|
Describe data analytics. And what that might encompass?
I would say the easiest way to describe it (data analytics and data science as a whole) is solving problems and answering questions, using data. That's the most succinct way to describe it. Whether we're identifying trends, forecasting future performance, or diagnosing process issues, the goal is always to use data to answer questions and improve outcomes. Get data and try to answer those questions.
How do you see data analytics being integrated into manufacturing? And product development in general?
Well, we're already trying to do some of that now. Especially on the R&D side, how can we take mill data and develop models, forecasts or predictions? Okay, we take all this data, we run modeling through it. This is how this process works, these are our assumptions, and we can help prove those out using data. That's a good piece of it.
Anomaly detection, how do we show that there's some sort of problem in a machine or a process? Use data, figure out when the data is acting abnormal. Use alerting and say, hey, there's issues, things happening, we need somebody to investigate and fix these problems.
So, there's a few different routes and it's interesting and there's still more we want to do, you know, predictive maintenance, time to failure, process improvements, that type of stuff. Ultimately, we’re trying to improve reliability, product quality and efficiency for our customers.
Most modern mills already produce and collect a lot of data. What kind of challenges does that pose?
I would say it's two-part. There's once we get the data, there's a ton of it, yes. It's figuring out what's noise and what isn't. There's also, you know, papers been around for 100s of years and these plants, some of them are 70, 80 years old. So the data availability, actually being able to collect that data, store it, and then analyze it is also a challenge. There are all sorts of challenges in the mill, let alone logistics, accessibility, security and all this stuff.
Then every mill is different. Some of them have one pulper or two, some have multiple lines, some are missing pieces of machinery that we have for other ones. It's kind of just understanding each one separately and understanding the whole process and how they play a part in each other. That’s also where data doesn’t tell the whole story. Without the context of the mill, or without the help of the engineers we have here, we wouldn’t be able to understand what’s happening. That context is important, and without we don’t know what the data means.
What are some common mistakes made by organizations when bringing the mechanical into the digital/analytical domain?
Well, I will say going too slow and too cautious. Organizations may wait for the perfect data and 100% certainty, but that’s not always possible. Analytics should be approached as a learning exercise. You're going to make mistakes. It’s just going to happen.
There’s always going to be some of those issues where you're going to find a better way to do it. But that's part of the scientific process. Right? You have got to try it. Sometimes with analytics you may hit a dead end or find something you or the mill thought is not the case. That doesn’t make the endeavor a failure.
Part of analytics is the learning process and understanding what works and what doesn’t and trying again until you’ve solved the problem. So don’t be afraid of failure and the unknown.
How are manufacturers adapting and excepting the dramatic changes this digitization brings upon them? Early adopters, fast followers, or wait-and-see?
A bit more behind if you compare it to some of the larger players and more digitally enabled industries, but again, some of these things have been out and around for so long. 30, 40 years and they worked just fine. A lot of it is why break what's been working for us for 40 years? And then there's all the, there's a lot of data security questions too.
The biggest challenge is probably security. Yep, you have security getting mills on the same page of what data we need and how to make sure it's secure, but still accessible for us. That's been a challenge that we've tried to start through. That said, we’re seeing progress. Mills are open to change, provided you show the value clearly.
What do you see as the biggest benefit in applying data sciences to a manufacturing environment?
The biggest selling point is we have extremely talented, intelligent engineers that are able to see data and understand everything that's going on in the mill. They can help them run better and more efficiently. And that's something that without the data, we can certainly do, it's just much harder.
Someone must go to the mill, must see it at the right time, see it acting up, understand exactly what's going on. Maybe they need 30, 40 years of experience to be able to do that. Our goal is to enable our experts here to be able to help diagnose problems in mills hundreds of miles away without having to step foot outside of our building.
Of course, you need to see the machinery and understand it firsthand, no technology will be able to replace that. But once you understand the system and the process, and you’ve seen a mill firsthand. When you have the data right in front of you, you can make much better, more informed decisions. That’s our goal and where I think data science can be most useful.
| 
