The Evolution of Industrial Data: From Sensors to Strategy - with Vlad Romanov
Show notes
Welcome to the Trend Detection podcast, brought to you by Senseye Predictive Maintenance – which gives you visibility and insights into all your assets, from single machines to full plants to help you reduce downtime, increase knowledge sharing and accelerate digital transformation across your organization.
In this episode, we're joined by Vlad Romanov, an industrial automation and data integration specialist with experience spanning plant‑floor engineering, systems integration, and enterprise strategy, who shares a practical view on how industrial data moves from machines to board‑level decisions:
What industrial data really is, starting at sensors and control systems on the plant floor and evolving into decision‑ready information used across SCADA, MES, and enterprise systems.
How data flows from machines to strategy, explaining the progression from standalone equipment, to production lines, to site‑wide and multi‑site performance insights.
Why digitalisation has accelerated in recent years, particularly post‑COVID, as manufacturers needed remote visibility, faster decision‑making, and more resilient operations.
The reality of IT/OT integration, including cultural differences, conflicting priorities, and why alignment and over‑communication matter more than technology alone.
Where AI and machine learning add value today—and where they don’t yet, highlighting realistic use cases such as analysis support, infrastructure modernisation, and decision assistance rather than full autonomy.
What separates successful data initiatives from failed ones, including mindset, patience, iterative improvement, and the willingness to modernise legacy infrastructure step by step.
You can find out more about how Senseye Predictive Maintenance can reduce unplanned downtime and contribute towards improved sustainability within your manufacturing plants, by visiting: www.siemens.com/senseye-predictive-maintenance
Connect with Vlad on LinkedIn:
https://www.linkedin.com/in/vladromanov/
Transcript
The industrial data starts at the plant floor. And so, again, we can obviously spend a lot of time discussing the, the different formats, but you have your analog and digital data that is collected by your machine, then it is usually sent to your PLCs for control. Systems purposes, right? And so this is where we can have like, I guess like deeper conversations as to what is control system data, what is data that will be used to make decisions, but ultimately, it gets, um, it gets used primarily for controlling the equipment. Welcome to the trend detection podcast powered by Siemens, and now your host, Nel Sullivan. Hello and welcome to the latest episode of the Trend detection podcast. I'm really pleased to be joined by, uh, Vlad Rinov, um, and we're gonna be talking today about industrial data, um, from sensors to strategy. So it's a, it's a really interesting topic. I mean, we've covered sensors in detail, we've covered strategy in detail, but maybe not in between as much, so I'm really excited for this, um, session today. Um, I'm gonna make a confession to start with, because we have already had this conversation. Um, Vlad has very kindly agreed to re-record. We had a bit of a technical issue where the recording corrupted or, you know, didn't, didn't come through properly. So unfortunately, our fantastic conversation we had before is lost forever, but I'm sure we'll do a better job this time either way, Vlad. So welcome back and thank you for your, for your patience. No, absolutely, thank you, thank you so much for having me. I think it's an important conversation to have in our industry, so. Um, happy to be back and, uh, let's make it happen. Fantastic, fantastic. So, I mean, that's, you know, right at the start, I'd be, it'd be great if, um, we could find out a little bit more about you, lads, your role, what kind of things you're working on at the minute and all of that good stuff. Sure, so formal education in electrical engineering as well as an MBA from Canadian universities. I have spent over a decade of my career in manufacturing and industrial automation settings. I worked for very large end users to start my career, Procter and Gamble, Kraft Heinz, Post Holdings, again, very plant floor heavy, so, you know, setting up machines, programming PLCs, HMIs, Scata platforms before making a transition to a system. Integrator where I learned a lot more about industrial data, what it takes to create enterprise networks, what it takes to, you know, extract, basically store and synthesize information for various stakeholders in manufacturing, and we'll get into some of those details a little bit later. Um, I've been in both, like I said, very large environments, but also in startups. I spent a couple of years building a DevOps solution. So I've also gained an appreciation not only um again from my earlier years, spending a lot more time on the OT side but also learning about IT in a lot more depth than I would say a lot of the manufacturing teams have the capability, at least in my experience and so currently I work on a variety of different projects. I still do a lot of very traditional integration and sort of modernization. Uh, type of initiatives, you know, pretty much starting with a plant assessment and then going through projects of upgrading whether that's PLC systems, servers, networks, tying in all of that data back into an on-prem database or a cloud database and then obviously funneling that into a scale MES type of a solution that allows. Basically manufacturing personnel to to drive better decisions. Um, I also have a few other projects, um, you know, on the consulting side and private equity type of due diligence work again based on my financial and business knowledge gained primarily from the MBA, um, again, happy to, to discuss any and all of those, uh, angles with you today. Fantastic, and thank you for such a comprehensive introduction. I think we might need about 3 podcasts to cover all of that, but no, we'll see, we'll see how we go, but in a good way, I mean, so you'll be definitely invited back again. But, um, but, I mean, let's start at the basics, and this might seem quite basic to people listening, maybe, because we're gonna, I'm gonna sort of ask you to start by defining industrial data, which everyone may have their own view or opinion on, but from obviously your, your perspective, but also that, how that's sort of changed over the past sort of 10 years or so. Yeah, so starting with, I guess, the plant floor, right, uh, again, those that operate machines will usually notice the sensors that are present for detecting various, I would say, materials on the conveyor belts. Inside the machines during the transformation, so the industrial data starts at the plant floor. And so again, we can obviously spend a lot of time discussing the, the different formats, but you have your analog and digital data that is collected by your machine. Then it is usually sent to your PLCs for control systems purposes, right? And so this is where we can have like, I guess like deeper conversations as to what is control system data, what is data that will be used to make decisions, but ultimately, it gets, um, it gets used primarily for controlling the equipment. Secondarily, it is used to to be funneled, as I mentioned, to make decisions, and that can be your, again, your Edge IPC, your Edge solution that will synthesize or I guess collect a portion of that control system data, store it for historical. Purposes, uh, draw certain, I would say, maybe definitions based on that information, add time stamps, right? So usually your control systems are going to be cyclical, whereas you want to make decisions based on a time period. So you need to create. Um, more information than what the control systems generally provide. And so if I want to take a step back maybe and define what is industrial data, I would say that it is the umbrella term used for any and all information that will ultimately touch or have an impact on the plant floor, but me as La, I would take that a step further and say that there is maybe control system data that is used to run the actual equipment. And there is data that is used to make decisions, and we can get into the distinctions of both, but ultimately they will be from a technical and a business perspective, uh, somewhat different. Um, to your question on how do things change, well, I would say that there was always an appetite for data. I think that, you know, scata systems have existed for many years and decades. And in a very broad sense, every manufacturer has always captured the information on how productive they are, right? And so in the early years, that would be maybe using a clipboard or a whiteboard, so a lot of times you go to a plant, they would have sort of a sensor at the end of the line. They would read how many cases are coming through in let's say an hour or a shift. They would write that down on the whiteboard. I think that the major change, of course, and I would say since COVID because everyone wanted to work more remotely or couldn't be at the facilities, everyone is trying to get as digital as possible, right? So now we see a lot more implementations of SCATA systems that are properly. I would say digitalized, a lot of um MES types of application manufacturing execution systems that allow you to again take that a step further and provide insights. Um, so industrial data, like I said, was always there in a general sense. It's just been a lot more digitized, I would say in the last probably 6 or so years at this point. And I guess maybe let's start there, you know, you know, you know, why has the, you know, digitizing started in the last 6 years, cause I mean that seems quite recent to me actually, but is there a reason why, why then it's the right time? Maybe it's a difficult one to answer, but. Yeah, so, so, so look, let me, I guess, take a step back, right? So I think in our industry there's quite a wide gap between sort of the cutting edge and bleeding edge, if you will, companies versus I would say the medium to small manufacturers, right? So it's not to say that there was never digital solutions. I just, I guess from what I've witnessed. During the time of COVID, at least, so I've been to, uh, as a result of COVID to many facilities during those two years, right? And of course we can talk about sort of the career shifts that people have made, we can talk about the fact that, you know, it was very difficult to get inside of a plant without proper clearance, proper checks. Again, they were maybe looser or more restrictive depending on the region and the world that you were in, but ultimately, um. Manufacturing needed to keep going, right? You had no choice but to manufacture, and I do a lot of uh food and beverage and CPG type of facilities, so as you can imagine, there was a lot of pressure to continue working while all of these shutdowns are happening. So of course personnel needed to be able to eva. Evaluate how well are we producing, where are the deficiencies, uh, where do we invest our Capital, right? So there were a lot of, there was a lot of pressure to make those decisions, um, and it was more difficult to access data that was manually created. So therefore I believe we saw this sort of roll up from. Uh, medium, at least manufacturers that didn't have or didn't have a big incentive to have a lot of those digital solutions to roll them out. Does that make sense? No, it does, it does. I think, I think it leads, it leads quite nicely onto sort of my next question, and it, and actually what you were talking about before. So you talked about, you know, different, you know, the flow of data from the shop floor, from sensors, you know, all the way up to sort of strategic or enterprise level decision making. But could, I mean, could you describe that in a bit more detail, because it seems quite, you know. Someone like me, maybe a little bit on the outside, a little disconnected, right, you know, you've got data here, which at the end of the day should inform strategic decisions at board level, for example. So how, sure, so I, I like to use, I guess, like props maybe a little bit to, to answer that question, right? And you see behind me there's, there's a couple of PLCs, there's one from Siemens, Alan Bradley, there's some. IPC from Siemens, there's basically a non-industrial computer, right, and so to paint you a clear picture of an architecture, and let's start maybe with a simple example. So when you have a stand-alone machine, You will typically have your sensors tied to a PLC and then that PLC talks to an HMI, right? So that data in a, in a sense is contained without that piece of equipment, right? And of course when I say a machine, that could be as simple as a, as a palletizing machine, a case packing machine, but it could be an entire process, right? So for some facilities, you will have a larger sort of quote unquote machine or. Area that is going to be controlled by that standalone type of system. And so in that case, your data is basically going from your sensors that are your outputs and inputs on the PLC side, and it's controlling your motors, it's controlling your different valves, it's actuators, there's going to be specialty modules, right, so your data is basically contained within that system. As you start basically growing those operations, you're looking to make decisions on a slightly higher scale, and as I mentioned before, you're looking to control more pieces of equipment, but you're also looking to understand more pieces of equipment, right? So this is where, and again there's a lot. of debates in our industry between let's say SCAta function versus MES functions, but basically picture a production line where you have a sequential step of processes where you want to see what is the filler doing, what is the case backer doing, what is the palletizer doing, and you're trying to understand. Um, at the end of the day you're trying to understand how can we better the throughput of the finished product, right, because if we take a step back, what is paying the, the salaries and the. Um, how would I say it, the making the incentives for any of these types of projects, it's obviously product created by the manufacturer, right? And so when we talk about data, we're trying to understand, well, where is first and foremost the bottleneck on our line? Why is that a bottleneck? Is it because it's going down? Is it because it's breaking? Is it because, you know, the operator is not loading the raw materials fast enough, so data allows you to start understanding that from an enterprise level, right? And so as we take another layer up right when we again talk about plant maturity, now you're trying to understand not only a single line, you're trying to understand the entire facility. So depending on whom you visit, they may have an area that runs a specific SKU that might be a lot more critical for that specific business unit, right? And so now you're trying to understand where do we invest. Do we invest in this specific line that runs SKUW X, Y, Z, or do we invest in a secondary line. That runs perhaps some different products and so then you're starting to again funnel that data from the plant floor into these, I would say more cloud native but ultimately more consolidated types of solutions where you're not looking at the sensor on the machine, you're now looking at how is my line performing, how is this side doing, right? So you're starting to have a much more higher level view of your entire operation on a multi-site basis. And I guess another sort of aspect of this, with all those different sort of data sources, what we're talking about is, you know, data integration between IT and OT systems, which is, of course, always sort of top of mind during sort of these large, um, well, large and small, um, projects, but, you know, how can organizations sort of, how should they approach that and, you know, what are some of the common pitfalls around integrating those different data sources? That is, that is the very difficult question, right, uh, for, for, for many years I've had conversations on, on both sides. And I simply believe that there is a difference in in and it's a generalization, right? There's generally a difference in KPIs associated for both teams, and you will generally see, again, engineers on the IT side have a different perspective on what's important versus the OT side. And there's not necessarily anything wrong with that. I just think that there's a, There's generally a need to orchestrate so that they don't clash during the implementation, right? And what I mean by that is the OT side usually prioritizes, like I've mentioned a little bit earlier, product out the door. So if you've ever been. In any commissioning project, you will very quickly find out that there's a very strict timeline, we only have access to, let's say the line being down for a very specific number of hours or days to implement whatever change, and you have a lot of pressure to make sure that product gets made. So of course you're going to not necessarily cut corners, but you're going to do whatever is necessary to make sure that equipment runs. On the IT side, the priorities are different for again different reasons, but usually they are trained to mitigate risk from external cyberattacks or ransomware. They're, you know, they're used to manage access to those systems. They want to make sure that everything is secure on the data side. They want. To audit, you know, the protocols that allow you to pass different data and so of course the two priorities need to align in a project that's going to touch both sides. So usually again going back to the point of data, if it's a stand-alone machine that gets deployed on the plant floor and it doesn't need to communicate. Anything outside of that environment, usually OT will be more than capable of handling that project, but as soon as it touches yoursCATA and MES layers, and again there's going to be, of course, a lot of auxiliary things, it could be on the quality side, it could be on the laboratory testing side, right, so there's going to be a lot of sort of like gray areas as well. IT needs. Get involved, right? And so usually they will provision hardware, so your servers, they will provision your virtual machines upon which, you know, a lot of the software is going to run, and so they primarily from their standpoint, they don't need to understand what the data delivers to the plant floor, but they need to understand how it's going to be secure, who, who's going to have access. Um, a lot of times, who is going to from the outside be allowed to remote in, you know, and make changes. How are they going to make those changes? How are we going to, uh, I've, I've been through these conversations many times, but who's going to upgrade your Microsoft. You know, directory service, uh, who's going to upgrade the server instances, so there's going to be a lot of questions that again concern IT, whereas again OT is usually not going to be trained on the server architecture of a facility. And I guess in amongst all that, because of course there's a lot of technical considerations, but what I sort of noted down, I was writing down there is more around that, you know, the, there's a lot of cultural differences, you know, maybe differences, but approaches, I'm gonna say, because neither approach is wrong, right, but it's just different mindsets, different goals, and I guess it's trying to align some of those goals together, I guess. I, I think for me it's um to be honest with you, over over communicate your objectives, right? So like I said, I believe that both groups are trained in to some extent like different disciplines ultimately they do have an aligned objective for the business, right? They are trying to make sure that the business will continue to function as expected. I think it's just maybe a misunderstanding of what the other is trying to do that scares the other group. Right, so if I'm putting on my OT boots and I'm just trying to funnel data, obviously I'm trying to understand how the machines are running and I'm trying to, you know, basically have a good overview of the, the plant floor, but that scares the IT team if I don't explicitly. Uh, tell them what I'm looking to have access to, how it's going to be stored, right, and having those conversations is going to help solidify some of that understanding. Similarly, if I'm on the, um, again on the IT side, I need to be very clear. What does it mean to have access, have permissions? How do I access different layers of the servers, because again, people on the OT side usually are not trained there, and they get worried when, you know, it becomes way too troublesome to support equipment that, you know, was deployed in a facility 30 years ago and requires a parallel port to connect to, and the IT is revoking that access, right? So I think it's being very explicit. Um, on both sides in those conversations that can and sometimes does solve some of the conflicts. And, and do you think that that relationship is maybe stronger compared to previous years or, you know, I know it's a general question cos it differs across organizations, industries, etc. but is that, do you feel that's stronger for those reasons you, you mentioned there? To be honest with you, I wish I could say yes, um, but what I see is I believe that the technical complexity is increasing on both sides, right, and we, we can talk as to, uh, as to why that is maybe in a in a separate question, but I believe that there's a technical complexity increase in both OT as well as IT environments. And at least from what I have seen, the projects are having a harder time now because of that, uh, sort of like mesh well, uh, between the two teams. Exactly. …