Status Go: Ep. 236 – Driving R&D in Manufacturing | Ali Shakouri

Episode Highlights

[00:00:00]: Interoperability is the Key

[00:00:43]: Introduction: The Evolution of the Supply Chain

[00:04:31]: Ali Shakouri’s Journey to Purdue

[00:14:42]: How the Pandemic Shifted the Supply Chain

[00:19:54]: Software to the Rescue

[00:21:44]: Shifting to a Data Mindset

[00:26:00]: Will Manufacturers Share?

[00:27:13]: How Purdue is Addressing the Problem

[00:31:56]: Where Can We Learn More?

[00:35:40]: Thank You and Close

 

Episode Transcript

Ali Shakouri [00:00:00]:

How the data could be interoperable…and that’s a huge challenge. And I think this is the reason manufacturing…how it has been developed, of course, each of the big automation companies had no interest to make them compatible. They wanted everybody to buy all their solutions. So, I think that element of interoperability, it is important. There are some…Manufacturing USA Institute headquartered at UCLA called CESMII is focused on this type of common data format and so on.

Voice Over – Ben Miller [00:00:43]:

Technology is transforming how we think, how we lead, and how we win. From InterVision, this is Status Go, the show helping IT leaders move beyond the status quo, master their craft, and propel their IT vision.

Jeff Ton [00:01:01]:

Welcome to another exciting episode of the Status Go podcast, where we dive deep into the world of innovation and the driving forces behind it. I’m your host, Jeff Ton, and today we have a special guest who’s been at the forefront of empowering small and medium-sized enterprises in the manufacturing space to embrace innovation and adapt to changing times.

Our guest today is Ali Shakouri. Ali is a professor at Purdue University and the lead of the Wabash Heartland Innovation Network. He’s also the co-author of a fascinating HBR article that examines the post-pandemic transformation of supply chain. It’s this article that we’re going to chat through today.

The transformation that is going on in the supply chains has been brought about by a renewed focus on regional and resilient solutions, opening up a world of growth opportunities for small and medium-sized manufacturers. But you see, they’ve been through a lot in that space.

When you think about the early two thousands and the offshoring that was going on specifically in the manufacturing space, many of these SMEs in the manufacturing space spent their focus on crafting lower volume, complex-to-ship components for industries like automotive, aerospace, and industrial equipment. They managed to survive, but the growth that has happened in other sectors was somewhat elusive.

So, we’re going to dig into some of the challenges that they’re facing when they’re confronted with “where do we go from here” and this changing of the supply chain.

They’ve been using the same processes that have been proven effective. How do you change when you’ve been doing the same thing for many, many years?

They may have a limited willingness to invest in the face of growing demand because they’re constantly getting cost pressures from their customers. These are thin-margin businesses, right? And foreign competition many times can undercut costs.

And then, finally, one of the things that we’ve talked a lot about here on Status Go is this digital landscape, that is so rapidly evolving and growing, can be intimidating because it requires new skills and new ways to approach work.

All of these factors combined make it really challenging for SMEs to keep up with the technologies that are reshaping the competitive landscape. Ali and I will discuss ways that these organizations can leverage technology to drive innovation.

And I will say to those of you who may not be in the manufacturing space these concepts will work for you as well.

So Ali, with that, welcome to Status Go.

Ali Shakouri [00:04:20]:

Good afternoon. It’s a pleasure to be here. I’m glad to discuss about such an important topic that has kept us busy in the last seven, eight years.

Jeff Ton [00:04:31]:

You’ve been, you’ve been very busy since we last spoke. I’ve in looking at some of your other work that’s available.

Before we dive into this, would you share a little bit about your background, your journey, and kind of wrap it up with your focus of your work there at Purdue?

Ali Shakouri [00:04:51]:

Yes, I did my PhD and postdoctoral work in the area of optoelectronic and semiconductor physics. That was a time when the fiber optics was expanding with internet, with multi wavelength networks, and so on, and I could see how the new technology was changing everyday life.

All of us know what happens to our computers, how they shrank, and what happens with our tablets and smartphones. But what is on the backbone is also huge advances, how we make chips, how we make fiber optic components.

So that is my background and that’s what I did my research. I had some work about studying some of these, and I had a couple of students. We commercialized some of the technology for integrated circuit characterization, specifically for thermal imaging.

I came to Purdue in 2011 to lead the Burke Nanotechnology Center. It’s one of the largest and cleanest university clean rooms. I felt there is really a great investment here, a lot of outstanding faculty. But then I told myself, what does nanotechnology Midwest mean? If we try to do the same thing where the center of gravity is in Silicon Valley or the East Coast, we do it the same. 30, 40 years later, we can never catch up.

So, I was already thinking, when I came here, there’s a lot of good resources, but we need an ecosystem nearby. I myself benefited from having my graduate students, when I was a professor at University of California in Santa Cruz, find jobs within a radius of one, one-and-a-half-hour drive, and then they come back, and help the existing students. And the same thing happened during my PhD and postdoc when I was in Southern California. And I think here in the nano area, a lot of our graduates find jobs under two coasts, which is good.

Part of our mandate is to train good students. But the question is, what could be the areas we take the niche. And we identified applications of IoT, Internet of Things to ag food. Healthcare is an area that is more nascent, and this is an area that maybe Midwest could be a center of gravity. And we started focusing on that.

That was my first introduction to manufacturing. What happened is that at universities, we are very good in demonstrating a new idea, a new type of a device, new type of an algorithm. And of course, we build companies, startups, and a lot of things out of it.

But the strength in Midwest is also how to make the same thing at the large scale, at the low cost. And that’s actually hard to do. At university. We always say we have a student-dependent effect, oh, this student can make this device work. Another student. Manufacturing needs to be more mature. So really working on manufacturing is more challenging. And we started developing some test beds for this, what we call smart films, scalable manufacturing of aware and responsive thin films, with an industry consortium.

And we learned that actually manufacturing can benefit from data in a way that was not possible before. Why? Manufacturing is hard because everything is about how you scale up something that you know how to make it, but now you need to make at thousands or tens of thousands, hundreds of thousands at the low cost. And how do you optimize everything?

And the key challenge is to get the specs down, how do you control each step? But with advance of data, we learn that we don’t need to make the specs as narrow as possible. Some of the variability that happens in earlier stage scale up can be corrected using data and kind of some sort of data analytics. So that was our kind of foray into digital and manufacturing, specifically for the manufacturing work we were doing at the university for IoT devices about six, seven years ago, we were fortunate.

There were study by Lilly Endowment about how to make our region more prosperous. That’s the project Wabash Heartland Innovation Network that you alluded to. So the genesis of the project was Purdue. And the ten counties around us has about 400,000 people in this part of, within a radius of maybe 50 miles or so. And this is among the rural areas. But we have 250 to 300 small, medium manufacturers in our region. We have a couple of thousand farmers in our region. These are the two main assets. We have one of the best farming lands, and we have a lot of manufacturing in our ten-county region. But between 2000 and 2010, like the rest of the US, we lost about one-third of a manufacturing jobs. So there’s already the pressure, and we see it economically in the counties specifically.

And so that is causing a pressure. So the Lilly Endowment gave funding, they wanted to give funding and they said, we want Purdue to work with the region to help in AG and manufacturing. And in some sense, it’s interesting coincidence all of this happened before COVID and so on. Nobody predicted COVID happened, but as a result of it, between 2016, 17, when we started thinking about this, the actual project was funded in 2018 as a five-year project. And we started working both AG people in the region with our College of AG, with extension program and also with manufacturers.

Let me focus on the manufacturing side. What we notice is we do have a couple of big manufacturers in our region. Subaru is here, Caterpillar, Babaj and so on. But like any other average us, I think 98% of the US manufacturers have less than 500 people. So really majority are small. And when we went there to discuss how Purdue could help them, and our idea was that data and IoT could help them, some of the small manufacturers said we had to Google IoT before you came to see what you’re talking about.

They are through this project, I think rather than sitting down and doing a lot of analysis, we said learn by doing. So we send student groups, do some project, install some sensors, with the idea that students spending time at a factory floor, they learn the environment and providing data to manufacturers will be hopefully helpful for them. So through the initial years of the project, we had workshops and we had students starting installing sensors and getting data and low hanging fruit. Everybody talks about is condition-based maintenance. How can by listening to the motors and all the machines we can predict before they fail.

And we started doing some of that, but we also realized that’s not where it’s hurting. Most of our manufacturers, as still is the case, they don’t have enough labor that they could hire. Productivity improvement is what they need. And so, the question is how we could help with that. And that’s harder. Correct, because each of them is a different workflow. And I think part of the lesson is that in some industries, startups and the startup ecosystem bring new technology. And what happens is that there are good efforts in the area of smart manufacturing, sometimes called Industry 4.0.

But it seems to me, and that was the impression we had, is that the entry point is too high. Is that meaning that a lot of companies, they don’t even know what will be an ROI, and they cannot spend tens of thousands of dollars just in a solution to get data. And because each of them is different. The scalability is a challenge.

I think a lot of the lessons that we highlighted in this article in Harvard Business Review has to do with the lessons we learned. What are potential areas we could work on that is scalable and where manufacturers could learn from each other? That’s another important point. Cohorts are really essential, but also not everything that we, as a faculty, we can write papers about.

It is not relevant to a small medium. So really matching the two sides was a good back and forth. So I think this is some of the elements of how this project came about.

Jeff Ton [00:14:42]:

Well, and I love that you’re sending students into the shop floor to work side by side, because not only are you able to help the manufacturers, but hopefully the students are catching the bug and wanting to stay and continue to solve problems.

I know, and we’ve talked a lot about this on this program in Agtech, that specific sector of technology and the intersection of agriculture. A good friend of mine is at Beck Hybrid Seed Company and is involved with Purdue, Brad Fruth, in looking for innovation. And I think that’s the other side of this coin, is driving innovation. When you’re focused so much on just doing things the way you’ve always done them, finding time for innovation can be difficult because of the things I mentioned from your article. “Hey, we’ve done that. These are our processes. They’ve worked for us. The margins are razor thin, so we don’t have money to invest.”

But one of the things you point out in your article is that the pandemic itself has caused a shift in the Supply Chain. Can you talk a little bit about what you’ve seen in that shift? And then, I’d love to dive into some of the solutions that you outlined in your Harvard Business Review article.

Ali Shakouri [00:16:23]:

Well, I think one of the lessons that we learned during the pandemic is that over the years, through a lot of innovations about how to scale things up and offshoring and so on, we have optimized our supply chain to be the most efficient, lowest cost, because who wants to pay more? Pandemic shows us, by the way, this is quite fragile. All you have is one thing happens in the Suez Canal, and then for six months, everything is affected.

Jeff Ton [00:17:01]:

Some guy crashes a boat, and it backs up manufacturing across the world.

Ali Shakouri [00:17:06]:

Yeah, and think about what happened with the cheap supplies for cars and how many factories had to. So I think the big lesson, which probably people have not appreciated, is this focus on best efficiency and the lowest cost has made our system fragile, and we need to do something about it because we know natural events happen and so on, and even pandemic. It’s not that every 100 years will happen. Who knows? It could be sooner. There are other things. So I think that’s number one. So as a result, there were interests in local manufacturing and how to kind of encourage domestic manufacturing. And I think this is an area in which us in a more fortunate situation than Europe because Europe kind of abandoned manufacturing earlier because they said service is better and so on.

They’re dirty things. Let’s kind of send it offshore for them. It’s much harder to come back. I think in the case of manufacturing, we do have some strength. My co-author, Willie Shih is actually one of the first people in 2006 or 7, when the first economic crisis happened, wrote some very influential articles about the importance of manufacturing commons. And he emphasized that there is certain know-how, certain relations, expertise that exists regionally. And once you lose it, it’s not easy. Just, okay, let’s bring the company back here because the supplier of the company, the maintenance, all of these go with it.

I think even though we had some losses, we have, at least in Midwest, a lot of good know-how the challenge is. Well, first of all, these companies are good because they survive the economic challenge of 2000 to 2010, but the pressure keeps going. And I think another comment that Willie made was that all the manufacturers that competition that was built in Asia has newer factory than us. So by definition, their lower barrier, they get data for free because that’s their newer machine. And here we have had manufacturing places that we visit and some of them, they have machines that are working continuously for 50 years.

Jeff Ton [00:19:48]:

Yeah, it’s working.

Ali Shakouri [00:19:49]:

Don’t touch it.

Jeff Ton [00:19:54]:

In the article that you and Willie Shih wrote, the first thing that you mentioned in here is near and dear to my heart because I’m a software guy and you talk about leveraging new simulation software tools. Are you talking about things like digital twins? Is that the kind of thing you’re talking about or are there different kinds of tools?

Ali Shakouri [00:20:17]:

Well, I think eventually people want to get to digital twins and so on. But our idea was that maybe the advance of AI and some of the kind of machine learning may have lowered the barrier. Because before, if you want to develop a digital twin for a line, you need to hire a couple of software people. There’s lots of back and forth to optimize it and make it production-ready and robust. So, the barrier is high. You will only do it for high-value-added product and things that you can put a couple of people to emphasize that. So one of the ideas is that now the computer user interface and the way you write program, all of them have lowered the barrier. So maybe more smaller manufacturer could introduce simulation tool.

The idea is identify bottlenecks in whole manufacturing steps or their supply chain or inventory. Eventually, I think by them learning how to do this and have more and more simulation tool, then they will have a digital twin of their whole maybe operation. But that’s even farther. I think there are a lot of low-hanging fruits, things that they can do.

Jeff Ton [00:21:44]:

Do before they get that involved and that sophisticated. I think hand in hand with that. And you’ve mentioned data several times. And I think when we talked the first time, Ali, you mentioned that many of these manufacturers that you walked into initially were doing things on paper. So not only didn’t they know what IoT was, they were on paper.

So, how do you begin to have them shift their mindset that data can help drive this innovation within that sector? What are some of the things that you suggest that they look to?

Ali Shakouri [00:22:29]:

That’s a very good question. So, first of all, it’s not that they don’t want to, but once it’s working and then adding a digital solution, it needs to be integrated into the workflow. And of course, if you have a state-of-the-art manufacturing execution system, MES and so on, it already has some of this digital. But the challenge is most of our small, medium manufacturers, they don’t have that. So I think there is a big component is at the end, who gets the data are the factory floor worker. And there is a cultural change and their engagement. You cannot just add something to them and you do your job and now you start typing and no, they need to be engaged. There is two elements to it.

One is the comfort with how to input the data. But the second part of it is really the buy-in. As we know, data could be used in many ways. And one of the first things could come to people’s big brother watching and then it’s become counterintuitive. I think one of the things we realize is a lot of small, medium manufacturers look at their workers, as their family, and they are together, but on the other hand, you cannot just force things. It takes some time for them to take ownership. And I think it’s a learning for everybody what is the best way of doing things. I think that’s one element of it.

The second part is how the data could be interoperable. And that’s a huge challenge. And I think this is the reason manufacturing how it has been developed. Of course, each of the big automation companies had no interest to make them compatible. They wanted everybody to buy all their solutions and if I was in their shoe probably I would do the same. Correct? Why would I make my market margins and more competition higher? So, I think that element of interoperability is important. There are some. A Manufacturing USA Institute headquartered at UCLA called CESMII is focused on this type of common data format and so on.

But in the end, it’s not a top-down approach. Industry people need to work together, one has to find ways. Of course, there is something common, something open, and on top of it, you can build your proprietary models and proprietary solutions. I think similar challenges existed in the computer industry. I remember in 70s DEC and IBM had outstanding systems and they have their own languages, their own operating system only with actually interesting some effort of universities such as UC Berkeley, MIT as well as the labs Unix came about and then Linux and then now we have an ecosystem that okay there is a Linux that is open and people can make it easily add compound but on top of it of course Google can make money out of it by making their own operating system.

Jeff Ton [00:26:00]:

Do you think we’ll see that same kind of thing come to the manufacturing space where they’re sharing at that level?

Ali Shakouri [00:26:07]:

That’s something really remains to be seen. That’s a very good question as we are discussing of how this data compatibility and kind of collaborative atmosphere could develop our ideas or our motivation is this. Look at Python, look at Linux. There are a few others. I think we need that but the question is who is going to do the groundwork and because that is the real challenge.

Jeff Ton [00:26:36]:

Well and I think you lay out in the article some other ideas, but one I want to highlight and that is leverage programs from universities and outside resources. I don’t know that we in business, in the business sector, always think of partnering with our universities to help in that.

Can you speak to a little bit about the program specifically at Purdue, but in general, how universities are helping solve some of these problems?

Ali Shakouri [00:27:13]:

Well, I think at a certain level, many universities have quite vibrant industry research programs, but the way it works, and that’s caused a matching issue, is that if you are a big company, Boeing, GE, Lockheed Martin, and others, you have some general challenges. You can give some big funding to university support, couple of graduate students, PhD students and I think there’s a lot of good work there.

What has been missing is how the small manufacturers could benefit from the strength of the research universities. But there is a level of okay, manufacturers propose senior design projects and this is good, okay you have a group of undergraduates for six weeks, ten weeks, 16 weeks work on a problem. But this is not research intensive. This is kind of get your feet wet on a good problem, and you don’t short term.

Jeff Ton [00:28:14]:

Right. Some of these problems are bigger than that.

Ali Shakouri [00:28:19]:

There is this wheel of the faculty who want to have an impact, but the difficulty is how to match this with tiny bits of here and there. And I think that has been the challenge. One of the lessons we learned, the project we have currently funded by National Science Foundation Future Manufacturing, involves five, six faculty at Purdue, three faculty at Harvard University, two at Tuskegee, and two at Ivy Tech Community College. One of our, a lot of thoughts went to it is to define projects that matches the expertise and the PhD level. But there are enough commonalities that multiple small, medium manufacturers could help. And I think that requires some thinking. It’s not let’s do whatever is possible to help them. I think this identifies some of the low common areas for us; one of them is privacy-preserving machine learning.

Again, AI has changed the way a lot what we do things, but that happened with lots of labeled data sets in a big data center. That’s what Google and Amazon, and so on have it. None of the manufacturers, even the big ones, have enough type of label data to do this. But on the other hand, nobody will put all of their raw data into some sort of data lake. Even with some anonymization, it’s very hard, you can always be anonymized, and so on. So, part of our focus has been these guarantees, which is algorithmically can be guaranteed that you cannot reverse engineer, you never share your raw data, but we can interrogate your data and get some information out. And that interrogation now can learn from multiple manufacturers without mixing their data and so on. So that was, I don’t have to.

Jeff Ton [00:30:18]:

Give up my IP in my data.

Ali Shakouri [00:30:22]:

I can still participate; IT can still participate. Again, we can interrogate the question, and before we get the result out of your manufacturing data, you can see what is the question we are not asking. So that is one.

The second part of it is a lot of the great AI advances have happened with big, centralized databases. In manufacturing, there are rooms, and we need to identify some of these common processes centralized. But a lot of the things happen at the edge, and if you just take all of the data to the cloud, it will be too expensive anyway in terms of the infrastructure required and what is the resources needed to keep it. So we have a focus on edge analytics, what we call tiny machine learning, tiny ML. So, these are some of the building blocks that we have identified, and we hope, again, working with manufacturers on a specific productivity quality challenges they have, we can grow it slowly.

I think one of the really important lessons for us is, and it takes time. After four or five years of working with the industry, we now have a couple of partners who are sharing their production data with us. And that has been a big plus because real-life data is so valuable. Actually, computer scientists are hungry for that. And now we get our hand on it. And that’s how this AI commons that we are discussing is shaping up and slowly growing well.

Jeff Ton [00:31:56]:

And it provides value to the researchers so that they can have a larger data set to learn more, gain more insights, and it’s providing value to the manufacturers.

So, Ali, we’re about out of time here, but I want to ask for those that are listening that we’ve piqued their curiosity, they want to learn more. Obviously, we’re going to link to the show notes, link in the show notes to your HBR article and they can read that, but where else can they go to learn more?

Ali Shakouri [00:32:34]:

So, we are actually trying to put together a collaborative web platform, what we call the Manu Future Today platform. And the idea is we will have a place where some of these ideas, lectures and so on is available. There is a forum for discussion. I think this peer network is so important. We learned a lot by having face to face meeting. I think now some of these lessons having in this manufacture web base could be an opportunity to engage broader people. And my suggestion is we hopefully have this out in the next month or two. And meanwhile they can look at the links.

The Harvard Business Review article has some links to mydatacam.org. That’s another. This is our partners at Harvard manage it. It has some algorithm; if it’s useful, they can use it. But I think as an umbrella, this Manu Future Network could help another opportunity. And we are working. One of the lessons in this article is for manufacturing competitiveness. We don’t have another 20 or 30 years to slowly do that.

The competition is there. If we are not quick to scale it, really, we cannot compete with other countries. So in order to do that, really we need the manufacturers to come together. So, manufacturing extension program, MEP is a resource. So, we are discussing how we could work with them, the advantages there is an infrastructure in many states and how the same way Ace extension is, we are hoping we can benefit from MEP.

Jeff Ton [00:34:25]:

Well, and you’re right, it is a problem that has quite a bit of urgency from a time frame perspective in it. I’m also assuming, Ali, that if anybody listening is from manufacturing within that ten-county area around Purdue. They could reach out to you at the university or someone at the university and get some additional information as well.

Ali Shakouri [00:34:52]:

Certainly, yes. We are very much looking forward to hear. We are not limited to the original win Project grant for us ended, but we are interested in manufacturers who are willing to share production data with us and help us develop better algorithms. We have a grant now called AnalyticsIN which is funded by the state of Indiana. So anywhere in Indiana, we do also have NSF project which is federally funded. We have partners in Alabama. We have partners in Massachusetts. But again, overall we are hoping there will be satellites.

It’s not that we can handle everything, but if you are interested, let us know. Either we can work with you directly or we can connect you to one of our partner universities.

Jeff Ton [00:35:40]:

Excellent. Excellent. Well, we’ll be sure and put ways that people can get in touch with you and the university in the show notes as well. Ali, I want to thank you for coming on Status Go today. I think this has been a great conversation. I encourage our listeners to dig deeper into this topic and the way that these SMEs can leverage universities, leverage each other, and leverage associations to drive that innovation. And as the supply chain continues to morph, take advantage of that.

So, Ali, thank you so much. I appreciate it.

Ali Shakouri [00:36:25]:

Thank you very much, Jeff. Really, it’s been a pleasure, and look forward to expand this work and see how we could leverage your network and your resources. Thank you.

Jeff Ton [00:36:36]:

Excellent. To our listeners, as I mentioned, we’re going to put links in the show notes. You can find those show notes on Intervision.com/status-go. And we’ll be sure and have contact information as well as links to this HBR article that Ali and I have talked about. This is Jeff Ton for Ali Shakouri. Thank you very much for listening.

Voice Over – Ben Miller [00:37:03]:

You’ve been listening to the Status Go podcast. You can subscribe on iTunes or get more information at intervision.com. If you’d like to contribute to the conversation, find InterVision on Facebook, LinkedIn, or Twitter. Thank you for listening. Until next time.