And an interesting follow-up:
From: nicolet.com (Doug Pfrang)
Subject: Re: Calibration/Pfrang/Volker/Pfrang
> Your example is not very illuminating for real-world problems. You
use
>a wrench and it turns the nut without damaging it-- fine. Your wrench
>is not an instrument, it's a tool. People don't calibrate "5 mm
>wrenches". You would probably want to calibrate a torque wrench,
>however. How do you "brute force validate" that??? Say you're
>installing motor cylinder heads. You do 100 motors and everything seems
>fine. "Brute force validated"? What happens if in a year some of them
>start warpping because they were unequally stressed? Would anyone ever
>even know that the problem goes back to your torque wrenches? Does it
>make any difference if no one figures out the root cause?
>
> Is it possible to "brute force validate" anything, if your process
>produces occasional non-conforming products?
>
> In the real world we often don't know why a product fails-- there
are
>too many variables in production to pinpoint the exact cause of
>failure. That's why quality systems such as ISO 9000 lay such great
>importance on constancy of process and calibrated measurements.
OK, maybe my example was a bit too simple, so let's back up and see if we can clear things up. First of all, let me repeat that the main point of my posting was that calibration is a SUBSET of validation. This is a really key point, because many people have been posting questions about calibrating things that obviously could be validated more easily by brute force; namely, things like tape measures and injection molding machines. My point is that IF you can validate a device more easily by brute force trial and error, then you do not HAVE to calibrate a device just because it has numbers on it or just because it makes a measurement. You CHOOSE whether or not to calibrate a device based on whichever method of validation makes the most sense in your particular process.
Second, when it comes to CHOOSING whether or not to calibrate a particular device for a given task, it makes no difference whether we label that device a "tool" or an "instrument." ANY device (even an "instrument") CAN be validated by brute force, because the only thing you care about is whether that device does what you want it to do in your process. A 5 mm wrench, a torque wrench, a voltmeter, a soldering iron, whatever. They are ALL "tools." If you want to label some of them "instruments," that's fine, but I think that tends to confuse people into believing that the decision process is different for "instruments" than it is for "tools." In fact, it's not, and that's another point I'm trying to make. Whether we call something a "tool" or an "instrument" doesn't change the analysis at all.
Let's take the torque wrench example. In your example, the engines fall apart a year later because of UNEQUALLY stressed parts. This problem is obviously NOT a result of the tool being UNCALIBRATED; it is a result of the tool being DEFECTIVE: the bolts are UNEQUALLY stressed because the tool's readings are not REPEATABLE. You cannot resolve this problem by CALIBRATING the tool; you must resolve this problem by REPLACING the tool, because, in your example, the tool is broken. It can't make repeatable readings.
Now, let's use the same tool, but let's use a torque wrench that isn't broken. You're building an engine, you want to torque the head bolts to a torque that will hold the engine together, and you have a torque wrench that is not calibrated. The question is: do you need to calibrate that torque wrench or not? Well, as I said before, the answer to this question depends totally on YOUR PROCESS, NOT ON THE TOOL. Let's say YOUR PROCESS is to build engines where SOMEONE ELSE has already determined what the torque should be. Ford Motor Company TELLS YOU that the torque on those head bolts must be 85 ft-lbs. OK, fine, FORD has VALIDATED that torque value by some method (possibly their own trial and error), so YOU don't need to VALIDATE whether 85 ft-lbs is right or wrong. All YOU need to do is ensure that you torque the part to 85 ft-lbs, and you know ahead of time that this value will work. How will you do that? Well, probably by calibrating your torque wrench so you can torque the head bolts to 85 ft-lbs.
But now, let's change the process. Let's say YOUR PROCESS is to build engines where someone else has NOT already determined what the torque should be. You're building a custom race engine, using your own head casting, and you don't know what the torque should be because no one else has VALIDATED the value for you. So, does it matter whether your torque wrench is calibrated or not? No, it doesn't, because you don't know if the torque should be 25 ft-lbs, or 90 ft-lbs, or 115 ft-lbs. Sure, you might use the numbers on your torque wrench as a REFERENCE, because in your experience you believe that the torque should be somewhere between 50 and 100 ft-lbs, but you don't know what the torque really needs to be. So, calibrating your torque wrench does NOT answer the question of what the torque needs to be. HOW do you determine what the torque needs to be? By brute force trial and error. You torque the part to some value and give it a try. In this example, it really doesn't matter if the torque wrench is calibrated or not, because you're not trying to hit a specific torque value that someone else as validated for you; you're just trying to hit a value that holds your custom engine together. In fact, you could just as easily use a torque wrench that has no numbers on it at all -- just a blank dial -- and you could simply put a mark on the dial that corresponds to how far you torqued the head bolts each time. Then, if you discover that you torqued the bolts too tight or too loose, then you just move the mark up or down a little until you get it right. Once you get the torque right, you just use your little mark on the dial as your personal indicator for your custom engines; and this method works just fine in YOUR PROCESS, because you're not trying to match some factory torque spec that someone else validated for you. Now, do you need to run out and calibrate that torque wrench? No, you don't, because you have VALIDATED that torque wrench in YOUR PROCESS by brute force trial and error.
And the same analysis applies to ANY device in ANY process, and it makes no difference whether the device is a "tool" or an "instrument."
> Once you start covering yourself a lot of documentation, then you're
>really just "calibrating" with your own home-made procedure; the
>distinction between calibration and brute force validation seems lost to
>me, then.
That gets back to the main point I was trying to make: that calibration and brute force validation are just two sides of the same coin. The mistake that I see people making all the time is that they see the record-keeping requirement in the ISO-9000 standard and they think they need to run out and hire a third-party to calibrate every piece of equipment in their facility. As a result, they talk about calibrating things like tape measures and injection molding machines, and some of them spend a ton of money calibrating things that don't need to be calibrated. That's wrong. What I'm trying to do is get readers to see that:
(1) they have an ALTERNATIVE to calibration -- namely, brute force validation -- which they may use as an OPTION in appropriate situations; and
(2) the decision-making process that people should go through to DECIDE which validation method is best depends ENTIRELY on each facility's SPECIFIC PROCESS and NOT on the TYPE OF DEVICE being used.
True, no matter which validation method you chose, the ISO-9000 standard says you need to have records showing that your equipment is validated, but HOW you VALIDATE your equipment is up to you. In some situations, it makes more practical sense to VALIDATE a device by CALIBRATION, but, in other situations, it makes more practical sense to VALIATE a device by BRUTE FORCE. Whether you call your equipment "tools" or "instruments" DOES NOT MATTER. What matters is that you validate your equipment, and that you have records to prove it.
To sum up: the ISO-9000 standard mandates that you keep validation records for every device in your facility that impacts product quality, but it does NOT mandate HOW you validate every device in your facility that impacts product quality.
-- Doug Pfrang
