What we talked about
John Hooker: In this episode, John Hooker, Director of Engineering at Statistics & Control, shares how his lifelong fascination with gas turbines led him from aerospace engineering at Iowa State into turbomachinery control and, eventually, advanced process control and large-scale simulation. He explains how controlling a turbine in an airplane isn’t that different from controlling one on the ground:an insight that helped him land in oil & gas consulting at a small company that grew into a global team with offices across multiple countries.
Show notes
John Hooker joined a six-person company that had never said no to a problem, and two decades later that philosophy turned a turbomachinery control shop into a global simulation firm operating in eight countries. The through-line: every client project revealed some physical reality the model hadn’t accounted for yet, from Siberian temperatures dropping to -54°C to a 30-year-old leaking steam valve that couldn’t be replaced.
What we covered
- John’s entry into oil and gas consulting happened because a gas turbine is the same whether it’s on an airplane or on the ground. His aerospace background at Iowa State translated directly, and the company that hired him, the sixth person they brought on, offered a job within two weeks of the first interview.
- The company started in turbomachinery control working with Honeywell, Allen Bradley, and Siemens PLCs, then kept expanding into simulation because clients liked their work and kept asking for more: first gas turbines, then pipelines, then electrical systems, then steam, then oil wells. The boss, John says, has never said no to a problem.
- A geothermal steam project in the mountains caused two weeks of head-against-the-desk frustration because the simulation was consistently five megawatts off. The culprit turned out to be relative humidity, which nobody had factored in for a turbine sitting at 2,500–3,000 meters altitude. Two weeks of work resolved by one variable.
- The simulation hit a boundary condition of -50°C in Siberia, and the ambient temperature outside was -54°C. The client called to report the simulation had stopped. John had to expand the temperature range on the fly, a reminder that you cannot anticipate every physical parameter before you deploy.
- A predictive maintenance project on remote reciprocating compressors cut site visits from 18–20 per month down to 2–3 per month by putting a computer on site to relay data to the cloud. The shift wasn’t about knowing what was wrong in real time, it was about knowing whether something could wait a week or needed someone to drive out immediately.
- On AI in engineering, John tests every new large language model by asking it to write a PID controller in C++. They always start well, then fail when he asks how the model handles dividing by zero in the time domain. It takes four or five iterations to get a correct answer, and he says his bigger concern is not the model itself, but people who don’t know enough to catch when it’s wrong.
About John
John Hooker is the Director of Engineering at Statistics & Control, a global engineering and simulation firm with offices in eight countries. He specializes in advanced process control, turbomachinery control, and predictive maintenance across oil and gas and power generation industries.
- LinkedIn: https://www.linkedin.com/in/johndhooker
- Website: http://www.stctrl.com
Episode 91 of the PreVetted Podcast.
Full transcript
Federico Ramallo (00:01) Welcome back to the pre-vetted podcast where we spotlight extraordinary people and remarkable talent reshaping our world. Today we are joined by John Hooker. He’s the director of engineering and statistics and control. He has spent more than a decade leading global engineering work across oil and gas and power generation with deep expertise in advanced process control and turbo machinery control. He’s helped teams deliver real world control solutions.
from controller upgrades and migration to monitoring systems, testing, and predictive maintenance, often in complex brownfield environments where safety, reliability, and stakeholder trust are non-negotiable. John, welcome to the show.
John Hooker (00:45) Thank you, that was a nice introduction. Far too kind.
Federico Ramallo (00:47) Thank you, thank
you. Yeah, I mean I’m truly honored to have you here today. So how did you get started in control engineering?
John Hooker (00:54) Yeah, that’s a great question. my background, I went to Iowa State University, as you can see in there, and it was mostly focusing on, I’m really fascinated with gas turbines and have been for most of my life. And one of the things that I really focused on was gas turbines and the control of them, as well as being able to aerodynamics was another big portion of mine. So those are kind of the two areas that I focused on when
I was at university. And when I graduated, I was looking for a job and I didn’t really know what specifically I wanted to do. I guess I saw myself going to like places like GE or Pratt Whitney, know, turbine manufacturers, Rolls Royce, places like that. I was hoping to, you know, get to one of the big companies, but I actually had a small company here in West Des Moines, Iowa that contacted me.
And the hiring process was so smooth, so great that they were the first ones that it was pretty much, it was, think, two weeks between the first interview and a job offer. And so it turns out that when you control a gas turbine, or a turbine is the same on an airplane as it is on the ground. And so that’s kind of how I landed in the oil and gas consulting business. it was pretty awesome for our small company. I was actually the sixth person hired with the company. And now we’ve, yeah.
Federico Ramallo (02:07) wow.
John Hooker (02:08) I know, and now we’ve grown into a pretty global company. We’re small, but we’re mighty, as I always like to tell people. And we have about 84 people around the world, I think, and we have offices in eight different countries. So it just depends on where the work is normally. And that’s where we tend to either spin up an office or we tend to go to for that. And so I’ve implemented lots of projects all around. And basically my company started off in the turbo machinery control business. So we were doing PLC control. So working with
Honeywell working with Alan Bradley working with Siemens and doing gas turbine control surge control capacity control Unit control just different things like that steam turbine control So we were doing those applications and one of the things that our company was wanting to do was we were wanting to pivot more towards advanced process control and simulation and
with my background in aerospace engineering, we also had a lot of software engineering or a lot of computer programming in school. It was actually one of the few things that kind of took me by surprise when I went to school because I was just like I didn’t realize that I would do so much computer programming. And so that helped with being here because we’re mainly a software development company. We don’t really deal too much with hardware. We just care about the software inside of it. And we also develop our own software. And basically we did we had a
gas process that was going on and they wanted a training simulator for it. So we had a gas turbine simulator and we had a compressor simulation that we did and they liked that so much that they asked, hey, can you expand it? Can you go to pipelines? And so then we expanded our simulation to that and added that into there. And then
Another thing that we were doing, since we were doing gas turbine control, it kind of got into the power control business. And so we were starting to do power control, generator control, frequency control, voltage control, different things like that. And then we built an electrical simulation for that. And so, you know, kind of as we keep going along with projects, we keep progressing and progressing. And so we built, you know, gas simulation, then we built an electrical simulation. And then after that,
It started to expand even more where they asked for a gas process gathering system. You know, can we simulate the whole network and simulate the compressors in there? And so that project went very successfully. And so then it kind of moved to steam and we did a great job simulating steam. And then of course they started asking, okay, well, can you simulate oil wells? And so it’s just one of those that throughout the years we just kind of progress more and more and keep adding functionality and features to our software that
just helps out. So normally a lot of times people like to stick to you know one area of the oil and gas industry whereas we’re just all over the place. My boss can never say no to a project or say no to a problem to solve. So that’s one of the fun things about being with the company that I’m at right now is that instead of being kind of pigeonholed into one area or one thing you you know I get to experience a lot of different projects and a lot of different processes that
keeps me learning, keeps me on my toes, and it makes for really interesting time.
Federico Ramallo (05:11) Wow, that’s amazing. So many simulators. ⁓
John Hooker (05:13) Yeah, it’s pretty cool.
Federico Ramallo (05:15) I’m sure that the complexity of the simulation, when you get started and you get into it, the complexity becomes much bigger than you originally anticipated, right?
John Hooker (05:24) Yeah, absolutely. Well, and it’s people keep adding stuff on there because like you mentioned in the intro, for the simulation, one of the things that you can do with a lot of simulations is you can do things like predictive maintenance. So you can throw in algorithms for how well the turbine’s running, how well the compressor’s running, how well your pumps are running, different things like that, as well as you can start throwing in algorithm, optimization algorithms on top of that. So one customer that we had,
wanted to do okay, let’s push as much gas through a pipeline as possible because at that point in time natural gas prices were high and so then you know next thing they come along and they’re like hey that was a great algorithm for or that was a great objective function for when we were When natural gas prices were high well now that they dip down a little bit We want to run as efficiently as possible and not break the bank trying to push
so much through. So, you know, what are the economic conditions of that? So it’s just kind of adding that optimization, adding in some of the economic criteria to things. And as I said, people, it’s always fun where one of the things that I’ve done is we’ve built so many wonderful solutions for customers. And I keep hoping that I can reuse that same solution for another customer. And I’m like, it’ll be easy. We’ll just implement it right here. And we’ve already done that.
And every single project that I’ve had has either had some slight small twist or some slight deviancy from the past thing that we’ve done to where we just keep, again, keep adding more functionality and keep adding different objective functions or different things on top of that. So it makes for an interesting line.
Federico Ramallo (06:56) Right, right, because it’s an interesting challenge that then you can, you you have to figure out how to solve it, right?
John Hooker (07:03) Yeah, absolutely. That’s, being an engineer, one of the joys is having a customer come to you and say, we have this problem. How would you go about solving it? And of course, since we’re simulation and software company, we’ll say, well, we solve it with simulation, obviously.
And pretty much it’s being able to design a solution, it integrated into their workflow, and have them use that and say, well, this is pretty handy. We especially enjoy it, or we save a lot of money with it. Then we appreciate it. And it feels really good to, again, see your solution or see your design in action.
I’m sure as a software developer you probably implemented something and you’re like, wow, it actually worked correctly and how it was supposed to. Maybe after several iterations, but it’s just one of those that it’s really handy when it works really well.
Federico Ramallo (07:49) Right, right. You go through a process where you question your life decision until everything works, right?
John Hooker (07:57) Yeah, pretty much. Yeah,
that is definitely true. I remember one time we had a project. It was for a geothermal steam project. so the whole idea is you drill a well, you have steam come up, you basically get free energy or free power to a steam turbine. And so you can sell that power off, but they were losing too much heat out of the reservoir. And they were wanting to, essentially the optimization was
We lose less heat, but we still produce the same power, or we try not to lose too much heat out of the reservoir and maintain the reservoir lifespan, or elongate the reservoir lifespan. And we were trying to solve a problem where we were always producing five megawatts more than what was happening in real life. And so it was one of those where I was slamming my head against the desk for two weeks. And then it was one of the places where we never took in relative humidity into account.
for the steam turbine and they were up in the mountains so they were about 2500-3000 meters up so the steam turbine was running a lot more efficiently than we thought it was and so that that literally took two weeks of just adding in the correct humidity into the equation to make it correct.
Yeah, it’s fun too. I was questioning a lot of life decisions those couple of weeks that we were trying to deliver that project and have everything work correctly.
Federico Ramallo (09:15) I’m
Right, right. I understand that the complexity of steam is when you have an over, what is the word, over saturated steam or non superheated steam, right. Yeah, yeah, yeah. I know the terms in Spanish, so I have to kind of translate it in my mind, so.
John Hooker (09:28) Yeah, superheated steam, yeah.
That’s alright. As I say, our company was founded by four people who were not original English speakers. I’m used to… Yeah, I’m one of the few people that is normally a native English speaker in our company. We tend to hire a lot of people. Well, again, we have about 30 people in our office, but most of them tend to be from around the world. I’m used to it.
Federico Ramallo (09:58) Right, right.
There is a lot of talent globally that is really good. ⁓ So it’s great that you give them opportunities. Yeah. ⁓
John Hooker (10:03) Yeah.
And it just
leads to a different perspective and different, know, especially when you work in a couple different countries, you realize that not everybody has the…
You know, the same either work style or work ethic or work, you know, like in the US we just work 24-7 constantly and everything just needs to be done immediately. Whereas you go to some of those nice countries, like in the Caribbean islands, it’s just like, well, we’ll get there eventually. You know, it doesn’t have to be done now, but just very, very island time-esque. And so it’s always, it’s always a fun adjustment for something like that.
Federico Ramallo (10:39) Right, right. Yeah, I don’t know why, but I have a theory, right, that on hot places, the culture is more relaxed on results. My theory is that that happens because you have abundance, right? You used to have abundance of fruits and whatever. So, you know, there’s no rush, right? But if you go to cold places, it’s either, you know, it’s survival, right? So.
John Hooker (10:47) Yeah.
Yeah.
Yeah, absolutely. That’s definitely true. And for some reason, it seems that oil is only found in really cold places and only really hot places.
Federico Ramallo (11:02) You have to up your game.
John Hooker (11:14) I keep trying to get my boss to send me to nice places around the world, but it’s always just the middle of the jungle in Indonesia or the middle of the desert in the Middle East or up in the North Sea and then the nice frozen cold or in Siberia where it’s really, really cold. Yeah, it’s a…
Federico Ramallo (11:31) no!
John Hooker (11:33) Yeah, that’s probably the coldest I’ve ever. And again, Iowa is pretty cold. But as I said, I remember we were doing a simulation and it was in Siberia. And I think one of the boundary conditions of our simulation was at minus 50 degrees Celsius. And they’re like, hey, your simulation stopped. It says it’s minus 50 degrees Celsius for the ambient temperature. And I was like, oh, what’s the ambient temperature outside? And they’re like, it’s minus 54. And I was like, we need to expand that, I guess. it’s things you run into.
Federico Ramallo (12:03) Right, right. Because you never know all the parameters that you’re going to run into,
John Hooker (12:07) It’s true, battery conditions are important.
Federico Ramallo (12:10) Right, right. Have you seen the, and this is off topic, a little bit off topic, but have you seen the tank simulator that was built in the 60s that it’s all mechanical? It’s basically a little camera that goes through a scale version of a town and basically you drive the tank with, you know, it’s all mechanical, know, zero electronics.
John Hooker (12:31) Interesting. No,
I haven’t seen that but that sounds interesting
Federico Ramallo (12:36) I’ll send you the link where they talk about that, but basically it’s fiber optics and mechanical gears that moves the camera simulating the tank. So it’s very interesting how much we were able to evolve from that to what you guys are doing today.
John Hooker (12:39) cool.
yeah, absolutely. Well, and one of the great things about, especially with our company is we don’t get embedded into technology that much where we always, customers always want the latest and greatest technology. And so they’re always asking for that. And that’s one of the nice things with working with a small company is that we can pivot quickly to something new and something fresh or something that a lot of people are doing. So it’s makes for,
Sometimes it’s a nice thing, but other times it’s also a pain because it’s just like, I just finally polished off this wonderful, a great example is just the way that we allow access to our software. You know, we started off with just our own, you know, kind of security where it was just a name and a simple token. And now everybody, you know, it kind of evolved into using Windows domain and Windows security. And then that kind of domain or that kind of
expanded now that we have cloud and we have a different thing. Well, just a great example of it used to be Azure Active Directory and now it’s Azure Intra or something like that. So again, they change names and change how they do stuff. So it’s things like that that we update quite a lot. it’s fun and kind of somewhat exciting, but also somewhat frustrating too. So I always ask Microsoft why they do the things that they do.
Federico Ramallo (14:09) I remember Kerberos at one point, they started building Kerberos. They moved to Azure. I used to work at Microsoft, so I was very familiar with that. Then I did a 180 and I went to Linux Ubuntu and Ruby on Rails, a whole different technology set. It took me like two weeks to be able to read one email.
John Hooker (14:17) Arco.
Yeah.
Federico Ramallo (14:31) But it was a good experience. And yeah, I’ve seen that that happened. Microsoft make changes on the software for no particular reason other than just selling a new software license, I guess. ⁓ And then everybody else had implemented on top of that technology they need to adapt quickly, right?
John Hooker (14:44) Yeah.
yeah, absolutely. Well, and it’s just, you know, you finally get everything working and then suddenly IT is just like, no, we’re going to upgrade servers. So let’s go ahead and upgrade servers. And it’s like, well, we just got everything working. So why are we doing this? But it’s, as I said, it’s part of the game.
Federico Ramallo (15:05) Right, right. they sunset Windows XP and then Windows, I remember the other one, Windows 10, I think, is the other one. But there are lot of systems that depend on that that have not upgraded, right? Because on the industry, you don’t need to update to the latest computer all the time, right? What you need is reliability, right?
John Hooker (15:19) yeah.
Yeah, that’s definitely true. Actually, I remember we were trying to do a report for a remote compressor station and this one was predictive maintenance of when it would shut down because they didn’t have a guy manning the compressor station. They basically just had, they would send a guy out there if they ever had any trouble or anything like that. And the computer that they had there was a Windows XP and I was trying to get our software and at this point in time it was already like 2000.
2013 or 2014, mean Windows, obviously 8.1 I think had already been out for quite a while and I think they were, I don’t remember when Windows 10 came out, but it was just one of those where they wanted it to work on Windows XP and we were struggling a lot to get all of our functionality working on a 32-bit, and that was another thing is it was Windows XP 32-bit. so it was, I think that was one of the projects where our development staff just gave up and said,
we’re not going to do this. it was one of those where we kind of stopped doing legacy programs after that project. But I get it, they had a nice reporting system that was working and had been working for the last, you know, probably 10 or 12 years and they didn’t want to abandon it. And I get that. But it’s one of those where, you know, at some point in time it’s going to force you to update either that or the computer dies and breaks down.
Federico Ramallo (16:41) Right.
John Hooker (16:50) at some point, now one of things I like is especially with simulation, a lot of people are going to virtual machines. So it’s a lot nicer to update, move, migrate and things like that because they’re just like, yeah, we have a backup copy online and we can just restore it with the latest version of whatever and everything’s already installed or everything’s there that you need to. And so it makes it a lot easier than what we used to do where it’d be like VPN into
company’s internet and then VPN into a remote computer or something like that or a different business unit to access a remote computer. So we’d have to go through a couple hops and a couple of different ways to communicate. And now it’s just virtualization and you guys just log in either through Azure or through AWS or something like that and everything works wonderfully.
Federico Ramallo (17:34) Right, right. mean, now you can have a much better stable operating system for that. Yeah. Yeah. I haven’t figured out why people haven’t migrated to Linux solutions for the industry. Because, you know, in my mind, it’s, you know, less resources, you know, much more long-term, you know, support. But I understand that, you know, Windows…
John Hooker (17:41) Yeah.
Federico Ramallo (17:58) Microsoft has a big footprint in the industry as well, right?
John Hooker (18:01) Yeah, it really is hard with especially we use a lot of Microsoft Communication Foundation and a lot of stuff is written in .NET and communicates in .NET. So it’s really hard for an industrial process to get away from that. to tell you the truth, know especially in terms of like control systems, I think most of the control stuff, so you start looking at your Siemens, S7s and your
Honeywell Experian type stuff those servers are all still in Windows. I don’t even know if they offer Linux I don’t think I’ve ever seen a Linux computer running a lot of the main control software, so I agree I really like Linux because it’s one of those where you can kind of put it on and then not have to worry about anything for as many years as you want to you don’t have to worry about either forced migration or forced updates or Anything like that and you can set it up to where you can tell it
what you want it to do. it’s also another problem too is if you lose that knowledge or somebody sets it up in not necessarily weird manner and you don’t document it then it kind of gets lost to everything.
Federico Ramallo (19:02) Right, right. Yes, the that ⁓ see on Linux is that you have a fragmented industry where you don’t have one leader that you can follow. And this is a way to build on top of Linux, right? ⁓ And everybody does it differently. And yeah.
John Hooker (19:13) Mm-hmm.
Yeah.
It’s true, but I think more people are moving to it, with, again, you forced people away from, just especially, you know, Microsoft servers are getting really expensive and also the, you know, the, especially some of the virtual machines that they’re running on, like I know VMware just got bought out by Broadcom. So again, some of those solutions are also really expensive. Now too, to where it’s, it wouldn’t surprise me if more people try and start it off with,
not being tied especially to Microsoft or trying to go with Linux a little bit better and have a little bit better distribution in it. But it’s one of those you kind of have to build up your own personal knowledge in-house first, kind of deploy it, see how well it works, and then see if you can standardize it in some point in time. But if one guy who keeps the standards and knows everything decides to leave, then the company’s kind of out of luck for things like that.
Federico Ramallo (20:13) Right, right, yes, yes. I can see that happening. And the risk for that doesn’t justify going over Linux.
John Hooker (20:22) Yeah, pretty much because you get people who have Microsoft certifications and then you’re just like, OK, they kind of know what to do. And then you should be good to go. Whereas I could be mistaken in this, but there’s not too much for Linux in terms of standard certifications or standard use ones. I don’t know, people in the podcast listeners can always correct me if I’m wrong. I’m sure there’s lots of people out there who would tell me that I’m wrong.
Federico Ramallo (20:44) Right, I’ve seen some certifications, but I don’t know how complete they are. yeah, and as I mentioned, if you don’t have like a leader that has the power to, know, so people will follow it, then it’s hard to, you know, even if you have a great team doing certifications, it’s hard for people to follow it, right?
John Hooker (20:54) Yeah.
Yeah, I agree. And also the Microsoft salespeople are pretty good too. So that’s also helpful as well. I don’t think there’s too many Linux salespeople.
Federico Ramallo (21:13) Right, right. Yeah, I was working on the pre-sales team and I would do demos and I was almost doing the consulting part until the consulting team was like, no, no, that’s our job, Right. Yeah. So I would collaborate with them. But then eventually they will say, you know, now this is consulting. So now we take over. Right. ⁓ So which was frustrating, but I understand. Right.
John Hooker (21:24) gotcha.
Okay.
As
a consultant, you don’t want to get into the consulting business. As I said, it’s really fun to have people pay you to argue with you about your solution and advice that they’re paying you for. It tends to be how it goes. That’s ⁓ one of the fun things.
Federico Ramallo (21:50) Right.
Well, I was a 16 year old brat, you they, you know, I didn’t have a lot of, you know, gravitas to to say I want to go into consulting. Right. So on presale, I was not I could not get involved in production, which was safe. But also, you know, it was a frustrating part because everything I was doing was on prototypes. And I got tired of that. That’s why I left that position. Right.
John Hooker (21:54) No.
Yeah, when you just get it to prototype and it seems to stop constantly, it gets to be frustrating not seeing it actually be put into either production or just keep reiterating a different prototype and a different prototype again and again and again.
Federico Ramallo (22:28) Right, right, I I spent four, three, four years there, but after building prototypes and prototypes, I had that frustration of I want to build something that people are going to use, Yeah, and consulting would not take me because I was too young, so I decided to leave, yeah.
John Hooker (22:39) Yeah.
Yeah, that’s understandable. That’s one of the things I like especially about working for a small company is like I said, we pivot quite a lot. being able to, know, again, we have one customer who wants this particular solution. We have another customer who wants another particular solution and being able to try and implement that. Normally it is a very iterative process and a somewhat frustrating process, but it can be one of those that, as I said, it’s really nice to see the solution.
Federico Ramallo (22:52) Right.
John Hooker (23:12) that does that but a lot of times the customers don’t even know what they want either so it’s just kind of throwing stuff at the wall and see what sticks. I remember I built a really wonderful it was for a steam system it was for a steam injection in a big steam flood field and they had 264 steam generators essentially in this field and we built a beautiful optimization system to minimize their fuel gas because they were spending
more than like a billion dollars on natural gas, because they had 260 some odd steam generators. And we implemented putting out the best steam quality, putting out fuel flow, adjusting the water pumps and the flow to the water pumps in there. And we implement this nice, beautiful solution. the guy was just like, it’s a great solution, but the water pumps that we have are only single speed pumps, so you can’t change.
the speed or we can’t deliver more or less water, can only just send whatever the pump is running at. And so I was like, well, that would have been good to know, you know, at the beginning, but okay, we can adjust. And then we found out on another couple of steam stations that they were just like, yeah, by the way, we actually have two different fuel gases coming in. So one’s an okay gas and another one’s a sour gas. So mixing those, doesn’t put the correct output in there. And I was like, so.
Federico Ramallo (24:11) You
John Hooker (24:30) we have two fuel lines coming in and mixing and that makes it work a little bit differently. it’s one of those I’m like, all right, well, we can reiterate that. having that kind of initial design and what they want out of it is an important aspect. And they just said that they wanted us to optimize it and that was about it. So we did and it eventually worked out for them, but it didn’t quite go as much as they wanted to. So that was a fun project.
Federico Ramallo (24:52) Interesting, Yeah, I mean, eventually you find the physical limits of what, you know, which is something that you need to incorporate into your simulation, right? When I build applications, my bug, the risk is a bug that would not allow the user to do whatever they need to do. But if you miss the…
John Hooker (24:57) Yeah.
Yeah.
Federico Ramallo (25:18) the simulation or the control, more particularly the control, it goes boom, right?
John Hooker (25:22) Yeah,
it can definitely have bad effects on that. Well, and that’s one of the nice things too, and kind of one of the reasons why we started getting into more of the simulation and then also still work in the control business, the PLC control business, is because a lot of times, a great example, were doing a water station, and this was a water injection field for oil, and they had a big plant and they had four really big water pumping.
pumps to push water out to the field to inject it into the reservoir. And they were shutting down all the time. And the main reason why is their control for their water plant was they would either hit the low on the suction side so they’d have to cut it off because they weren’t getting enough suction because they were trying to grab too much water. And then one of the things is they were trying to push out as much water as possible.
And so that was, they were getting a lot of high discharge too on the other side. So it kept shutting down and it just kept doing a back and forth, back and forth type thing as we were trying to implement an optimization on it. And at one point in time, I told the plant manager, I was like, you can’t really optimize it if you can’t even control it very well. And so then we were able to do the actual control for the pumps and put in a much better pumping aspect or pumping algorithm for them.
And I think that was with an Allen Bradley PLC. And once we had that, then we could finally implement the optimization. then they could save money on the energy for the pumps. So like I said, that was kind one of those one, two punches of, well, if you can’t control it, you can’t really optimize it. And so you need to make sure your control works really well. And then we can start working on the optimization. So that was kind of a nice little introspective of you got to be able to control it and then drive it to where you want it to go.
Federico Ramallo (26:59) Right, right, because optimization is a higher function of control, basically.
John Hooker (27:03) Yeah,
yeah, pretty much.
Federico Ramallo (27:05) Interesting. I work with my godfather. used to build the automation and control for these gas compressors. 200 atmospheres for natural gas for cars, for vehicles. ⁓
John Hooker (27:18) Okay.
Okay.
Federico Ramallo (27:27) the whole cycle of getting the gas compressed and the banks and then getting ready for the cars to fill in and kind of based on the demand kind of turn on the compressor, right? It wasn’t turbines, it was pistons and then they could, if you do it wrong, it could blow up, right?
John Hooker (27:40) Hehe.
Yeah, that’s we we’ve had even several situations of shutdowns where that’s happened where my favorite was there’s a backup power system for most power plants and we were doing one where we asked him if they wanted us to upgrade the power plant and they said no and it was like okay that’s fine whatever and so we were upgrading the power station but at one point in time they had a shutdown an emergency shutdown because they lost power and so then they were trying to use their backup
power system and we opened up the control cabinet and a whole bunch of rats nests and other things fell out and so the main reason why it wasn’t working was nobody literally opened up the control cabinet and looked at the backup power system control and rats had to chew through all the wires in it and everything like that and it was yeah it was pretty bad it was like she probably opened this maybe you know once or twice a year just to see if everything’s still working
Federico Ramallo (28:36) no.
John Hooker (28:44) make sure that it does that. Then we got the backup power supply project after that. So that was pretty fun too.
Federico Ramallo (28:51) Right, right, interesting.
John Hooker (28:53) Yeah, I have a theory
that the whole world is held together on duct tape, band-aids, and a prayer. So it’s our power system, our water system, pretty much our natural gas that gets delivered to the house. It’s all held together on band-aids, duct tape, and a prayer.
Federico Ramallo (29:08) Right, right. I went to Argentina mid this year and I went to one of the old part of the city. They have one elevator of…
I remember the name of the creator of the elevator, the lift. Oris, so they had an original Oris elevator, right? And everything is mechanical, right? Electromechanical. And it’s very interesting to see how everything works. You have an open cabin, right? So I find it fascinating, right?
John Hooker (29:22) Otis?
Federico Ramallo (29:38) Fascinating how it works and fascinating that it’s still working, right? So we reached the peak of electromechanical back in what, the 40s, right? Maybe the 50s, right?
John Hooker (29:41) Yeah.
⁓
Yeah, my grandpa used to say that he really liked cars before they introduced all the electrical systems into it because one of the biggest issues in mechanical things, the only thing they do is break. So all you got to do is go in and repair it or go in and fix it. You know where the break is, you know what’s going on. Pretty easy and straightforward. problem with electrical systems is they tend to degrade over time. so, and especially as you start getting more complex electrical systems.
You don’t necessarily know where that wire is that either got degraded or got there. So he always talks about how mechanical failure is much easier to figure out and find than electrical failure. always go more mechanical than electrical. It’ll last longer.
Federico Ramallo (30:31) Right, right. I think I see the benefits of using mechanical electromechanical controls, but I think that the issue there is, you know, you cannot have remote viewing of the status, right? ⁓ Right, and then you cannot optimize it at the level that you can with with electronics, right?
John Hooker (30:43) Yeah, mm-hmm.
Yeah, well and I was just saying now especially with in terms of automation one of the things that we’ve seen quite a lot as I started or you know as I started out you normally had people at the plant running the plant being good to go with that and you normally had an engineer or two on site to help deal with issues whereas now they tend to focus a lot of the engineers in one big city where they can find a lot of engineers and hire them and that’s where they want to live.
And so plants out in the middle of nowhere are just mostly remotely monitored. And you have a couple of technicians or you have a couple of operators that only go out and then just call into the main place when something goes wrong. And most people just tend to watch it remotely nowadays. I don’t know whether that helps or not. As somebody who is in, especially remote SCADA and remote
solutions, I highly recommend it, but it’s also one of those where we can’t catch everything and also, you know, again, when signals break down or you can’t get the signals in, you never know what’s going to happen.
Federico Ramallo (31:53) Right, right. that’s, I think that the benefit of using electronic controls is that you can have this stability on unstable environments, right? ⁓ know, like having the center of gravity above the point of contact, can, you know, kind of with electronics, can balance that out with mechanical, you cannot, right? ⁓ Or it’s very difficult. But I think that the key for electronic control is
John Hooker (32:01) ⁓ yeah.
Mm-hmm. Yeah.
Federico Ramallo (32:17) the reliability of all the components within it, right?
John Hooker (32:20) Yeah, well and redundancy too because that’s another thing that we’ve seen a lot where it’s expensive and most of the time people will eventually go to redundancy because we normally ask if they want to put in a redundant system and a lot of times they’ll say no to start off with but when you have that one either blackout or shutdown or something then normally they’re like never mind we’ll pay for redundancy now so it makes it pretty handy.
Federico Ramallo (32:40) You
So with redundancy, can switch sensors and verify the sensor as failing and just switch to backup.
John Hooker (32:53) Yes, it’s not so much on the sensor part. It’s more because the sensors are normally pretty reliable It’s more of the data communication and how it’s communicated. So it’s normally two switches essentially and Two different types of controllers gathering the data. It’s two different. So say for example, we have what we call a virtual tag system and so we can have a You know a data collector essentially running on site and then it goes up to the cloud and then we can have another one
that’s in a different computer on site and just different communication and you know when one fails okay we’ll switch to the other and kind of go from there so we kind of put redundancy there that’s one of the things that we just finished with our software this year was we had a power project that wanted essentially they wanted two different areas with redundancy in each area but then say for example if something went down
we could then switch over to the other area and it would still get full data, full everything. And again, one of the things that we do is we also, if communication fails, we save data locally and we can kind of be able to bring back that data that was either lost or that if for some reason both of those systems failed, it would still save it locally. And then when it’s brought back up, we can get data.
back up to where everybody can use it and everybody can see it. So that’s where kind of the redundancy part starts to come in and click a lot. So yeah, there’s a lot of different areas that you can be redundant in. Normally it’s not necessarily the sensors, but it’s more of the from the controller to the server that is grabbing the data through OPC or Modbus or whatever communication that they’re using through and then up to our
cloud server or up to kind of the main area or the business level of the network. then other people can grab data from that. So it kind of helps with having redundant data and also never failing to get a lot of data. That’s one of the joys of, kind of now where we moved to is when I first started out, it was most of those of like, well, we don’t have enough field data in order to make decisions or do stuff. And now it’s kind of moved to, well, shit, we have too much data.
We don’t know what to do with all this data and how do we make decisions on it. So that’s another fun thing that we’ve kind of moved to.
Federico Ramallo (35:01) Right, that’s
why the machine learning role became more relevant because we have too much data and we don’t know what to do with it. Now we need somebody that can interpret the data, right?
John Hooker (35:05) Yeah.
Yeah, absolutely. And that’s, you know, again, one of the nice things about what we do for like predictive maintenance, it has a lot of the machine learning algorithms or just a lot of different aspects of, okay, this is going to happen and say, for example, 30 days instead of having emergency shutdown, why don’t you find a good time to stop it, take care of it now and prevent that, you know, prevent that emergency shutdown and kind of go from there. A great example was that we had a
bunch of reciprocating compressors that were kind of scattered out in the middle of nowhere and people were driving out you know two three hours to the compressor station to try and troubleshoot stuff we put remote sensors out there and well actually just a computer to talk to the local control panel there and then the remote computer can send data up to the cloud and it changed their you know they saw some problem where it was just like
instead of it being, know, we need to drive out there now and try and fix it, it’s one of those of, we can wait to do that in a week or two when we can schedule somebody out there. And they went from having, you know, normally about 18 to 20 site visits a month to, normally it was just about two to three a month, which, you know, obviously helps with manpower and saving on time and saving on effort. Is this a catastrophic thing that we need to fix now or is it something that can wait to where we can look into it when we actually need to go out there?
Federico Ramallo (36:33) Right, right. Yeah, I’ve seen those people talking about these sensors that detect vibrations on the engines and they presented as a magic solution of you put that and then we’re going to predict everything, which it’s not necessarily true. think it’s more of a, it works, right? But I think it’s more of a, you need the full life cycle of the engine in order to be able to predict what’s going to happen, right?
John Hooker (36:41) Come here.
Yeah, absolutely. Well, and also that’s one of the biggest issues with, I think, large language models now is that they just constantly need data. And one of the biggest problems is you don’t necessarily know if the data that you’re getting is correct and then the predictions that they make are correct on it as well. Because just a great example that we had was it was a
we had a flow sensor on a oil platform, oil production platform. And one of the biggest issues was they didn’t put any dampeners on the flow sensor. And so they had a big compressor package that was shipping the sales gas to the shore. And so it was really fun looking at the flow, because they had these really nice three-phase oil sensors. And I mean, they were just bouncing up and down, up and down, up and down, up and down, because the compressor was vibrating everything. And it was just one of those of like,
You know, they’re just like, well, we can’t do anything with this date or, you know, we’re getting, you know, inaccurate results with all of this data. And I was like, yeah, I can imagine it’s going crazy. And so they were just like, how would you do that? was one of those of like, well, first you should have had dampeners on it or put it not so close to the compressor for on the way out. But I don’t know why people didn’t tell you that. And then also what they were doing with the data essentially was they were just averaging it out.
Federico Ramallo (38:02) you
John Hooker (38:16) And I was like, that’s not a great way to make good predictions because all you’re doing is just taking a whole bunch of noise, averaging it, and saying, yeah, this is kind of close. And that was about it. It was funny. They were calculating like 50 % inaccuracy on their live values. they’re like, if you can get it to 20%, we’d appreciate that. And we eventually.
Federico Ramallo (38:30) Ha ha!
John Hooker (38:35) We normally like to have error in less than 1 % range for saying, yes, this is a nice good calculation. So we eventually got it down to less than 1%. It took a while, but luckily we were able to get everything kind of helping and moving along there. But again, just having that data when you have garbage in, unfortunately you tend to get garbage out as well.
Federico Ramallo (38:57) Right, right. That’s why you need the holistic approach of understanding not only the software, but also the hardware and the physics, right? Yeah. I used to joke that, you know, the ISO 9001 certification was, know, well, we’re making, you know, whatever we’re building this, plus minus a lot, right? And we’re certified, you know, which becomes useless, right?
John Hooker (38:58) you
Yeah.
Yeah,
it is. That’s again one of the things that is always really interesting and why I don’t think that AI is going to replace my job too soon is a lot of times when a new model gets out, I always ask it just to program a PID control in C++. so every single time that they do that, they start off with a good basic level of it. But again, it’s one of those where it forgets a lot of the physical constraints and other things that I need to
deal with that. like a great example is normally it takes me about four or five iterations to do that so like I’m just like okay well you know how would you do a PID control in an actual system and so then they give the equation and then they calculate the error and then go from there and I was like okay well it’s in the time domain so now you know can you can you put it in the time domain and they said sure yes we can put it and then they do the formula again with the time domain on there.
And so I’m just like, well, what if you happen to divide by zero? That’s not a number. So how do you handle that? And it’s just kind of all of those things that if you don’t really have experience on it or don’t know what you’re doing, then it starts getting into issues like that of like, can’t do my job really well. So it’s one of the things that I’m not too concerned about.
Federico Ramallo (40:32) But it responds with so much certainty, right?
John Hooker (40:36) the truth is it’s not that I think the thing is really cool and really neat and wonderful especially for what you can do for large language models. It’s unfortunately the people who don’t know what they’re doing that are asking it and then believe them or believe the AI unfortunately is kind of the biggest thing. So when it comes to designing stuff it doesn’t work out very well unfortunately.
Federico Ramallo (40:57) Right, right. I mean, reminds me of the movie WALL-E where the humanity doesn’t know how the computers work, right? Or the technology works, right? I think what is going to happen with AI is that the people with core knowledge are going to become much more valuable in the industry because of that, right? Because when you build something using AI, you don’t learn how to do stuff, right? You get the output and you put it in and if it works, works, right?
John Hooker (41:20) Yeah.
Yeah, think eventually it’s, sorry, you can go ahead and finish your thought. Okay, I think at some point in time, it’s going to do, you’re have prompt engineers to where you’re very good at prompting it for what the outcome is that you want. But the thing is, is you still need to have knowledge of the outcome that you want in it in order to go about doing that. So that’s, would say, kind of one of the hard parts about it. So I think it makes your already your…
Federico Ramallo (41:25) Yeah, and… ⁓ go ahead, go ahead, go ahead.
John Hooker (41:51) you know, your top performing people, it makes them a lot better. It makes them able to respond and do a lot more stuff. But unfortunately, if you, as I said, if you don’t know what you’re doing, it’s not as great of a useful tool and you can kind of make errors with it. We’ve seen a lot with companies that outsource automation, unfortunately, to different parts of the world, because it’s a lot cheaper and it’s still one of those where, you know, I can’t complain about it too much because they normally come to my company and say, hey, can you help fix this?
So it’s, but it’s one of those of, I remember we were doing a design of, it was an LNG, liquefied natural gas plant. And they did the liquefaction part and I didn’t understand why they were doing what they were doing. And one of the biggest issues that they had was it was not running correctly because they had the wrong kind of coolant in there.
Federico Ramallo (42:19) Right.
John Hooker (42:42) They were trying to use, it was an ethylene glycol cooled plant and they were using propane in lot of their either calculations or studies or something like that and I was like have they even been on site to look at it? You know that it’s ethylene glycol and apparently there was one automation director who didn’t know what he was doing and then he was relying on a lot of people somewhere overseas that were trying to tell him what to do with it that also was not working out very well.
It was one of those interesting situations where it was just like, okay, well, we’ll just take everything from here. Don’t worry about it.
Federico Ramallo (43:13) Right, and that’s what I’m thinking, that the core knowledge is going to become much more relevant. Because the agents are going to become better at bullshitting, And they don’t know what they don’t know, ⁓ But it will provide, when you have the experience, will provide you productivity boost. Because now, instead of.
John Hooker (43:27) Yeah.
Federico Ramallo (43:34) like spending hours fixing the syntaxes, then you can just run it through the agent, fix those things, and then move forward.
John Hooker (43:42) Yeah, as I said, depends. know for, especially for, say for example, compressor control, one of the big things is where the flow device is or also just where the pressure and temperature sensors are. Luckily now they’re getting so cheap that you can set up pressure and temperature on both sides and you normally have a flow either going in or a flow coming out. But again, the control in it is different depending on what sensors you have and where it is. And so it’s,
Again, also knowing that and knowing to ask the right questions of, where are the sensors on this? Or even just reading the P &IDs correctly is another important thing. Although, again, one of the biggest problems is that we also sometimes don’t have the most up-to-date P &IDs either. It always amazes me when somebody will give us something that’s eight years old and you’re just like, is this the correct thing?
A great example is they’re like, no, we took out a couple valves and we took out some sensors a while ago, so we don’t have those anymore. And this is an old one. And I was like, well, do you have an updated one? It’s like, we’ll have to get that to you. So, and it’s like, man, you don’t even know how your plant is working, but apparently that’s pretty common. So.
Federico Ramallo (44:51) Right, right. And those sensors degrade with time. So you have to take that into consideration with equations. Yeah.
John Hooker (44:56) That’s also true too, where
we’ve been on site and we’ve asked how, when was the last time that they were calibrated, essentially, and that’s another fun issue that we run into quite a lot. Or even, we were actually just doing a turbine project, and so normally the way that you control a steam turbine is just through the flow control valve for the steam that goes into it. And their biggest issue was the valve was 30 years old, so it was leaking like none other.
Again, normally you have, say for example, 100 % open, you expect full flow through. And they had it at, I think it was only 30 % open and we were still getting the full throughput of the completely open valve. And so I was just like, I think your valve’s leaking, so you either need to take it out or upgrade it or fix the seals on it and just help it out. And they didn’t want to do any of that. They’re just like, nah, just control it. And I was like, I don’t know how I can control something where…
I have from and that’s what we essentially did was instead of from 0 to 100 % we were just like okay at 30 % it’s full open so we’ll just do stuff from 0 to 30 % and then we’ll just say yeah open up to 100 % after that and do whatever but yeah that was that was kind of a fun aspect. Luckily it wasn’t too complicated they didn’t need to change the flow all that much so they just mostly wanted it 100 % open but starting it up was a was a real fun joy and trying to tune it was also a real
joy as well. You get all kinds of things.
Federico Ramallo (46:21) That’s amazing. It’s a fun part of having those projects, right? Where you can find these challenges.
John Hooker (46:26) Yeah,
it’s true. Well, it’s always really funny where they’re like, solve this problem. And it’s like, well, it’d be best if you do it this way. And they’re like, nah, that costs either time or money. And we don’t want to do that. it’s just like, OK, well, we’ll see what we can do after that. So as I said, it’s always pretty fun. But that’s what makes the job enjoyable.
Federico Ramallo (46:40) Right.
Amazing. before we wrap it up, I wanted to ask you what advice would you give young engineers that want wants to get into control business?
John Hooker (46:52) Yeah, well there’s lots of different ways that you can do that. One of the things that I’d recommend is ISA does a great job. I’m part of ISA and they do a great job of teaching people fundamentals of control and being able to do that. Another great thing is just finding what industry you want to do and how the controls work for that or finding a specific industry that you want to work in and then just trying to meet people in that industry, especially in this wonderful day and age with
how easy it is to get people on LinkedIn and Twitter and different things like that. If you have any questions or want to reach out to someone, I’ve normally found that most people are pretty responsive if you have just a legitimate question and reach out to them asking about their opinion on it or what they would do. there’s lots of different resources to use, so use your resources. And I would say instead of waiting to…
have everything perfect or have every condition be perfect just try new things, try new control schemes, try and do stuff yourself. You can literally program a PID control algorithm here pretty easily. You can download free simulators for simulating either like ladder logic or simulating the sequential function blocks or something that you’d find in controls. So just try and do an example of tank level control, try and do an example of motor control. And it’s…
A pretty fun way to learn is by doing, so just start doing.
Federico Ramallo (48:11) Amazing. John, thank you very much for joining us today. I really appreciate it. We learned a lot about the control business and I’m looking forward to what you’re going to do next on statistics and control.
John Hooker (48:23) Yeah, absolutely. Thank you for having me. I appreciate it. It was really fun.