Dear maintenance and reliability managers!
Please, be welcome to listen to our Podcast to learn more about Digital Reliability:
Reliability Radio Podcast – Full Text of Andrey Kostyukov From Dynamics Scientific Interview
00:00
digital reliability, which means that you know exact health of every piece of equipment which operated on your facility. The voice you just heard is that of Andrey Kostyukov. Andrey is the CEO and president of Dynamic Scientific. And what he just described there is a concept they call digital reliability. Now the word
00:29
Digital is not new to any of us, nor is the word reliability. But when you combine those two, you have a unique and different approach to how you look at the health of your assets. My name is Blair Fraser. I am the host of Reliability Radio. And this is my interview with Andrey Kostyukov of Dynamic Scientific. I hope you enjoy.
00:56
Andrey, welcome to Reliability Radio. I’m so glad to have you on today’s podcast. Hello, Blair. Thank you for having me. So, Andrey, we have been filled over the last few years with these buzz and hype words, industry 4.0, smart manufacturing, digital transformation. And when I did my research prior to this interview, when I looked up Dynamics Scientific, it’s hard not to come across the words digital reliability and think
01:24
you know, what is that? What is digital reliability? And as I said, you know, we know reliability, we know digital, but what happens when you combine those two? And from my understanding, you know, it’s a concept that dynamic scientific has really put together. Can you explain what digital reliability is and what it does? Yeah, thank you for the question. It’s a really interesting thing because digitalization itself is like a buzzword right now.
01:52
Of course, I definitely need to clarify it. So digital reliability means that concept from the predictions and from a probabilistic assessment of some events such as failure, for example, or maintenance of machinery, we move from there to another, absolutely another concept.
02:22
digital reliability, which means that you know exact health of every piece of equipment which operated on your facility. So you have to you have to change this digital reliability, change paradigm shift, like makes paradigm shift from probabilistic assessment, from guessing does it have or not?
02:51
Should I maintain or not? Should I improve reliability of this piece of equipment for another one? And you have exact, unbiased and timely information about machinery health. And digital reliability means that you utilize the computer power for three main domains.
03:21
you collect proven, scientifically proven information about machinery health from the equipment. Second, the artificial intelligence, which has physics rules of degradation, physically proven rules of degradation of different parts of equipment on one hand, and the bunch of algorithms which…
03:47
could be chosen to analyze the particular machinery health. And this physics-based artificial intelligence just analyze the entire data which field bus collected from the sources and recognize almost all failures and defects with high probability are very close to one.
04:17
to 100%, like 97-98%. In that case, the artificial intelligence forms prescription, which could be delivered simultaneously in several points of decision making to several people in the same time. And of course, in that case, in that case, we have the third point of digital reliability.
04:47
when the information could be displayed simultaneously in several places and for several people who are involved in decision making in the reliability process. Excellent. No, I think, so you said a lot. And I think when our listeners listen to that, there’s gonna be a few key things that I specifically wanna dive in deeper and specifically around the scientific knowledge of degradation and what you call the scientifically proven parameters.
05:16
And I think we’ll dive into that a bit deeper, but just holistically, staying on the digital reliability part of it, when you’re talking, when you give your customers digital reliability, what are the benefits they’re achieving from that digital reliability? Yeah, it’s, thank you for this question as well. So the, there are three, at least three important benefits you have. First of all, you could achieve the almost,
05:46
completely, you could completely eliminate the any failures and breakdowns due to equipment shutdown and due to equipment breakdown because the digital reliability approach provides the identification almost 95 percent of defects with high probability like 97-98 percent. Second, this is a scientific knowledge of degradation processing equipment.
06:16
which can be used in real time by layman. You don’t need to be an expert in vibration analysis in order to use it. You just need to have this unbiased information timely. And third one, last one, maybe last but not least. So the transparency of the relationships between all departments in the facility, between operators, maintenance team and management in terms of…
06:45
in terms of machinery health, the safety and reliability improvements. Excellent. So on point number two, you mentioned the scientific knowledge of degradation process and equipment. And you mentioned that it’s in layman’s. What is the example of that scientific knowledge of degradation process? Are you using like unique failure patterns or or the physics of failures? Yeah, absolutely. For example, when you know exactly the physics of
07:14
in machine support, you know what kind of parameters of vibration signal strongly related to this defect and when the physics-based AI recognizes these parameters and assess them as they’re dangerous or hazards parameters, I mean, in terms of their level. So…
07:43
it could identify and transcribe this technical data to layman, to people who don’t have any knowledge in vibration analysis. So you don’t need to be a next level… That would be me. Yeah, exactly. So you don’t need to be a cat 4 level of vibration…
08:12
Absolutely, absolutely. Yes. This is a main one of three main advantages that artificial intelligence provides unbiased information and very deep information which is based on knowledge, on scientific knowledge, not just on some statistical assessments.
08:40
Okay, I just wanted to clarify that in the first one, I think, and I don’t think there were in any order in how you listed those three, but the first one, I’m just going to repeat it because it’s a pretty outstanding number. 95 to 97% of defects and malfunctions are detected in real time as they emerge. Yeah, 95 to 97%. That’s a lot of defects that you’re covering. Yes, because I’m talking, look, I’m talking not about root causes of defects.
09:10
because of course, root cause failure analysis, this is a very complex and completely different option and completely different point. I’m talking about particular defects in particular machine. So there are several parts, spare parts in every machine, and all these parts have own mechanism of degradation. And if you know this mechanism,
09:40
identify the main rules or main law for degradation for every piece of this spare part. And when you know exactly these rules, you could compare parameters of vibration signal which are related. You should not compare. I mean, you could choose specific parameters of vibration signal.
10:07
Of course, it’s not just vibration velocity. Vibration velocity shows just like 10% of malfunction in the early stage of degradation. And all other bunch of malfunctions and defects in the late final stage, when you cannot prevent the problem, you just can shut down the machine and reduce the consequences of the breakdown.
10:37
or so yeah and in that case in that case the 95 97 percent means that if we list all malfunctions and all defects of this machine according to its its part yeah I mean it quantity of parts of this machine we could identify 95 percent at least all mechanical problems
11:05
which could be identified by eight most often utilized the non-destructive testing methods, such as vibration analysis, thermography, electrical current, and etc. So how do you determine what sensors do you need on pieces of…
11:33
Good question, thank you. It’s a very important thing because of course according to our experience, actually our name, our title of our company name is Dynamic Scientific. Because we are deeply involved in scientific research about machinery health and diagnostics of machines since the beginning of the 70s.
12:02
So I mean that, of course, depending on what kind of defect could appear in particular equipment, we have to create the list of parameters which we should identify for diagnostics of this particular machine. So you’re going back to that physics of failure. How can it fail? How can we detect it? Absolutely. Yeah.
12:32
I mean, so the other thing that I came across when I was doing some research was this concept called physics-based AI. Now without going, cause artificial intelligence is obviously, is going to play a big role in terms of operational excellence, maintenance forward, all those other things. And so first time I’ve come across physics-based AI, without going into too much detail, cause we don’t have all day to talk about this, is what will…
13:01
And generally, what is physics based AI? And do I need to be a data scientist to be able to, I guess what we’re talking about here is your compact system from USA Dynamics. What is physics based AI? You know, it’s a very good question because do you know difference between physics and mathematics? Mathematics can calculate everything. Physics can calculate only something which exists. So…
13:31
physics has a focus on existing events and existing issues and existing problems. In that case, we have a chance to not just to imagine, but to confirm the particular rules of degradation of particular pieces of equipment. And knowing these rules, providing…
14:00
deep and wide scientific research in order to learn more about degradation of, for example, bearings in operation and analyzing their root cause of bearing fails and so on, we could identify particular parameters in different non-destructive testing methods, which
14:30
should be utilized for particular defect identification. So, it means this is a physics way, physical way. A mathematical way or statistical way. It means that you can measure everything you want. For example, how bright the sun is shining today. And you just create a correlation between
14:59
such parameter and the particular defect. And if you find some correlation, you can say, hey, I suppose that if sunshine will, yes, sun will shine like 23 hours, then pump will fail. But it’s, this is the reason why statistical-based artificial intelligence really doesn’t work. Because even if you measure,
15:28
All 8 billion people on our planet, and I mean the toll of every person, of every individual, you cannot identify particular health of particular men, because health doesn’t correlate and doesn’t relate, I would say, with the toll of the people. So, and in order to…
15:58
How can I say to distinguish the difference of real artificial intelligence which could identify malfunctions according to scientific research and All other artificial intelligence because right now it’s also buzzword, you know, everybody say hey I have artificial intelligence. I create the algorithm which calculates
16:28
artificial intelligence, but it is not. And in order to, how can I say, to distinguish the real artificial intelligence from a bunch of others, we decided to call our artificial intelligence like a physics-based artificial intelligence. That’s where it, okay. So that’s where it came from. That makes sense.
16:57
And I think it’s, you know, use the analogy of people, the billion people across this planet, and if you were to take data for every one of them. Right, and I think that when we look at, and you made a statement early on, is knowing the exact health of every piece of equipment. Right? So do you need to monitor every pump? Like say you’re working refinery, and you obviously there probably have more than one pump in a refinery.
17:25
Do you need to use an online system for each and every pump? Actually, I don’t. But the customer does. Why? Because just imagine that you would like to be safe and healthy and you would like to be alive actually, yes. And you monitor your heart. You monitor your heart, you know exactly that your heart is well, but you don’t monitor your knees.
17:55
your legs, your brain, and some even finger, for example. When you fail your finger or you damage it, it’s not just very painful. It also could confine your operation ability. The same situation with the facility. On the facility, there are four types of equipment.
18:25
First one, and all of them could be split according to its possible failure consequences. For example, first category of equipment. This is equipment due to fail, when the explosion of fire could happen. The second category, this is equipment
18:55
in case of breakdown, which the operations could be confined or shut down completely. This is the second category. Third one, this is a category where just increase quantity and price and cost of maintenance and
19:21
Fourth one which could be operates until breakdown. It doesn’t matter. For example, we change The light bulbs only when it’s gone We don’t we don’t we don’t change them in preventive for like preventive maintenance so the same situation with equipment so in order to be able to To have a Payback for
19:50
investments in reliability, you need to provide at least safety, so you have to equip the equipment which could cause the explosion of fire on one hand. And on the other hand, you need to equip with the real-time diagnostics equipment of second category which could confine the operations.
20:19
capacity of the facility. And those two categories consist of around 80% of all machinery of the unit. And even small pump which pumps water for cooling some very high temperature
20:47
part of the facility. If this pump is broken, in that case, the more hazard and more wide problem could happen. And this problem, not just because of this pump broken, but because this pump stopped…
21:15
stops to perform a function which it is responsible for. So, this is also a very interesting question. So, when we offer our solution, first of all, we do engineering study. When we assess all equipment in the facility, we discuss this equipment with the particular…
21:44
maintenance and reliability team and then we create the project which provides ability to operate as long as possible and technically safe. So that’s it. So I think a lot of the listeners are you know obviously we’re gonna direct them to your website USA Dynamics.
22:11
to get more information on the compact system we’re talking about. But I think one of the things that I struggle with is separating systems like this from, say, just connecting an online vibration system and interfacing through that through the plant network. How does digital reliability offered through the compact system different than just, say, putting an online vibration monitoring system on there?
22:38
Yeah, it’s a very important question because a lot of people don’t distinguish the difference between unbiased information which is available for everyone who is involved in reliability and vibration data which is available for vibration experts. Actually, this is a…
23:05
main reason why all accidents and fires and breakdowns happen due to night shift or weekend shift when no one expert is on duty. So the digital reliability concept means actually three main
23:28
I would say three, it requires three main approaches. First, it requires that operators should be involved in reliability. So when we provide the digital reliability approach, we need to provide, we need also provide the operators driven reliability approach for sure. Because only operators are 24-7 in the facility and only they could react timely.
23:57
in order to prevent any failure or any breakdown. Even if they could eliminate the destructive forces, we just began to influence on weakest part of equipment, they could prevent even redundant maintenance. So, operators-driven reliability first. Second, they need to organize, I mean customer, need to
24:27
the on-demand maintenance. It means that maintenance should be ordered by operators for a better reliability of the facility. So it’s completely different paradigm, it’s paradigm shift for entire relationships of the different departments in the facility. So in…
24:57
This is really the biggest challenge when you implement digital reliability. And of course, the last one is the management support. Without management support, it cannot be changed. So, digital reliability means operators-driven reliability on demand maintenance and management support. There are three pillars.
25:25
which digital reliability stands on.
25:31
Great. So when we look at this and my questions, I think it’s the way any of us in this community, we think, we live, we breathe rotating pieces of equipment, things that move in a circle. Because they move, they’re more likely to break. They have more failure modes. But you look in a chemical plant, a petrochemical refinery, yes, they have stuff that spins in a circle, but they also have assets that doesn’t move, fixed assets, heat exchangers, things like that.
25:59
Can you apply digital reliability, that same concept around physics-based AI, the scientific way things fail, can you apply that same technique to say heat exchanger? Yes, of course, exactly. So we are working, for rotating equipment, we are working approximately for 50 years, but approximately 20 years ago in…
26:28
to be exact in 1998, we started to develop the physics-based AI for fixed equipment, such as columns, pipelines, reactors, heat exchangers, cog drums, etc. And right now, our artificial intelligence could identify cracks in fixed equipment
26:57
could locate cracks on the body. And also artificial intelligence identifies the malfunctions of the process mode, of the mode of operations of fixed equipment and contaminations and heat exchangers and leakages in pipelines. So there are not too many defects and malfunctions as we have in rotating equipment.
27:27
But these four, this is a main, maybe hazardous of defects, especially in terms of refining in petrochemical business. Perfect, that’s great. So we are running out of time here. I think we could talk about digital reliability all day, at least I could in my mind. So I think, you know, when I started doing some research prior to this interview and I typed in digital reliability, it actually came to USA Dynamics or dynamic scientific.
27:55
So what I’m going to encourage anybody that wants to learn more is to go to USA Dynamics.com and Andrey is doing a lot of webinars spreading the word around digital reliability and if you go there you can start to find more information on the Compact system. And I would highly recommend you attend one of his or the company’s webinars coming up. There seems to be a lot of them if you go to the news and events page.
28:23
Anything you want to add, anything I missed in terms of key features or key information you want to get out there, Andrey, to the audience? Yeah, I would like, first of all, I would like to appreciate the time which people spent, I mean, I hope to invest, they invested in this interview. Right, they got something out of it, yeah.
28:48
Yeah, and also I would invite also thank you, Blair, that you remind me and I’d like to invite everyone to join our free webinars, Digital Reliability 24-7 Real-Time Machinery Diagnostics. And one of them will be soon on June 2 and another one will be later in middle of June.
29:16
Please follow our website folder the news folder and you can find a lot of information in This in this part of our website. Thank you very much Blair. Thank you I’m sure we’ll be connecting and have you on again. Thank you so much Thank you. This podcast is copyrighted by reliability web make all rights reserved reliability web.com and reliability radio are trademarks of reliability web
Explore Our Case Studies
Reciprocating Compressor Monitoring Case Study
06/12/2025
Introduction: Compressor Monitoring Reciprocating compressors are widely used in critical processes at oil refineries, such as diesel hydrotreating and dewaxing...
Read More
Pump Monitoring Case - The COMPACS System Resolves Critical Failures
11/20/2024
Industrial pumps are vital for operations in industries like oil and gas, where equipment reliability and efficiency are non-negotiable. Unplanned...
Read More
Case Study: Condition-Based Maintenance Of an Oil Pump
06/28/2024
In this case study, we present a condition-based maintenance case using the real-time diagnostic COMPACS system. Pump Monitoring In the...
Read More