What Data Tells You - or Doesn't (Part 4)

I must figure that the guy that sent me the “we’re so happy to have you as a customer another year” email yesterday did not have access to current information or had an automated email campaign that failed to recognize what we would call a non-conformance in manufacturing. Two weeks ago, I received a renewal reminder from their team for a tool that we no longer use. I’m one of those people that get 200+ emails delivered a day (this is a small subset of the 2,500 or so that are sent - after 26 years with the same email, I’m on all kinds of lists) and so I have a process to take care of them quickly.

An email like this goes into my active folder and gets taken care of quickly. I logged in and stopped the subscription so it would not renew in March. Two weeks later, I received another email from him welcoming me back. Hmmm. An automation failure on their end.

What is automation really?

First, automation is not a single thing. You have an automation to do a thing, but you need a wholistic approach to know when that cycle should stop, pause, or shift in some way. We look at automation in manufacturing as an approach to one or more key processes you have that you want to perform, measure, inspect, or record.

Automation should make your typical workflow easier. It can perform repeating work or work that does not need any care. Think about filling bottles on an assembly line. If the empty ones are flowing, so should the liquid. And you don’t need a human to do it – you can safely automate that. However, that machine ought to have a stopgap measure for when one in a million bottles turns over on its side or doesn’t feed correctly. If you don’t stop the fluid at that point, you are going to be cleaning that machine for a while.

Automation is often cited as having started in the automobile industry in 1946 at the Ford Motor Company by an engineering manager, D.S. Harder. Today, it can also refer to the use of a machine to do work that a human cannot. We see that all the time with manufacturers that want to complete a routine in a set period and no human could make the calculation in the time allotted so a machine does the work.

Where automation fails

In most manufacturing, automations must be flexible enough to do one or more workflows and have set and on-demand escapes for when things go wrong. One of the most obvious ways I think of this is when a member of your team must measure every n^th item in production (let’s say every 10 for argument’s sake), and each one after a measured failure until there are 7 in a row that measure properly. The automation provides a cue to measure the 10^th, 20^th, 30^th, etc. until the job is complete. When the 12^th one comes back out of tolerance, they need a way to measure items 13 through 20 or so.

Building a system that automates and un-automates (I know that’s not a word) is difficult. You need to identify when things are “going right” and when they are “not” to execute on the proper thing. We encourage manufacturers to seek out flexibility in their toolsets; you simply don’t know that you are going to do things the same way forever and you will eventually need that flexibility if you don’t already.

Automations fail where they are built to do a certain task or set of tasks without that flexibility. If you don’t need it today, eventually you will, as every business has to shift for one reason or another. Flexibility allows you to change things as you need, and where automations fail, they often fail to be flexible.