Statistical process control (SPC
) offers many advantages for manufacturers, no matter what industry you may be in. I love talking about SPC because it is the bread and butter of today’s manufacturing.
It’s nothing new, been around since 1928, so there’s really no need to sell it to anyone. It sells itself with its proven track record. The only difference between SPC today and back “in the day” is that now all the calculations are done by computer. But it’s still the same in its most basic way: the focus of SPC is, and always has been, on analyzing results and selecting the right sampling strategies. More on that later.
Right now, for this blog, I’d like to talk about what I see as the top advantages of SPC
in the modern manufacturing environment. Like I said, SPC sells itself, but who doesn’t love a good top-10 or top-5 list? I know I do…
See into the Future
My number one advantage for using SPC is that it is a way in which you, the manufacturer, can see into the future. That’s right! You can predict the future with SPC. That’s why this is advantage number one for me.
Way back when, I worked for a large aircraft manufacturer and one of my responsibilities was teaching SPC classes. I loved it. I had machinists, operators, and professionals of all kinds in my classes. I would always ask the question, “If we had a 30-second peek into the future, would any of you be sitting in this class right now?” Heck no! “OK. Make it 10 seconds! Would you be here?” In one form or another, they would all say, “No way. I’d be in Vegas. I’d be a millionaire in one day.” I love that.
Head ‘Em Off at the Pass
If you had a tool that could offer you that 10-second peek into the future, then you would know what was going to happen. Sure, you could make yourself personally rich, that goes without saying, but put it into a manufacturing context—and the result is you could make your business more profitable, right?
Think about it. If you knew what to expect from a process, you could easily make the minor tweaks necessary to improve your manufacturing. Leave it alone. Adjust here. Adjust there. You could predict the costs associated with rework and plan for it. You could order enough material ahead of time to offset any issues.
A 10-second glimpse into the future would be invaluable.
Save Machinery Wear and Tear
In the long term, SPC can save your machinery unnecessary wear and tear. If a production line machine is just a little out of whack, it can create defects that you have to address at some point. If properly maintained, it’s one less headache for you to worry about.
In the short term, SPC can increase the efficiency of your machines. When my wife and I built our home, it was so great to watch the contractor work his magic with the schedule. If he knew the tile guy was a few weeks out from getting to our job, the contractor would plan his schedule around that and get other things done in the meantime. But if the tile guy, or any other subcontractor, showed up at random, then that throws a monkey wrench into the whole deal.
Knowing exactly how your machines are working enables you to look ahead in the schedule, plan accordingly, and avoid major setbacks. Your machines make your products, so knowing all you can about how they are working is precious information.
Manufacturers make what are called “machine matching” and “make-or-buy” decisions all time. Machine matching is just what it sounds like: matching the job to the right machine.
I think it’s common knowledge by now that if you have 20 of the exact same machines lined up, they will each output something a little different. Machines have personalities. They have strengths and weaknesses. Matching the right machine to each job is important. It’s another way to save machine wear and tear. And it’s a sure-fire way to ensure that the job is done right and on time. But to do so, you need to have information about your machines. You need data to show you how each individual machine is performing. That’s where SPC analysis comes in.
Statistics and Machine Matching
To explain, I need to step back a moment. Let me begin with this: The two statistics that drive SPC are mean
. That’s where we derive all the knowledge from SPC. So, based solely on the history of a given machine, I can make the determination that machine A is better suited than machine B for the job we’re looking at.
Let me put it like this: machine A produces rework and machine B produces scrap. Machine A has a feature that we’re monitoring that has a tendency to have “fallout” on the high side (out of spec)—I can work with that; I can rework that. For machine B, the feature I’m keeping an eye on falls out on the low side, which is scrap. If machine A produces 10% rework and no scrap, versus machine B’s 1% scrap and zero rework, I might want to rethink how I’m using the machines. This is valuable information. Is it more profitable for me to scrap 1% and not have to spend any money on rework? Or for me to rework 10%?
At some larger companies, machine matching is a full-time job. SPC can help you make decisions like this quickly and easily. Never again use a machine that can run tight tolerances on a job that doesn’t require tight tolerances.
Human nature is, "Well, heck. I'll run all my jobs on the best machines that I've got, even at the risk of other machines sitting idle, because I don't want anything bad coming out of my area." And by making those decisions, my throughput in that area goes down because I'm wasting some resources in there, and I'm using these highly capable machines to run jobs that don't need that capability. Make sense?
[My colleague, Doug Fair, InfinityQS COO, wrote at length about the “personalities” of machines and the operator-artists who run them. Please check it out here
Lots of companies have customers who call them up and ask, "Hey, can you do this for me?" And so, the estimators will look at what they're being asked to produce and give them a quote. And the quote is based on: what is it going to cost our shop to make that part, plus some profit? And if the estimator assumes that the machines they have can produce those types of tolerances on those types of features, but your machines can’t handle that, then all they're doing is introducing a big bunch of headaches to the guys on the plant floor who are being asked to make these parts.
And the operators are wondering, “Why do we keep accepting these jobs? They know we can't make these things. We're going to have 20% scrap and all this rework. But they keep volunteering us. They want the work, so they'll take these in. But we're losing money on them." You hear all these complaints on the plant floor because of all the scrap and rework.
And so, by the right people understanding what the true process personalities are, and the real capabilities of machines that they have on the plant floor, they could do a much better job of what are called “make-buy decisions.” Then, when they get a customer asking about a job, they decide, "What parts of this job am I going to make myself in-house, and what parts am I going to ship out to somebody else—a third party–to make for me, as my supplier." Do I make it? Or is it more profitable for me to buy it?
SPC makes those decisions clear and decisive. Because you have all the information you need to determine whether to make or buy.
Understand Your Data Better
As I mentioned early on, SPC has been around for a long time. Today, most of the work in SPC is performed by computers. That leaves you and me free to focus conceptually on what we're trying to learn from this data and not worry about the time and effort and knowledge and skill needed to do the number crunching. We can focus on analyzing the results.
SPC takes mountains of data—and that can be billions of data points—and turns them into very simple graphics. We’re talking about simple time-ordered plot points, histograms, and just simple graphical representations of the data that expose everything and more about what you’re focusing on.
Simple graphical representations of complex data streams, so we can quickly evaluate the data.
Breaking it Down
In SPC, it’s all about variation. Somewhere on your production line something is causing variation…an unwanted defect.
Your first step is to find where the defects are coming from, which was the topic of last week's blog
. Once you find that, then you look at all the inputs into that process that are producing those defects. What's the defect rate? What's the nature of the defects? The variation in the inputs quite likely are what's causing the defects on the output. Often, once you find out where the defects are coming from, it's not so difficult to know how to fix it. But sometimes it is difficult. Sometimes we know exactly where things are coming from, but we have no idea what's causing it. So, you’ve got to dig deeper.
You must look at all the process feature streams that are coming into the process, analyze those, and see which of those aren't set properly or have too much variation, and are creating those defects. And SPC is the tool to do that.
Because the graphical representation you’re looking at—the output from your SPC tool—is time-related, it’s easy to spot trends and anomalies, and it can help you pinpoint root causes. A simple, time-related chart can give you so much more information than a snapshot in time.
SPC—the Great Correlator
SPC realizes that all data is not equal. Let’s face it, some data are more important than others. Even if it's from the same data stream. So that's why so many of the tools of SPC are exception-based triggers. SPC tools hunt for variation, and when unexpected or excessive variation is found, alert the user. So, what that means is that you’re not wasting your time combing through stacks of graphs and charts to look for issues. Let the software tell you, "Oh, here's something to go look at. And if you want to see it, yeah, here's the graph for it." So SPC provides evidence of whether something has changed, or something has not changed. Do something or do nothing.
That’s the correlation power of SPC. Let’s take the example I always use of paint thickness. My job is to paint a particular part in the manufacturing process. And I want to get the thickness of the paint just right. When the paint is wet, I can’t touch it with anything, because I’ll mess up the finish, right? So, we wait for the paint to dry before we know the true thickness.
Anyway, paint is a combination of solids and volatiles. The volatiles, like alcohol, evaporate off. And once that happens, all you’re left with are the solids. Then you can measure thickness. Makes sense to me. So, rather than controlling the paint booth by tracking paint thickness, the operator correlates the percentage of solids in the paint with SPC correlation methodology. The control charts here verify the percentage of solids in the bucket of paint you’re using, controlling how long the sprayer stays on the area you’re painting—thus controlling paint thickness.
You take the information that SPC correlation provides and adjust and compensate until you get the thickness just right.
So, there they are, my top advantages of SPC. To summarize:
- See into the future
- Save machinery wear and tear
- Understand your data better, and
- SPC as the great correlator
This list can go on, obviously, and it will. I didn’t get a chance to go into any depth regarding sampling strategies, so I’ll do that in a future blog. Thanks for joining me in this SPC blog series. Next up? Zero Defects! Please come back.
To read the first blog in this series:
Take advantage of the technology at your fingertips today: contact one of our account managers (1.800.772.7978 or via our website
) for more information.