George, please outline your work with the University of New Hampshire to demo automation and IOT adoption.
We know that there is a lot of apprehension by small and medium businesses to really jump into Industry 4.0, IoT and automation. There are a lot of reasons why, including the big financial investment. So we took up a project at the University of New Hampshire’s John Olson Advanced Manufacturing Center to create a demonstration cell that would show how a small/medium manufacturer could embark on the journey toward getting to full automation and IoT data utilization. The idea was to demonstrate an approach that would significantly reduce the financial investment risks. To be as relevant as possible, we took a sample of the most widely used CNC machines. We have a high-end, simultaneous 5-axis DMG Mori milling center, a lower-end, 3-axis Haas milling machine and a middle-grade Mazak CNC vertical turning center. We have about 60 feet of track fitted with an ABB 150kg tender and stationary 150kg capacity KUKA for cleaning, de-burring and drying. We essentially created an example of a new kind of automation cell. The idea is to be able to see it firsthand in operation rather than on charts and in pictures. That was the goal—to first implement the system and then be able to explain what costs were associated with creating it.
When will it be open to the public?
It is still a work in progress. We’re pretty close. We are in the phase of industrializing it: We’re using real-world parts from manufacturers and plan to run them through the system to make it more robust.
What were the biggest takeaways from this work?
The purpose was to minimize the risk to the small manufacturer, financially and technically—because that’s really where the biggest apprehension exists.
The first part of it is the way we laid out the cell. The new cell concept goes away from typical automation layouts that limit access and instead you’re able to approach each of your individual machine tools as if there was no automation. There are no cages, no barriers. You can walk right up to your machine. You can bang out a single prototype or do a lot of 20 or 50 of some part and walk away and not even consider the automation’s existence. At the same time, the machine is available for automation because of the way that the cell is layed out. So, if it makes sense and you do get a high-volume contract, you basically use the automation features to have that done. The fact that you can use these machines in both ways simultaneously, essentially a multi-purpose capability, reduces the risk to the small manufacturer because the utilization of the equipment is always maximized.
The second big takeaway from this is the fact that the system that’s actually doing all of this is based on open standards. It was written against the MTConnect standard, and is compatible with other standards like OPC UA, MODBUS, PROFIBUS and QIF and, while supporting all these capabilities, it’s based primarily on MTConnect. It takes MTConnect in two directions: It not only receives MTConnect data but also integrates separate sensor inputs and provides controls. The software is $100 per license, and it’s a one-time cost; there’s no requirement to have continuous annual license fees. The fact that it’s extremely low cost reduces the risk a lot.
Thirdly, just by being able to put this system together, you automatically get access to all of the data points that it’s functioning on. So you have an entire IoT system, and all the data is being put into a open SQL database in a temporally consistent fashion. That means you now have a lot of context: Everything’s time stamped. And it’s in a form that is conducive for machine learning and artificial intelligence.
What new advice does it cause you to give?
The advice is to get past that initial reluctance of getting into Industry 4.0 and some of the advanced technologies that are out there. The risk is relatively low if you go in with low cost COTS hardware and open systems. You don’t have to have an extravagant ROI established by somebody, and you’re not really taking a leap of faith here. There is some immediate, low-hanging fruit in implementing a system like this. A single box is $3,000 and can handle three or four machines. It comes with a $100 software license. If you wanted another copy of the User Interface to live on a separate workstation or device, that’s another $100. So it’s very low cost. With regard to low-hanging fruit, you could, for example, have a Bluetooth caliper at the machine tool as you’re machining your part, basically checking KPPs, hitting a button. The measurement is automatically being loaded into a database and tracked. You can set upper and lower limits, and as soon as those KPP measurements drift over time and reach one of the limits you can create an automatic tool offset in the CNC machine without the operator even being involved. That’s a pretty immediate value add you can get from installing a system like this. And on top of it, you get all this back-end expandability that you haven’t even thought about how to implement.
How difficult are the logistics of adopting automation?
Well, you still need to program the robot paths. You still need to write G code for machines to make the parts. And those things, while they may be automated in the future, aren’t disappearing anytime soon. So the components of this are already being done. You either have this type of expertise in-house or you don’t. If you don’t, there are a bunch of integrators out there. And that’s really how you want to approach this: You want to be implementing it on an independent service basis rather than paying some long-term license fee to the developer. So it’s really the ultimate model. You get in with a low cost, at low risk. You’re not pinned down through a costly licensing-type mechanism, and your people evolve over time to get more and more value out of the system.