Industry 4.0 in the semiconductor manufacturing industry. Overall, what is the relevance of data with regards to scheduling and dispatching?
James: OK. Well, smart manufacturing is a really large topic that covers a lot of different areas, not just scheduling and dispatching, but also advanced process control, predictive maintenance, virtual metrology, pretty much all of those high tech things that you see going on in semiconductor manufacturing today. And it’s starting even to get into the supply chain where we’re talking about managing the upstream supply chain as well as delivery to customers and maybe doing things like latent yield where we’re finding a problem, let’s say in an automotive facility where they’re shipping vehicles and they’re starting to fail and we have to trace this back to the chip production and what the problem is. So all of that really falls under the smart manufacturing and one of the big enablers is the machine learning, the artificial intelligence, the analytics, if you will.
Scheduling dispatch is just one of those areas. But scheduling dispatch is going to benefit a lot from the fact that a big part of smart manufacturing is the integration both vertically and horizontally. Vertically means from the sensors up through the machines, up through you know the station controllers, the manufacturing execution systems and even the ERP. Horizontally, we’re talking about integration from the upstream supply chain through the four walls of the fab and then downstream into the customer base.
And so, if you think about it, traditional scheduling and dispatch basically took an order from an ERP and then it would go down to the manufacturing execution system. It would figure out how to allocate the resources to deliver on that order, OK. And it would do the scheduling and dispatch. And then it would have rules like if this machine starts underperforming, then maybe switch over to this other machine. Or you know when there’s when the queue length gets greater than X at a particular machine that maybe we want to make a change there.
It’s largely rules based and it’s largely kind of looking at the factory as a set of rules within the four walls. OK. It doesn’t really look too much at how is this machine performing in terms of it’s run to run control or it’s fault detection or it doesn’t really look at what’s the upstream supply chain doing. Are we going to run low on this particular component, which if we overdrive this piece of equipment, suddenly we won’t have this equipment available? And it’s definitely not looking downstream at the customer base saying is there any push back from the customer like we’re not producing good parts?
So if you can think about smart manufacturing in the future, all of these things are going to be integrated horizontally and vertically, and that is driven by data and data interfaces. There’s a lot of issues that have to be addressed though.
First of all, the data itself, right? You have to get the data, you have to consolidate the data. Data that is normally used for things like fault detection down at the equipment level will not have to be consolidated with data that might be coming from the supply chain, because maybe you’re using that data together to solve a problem which might impact your scheduling.
