Industrial autonomy goes beyond automation by adding layers of smart sensing and machine cognition to anticipate and respond to unforeseen circumstances, ultimately culminating in systems responsible for all aspects of operation, from startup, through safe shutdown. In this episode, Keith Larson and ARC Advisory Group's Mark Sen Gupta discuss the process industry's path to autonomous operations.
Transcript
Keith Larson: Already highly automated, the process industries are poised for the next advance, industrial autonomy. Autonomy goes beyond automation by adding layers of smart sensing and machine cognition to anticipate and respond to unforeseen circumstances, ultimately culminating in systems responsible for all aspects of operation, from startup, through safe shutdown.
Hello, this is Keith Larson, editor of Control magazine and controlglobal.com. Welcome to another episode of our Control Amplified podcast, brought to you today with the support of Honeywell, a global leader solving industry's toughest problems for remote and autonomous solutions. Joining me today to discuss the process industry's continuing path towards autonomous operations is Mark Sen Gupta, research director, ARC Advisory Group.
Welcome, Mark. And thanks so much for joining me today.
Mark Sen Gupta: Sure. Thanks for having me. Looking forward to this conversation.
Keith: Yeah. Just for our listeners out there, maybe you can tell us a little bit more about your background and what qualifies you to talk about automation and autonomy a little bit?
Mark: Oh, that's a great question right there. What does qualify me for this?
Keith: You were available.
Mark: Yeah. Sure. So, I graduated college, I've got a master's degree in electrical engineering, and I got sucked into process control. My degree is actually more toward robotics and machine control, but it was a hard time in the economy, and I got sucked into the chemical industry instead. So, I spent a couple of years on the plant site and then I moved to a supplier. I spent about 12 years with them in process control programming, installing systems, moving from DCS to SCADA. I did stuff upstream, downstream, offshore, and then these paper plants, I have gone into chemical plants. I've done specialty chem batch systems. I worked for Fieldbus when that came out, back in the late '90s. I've been all over the automation, industrial automation space, especially in process control.
For ARC, I actually cover DCS safety systems and SCADA systems. And so, yeah, this is actually fascinating to me because I think most people my age, and your age, our age, we grew up with "Star Trek" and all kinds of fancy automated, futuristic stuff and now we're looking at what's going on in the consumer world with autonomy. And it's really getting exciting for us, older automation guys, and now we're talking about what about our chemical plants? Or what about our processing plants? Yeah, it's a great topic.
Keith: But I think your grounding like you have in robotics and some interest there, it's kind of interesting, because you see, it seems like in order to make plants more autonomous, there's going to be some robotic aspects to that operations. I always picture the classic R2-D2 in the back of a fighter and fixing stuff when they're on the flight, flying through the air in "Star Wars." And that's the kind of, I think, some kind of cognitive abilities that we'll need in our plants to really succeed in some sort of autonomy. Maybe that's an extreme case, but that's the kind of vision that I think we did experience in the movie theater if not anywhere ekse. So, from your perspective, then, is there something different about autonomy than industrial automation? Is it just a marketing blitz or is there a real qualitative difference in what we're talking about, from industrial automation to industrial autonomy?
Mark: Right. Well, when you first approached me with this, it's a really fascinating question because my first approach at one take, and then I took a second to think about it, so this really depends on how you frame it, and where you're coming from, right? Because I could argue, back in the day I was working in the plant, I put in a pneumatic PID controller for temperature, and nobody touched it after I put it in there. I could argue that was autonomous, but it's also industrial and it's pneumatic. It had absolutely no electronic, nothing.
Keith: Nothing digital about it at all.
Mark: Right. So, in my mind, autonomy is like, well, you could say industrial automation is the basis. You could say, industrial autonomy is a subset of industrial automation. Or you say it's a logical next step. I could go with any of those three approaches, but if you look at who's pushing industrial autonomy, who's leading the charge, it's industrial automation companies. And so, I would argue that probably the best definition is, it's a subset. Industrial automation is a major umbrella, and autonomy is just, you know, adding to the industrial automation.
Keith: Almost a matter of degrees. It's like, okay, a PID loop is autonomous, does its work without interruption or without needing other things. But as you get more complex and more of what we would call machine intelligence required, that's more what they're talking about with autonomy, maybe. Does that make sense?
Mark: Right. You're trying to get...what you're really getting the autonomy from is a person at the board, right? You know, I used to work... In the summer, I had a job at a zipper company. And we would make machines to automatically feed the zipper onto the tape, which is the stuff that has the teeth on it, in case you don't know. But anyway, you would dump a bunch of zipper sliders, which is this thing that goes up and down, and you put a bunch in the hopper and turn the machine on, and the cams and the little microswitches would all go and that thing would run until it ran out. That was autonomous.
Keith: Yeah. Sure.
Mark: I mean, it's not that autonomy is new, it's the computing power. It's just like digital transformation. It's not that any one of those technologies that enable digital transformation is new, it's that we get them all at one time, and they come together and they intersect. Industrial autonomy is kind of the intersection of high-performance computing and AI, etc., coming together and melding with what we would call the industrial automation sensing and control.
Keith: Yeah. But I think if you look at the extremes of continuous process, automation refineries, other very highly automated facilities, you know, there's not that many workers there as it is. There is the board operator and there's maintenance techs that are running around, doing other things. But what are driving forces that are having industrial automation companies and maybe major process manufacturers really look at autonomy as being necessary to think about in terms of what's next for them?
Mark: Right. Well, this is a key question because what we've noticed at ARC is there's been a maturing in the way industrial automation is applied. So, you know, in the early days, people just got excited about technology. And in some cases, you'd find people applying technology because they could. Now, we're in a phase where it's like, okay, what problem does that solve for me? Okay? And so the factors that are driving companies to look at autonomy, it's key. Because I think, again, like we said just a moment ago, we're looking at automating the person, right? And so it's the...
Keith: The judgment as well as physical actions, right?
Mark: Right. And, we keep falling back a lot on this stat where, like, 40%-some odd of abnormal situations are caused by human error. So, we're looking at the reliability of a person as being a possible contributor to autonomy. But the other thing that we've also been talking about, of course, is the graying of the workforce.
Keith: Yeah. I thought that was your 40% when you first started saying it was going to be 40% retiring. But, yeah. Yeah.
Mark: But right. I mean, we've got folks retiring out, and we have millennials saying I don't want to work there, it's filthy, dirty, I want to work for Google. I want to work somewhere where I get a pool table in the office, or I can go play video games, or whatever it might be. But there's a lifestyle shift in mindset for replacing those people. And then you've got the training, the heavy training that gets involved. Because people are in short supply, it may be more difficult to keep them. But then there's also the aspect where we talk about, if a person creates an abnormal situation inadvertently, that's a safety issue. So, the fewer people I have on-site reduces my risk.
Keith: Sure.
Mark: The other thing that it would do is it would reduce my environmental impact or could if I had a spill or some sort of release. One of the other things that autonomy does, or industrial automation, in general, is it also allows you to do the same thing over and over again. And if you want to make it better, you make it better once, right? And so, with personnel, you'll talk to people, and they say, “Oh, Shift A, man, they really got it down, Shift B, man, everything goes to pot,” and that's an extreme. But you know what I'm saying? It runs better when shift A is on the board than when shift B is on the board, shift C is okay. When you go to more and more automation, and this includes autonomy, you have the ability to tune that control. And it's tuned 24/7, as long as you can keep the maintenance on the instrumentation and keep everything running from an automation standpoint, but those are the drivers. And one of the things that you see, again, we talked about automation and repeatability, but that's also a quality issue. And so, if I can maintain a certain level of quality consistently, I can also increase my price for that product, depending on the market. Obviously, gasoline, you get 97 octane, you get 97 octane, but some of the other chemicals, you can get a premium for a better quality product. Now, one thing that we haven't touched on is regulation. And that's going to be very interesting, because you know, I remember, I don't know if it's still true, but I remember talking to one fella, he said that Germany actually had a regulation that said you had to have so many people per square foot, okay? And if you've got a regulation of that nature, you're not going to be able to get rid of that person with autonomy, but again, you still get the other benefits.
Keith: Put them in the more secure part of the plant at least so he's not as vulnerable to a spill or another incident, that sort of thing.
Mark: Right. I would expect the regulation doesn't mean that guy has to be in that square footage.
Keith: Right. It's the headcount. That would make sense. That would make sense.
Mark: Yeah.
Keith: Yeah. But it does seem like in talking autonomy, the use of drones and robots and those kinds of things that are typically things you think of in discrete manufacturing versus industrial process automation, what role do you see them playing in a more autonomous plant, or for maintenance tasks, or inspection tasks, or things like that?
Mark: Well, it's interesting, because they are beginning to be used, you know, like the...was it Boston Dynamics got that Spot robot?
Keith: Yeah.
Mark: That it looks like dog, I guess. But you could put all kinds of sensors on that, and you can put an articulating hand. So, you could envision where you put acoustic sensor on that robot, he goes to the plant, and just like how you had that old maintenance hand, Bob, he'd walk around and say, "You know that doesn't sound right." But you could use analytics on the acoustics and you could pick up, “hey, I think that bearing is showing where, because it's vibrating different and I'm picking it up on this microphone.” You could also look for leaks. If you needed to go out and crank a manual valve closed, are you going to be able to do that? You know, and may also need to say, “hey, I need another robot over here, because I can't get this thing closed because it's stuck.”
Keith: I need a spare hand, like, yes, I don't know what you'd call them, but a spare actuator. Yeah.
Mark: Right. But, you know, they're already experimenting with that in several cases where they send drones out to inspect automatically, that those pictures come back, and they get analyzed by the software. Yeah, and they do geofencing to make sure those things don't leave the site or...you know.
Keith: Yeah. But then it also seems, you mentioned the quality thing, but, like, consistency of operation, too, can be very important for the other objectives, like sustainability and efficiency, and those stuff too because if you're doing things more reliably, more predictably, that takes into things like procedural automation for doing startups and shutdowns that you may only do once every seven, eight years nowadays, so the last people who did that shut down are no longer with the company. So, if you can do that more programmatically, through these more autonomous systems, to do those less often performed procedures, that would seem to me a good aspect of, we're calling it, autonomy or more autonomy.
Mark: Correct. Right. Right. Yeah, I think procedural automation is undersold. I think there's a lot of value in capturing and using that automated procedure because it gives you a step. Because if you're going to go to autonomy, you're going to need that step anyway, you've got to know what they're doing. But then, it also helps you with the quality, and I know that when I step through this procedure, I'm going to shut down the safest way possible, all right? And I didn't wait too long on something and ruin my batch, right? And I think, you know, batch processes tend to be a little more ahead in that area because that's how they make their money is automating. Yeah. And that's why I think a lot of the pharmaceutical, you see a lot of this automation in pharmaceutical. They have a double whammy because you don't want to mix it wrong. But in some cases, you do not want to contaminate it, so you want to keep people out. It's kind of like the semiconductor industry, you keep people out of the cleanrooms.
Keith: Yeah. And the more reproducible one batch is to the next, the higher quality and all those same things. It's not quite the same when we're producing gasoline as it is when you're producing something that can be injected in somebody's body, yeah.
Mark: Right. The car doesn't care.
Keith: Yeah, it handle a few octane swing here, too, no doubt about that. So, what do you think is next over the next several years? Where will this autonomy movement be focused do you think, in terms of priority, in terms of new products, new solutions, as well as implementations on the process industry side of things? Any particular focus that you predict?
Mark: That's a really interesting question because it comes back to that needs-based thing. Obviously, we haven't mentioned it yet, but COVID was a big driver in pushing industries in ways they were really dragging their feet. A lot of the technologies and approaches that many suppliers and many analysts firms and press companies have been writing about suddenly became high priority last year.
Keith: Guilty as charged. I wrote some of those articles myself.
Mark: Yeah. I mean, now that we have 20/20 vision, haha, we can look back and say, “oh, shoot, I should have invested in that networking, that would have allowed me to implement a lot of these remote solutions.” We saw a lot of suppliers step up and say, "Hey, I know you've got a trial on this remote solution that we sold you, you're now in need, let me just give you full access with no limitations on the license so you can get through this." And suddenly, a lot of the remote access technologies became high priority for almost everybody. And the nice thing is, now that they have a year under their belt with that technology, they've seen the benefit.
Keith: Sure.
Mark: You know, it's not just us talking about it, but they've actually got boots on the ground using it, and say, hey. But the other thing that we found that is interesting is companies said, “Hey, I've got this technology in place, but I didn't realize I could solve this other problem I had.” And so you're seeing a lot of innovation at site with technologies that we didn't expect. And then the other thing that it's kind of strange, so, if you look at like subsea, offshore and mining, there's a good deal of autonomy implemented in those already, right, especially subsea, where they were doing, and it's very small pockets, but there was an approach of doing undersea or subsea processing. So, you're actually separating the water, sand, gases, whatever, and you're processing it under the sea and sending it up. Whereas, before you had to bring it to the surface, and you have people on the surface, you don't have people under the sea. That's totally autonomous.
Keith: So, that is autonomous. Yeah.
Mark: You know, there were some offshore rigs that were becoming autonomous. I think that's getting accelerated. Again, we talked about pharma and some of the needs that they had. But again, a lot of the autonomy, they're still going to be dragging, because there's going to be this cost versus benefit analysis that has to be done. And if the industry doesn't do a risk analysis as well, it's going to be even slower. But I would expect some of the offshore land-based sites to adopt more autonomy. So, yeah. I'm thinking of somewhere like Fort McMurray where in the...or Prudhoe Bay in the winter, you have four minutes outside, and if you spend more than four minutes, you're dead, type of thing. You know, some of those really harsh areas are going to be interesting and probably first adopters. But you're also going to see places where, you know, I just can't hire somebody to go to the site anymore. So, you see that too.
Keith: Yeah. So, you'd have more maybe unmanned operations for periods of time and you have enough capable, enough predictive diagnostics to monitor the machinery so that you have time to fly somebody up there if you really have to, in order to do a turnaround on some maintenance on some equipment but routinely have it operating for days or weeks at a time with nobody there, I suppose would be a fair transition to more full autonomy where you might have a self-healing plant that could repair itself. That's going to be a little ways further off. That's a little ways further off.
Mark: Right. But what I'm expecting is that autonomous operations will follow remote operations.
Keith: Yeah. That makes sense.
Mark: So, anywhere that they've pulled back people off-site, those are going to be the sites that are going to lead in autonomy.
Keith: Sure. That makes sense. And some of the mining operations are very sophisticated, as you were mentioning as well, in terms of autonomous vehicles and autonomous mining machines underground to keep people off dangerous places.
Mark: Right. Right.
Keith: Makes a lot of sense. All right.
We've come to the end of our allotted time here. I just want to say thanks so much, Mark, for taking the time to share your insights with us today. And for those of you listening, thanks for tuning in. Thanks also to Honeywell for sponsoring this episode.
My name is Keith Larson, you've been listening to a Control Amplified podcast. Thank you again, Mark Sen Gupta, with ARC Advisory Group for joining us. We really appreciate your sharing your perspective.
Mark: Thank you. It's fun.
Keith: Great. And you listening, if you've enjoyed the episode, you can subscribe at the iTunes Store or Google Podcasts. Plus, you can find the full archive of past episodes at controlglobal.com. Signing off until next time. Thanks again, Mark.
For more, tune into Control Amplified: The Process Automation Podcast.
