NDE 4.0 Podcast Transcript
Episode 19 — Digital twins in NDT
Our Guest: Nandu Vellal, CEO and co-founder of nAurava Technologies
Nasrin Azari (00:00) Hello and welcome to Floodlight Software's NDE 4.0 podcast. In this series, we interview various experts in industry 4.0 concepts, ideas, issues, and technologies as they relate to non-destructive testing and inspections. This show is designed to explore the biggest challenges and opportunities for the future of NDT guided by some of the smartest people in the industry. So be prepared for a thought-provoking discussion and to learn something new in the next 30 minutes. Hope you enjoy the episode. Today we are joined by Nandu Vellal, CEO at N'Aruva and an expert in digital twins and industrial 4.0. Nandu has more than 20 years of experience in the high tech field with experience in CAD, CAM, 3D printing, 3D visualization, AR, VR, robotics, database analytics, cloud software and project management. He designed and developed an interactive 3D visualization exhibit for space education at Krista McAuliffe Center in Framingham State University. And he has a background in mechanical engineering. In today's discussion, we are going to focus on your area of expertise, Nandu, and talk a lot about digital twins, how they work, their value, and how they relate to industry 4.0. I'm super excited to have you with us today. Welcome to the show. Nandu Vellal (01:15) Thank you very much, Nasrin. Nasrin Azari (01:16) The format of this podcast is that I'll pose five questions to you. And these questions are designed to dig into some of the most meaningful and interesting aspects of NDE 4.0. And I'm excited about today because digital twins is one of those areas that is really, really exciting in non-destructive testing. And our podcast was developed to help educate and expand conversations around the possibilities, challenges, impacts, and opportunities surrounding NDE 4.0. So let's jump in. Our first question, Nandu, is how can digital twins create a competitive advantage in NDT and what kinds of ROI should we come to expect from these types of investments? Nandu Vellal (01:55) All right, so first of all, thanks for having me on the podcast. yeah, I think digital twins are kind of like becoming the talk of the town, especially on the industrial sector. So now specifically for NDT, I think, you there's a lot of work that has happened within aerospace and DOD sectors. And, you know, we see that slowly proliferating into other industrial sectors as well. So when we think of the competitive advantage that digital twins can provide for NDT, with NDT we typically compare two parallels with respect to the medical industry. So if you think of in medicine you have You go for a health checkup and the doctor checks for your pulse. And then if need be, he's going to ask for x-ray or CT scan for more detailed information. Something similar is happening with NDT. So it's not just for the health of an asset. Now you have, apart from centers and gauges where the data is captured, now with NDT, you get more detailed information about the structural integrity on the internal structure of the asset. this is where for digital twins, publishing that NDT data can provide a more comprehensive view. And that's a significant advantage in terms of knowing in a much better fashion, in a much better manner, the health of the asset. Nasrin Azari (03:22) Yeah, that's interesting. Nandu Vellal (03:23) So yeah, so I mean, also when it comes to the sustainment of the assets, it kind of gives the ability to remove a lot of uncertainties that you may not know from a traditional inspection procedure and looking at the NDT gives you much more insights. into the health of the asset as to what's happening. And also with the evolution of advanced analytics, you can not only take the NDT data, publish it onto a digital twin, and you get to build like a predictive model that is much more accurate as compared to just using sensors and gauges of what traditional digital twins have. So you have a much better, much accurate predictive models where you can predict the health of the asset and when the asset is going to degrade. Nasrin Azari (04:13) Yeah, I feel like the one of the kind of a dreams or their nirvana around this concept of using digital twins in non-destructive testing is for that purpose of being a lot more predictive about what could happen down the road with these assets versus today. It feels very much like non-destructive testing is just a box that's checked in a process or or it's kind of a gate of getting from one step to the next. And it's just a point in time. It doesn't really help create a better understanding of your asset and its overall health and its potential, like a more accurate prediction on when it might fail and what things might cause it to fail, which I feel like digital twins are, I feel like that's one of the main like goals of having a digital twin representation of an infrastructure type asset, right? Nandu Vellal (05:11) Exactly. I mean, I think the other thing is like also the, you know, with all the advancements that's happening with AI, right? So what you're going to see is like, you know, a combination of, you know, physics-based modeling as well as like AI-based modeling, which can give you much more better prediction in terms of, you know, how, you know, not just when an asset is going to fail, but also in the near future how an asset is going to fail. these are exciting things for the NDT industry. The other important thing that most of us miss is like, because the NDT inspection data is kind of like, as you said, once the inspection is performed, it's kind of like boxed and it stays inside some closet, right? So now if that data is published onto a digital twin, imagine what kind of value it can provide for an inspector who wants to perform an inspection on an asset. He wants to understand what has happened in the past. publishing the data to digital twin is going to give the inspector, he can go back and look at what has happened to this asset. He can pull up that information and visually see where the problems are for this particular asset. So it takes so much of burden off of the inspector because now he can visually and representatively rapidly gauge that information and make better decisions for his next inspection procedures. Nasrin Azari (06:40) Yeah, I think I read somewhere about the difference between how much energy it takes to process the written word versus visual. Like it's almost like a 10x difference where if they can, and that's, think one of the reasons why I find your research so fascinating is that having that visual representation in a digital twin, mean, sure, you can have all of that data available in text format. But it takes time because your brain, have to basically read through all of that data and then you kind of have to construct the picture in your head before you can really understand what's happening. And what you're doing is you're kind of short-cutting the process by creating that visualization and having the individual skip all of that requirement of doing it themselves. versus just getting them there and also getting everybody on the same page so everybody's looking at the same thing versus interpreting the data in different ways. Nandu Vellal (07:31) Absolutely. So, you know, the industry refers to this as the cognitive overload, right? So what we are doing here is like reducing the cognitive overload on the inspector, right? So, yeah. And, you if you look at, you know, some of the, you know, so that's one of the significant benefits and the ROI that digital twins can provide. And the, you know, the other important thing is like, if you think of, having all the information accessible readily can help in terms of reducing the uncertainties so that people who need to make the decisions can have all the information readily available for them so that they can make right decisions at the right time. So the other important thing is we also need to consider is like So for asset owners, in the sense like who are processing, let's say for example, you take an oil and gas industry, right? So they are processing chemicals and they're at the end of the day, so they have an output and maintaining these assets in a better manner enables them to maintain that a decent level of quality of service so that they can meet their customers' demands. And I think the asset uptime is a key factor. And I think these digital twins will enhance those capabilities to ensure that they can maintain the asset in a better manner and keep the asset uptime to as high as possible, right? Nasrin Azari (08:56) Yeah, absolutely. Nandu Vellal (08:57) You know, there is another interesting thing when it comes to the ROI of, know, provisioning digital twins for, you know, NDT, NDE, which is like, the cost of technology is coming down. And what this enables is like, you know, as, you know, Michael Greaves from Florida Institute of Technology had mentioned, the cost of the physical asset, you know, goes up at the rate of inflation. whereas the cost of digital twin is coming down due to the technology advancement. So which is a fascinating thing. And I think this, are at the right juncture in terms of like how we can enable NDT, NDE to be a significant value driver along with digital twins. Nasrin Azari (09:39) That is fascinating. That's a great point that you make about the physical asset cost or the cost of the physical asset itself are growing, whereas the digital twin costs are probably going to continue to decrease over time as technology improves. But let's move to question two so we can move through some of these questions. I'm sure we'll hit some of these things again. as we go through these. But what do you think, you we've talked about a lot of the benefits and the ROI and how valuable having digital twins is for these physical assets. What do you think are the biggest barriers to scaling digital twin technology across industrial operations and non-destructive testing? And how can asset owners and operators overcome them? Nandu Vellal (10:27) Yeah, you know, this is one of the topics that I keep hearing from my colleague, Dr. Ripya Singh. You know, one of the things that's always at the top of the chart, the top of the list for such, you know, any new technology, you know, kind of like implementation or, you know, needs to be brought over in any organization is so. it's the organizational and the cultural change, right? So it's the people. And I think a lot of these things has to happen with all the key stakeholders, the buy-in from the key stakeholders. And it's not just about the buy-in. I think there's a lot of education that needs to happen in terms of the kind of value these kind of systems like Digital Twins can provide for NDT, NDE, as well as all the industries that leverage NDT-NDE. So that's kind of like the key barrier that needs to be addressed. Nasrin Azari (11:21) Do think there are any barriers around? You know, I definitely agree with you. feel like in a lot of a lot of the conversations I've had with folks, there's cultural barriers and and what about education and other kinds of resistance to technology in terms of of risk to companies? Like like, do think that there's a distrust of the data that might be collected in digital twins? Do you think that there is? any concern about, I've got all this data stored in my digital twin, somebody might steal it or take advantage of this information that I didn't previously have to store? Do think there's any of that kind of resistance? Nandu Vellal (12:00) I think from a NDT and EE perspective, mean, if you look at it, think most of the time, the folks are concerned about like ensuring that the asset is in a healthier condition most of the time or all the time, right? So you are right. So in terms of the sharing of the data, right? So there is always this hesitancy in terms of like, if somebody else comes to know about the nature of the problems and challenges that our assets are facing and how that can be a of like a negative aspect in terms of our competitive advantage, right? I think that is an issue. And I think what the companies need to do is educate all the personnel about why such systems are needed. And what is the benefit at the end of the day? You're trying to keep your assets in a much better, healthier condition so that you can ensure that you can fulfill your customer's needs, right? So, and I think companies need to put together policies and practices in place to ensure that the data is shared so that like across teams and so that they can provide a better product throughput at the end of the day. So yeah, now. Some of the, you know, when you talk about, you know, the data, the significant challenge is also, you know, compared to, if you compare like, you know, traditional digital twins, which has like, you know, sensors and gauges with respect to NDT, the kind of data is different. So in the sense like the inspection produces vast amounts of data and these data are also complex. and complex in the sense different techniques like artisan, eddy current, x-ray. So they all produce different kinds of data in different modalities. So there is a challenge in terms of how do you store all the data? And then how do you make sure that is provisioned as part of a digital twin? Is it going to be provisioned as a signal information? Is it going to be some kind of a physical representation. So all of these are, you know, research that's happening at this point at various universities. you know, the data itself is, is, you know, a challenge in the sense it's coming from multiple modalities, you know, different format. And as such, you know, the, a lot of the data is kind of like equipment or OEM, know, equipment OEMs. locked in, right? So now only the industry is waking up to it, which is fantastic. You have formats like Dicomd or .nde. So where now you can take the data and use it as a much more, in an open format, you can represent that onto a digital twin. And that's a great, kind of like a, work that is being done so that which will help the development of digital twins and specifically publishing the NDE data onto digital twins. these are the barriers. Yeah, please go ahead. Nasrin Azari (15:09) Speaking of the conversation that you just, or the items that you were just talking about with the data, yeah, I mean, hear that a lot about how much data there is associated with sensors and collecting data. And our question number three today is around that area. So with NDT generating such massive amounts of sensor and imaging data, How can organizations turn this data into actionable intelligence rather than storage overhead? Because there's this, just like we were talking about, there's this kind of barrier or there's this problem that needs to be solved in terms of storing the data, but it can't, it's not just storing it, it's also being able to analyze it and understand, be able to wade through it and understand what kind of insights you can act upon, right? Nandu Vellal (15:58) Yeah, no, I think that's kind of like also one of the heavily discussed topics too, right? So because this is where AIML is going to come in, right? So now before I jump on to answer that question, so I think the other challenge, you brought up like massive amounts of data. So the other challenge companies have is like, they have a lot of historical data. So what do do with that, right? So, especially for assets that are still in service, companies need to think about how can they digitize the data if it's not even in digital format. And once it's digitized, how do you make it accessible? What kind of IT systems to put in place so that all the data is captured and it's accessible to the right people, right? So I think one of the... topics that we already talked about was the data formats, but also the other aspect is like data integration, right? So there needs to be systems that needs to be put in place for data integration. coming to the massive volumes of data, think what we see in the industry is that there are a few things that are emerging. Excuse me. One is, of course, AIML, right? So you want to make, once you capture the data, you want to make a meaningful insight. You want to generate meaningful insight from the data. And AIML is going to be the key there. Specifically, a lot of work that's going on in terms of X-ray or UT in terms of like, automated defect recognition. Is there a way the moment an indication is captured or is visualized on an instrument, is there a way we can say, it really valid or is it just a false positive? So there are lot of these kind of things that are emerging in terms of the AI. Now, the other important that is also emerging is like edge AI. So when we talk about, recently I was at this NDE 4.0 conference and now companies are trying to build add-on devices for your UD instrument. where it can, and also along with that, techniques of NDE inspection that are being performed. it can sift out the noise and remove like 90 % of the noise and retain only the indication or the right signal information. So these are some of the techniques that are emerging, but I think AIML for sure is going to play a significant role. The other important thing is also what you mentioned about earlier in terms of the visualization. So what happens is like, think of like a large asset, like an aircraft or a ship. And if you have a massive amount of data, and then if you're able to contextualize and then say like, within the specific, like say hull of a ship, right? So within specific areas, so it becomes a lot more easier for people to view and understand. I think digital twins are gonna play a key role here. But then again, I think a lot of the work is going to be done by AI in terms of helping people, helping inspectors to sift out and then say, what are the right indications, right? So these are some of the things that we are gonna see in the near future. Nasrin Azari (19:04) I like that idea of putting smarts into the devices themselves so that as they capture the data, they can remove stuff that's unimportant so that it doesn't even get caught up in the analytics part, right? And it's data that may or may not even need to be stored, for example. Nandu Vellal (19:23) Exactly. yeah, but it's still like, you know, it's still only for certain techniques. So, you know, it depends on, you know, technique to technique and, know, also the, you know, the level with which the researchers are working on these kinds of, you know, technologies of building the AI to evaluate these things. Nasrin Azari (19:44) Right, right. Yeah, that makes sense. just kind of moving on to our question four, which is around the topic of artificial intelligence. for as AI and automation enhance NDT, how should executives rethink workforce strategy, know, balancing that automation with human expertise? How do we think that this whole process of digital twins will affect that? Nandu Vellal (20:06) Yeah, so I think this was also one of the most discussed topics within the NDE4 conference. So I think the first step is to ensure all the employees are trained in these newer technologies. education is the number one thing that needs to happen. So in terms of whether it's, I think the, one is like the education of the employees, the other is like the education of the executives in terms of like what technologies will really make sense for the companies to implement as part of ND4 and also like how AI and diesel twins can benefit their companies. when it comes to, you know, specifically for employees, I think the key question everybody's trying to ask is like, okay, our robots going to take our jobs, right? So now our AML is going to take our jobs, right? So now these technologies are more like tools where they're good at doing certain specific tasks, right? So If you look at robots, so if you think of repetitive tasks, so they are good at it. Or if you don't want to, if there is a need to inspect an hazardous area, I think that's where robots are good at. so similarly for AI, right? So AI is good at processing massive amounts of data and look for patterns or look for you know, specific anomalies. And what needs to happen is the inspectors and all the, know, NET personnel need to be trained not just in using these technologies, but also in some of the, like, you know, higher level thinking in terms of, like, say, critical thinking or decision making. those areas, we need to train them so that they become kind of like the owners, or they become the ones who tell the AI what to do or tell the robot what to do, right? So I think the other big challenge we're gonna see is how do we trust what the work the AI is doing, right? So I think this is where the explainable AI and those technologies will come into play because You want an inspector who can understand what the AI is telling, right? So I think the key aspect here is, we need to train inspectors and the NDT personnel to understand how to use AI and how to use a lot of these NDE 4.0 technologies for their own benefit. think that's going to be the key because you're always going to see a human at the end of the day is going to make a decision based on what the AI is recommending because we are not there in terms of completely trusting the AI yet. So I don't think in the near future that's going to be there. think there's always, especially with respect to NDT, ND, there'll always be a human in the loop. So I think these are some of the things that the executives need to think about in terms of how to train people to use these tools and technologies effectively. Nasrin Azari (23:10) Yeah, I think in some ways, you know, the future is is a little bit of unknown still, right? I mean, we don't really know how AI is fully going to affect the industry. But the reality is it's important to kind of like you suggested, it's important to stay on top of the tech and the latest technologies, understand how they're being used and what's most effective and what's not. And if, you know, folks are concerned about you know their particular positions or their jobs, having that knowledge is always going to be valuable because there's always going to be something, some work that needs to be done. And a lot of the conversations I've had with other folks are along the same lines that where, you know, AI and robots and automation will take over some of the rote tasks, some of the things that are difficult for humans to do, things that are, you know, very repetitive and non-creative and and kind of difficult for a human being to stay focused on because it's, you know, potentially boring or, you know, not very exciting and then have the basically enable the humans to be able to take the data that comes out of these, you know, technologies and focus them on the basically the higher that the higher value work of analyzing and coming up with recommendations and interpretations and AI can certainly assist with that but like you said at the end of the day it's a human that's going to make the decision not an AI at least as far as we can say today right. Nandu Vellal (24:39) Yeah, exactly. Right. So now also, also one other thing we shouldn't forget is like, know, we need people to, you know, fix, operate and maintain these robots, drones, all of these technologies too. Right. So, and I think we also need to, train NDU personnel because like, you know, the lot of effort goes in terms of like, know, how do you make sure these robots and drones are capturing the right data? there are a of those kinds of challenges exist as well. Nasrin Azari (25:02) Right, right, yeah. And also, writing the software programs that do the analytics, right? mean, only somebody that understands NDT can do the best job at building those algorithms that are eventually going to give us that assessment of a particular anomaly on whether it's okay or whether it's not or what the... or grade it or whatever needs to be done for a particular method. But certainly building and refining those algorithms, that's something that's gonna happen continuously too, either removing or adding bias depending on the situation. So I agree with you, I feel like there's a lot of potential for different types of work to be done, but it certainly is going to change. change how people are working and maybe what their responsibilities will be. Nandu Vellal (25:53) Definitely, yes, yes. Nasrin Azari (25:54) So let's move to our final question of the day, which is what strategies should executives adopt today to ensure that their NDT and digital twin investments remain relevant in a rapidly evolving industry 4.0 landscape? think part of where that question stems is that these investments are not gonna be cheap, right? They're not gonna be inexpensive. So it is going to be important for folks as they take on. And obviously we want to encourage executives to move forward, you know, with as little risk as possible. So I think it's an interesting question and I'd love to hear your take on it. Nandu Vellal (26:33) Yeah, yeah, no, mean, absolutely, absolutely. you know, that's what we work towards. let me start. So now when we think of a digital twin, you know, we look at it as like, so it's a replica of a physical asset and it is giving us, you know, whatever the frequency with which the data is being updated. you know, all the insights of the physical asset, right? Now, if you think about you know how you know this can add value to NDT. We already talked about how we can publish the NDT data and you you can look at it you know from a visualization from predictive analytics and I think one of the things that you know in terms of you know how you know so so so when executives are thinking of making an investment it's not just that you know they built a digital twin and you know they are publishing the data they are looking at this and building predictive model. Now if you think of or an asset, right? So if you think of a lifecycle of an asset, right? So it goes through a stages of inspection, repair, and maintenance. So how can this data that exists in the digital twin can help this inspection, repair, and maintenance cycle, right? So if you think about it, so the data exists in a digital twin. So you take the data and you infer, and then you make a decision. You need to, you use that information to fix whatever the problems or challenges the asset is having. And then you bring the information of the exact physical replica back to the digital twin. Now, that's the process of that sustainment of doing the inspection, repair, and maintenance. Now, that process, if you think of today, It's such a laborious process in terms of looking at all the records, historical records you repair, you maintain. And a lot of times you don't even know what the previous maintenance was done and there are problems and challenges. Now this digital twin is going to streamline all of that. So now you have all that information available in a single place and you can go and look at historically what has happened to it. So that's the number one advantage for such systems. How do you make sure that what you invest in is sustainable or is kind of like, you can live with it in the long run, right? So one of the things our company and Ottawa is working towards is we are a member of a consortium called Alliance for OpenUST. Now to give a little bit of a background, OpenUST stands for Open Universal Scene Descriptions. Now, This is a technology that was developed by Pixar for building animation movies, which typically contains tens of millions of assets in 3D scene. And that technology is being adopted by all the industry leaders like NVIDIA, Ansys, Microsoft, Autodesk, everybody who's trying to build and they're trying to use that technology to build 3D representations of the virtual worlds. Now, one of the things that our company is, is we are a member of the industrial digital twin working group. And what we are trying to do is like, geez, how can we bring in NDT data into digital twins? How can we make it as a, this is a open standard. in the sense like if you represent this in OpenUST that is accessible, you can currently use NVIDIA's tools to view the digital twins, but in the future, you can anticipate that there are other tools will emerge. You are not locked into any particular vendor. And that's, if you think about this kind of a framework and an ecosystem, this is going to be a significant game changer because you're not locked into any particular vendor. You're representing using OpenUST. You're using it for digital twins. And we anticipate this for other digital engineering applications in the future of how you're going to represent this data. Nasrin Azari (30:15) I think it's probably important for folks as they start to investigate different technologies to really think about the future when they do that. So ask questions of the vendor on future compatibility and maybe their use of these open standards like OpenUSD, particularly for digital twins and visualizations because there's no doubt that technology is going to change. and get better over time for sure, it always does. So being able to have that confidence that the vendor is at least focused on that and focused on providing you with the ability, you as a customer, the ability to grow your business without having to throw away a product and buy something brand new and then start over again. my goodness. Yeah, that's, yeah. Nandu Vellal (30:59) Exactly. Right? Yeah, so the investment that you make today, you want it to last as long as possible, right? So, and I think that that is the kind of like the Holy grail for sustainment. you know, I think one other thing that I would like to point is also like, you I suggest for executives to think about in terms of, you know, how they can establish, you know, strategic partnerships with, you know, technology providers, you know, with companies, you know, like cloud infrastructure providers, right? So because I think more and more, this is where things are moving towards, the cost of accessing data and then processing data, communicating, all of those are going to come down. And then fostering an environment of collaboration, innovation and continuous learning within the companies is what is going to take them. in the long run is more helpful in the long run. Nasrin Azari (31:48) Yeah, and that's really true for any technology that, for the listener's sake, for any technology that you're looking to employ in your organization. It's always good if you can find a vendor that is really more of a partner and wants to collaborate with you, wants to work with you, and wants to help you grow your business and use their technology to do that. There's a lot of value in that, I guess. That's kind of what you're saying too, is there's a lot of value in building that strong relationship with the vendor so that you feel confident that your vision is aligned with their vision for their product. Nandu Vellal (32:24) Exactly. Yeah. Nasrin Azari (32:25) Awesome. Well, this was such a great discussion today. I'm really glad to have had you here today, Nandu. And this is our first real deep dive into digital twins. So this was super great. And thank you so much for being here today with us, especially as I know you've been traveling quite a bit lately. So thanks for making time for us today. We really appreciate it. Nandu Vellal (32:47) Thank you so much, Nusreen. I greatly appreciate it. And it was glad to be on this podcast. Nasrin Azari (32:52) Thank you. for the listeners, you'd like to learn more about Nandu's work at N'Orava and their state of the art technology platform for 3D visualizations, visit naurava.com. Again, that's naurava.com. For more information about N'Orava and about Nandu, are there any other places, Nandu, where folks can find you and get information about you? Nandu Vellal (33:18) Yeah, sure, absolutely on LinkedIn. Nasrin Azari (33:20) Awesome. Yeah. And we'll post that when we post our this podcast to or this episode to the podcast website. We'll include some links so that you can get in touch with Nandoo if you're interested in learning more. And I'd like to remind all of our listeners that we welcome feedback as well as nominations for future guests. To do that, you can send a message to us through the contact us form on our website, www.floodlightsoft.com. Thanks again for joining us and see you next time. And that's a wrap for today's discussion. Head on over to our NDE 4.0 podcast page for more interviews like this one and reach out if you have any questions, feedback, or ideas that you'd like to share. Thanks very much and have a great day.
For more expert views on NDE 4.0, subscribe to the Floodlight Software blog at floodlightsoft.com.