Manufacturing Automation Requires Lots of Data To Work Properly
As crazy as it sounds, manufacturers that move towards more and more automation often start noticing a strange problem. They begin missing the benefits of non-automated processes and realize that they may have actually relied on manual human efforts more than they may have realized on the surface. For those that often read our blog, it may come as a surprise to hear us touting the benefits of anything manual vs. manufacturing automation. However, the truth is that simply automating manufacturing processes, cells, or machines does not inherently make things better or more cost-effective.
Manufacturing automation requires good data to work properly, and in most cases, more data than you had before. This is because — whether you knew it or not — humans were actually collected data all along. They just weren’t sharing it with anyone or storing it anywhere…
Will Robots Replace Humans in Factories?
The truth is that the more you automate in manufacturing the more data you need to prevent disaster. As manufacturers smartly move away from manual human processes — toward a digitally transformed shop floor — they begin to lose the smell, feel, and a touch of how different machine lines are running. They lose the human element that is seeing when something doesn’t look correct, sound right, or is operating in a generally abnormal way.
This sounds sort of strange, but anyone that has worked on the shop floor can tell you that lots of issues and problems are identified and solved proactively — or in real-time — by employees on the shop floor that leverage their experience to prevent major downtime. This might sound like a benefit of manual human processes vs. automation, but it actually isn’t. Organizations that are unable to capture this process of predicting, diagnosing, and solving problems on the shop floor will always be held hostage to these manual processes.
What happens when the human responsible for preventing disaster on a certain machine line is gone for a week, finds another job, or retires? Will his knowledge be captured properly, will the machine line run the same way when another human steps in? If not, how do we know what the difference was if it was never recorded?
Replacing Intuition with Analytics
This is the imperative part of automated manufacturing lines, cells, and processes. There must be a point where manufacturers close that gap on the knowledge and insights that humans provided and the value that machines bring to the business. This means one simple thing… more data collection and more analysis of the data that is significant to every cell, machine, and job role.
The more manufacturers automate, the more they need data analytics. Humans and automation should go hand in hand, one should not be without the other. (Don’t worry, human jobs aren’t going anywhere. Robot automation can actually create more jobs!) Manufacturers need good analytics to be able to understand if something is right or wrong. Robots don’t inherently understand when something is not performing well or is trending towards a shutdown.
Humans monitoring the same things do a much better job of this “out of the box” (humor me with that one). Toyota even references automation with a human touch when describing The Toyota Way. “In the Toyota Way,” author Jeffrey K. Liker explains, “it’s the people who bring the system to life: working, communicating, resolving issues, and growing together.” Even further, “It encourages, supports, and in fact demands employee involvement.”
The value for manufacturers is in the ability to leverage the power and benefit of robots and automation; while maintaining the predicting, diagnosing, and solving power of humans. For manufacturers, this makes data collection in the lines that leverage automation critical.
How to Deliver the Right Data, at the Right Time, to the Right People
The best way to accomplish this marriage is to start by collecting all the data and centralizing it (we talked about how to do this a lot here). Having the data centralized is the first step towards being able to see how things affect each other; i.e. seeing trends, identifying key causes of downtime, and seeing why certain lines perform differently from others. Even if some of this data is manually collected, centralizing this data is the first step that manufacturers should be making before automating a line, cell, or process.
Speak of the devil, once the data is centralized, I’ve personally seen a lot of manufacturers directly benefit from automating all of their machine data collection. Every manufacturer has a reason why they can’t do this, but universally, it is almost always possible. And, I’ve never met a manufacturer that regretted this project; nor claimed they didn’t see strong ROI for their efforts.
Finally, this data usually becomes truly valuable once it is made actionable. This rarely happens unless the people that need the data the most can see it and understand it. Traditionally, this has been a big headache, a massive project, and an expensive undertaking. That’s not really the case anymore. businesses like Mingo allows manufacturers to access all of their data in role-based dashboards without lengthy implementations or costly software.
Get a Leg up and Save Some Dough
Solutions like this are making it easier and easier for businesses to pursue true automation on their shop floor. And, the time for them to be starting all of these projects is now. The cost and discomfort caused by massive digital transformation projects in the future will be significant. I have preached this at every conference I’ve spoken at this year. There is a lot of money to be saved in small, practical projects now that will set manufacturers up for a strong modern manufacturing future.
Manufacturers that put in the effort to centralize their data, automate the collection, and contextualize it for everyone inside the business will be able to maximize automation projects, prepare for a digital shop floor in the future, and prepare themselves for a future that includes fantastic innovations like machine learning.