32 mins 47 secs | 11 July 2023
Announcer
Welcome to the Decoding Innovation Podcast series brought to you by the EY-Nottingham Spirk Innovation Hub, where we explore the innovative technologies, business models and ideas that are shaping the future of industries. During each episode, Mitali Sharma, a principal in the EY-Parthenon Strategy practice, meets with stakeholders at the cutting edge to discuss innovations in their space, challenges they need to overcome and their outlook on the future.
Mitali Sharma
Hello, I'm the host, Mitali Sharma, and today's topic is industry 4.0. Our guest is Ajay Khaladkar from the Golisano Institute of Sustainability (GIS) from Rochester Institute of Technology (RIT). Hello Ajay, welcome to the show.
Ajay Khaladkar
Hi Sharma. Thanks for having me.
Sharma
Ajay, before we start, would you mind sharing a little bit about your journey so far and what got you to the Center of Excellence in Advanced & Sustainable Manufacturing at Golisano?
Khaladkar
Sure. I started my career as a mechanical engineer and I started working for a large steel company in India back in 2006. Moved to a larger European company after that. I was a project proposal engineer in the material handling equipment division. After that, I was fortunate enough to get 100% graduate merit scholarship to a master’s program in engineering, management and industrial engineering. So came to United States in 2009, completed my master’s degree and then started working for an OEM that was into gear production machinery business. I worked there for around 10 years and until that point I'd gathered enough experience in discrete and continuous manufacturing with significant years dedicated to digitalization. So, the State of New York wanted a program manager to lead an Industry 4.0 Transition Assistance Program at RIT. RIT has several different entities and Golisano Institute for Sustainability has six different applied research centers. And I thought that was a perfect segue into what I wanted to do next. I applied for the role and I began at RIT in 2020 as a program manager for Industry 4.0 Transition Assistance Program.
Sharma
Thank you for that. And that's a perfect segue to talk about, Industry 4.0. So, if you could give us a high-level overview of the progression of the industry from 1.0 to 4.0, basically what are the differences? What is it addressing?
Khaladkar
Sure. Industry 1.0, 2.0, 3.0, 4.0, we made up these milestones. Humanity has this relentless pursuit for improvement and higher degree of comfort. So, we began with the First Revolution which introduced mechanization in general, where we had handcrafted manufacturing before Industry 1.0. We started using steam and basic mechanization in 1.0 and that was back, I think, in 1700s. And that assured in a large change in society. And that was followed quickly by Second Revolution, which was probably 100 years after the first one, which assured in use of moving assembly lines and mass production. That was Industry 2.0, which was followed maybe 50 years after that with Industry 3.0. That had the first advent of robotics, first usage of CNC [computer numerical control] machines, embedding programmable logic controllers within the machinery. So, we did that back in 1960s and now we are into this Fourth Industrial Revolution which is seeing widespread integration of information and communication technology within manufacturing operations. Now one thing to note is maybe we took 1,700 years to get to the First Industrial Revolution. We took 100 years to get to the second one. Maybe 60 to get to the third one, after the second one. So they're getting quicker. Before we know it, we might hit Industry 5.0.
Sharma
So, if you think about the major difference between 3.0 and 4.0, how would you describe that?
Khaladkar
Well, we saw a massive exponential shift in three primary areas between 3.0 and 4.0, commonly known as traders’ law, Butters’ Law and Moore’s Law, where we've doubled the information processing speed every 18 months since 1970s. We've doubled how much data we can transmit every nine months. And we've doubled how much data we can store almost every 13 months. So, Industry 3.0 preceded these exponential changes in my opinion. And once we had this information explosion, Industry 4.0 was natural progression. While digitalization is not new, the scale and reach of digitalization and the second-degree, maybe third-degree, benefits of digitalization are unique to Industry 4.0 in my opinion.
Sharma
And how is 4.0 linked with the broader themes of sustainability, social equity and other similar things?
Khaladkar
It's going to be quite interesting in my opinion. It's hard to predict future here and the prognosis seems a bit hazy to me right now, but we are going to see massive shift in how we are utilizing labor and that's going to be a societal shift as well. Things that are rules-based, if you're doing X after Y after Z, and if you have a set rhythm to your life, that's going to be disrupted for sure. You might be an office worker. You might be an industry worker. Automation is probably going to be disrupting those job profiles. Universal basic income is probably going to be a topic of discussion in years to come. Things like crypto and blockchain are just scratching the surface at this point. They might take a different shape as well. Broader societal themes are quite complicated and hard to predict. There's definitely a huge toolkit at our disposal, to be more sustainable. If you care about sustainability, there are a lot many things that you can do with Industry 4.0 toolkit, like IoT [internet of things] holds a lot of promise for waste reduction, for enabling new cases, for remanufacturing and reuse. It's almost like what can you dream of. If you have a goal of becoming more sustainable, Industry 4.0 gives you a lever to push that.
Sharma
And we'll come back to that in a minute. If we could shift gears a little bit and talk about the center of excellence. How did the idea come about?
Khaladkar
I believe the Empire State Development (ESD) Fund was the driving entity for these centers of excellence. I work for Center of Excellence in Advancement & Sustainable Manufacturing. The ecosystem is massive. I think they have a portfolio of more than 70 entities with over US$55 million in funding for all of these entities. And they include centers of excellence, centers for applied technology, manufacturing extension partnerships. I think our center was established back in 2012. Since then, we've done quite well and we've had significant impact on different sectors within the state. So, all projects are subsidized by state money. We have done projects that are 100% funded by the state money and there have been projects that have been 100% funded by the company.
Sharma
What's the criteria for selecting a company?
Khaladkar
For our program, generally anyone that is based in the State of New York can avail the funding. The more recent grant from Economic Development Administration (EDA) specifically states that business needs to be small- or medium-sized under US$500 million to receive that EDS assistance. Depending on where the money is coming from, the criteria might differ, but anyone from the State of New York is welcome to avail our assistance.
Sharma
And could you talk a little bit about the kind of programs you've been involved in?
Khaladkar
Typically, my time is spent these days, some for the state program funded by the ESD for Industry 4.0 transition assistance. The EDA program is on similar lines and the third program is by MxD (Manufacturing x Digital), previously known as Digital Manufacturing and Digital Innovation Institute. The nature of projects varies quite a bit based on what company we are working with. So, there are companies that don't even have a simple enterprise resource planning system. We might go and do a material information flow analysis to see where the digitalization opportunities lie. There are OEMs in our area that have a ton of data that is generated through their machines, but they don't have a greater value proposition that can have all that data into the package. That's a separate add-on entity on their machines.
So, we've worked with some companies like that. We are working with a few continuous manufacturing kind of plants and chemical in food and beverage industry who do not have things that you would take granted for, like SCADA system, for example, something these companies need to have so that they can control their lines effectively, but they don't have that level of digitalization. So, we work with them to understand where those opportunities lie, what can they expect as a return once they do that investment. It's quite a wide variety of projects under the EDA and the ESD program. MxD project is more focused, so that's related to developing a web application that can help companies create a digital maturity assessment on their own and then define their own roadmap. So, it's the Q&A session that can spit out a digital investment strategy for any small- or medium-sized company out there. So those are three areas roughly that I've been focusing on.
Sharma
What's definitionally a small company and what's medium? I mean, you don't have to give me exact numbers, but what would you describe as small company because the definition changes?
Khaladkar
I agree it should be simple, right? But it's not. So, I define medium-sized company as anyone who has top-line revenue between US$100 million and a US$500 million range, small-sized companies, anybody below that. But our program, if you go by the textbook definition, takes into account things defined by Small Business Administration (SBA) and those guys have different thresholds for number of employees, the top-line revenue, etc. And it can wildly differ, like a semiconductor business can have a different threshold compared to food and beverage industry.
Sharma
Great, but these are all established companies. Do you also work with startups?
Khaladkar
We do. We support them for their hardware scale-up program. So we work with startups as well, especially startups that have a value proposition in the Industry 4.0 area. They're making something new that can help our manufacturers implement Industry 4.0. We are always keen on working with them.
Sharma
So when you're working with different levels of these companies, obviously funding is different, right? So how do you think through return on investment that you talked about a little bit before and also your own internal metrics of success?
Khaladkar
It's tricky because we spoke about Industry 3.0, 2.0 and 1.0, and the return on investment with those three revolutions was, in my opinion, a little simpler than Industry 4.0. There was ROI on the replacement. You're replacing a human being with certain technology and you could find out pretty quickly that what is the equivalent investment you need to make for replacement. But this particular thing gets more tricky because you are talking about a return on investment for empowerment. If you have new and advanced tools for your leadership and you are empowering them with things that are going to help them make more reliable decisions in a more agile way, it becomes tricky to calculate return on investment. But most of the leaders understand that the world is getting more volatile, uncertain, complex and ambiguous, and Industry 4.0 is going to help them navigate that world better. Typical discounted cash flow models kind of fall apart sometimes when you think about these things. The other tricky part with ROI is you can fall pretty quickly into the trap of shiny objects if you don't have a cohesive digital strategy. You put the tool before the problem and implement the tool expecting that it's going to solve certain business problem. That's in my opinion the wrong way to go and might not always yield results.
I guess lastly, the third thing I mentioned, the cohesive digital manufacturing strategy, Industry 4.0 seems to be different compared to the previous industrial revolutions because it has potential to unlock second- and third-degree benefits. Just simply being digital gives you significant benefit, but being integrated and not having silos within your organization, and enabling information sharing between different functions can help you be a lot more productive and pave your path for being more autonomous. In terms of our internal metrics, there's obviously program-related metrics around how well are we doing with budget, how well we are doing with on-time delivery project completion. The metrics that we generally report to ESD relate to new jobs created within the economy, how many jobs we have retained, how much cost savings are our customers reporting, how much new revenue are they reporting as a direct result of our engagements — those four generally cover the areas that we report on.
Sharma
So, this is interesting. You said jobs created. Generally, people think of industrialization, especially Industry 4.0 with automation, robotics, which means job loss. Can you talk a little more on that piece of it?
Khaladkar
Yeah, they seem to be a bit conflicted, right? Revenue generated and cost savings can be jobs lost. But typically, the people that we have helped with on automation and IoT side have added jobs because things get more complicated, and they need more qualified people in the right areas to do those jobs. So, they have taken an approach of retraining the existing employees that show the inclination for a different kind of job and repurposing their existing staff and appropriate location. And generally, thus far we haven't really seen people firing employees and replacing them with robots. It's just different kind of jobs for people. So, there is a lot of retraining activity. We just started working on a tool called GIS Right Select which is a compilation of different resources for retraining people and paving career paths for people with certain kind of aptitude toward those technologies. That generally has been our experience.
Sharma
Could you also talk about other criteria or things that the center of excellence is uniquely positioned to provide to small, medium-sized companies?
Khaladkar
Yeah. Usually there is a lot of cross-reference between the research centers. They always want to build the talent pipeline. We see the medium-sized companies especially have significant shortage of qualified labor. We being a university, we generally deploy co-op students or interns on these projects and they have gone on to be full-time employees with the companies that they have worked with or full-time employees within GIS.
Sharma
And as you think about the adoption rate within again small- and medium-sized companies because that's who you're working with, what has surprised you in terms of who's more willing to adapt and what kind of solutions are easier to be implemented?
Khaladkar
If we look at manufacturing as a continuum, you know on one side you have engineered to order or discrete manufacturers. They might be engineered to order, make to order or make to stock. And then on the other side you have continuous manufacturing, things like chemicals and paints and food and beverage. The problems differ quite a bit when you go left to right on that spectrum. If you are a continuous manufacturer of your equipment uptime, your equipment OEE [overall equipment effectiveness] becomes a lot more important if you're on the far left on that spectrum. If you're a discrete manufacturer with engineer-to-order kind of business model, you're always struggling for on-time delivery and the right estimates. So, discrete and continuous show different traits, but overall, I've seen that pharmaceutical companies, companies in chemical industry, food and beverage are a little bit ahead of the curve when it comes to Industry 4.0 adoption, probably because anything related to equipment downtime is extremely expensive in these industries. And discrete manufacturers, I've seen, are a little behind when it comes to adopting IoT or analytics or even manufacturing execution systems. They do have computer-aided design (CAD) packages, maybe product lifecycle management (PLM) solutions, but there is a lot of opportunity in discrete manufacturing for Industry 4.0 implementation.
Sharma
Most of the softwares that are built are done for the bigger implementation, bigger companies. How does one scale it down and make it cost-effective?
Khaladkar
The solution seems to be extremely busy. We haven't encountered a situation where we cannot match a customer with the right solution. So, if you're talking about leading ERP systems, even larger names have products that are suited for small- and medium-sized businesses. If you look at some of the manufacturing execution systems, the IIoT [industrial internet of things] platforms seem to have some players that are targeting the niche market for small- and medium-sized business. Scaling things down and making it more affordable for small or medium-sized business is going to be honestly a bit cumbersome for larger players. We haven't seen people being too responsive to (small and medium-sized enterprise) SME needs because a) as we mentioned earlier on that they tend to be price-conscious, and they just don't have the wherewithal to implement all those systems with broader connection in mind.
Sharma
Generally, the solutions are geared toward big companies, so how do you scale them down? You referenced that these are small, medium companies with constraints on their resources. How do you get the solution to be viable for them?
Khaladkar
I can give you an example. We recently started a project with a semiconductor company that makes equipment related to gas leak detection and other allied areas. This is a US$300 million company. They have a problem of having an archaic electronic work instruction solution. So, they do have PDFs and work documents, but they don't have a cohesive nature of that solution that ties in with the engineering change orders or their care solutions or PLM solution. We started scouting for a solution and there are modular options in the market. You don't have to buy the whole package, but you can buy modules of that package. You can just buy one that you need and that modularization has proven quite effective for solution placement within these companies. The other thing that I mentioned previously is we haven't encountered a situation where we do not have a solution for a company that is small or medium-sized just because it's priced for larger company. We don't see larger companies being more responsive to the small and medium-sized businesses because honestly it's not worth their time probably to make changes to their solution, but they do have a different solution. The other part is if you look at IIoT, for example, I mentioned customers don't have SCADA systems all the time.
If you want data from your actual shop floor, if you have say six different production lines and you need data from your shop floor, but the other way is to build that stack yourself to go for low-cost sensors. I can name a few websites that can give you wireless sensors. And build that whole stack of unified architecture for drawing information from that production line yourself. You could have your own messaging brokers, you could have your own time series database, you could have an open-source version of those things so that you can prove value to your management. And show some analytics related to equipment uptime after implementing a pilot project. There are a few open-source technologies that can help you do that. So, I guess long-winded answer is modularization and some open-source solutions that can help you prove the value has been our approach.
Sharma
Interesting. And so when you are working with open source and modules, is there a lot of integration work that you have to do to make sure that everything works?
Khaladkar
Yes. Great question, because once you develop your own stack, you are liable to maintain it, so there is a trade-off there. You don't spend much money, but you do need qualified people who can understand the nuances of the information stack; and can not only just maintain that information stack, but if you tomorrow add a software system or an asset on your shop floor, you need to integrate that within that information architecture. And internal employees are going to have the responsibility of doing that. If you are going with a larger company, you have some kind of maintenance contract with them.
Sharma
When you work with companies which might be in different stages of development or history, is there a lot of retrofit involved on the shop floor and how do you deal with that?
Khaladkar
People don't want to throw out assets that are working just fine just so that they can have an IIoT solution. So, we do look at the native capabilities for asset if they have an RS232 serial communication or they can at least communicate certain basic information with larger information IIoT solution. There is a lot of retrofitting, especially Edge devices can help you translate some of the native protocols to something that your analytic solution can understand. So short answer, yes.
Sharma
Give us more understanding of Edge devices.
Khaladkar
The ones that we have experimented with can go on the machine that you want to draw some data from and it can give you certain number of tags. And sometimes, they do have the native capability of translating the information that is coming from your machine. So, there is a huge debate on what kind of communication protocol is best for your information architectures. Older machines seem to speak Modbus and CIP and there are plenty of machines out there with serial communication ports, but the debate seems to be between three different communication protocols. One is MTConnect, the other is OPC UA, unified architecture, and third is MQTT. I believe MQTT stands for message queue telemetry transport. What languages are your operational technology (OT) assets speaking? How do you translate them so that your IT side or your management can quickly glean information from them? These Edge devices can help you basically allocate an IP address for machines that were not on the network previously, bring them on to establish connectivity, they can help you establish interoperability. They can give you basic security-related things. They can make the communication more real time and ultimately, you know, break down that silo that this information is going to stay with this asset and you can bring in that information to a unified namespace that can be used by other applications. It's a hub and spoke kind of architecture where each node is generating information and each node can consume information that it needs to do its job. And Edge devices is something that can help your dumb machines or older machines be online, export information in some cases, also get information.
Sharma
As you're working with different sizes of companies and in their journey toward Industry 4.0, what should be their expectation? How long would it take them? What are the milestones they should be ready mentally to think about as they start on the journey?
Khaladkar
Firstly, it's not a project, it's a strategy, right? So, it's not going to end anytime soon. Our recommendation is if you do a three to five-year roadmap and if you're embarking upon that journey, different technologies can come to fruition on different timescales. Connectivity is one that seems to have a bit of a longer lifecycle compared to automation. So, if you want to automate something, this cycle seems, in my opinion, a bit shorter than connectivity and analytics packets. If you have no connectivity whatsoever establishing a network Wi-Fi, either with some basic firewall between IT-OT, step number one, you're not going to see ROI immediately after that. Step number two is getting digital and drawing some data from your equipment. Step number three is integrating that. Once you start integrating, that's when you're probably going to see some benefit on the key performance indicators you care about, like uptime or quality that in my opinion depends on the talent that you have in-house and the right partners that are helping you with it. It's going to take anywhere between six months to a year to see some initial benefit of IIoT implementation.
Analytics implementation can go hand in hand with that you choose an analytic solution that can give you information, can help you analyze what-if kind of scenarios. It heavily depends on how in tune your management and your supervision is with these solutions and the capabilities they bring to the table.
Initial benefits for IIoT and analytics probably six months to 12 months from implementation. Software systems can be a bit tricky to implement, especially if you are replacing an older version with a newer version. You need to migrate all your data, have plans for workflow reassessment if you are implementing a newer solution. These software implementation lifecycles can be six months to 12 months and your benefits are probably going to take a little longer than that.
Probably automation is going to be three months, IIoT is going to be eight months to 12 months, analytics is 8 months to 12 months, software systems is probably 12 months to 24 months.
Sharma
As we wrap up, could you sort of talk about things that have surprised you as you've worked in this journey over the past few years in a good way or a bad way?
Khaladkar
I'm surprised at how little data that we gather is used for decision-making. Probably an average manufacturing company generates maybe around one to three terabytes of data every day. Maybe less than 2% of it is used for decision-making. That was a surprise for me. We have a long way to go before we have every company in US doing data-driven decision-making.
Sharma
That's interesting. I'd love to double-click on that. So, could you give me an example and talk more about the kind of data that is currently being used and what's being generated by an average company?
Khaladkar
So your systems of record, your maiden software systems, your CAD, CAM, PLM, ERP, MES, QMS, manual, logs, Excel sheets generate significant amount of data, but very rarely do we find integration between these systems. So what is generated by a system stays within that system. Secondly, OT assets are not connected. So I previously mentioned like you might have an asset that has RS232 that can communicate some data. I don't think people are utilizing that data even to calculate basic things like overall equipment effectiveness, yet those KPIs tend to be connected to their bottom line, very closely. There's that aspect. The other aspect that we haven't really touched upon is the broader supply chain. If you have many different suppliers and your on-time delivery is predicated upon your broader supply chain, the inventory levels that they have, you have the capability of having a unified supplier portal that can have horizontal integration capabilities. So, in an ideal world, you want that data exchange seamless and frictionless. I don't think people collect that data, or if they collect, act upon that data.
Sharma
So, there's two sets of problems. One is collecting the right data and then using that right data and decision-making. So, if you were to qualify those two, which one is underutilized?
Khaladkar
I would say the existing data is definitely underutilized. You can go from three terabytes to 30 terabytes pretty quick, but without utilizing the right kind of data for making decisions. So, every time a manufacturing leader goes through a crisis situation, they have to go through this OODA cycle: observe, orient, decide and act. So, you need to observe what is happening. So my HVAC system is going down. How do I safeguard my assets? My key supplier has reported a shutdown because of certain pandemic. How do I ensure that my customer orders are fulfilled? They have to observe what is going on in the field, they have to orient themselves, they have to make a decision and then they have to act on the decision. And many times, when they are in the action mode, their decisions might change. All in all, Industry 4.0 enables the faster and more agile response to the crisis situation. This whole OODA loop concept has been something that I've used on and off to sell Industry 4.0 to small- and medium-sized companies that this is what is going to enable improvements on your shop floor. We previously spoke this is ROI and empowerment, not an ROI and replacement. Without data, none of this is going to happen. Using the data that you collect today is probably the first step to faster OODA cycles compared to previous times.
Sharma
Well, thank you, Ajay, for a very interesting conversation and good luck with everything that you're doing.
Khaladkar
Thank you so much, Mitali. It was a pleasure talking to you. Thanks for the opportunity.
Announcer
The Decoding Innovation podcast series is a limited production of the EY-Nottingham Spirk Innovation Hub, based in Cleveland, Ohio. For more information, visit our website at ey.com/decodinginnovation. If you enjoyed this podcast, please subscribe. Leave a review wherever you get your podcasts and be sure to spread the word.
The views of third parties set out in this podcast are not necessarily the views of the global EY organization or its member firms. Moreover, they should be seen in the context of the time they were made.
