What Does Statistical Process Control Really Do? (Part 3)

Posted by John Clemons on February 21, 2013 @ 9:16 am

SPC Really Do Part 3I think this is probably my last post on the topic for SPC, at least for awhile.  I really didn’t plan on having this one, but in putting together my first two posts, a several people asked me a lot about what the SPC really look like and what you really need to implement SPC.  And, since I mentioned the concepts of real-time SPC and historical SPC in my previous posts, they asked me about those as well.  So, I thought I would do one more post about SPC and try to give you an overview and what an implementation of SPC might look like.

Probably the most basic component of SPC is the data. You have to have data and you have to get it from somewhere.  Now, it can reside almost anywhere but most commonly it resides in a database or a historian or both. Now, how it got in there is not really important, but you have to have data. You can get it in there automatically (which is probably preferred) or manually (which also works) but you just have to get the data and get it in the historian or the database so you can do something with it in SPC.

You’re probably going to want a couple of different types of data, what I call process data and descriptive data. The process data is the data directly from the process and it’s what you’re going to analyze using the SPC tools. It’s going to be data like temperature, pressure, speed, flow, weight, volume, and so on.  You’ll want to capture it pretty frequently (hence the notion of the historian) and pretty consistently. The descriptive data describes the batch or the process run or whatever that is going on when the process data is collected. This descriptive data is like batch, order, product, material, manufacturing order, process order, supplier, source, destination, or whatever else you need to describe the process data and give it context.

Once you have the data the most fundamental aspect of SPC is the real-time data visibility. That is, you can see all this data in real-time as it’s being collected. So, even before you start to apply all the SPC rules and all the math, with real-time data visibility you can see the data, and right there, just seeing the data gives you and the operators a lot of important information.

And, once you’ve got the data, and the real-time visibility, you can then apply all the SPC rules and all the cool math. You can set up specification limits and set up real-time alarms and notifications based on the spec limits. You can apply the basic SPC rules and set up real-time alarms and notification based on them as well.  And, you can even set up the pattern rules or run rules with real-time alarms and notification on them, too. And, with all of this displayed in real-time, the people who can actually do something about all this, namely the operators on the shop floor, get to see the information they actually need – in real-time.

Historical analysis takes the SPC data to a whole new level. It’s also where the descriptive data I just talked about comes in.  Historical SPC allows you to analyze batches over time, or compare products over time, or compare equipment over time, or lines over time, and so on. And, there’s really no limit to the time spans you can look at as long as you have the data. So, if you wanted to, you could actually analyze like every batch that made a particular product on a particular line for the last year. Historical SPC is where you see if your continuous improvement initiatives are really working over the long-term.

Anyway, I could probably go on all day about this stuff.  But, you get the idea. SPC is all about the data, about the math and the rules, about making real-time course correction, and about long-term analyses. If you do all this stuff with SPC, you’ve done a lot.  And, I’m absolutely certain that you’re well on your way to achieving the real business benefits I mentioned in my first post on SPC.  So, I think this is probably it for SPC.  Good luck!

 

VN:F [1.9.22_1171]

Share and Enjoy

Tags: , , ,