How Machine Learning Saved Thanksgiving

Thanksgiving is next week already and the plans have been set for family, turkey, and kicking back watching my Detroit Lions disappoint me in the 4th quarter.

Managing a manufacturing company with a backlog of work always meant to me that we would be running at least two shifts on Thanksgiving day.  And without hesitation – I would get the phone call from the plant that one of our key production machines was down and needed repair.

Now if your wife is anything like mine she would give me that look that said, “Don’t you dare leave this table right now. ”  And like every other year, I would have to step away from the family and start scrambling to contact our maintenance team, manufacturer rep of the machine for parts, and try to find a way to get the production team back online.

Then I discovered Mtell.  They educated me on the fact that my machines were already “smart machines” and were trying to communicate with me, but I just hadn’t taken the time to listen.

You see, all of the key production equipment we have on the shop floor was already sensored and the data produced from those sensors was just sitting in our historian.  Yes we had implemented a CMMS and an EAM to help us schedule routine maintenance, but frankly the repairs sometimes were the cause of the machine breaking down shortly thereafter.

What I discovered through implementing Mtell Advance was that the cost of getting started was directly related to how long I had waited to purchase the software solution in the first place.

Installation took just a couple days to get everything talking to the smart machine learning engine.  Within 8 days Mtell had consumed 2 years of sensor data and connected the failure relationship to my Maximo system.  The maintenance dates, along with the repair records was now a part of the body of data Mtell was comparing the real time feed to.

Normal operating conditions, failure signatures, and outlier anomalies became the “new language” of the plant.  When something was out of the normal operating condition – a text message was automatically sent to myself and the maintenance team to go investigate.  The more we learned about what our machines were saying – the smarter our team and our machines got.

It’s now been two years of running Mtell within the plant and we have avoid over 30 machine failures because of advanced warning from the top ten list of sensor signature contributors.

I have to admit – I’m actually looking forward to sitting down to the table this year knowing that my turkey and mashed potatoes will be warm and delicious – and that I won’t be getting that look from my wife.

Happy Thanksgiving everyone!