In this episode of the Digital CxO Leadership Insights video series, Mike Vizard interviews Ara Surenian, vice president of product management for Plex, a unit of Rockwell Automation, about how digital twins can be used to solve supply chain issues.
Mike Vizard: Hello, and welcome to the latest edition of the digital CxO Leadership Insights video. I’m your host Mike Vizard. Today we’re with Ara Surenian who’s vice president of product management for Plex. They’re part of Rockwell, and they focus on ERP and supply chains and all kinds of good stuff. And we’re gonna be talking about digital twins. Ara, welcome to the show.
Ara Surenian: All right, thanks, Mike. Good to be here.
Mike Vizard: We have seen these massive disruptions to supply chains, but it’s not abundantly clear to me how people might apply digital twins technology to kinda help manage those supply chains better. So, what is the opportunity here to do something new and interesting?
Ara Surenian: Yeah. So, as the name sort of, it’s within the name, the idea is to use technology to mimic, create a twin of what your physical world is. So, your machines, your supply chain processes, your customer behavior, your supplier behavior and so on, right? It’s using digital methods, modeling AI/ML to mimic the real world, and use that to project what could happen and all the alternates in terms of how best to deal with disruptions and issues.
So, it’s about collecting as much information as possible in order to determine different outcome strategies and tactics to meet potential challenges the business is facing. And in today’s world where it seems every day there’s a new surprise, having that added insight to mimic your real world and make decisions based on the information and the data that’s collected is highly valuable and important.
Mike Vizard: How hard is it to set something up like that? ‘Cause I think a lot of folks would assume that this is super complex and requires a lot of horsepower and expertise. So, what’s exactly involved?
Ara Surenian: Yeah. I’m not going to sugar coat it. It isn’t easy. It first starts with a keen understanding of your current state and modeling out your current state. So, getting in, pulling in the data, making sure that data is accurate is critically important, that if you’re modeling out your production process, that your machine data, your routings, your bombs, your rates, your set up, all of that is accurate. And then building it out in terms of the chain of events that need to occur to ensure that when you run the models that the inputs and the outputs make sense.
So, it does take time to set up because you’re really trying to ensure all the data that’s coming in is accurate and the models you’re building are also represent that reality. So, it does require a level of expertise to build those models, and the understanding when the data is incorrect, how best to address it and correct those anomalies as appropriate.
Mike Vizard: Do you think people, once they build this, will also be able to engage more in what-if scenarios and what are the implications of being able to do that?
Ara Surenian: Yeah. Imagine you have a production line, for example, that may have challenges in maintaining up time, right? Older machines could break down or don’t run at rate. You can model that and you can say, understand the impact that’s going to have on your throughput, your ability to produce based on the demand requirements. And then from there, make assumptions about, well, what if we move this production to a different line?
Or what if we change this particular machine or work center to run 24 hours a day? And so on. How do we accommodate different performances between shifts, so second shift doesn’t run as efficiently as the first shift and so on? And you’re able to model the what-ifs that looks at different scenarios, looks at different options, look at outsourcing versus insourcing, increasing over time, looking at alternative work centers to produce the goods.
You can also look at market conditions, right? What if there’s a disruption in your supply chain? What alternate supply could be available? What impact will that have on the business? So, just like any model, once you build a model, you can go through the what-if scenarios.
Classically that was always done like in Excel spreadsheets, so you create these really complex Excel spreadsheets and you plug in the different variables and you see the impact of that in terms of revenue, cost of goods sold, margin and so on. A digital twin does that at basically at a supercharged level but focused on the different components within the business. The entire supply chain from customers through suppliers.
Mike Vizard: How smart can all this get? Can I start throwing machine learning algorithms at this and have some AI capabilities that will tell me, hey, here are three things that are likely to get you fired if you don’t fix them?
Ara Surenian: Yeah, absolutely. So, like any model, if you can now throw AI/ML into it to add a level of smarts, for a lack of a better term, on the process; you’re going to get deeper insight into what’s going on. So, you know, an area that we’re working in really heavily right now is in the realm of forecasting. We’re applying machine learning to the forecasting process and the more data we’re able to collect, the more appropriate related data and other types of data that feeds into the model, the output becomes more accurate and the actual projection becomes smarter because you’re adding all this added insight that allows the engine to produce a much more accurate result.
So, that’s the beauty of machine learning is understanding and being able to provide it more data to allow it to provide better, more predictable outcomes, more accurate outcomes than a human being could otherwise provide. So, there’s a whole host of opportunities available to apply to ML and AI, and it’s a really rich area. And really for companies, the challenge is you do need folks that understand these algorithms and have, maybe not necessarily data scientists, but have more of a data scientist bent to ensure that the right models and approaches and inputs are provided to provide the best outcome, the best result.
Mike Vizard: How hard is it to find those people these days? And is there some way to train them? ‘Cause it seems like the technology isn’t the issue so much as it is the people we don’t have to manage it or run it.
Ara Surenian: Yeah. They’re not easy to find. They’re often, they have degrees in data science. They could be through traditional computer science programs or industrial engineering types that have learned data science. So, if you look at that entire population and the number of companies that are competing for these people, they’re not plentiful. That’s the reality. And I think particularly for smaller manufacturers, mid-size, and even for some large manufacturers, getting these people is going to be hard.
So, our approach to this entire process is really about how do we build a solution that makes it easier for people to get the results that are provided by a data scientist without needing to have data scientists on staff? So, as a technology provider, to manufacturers really are challenge, and what we’re focused on is how do we make it easy? How do we allow the inputs and how do we ensure the data is structured in a manner that we’re doing all the hard work for the end user to provide that better outcome and allow them to leverage capabilities that would otherwise not be available to them if they didn’t have the right people on staff?
So, that’s the balance that needs to be attained. So, because digital twins are relatively, I wouldn’t say it’s a new technology, but it’s a newer technology that has really been supercharged by all the cloud solutions and the capacity that’s available, the data processing capacity that’s available. They’re becoming more apparent, but they still require that extra effort, and you’re going to see a lot of this becoming more mainstream over time as the technology providers like us figure out how to make it easier for end users to leverage that solution.
Mike Vizard: A lot of the incidents that we’re trying to work through today are what people call black swan events, right? The COVID issues and earthquakes happen and things like that. Can we actually factor those things into our models or will they always be somewhat exceptions to the rule, but at least we can see what would happen in normal times?
Ara Surenian: You can absolutely factor those into the model. So, there’s two ways of approaching it. One is you need to account for those events so that you can distinguish between normal demand versus abnormal demand. So, if you normally see a demand pattern that looks a certain way or supply coming in a certain way, you basically, the engine has to understand that when a black swan event happens, that it is an exception, it’s not going to be part of a normal pattern going forward, potentially.
So, that’s part of the input. You have to distinguish those events from what is typically going to happen, and then you can use that to model out the potential for something similar happening. So, you can test your business and saying, if there’s another, so now everybody’s talking about the monkeypox or whatever they’re calling it today, right?
If for some reason there’s projections that it’s going to be similar to what happened with COVID, you can use COVID as an input to building out a model that says what if this new pandemic behaves in a similar way or in a different way? You use that historical pattern to predict what’s going to happen in the future. Now, in the case of COVID, it was really hard to do that because it was unprecedented. No one knew what was going to happen, but the more you can draw similarities between past events, put those into your model, the better you are positioned to understand the impact it’ll have on your business and you can prepare accordingly.
And that’s the key thing. If you don’t have a plan to address even an exceptional event, you’re always going to be in a reactive mode and you can’t break out of it. And what we saw is the companies that had adopted supply chain planning as a component of their over business process, they were the ones that were more able to navigate all the uncertainty because they had the data, they were building out the models. They were seeing the changes in business patterns and shifts from one market to another market, and they were able to make decisions collectively to best meet that uncertainty. Even if you don’t have it, the very fact you’re planning and as you’re gathering information, you’re more adept to changing and modifying your practices to best meet the change as it’s occurring.
Mike Vizard: Being a supply chain manager is one of the toughest jobs there is right now.
Ara Surenian: Oh yeah. Without question.
Mike Vizard: So, what is your best advice to these folks? What should they be focused on?
Ara Surenian: Yeah. It’s tough because when things go well no one really notices, but when things go bad, you’re blamed for everything. And as a guy that came from an operations role and ran plants, it’s very hard to say no, right? It’s very hard to make excuses and no one wants to hear it because the business is at stake and revenue is at stake and customers are screaming and there’s a lot of stress. And a lot of folks don’t have the information they need to make informed decisions.
So, what I tell people is start planning and don’t plan using a spreadsheet. Really invest in better tools because those tools provide you the insight you need to, A: make those decisions and make the adjustments as needed, and B: better able to communicate to the rest of the business this is what’s going on, right? Because if the company doesn’t have a sales and operation planning process, for example, you should implement one. There needs to be a mechanism in place and the technology to support the process of collectively reviewing what’s happening, collectively making decisions based on the input and recommendation of the different vested interests in the organization to determine how best to move forward.
My recommendation is get yourself out of your silo. Put in processes that allow you to engage with the other members of the organization that have an equal interest in the success of the business as you do. And then find ways to implement technology to help the company and yourself make better decisions. And then from there prioritize. The Pareto principle has been around for a long time, but in times of uncertainty and stress, the Pareto principle where you invest your time and resources on the significant few over the trivial many really makes a difference.
Because those significant few is where the revenue is, where your margin is, and without those you’ll be out of business. And you can spend a lot of time trying to address all the noise. And if you address the needs of your big customers by identifying the items they need and building policies to address it, the more successful you’ll become.
And that’s also going back to your original point of the digital twin is that’s where modeling using these twins based on like the Pareto, you get the biggest bang for your buck. Because you can’t ensure all the data is correct, but if you focus on that significant data first and the significant items, it just makes things easier as you go downstream to less important customers and items. Not that any customer isn’t important, but focus is critical in times of uncertainty.
Mike Vizard: All right. There’s no substitute for digital focus. Hey, Ara, thanks for being on the show.
Ara Surenian: Oh, my pleasure. Thanks.
Mike Vizard: And thank you all for tuning into this latest episode. You can find this edition and others on the Digital CxO website. We invite you to check them all out. Until then, see around next time.