13
$\begingroup$

I am involved right now in the fabrication of steel structures on the order of 5-50 feet long. Presently, we measure these structures with garden variety commercial tape measures. Most of the time, we work to a tolerance of about +/-1/16" and don't have any problems. Recently, we're trying to make some items to a very high tolerance (at least +/- 1/32", approaching 1/64" or .016") This tolerance is due to visual not mechanical criteria, but it is still very important to our management.

My question is, how can we reliably measure these sorts of distances with that level of precision? I'm prepared to order some NIST traceable tape measures, but it's not clear to me if they'll really improve the situation. Are there other technologies or techniques that could practically be applied in a fabrication setting? Are any surveying tools accurate enough to solve the problem? Cost is obviously a factor, but we're more concerned with repeatability and robustness than price.

I realize this tolerance will sound ridiculous to most, but I imagine that some other industries may have to perform similarly, perhaps large engines for ships or power plant components?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ Locations of space probes at outskirts of the solar system are measured with several cm accuracy... so it depends how much you're willing to invest and how precise you want to go. $\endgroup$
    – SF.
    Commented Feb 6, 2015 at 10:02

3 Answers 3

Reset to default
6
$\begingroup$

To get that sort of accuracy over that scale is not trivial and probably won't be cheap.

For smaller size up to a few meters a portable CMM would be an option (here's an example). These have accuracy on the order of 10 $\mu$m and are used for things like high end/F1 car manufacture.

However, CMM type instruments wouldn't be useful for anything larger than a few meters as it is limited by the arm length and must be fixed in place when measuring to get a sensible result. For larger pieces the best performing option would be so sort of laser tracker (example). These also have pretty good repeatability (~20$\mu$m).

The versions I've seen are phase shifting interferometers, which fire a laser onto a retro-reflector in a metal ball which is placed on the surface. They are used by people like Boeing/Airbus to check tolerances on plane manufacture.

These aren't cheap the laser trackers are approximately $80,000. Conventional surveying laser rangefinders are much cheaper but I'm not sure they would have suitable accuracy.

Improve this answer
$\endgroup$
4
  • $\begingroup$ The portable CMM you listed is limited by the arm length too though. You can move it around, but it's only useful when the arm is stationary. You start moving it in between measurements of the same part and you're going to have useless readings. $\endgroup$
    – Trevor Archibald
    Commented Feb 6, 2015 at 13:18
  • $\begingroup$ Thanks Nivag, I think that may be the kind of thing we're looking for. @TrevorArchibald so are you saying we would need to set the tool up in a stationary position, against some sort of reference surface and always take our measurement from there? That's not ideal, but could be done. $\endgroup$
    – Ethan48
    Commented Feb 6, 2015 at 14:09
  • 2
    $\begingroup$ @Ethan48 Yeah, that's how the Faro arm works. It is still fairly portable, it doesn't need to be calibrated every time you move it, but it does all its calculations in reference to its base, so if the base moves, your measurements are off. $\endgroup$
    – Trevor Archibald
    Commented Feb 6, 2015 at 14:19
  • $\begingroup$ @TrevorArchibald Yes that was my point. Reading it again it wasn't very clear. Will edit. $\endgroup$
    – nivag
    Commented Feb 6, 2015 at 14:38
6
$\begingroup$

Even the best metal tape measure is susceptible to significant thermal expansion over large distances.

Try a laser measurement device ('electronic tape measure') instead: http://www.engineersupply.com/Laser-Measurers.aspx

The laser distance measure, flat plates clamped to the object, and some shims of known thickness, should be all you need to precisely locate something over a large distance.

Improve this answer
$\endgroup$
5
  • $\begingroup$ Thanks! We certainly use laser tapes for very long distances, but as far as I know, even a Leica only promises +/- .040 which isn't much more accurate than we get with a tape. Thermal expansion definitely becomes significant around here. Luckily, temperature is fairly well regulated in our shop, so it might be manageable. I read that the coefficient of thermal expansion is 6.7x10^-6 which I think mean .004" per degree over 50'. More importantly, since both the tape and the structure are steel (and oriented in the same direction,) wouldn't they expand similarly? Is that a naive statement? $\endgroup$
    – Ethan48
    Commented Feb 6, 2015 at 4:45
  • 1
    $\begingroup$ If they're both similar grades of steel, they should expand by very similar amounts. I agree with your ~0.004"/degree F number. If the laser devices aren't accurate enough, all I can think of is some kind of crazy Fizeau experiment done in reverse. $\endgroup$
    – user2790167
    Commented Feb 6, 2015 at 5:00
  • $\begingroup$ Yeah, that's the best I've come up with so far too, but obviously not very practical. From what I understand that's how they calibrate the master tapes in the first place. $\endgroup$
    – Ethan48
    Commented Feb 6, 2015 at 5:06
  • $\begingroup$ How do you know what is the accuracy/error of your measurements? $\endgroup$
    – SF.
    Commented Feb 6, 2015 at 10:05
  • $\begingroup$ @SF, I don't in a traceable sense, but we have checked a selection of tapes against a 15" micrometer standard, and also against each other on large pieces of steel to get an overall sense. $\endgroup$
    – Ethan48
    Commented Feb 6, 2015 at 14:22
4
$\begingroup$

For high accuracy over long distances it's typical to use general surveying techniques. You use a total station (like this) which will get you 1.5mm accuracy in a single shot. They're no laser tape measure. Repeated set ups /readings with some statistical corrections should get you easily below 1mm. Note too that these are immune to thermal expansion of any tape, but of course the subject will be. This too can be allowed for though with come mathematical compensation.

The good thing is that you don't have to buy a total station. You can rent it for a few days. Better still (contingent on budget) is to get a survey guy in to do it for you. This may be the way to go considering the experience required to use this kind of kit. It'll take a beginner a day just to stand it up accurately over a mark. I think that the accuracy you're asking for can be achieved at reasonable cost.

Improve this answer
$\endgroup$
4
  • $\begingroup$ OK, interesting. So the instrument itself is good to about the 1/16" tolerance I'm trying to beat, but you're saying that with multiple measurements there are ways to get more accurate? $\endgroup$
    – Ethan48
    Commented Dec 15, 2015 at 21:08
  • $\begingroup$ @Ethan48 Yes. It's the same as measuring the length of a cigar. Repeat the measurement 10 times and average. You'll get a better accuracy than that from a single measurement. I once spent a whole day up a mountain taking a 100 readings with something akin to this kit. $\endgroup$
    – Paul Uszak
    Commented Dec 15, 2015 at 23:04
  • $\begingroup$ Your problem will be technique and experience. The total station is very complicated and in all likelihood cannot be set up over the fabrication. It will have to be offset. This leads to a shed load of geometry and error dispersion techniques. You might achieve something like a root(n) accuracy improvement where n is the number of readings (rounds). Then there's allowance for thermal movements of the fabrication. Might be best to get a guy in for a day. $\endgroup$
    – Paul Uszak
    Commented Dec 15, 2015 at 23:09
  • $\begingroup$ Yeah, I'm all for bringing in someone with the knowledge. My previous understanding had been that statistical methods could increase confidence in a measurement, but not actually increase accuracy, but it isn't an area I have studied really. $\endgroup$
    – Ethan48
    Commented Dec 15, 2015 at 23:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.