Reverse engineering complex flanges?

[peterblais1]

I wasn't really sure what the best forum would be to put this in, but since I use solidworks- and I need to get this info into solidworks- I figured this would be the best spot.

I've got some complex flanges with odd, asymmetrical bolt patterns to reverse engineer. I do not have access to a faro arm and have been unsuccessful in finding somebody with one local to get some time on it.

One of these parts is like 2 feet long, has an odd border all around it, and about 35 bolts on a funky pattern. I'm not sure that I'd want to sit down and try to measure it all up with calipers.

Ideas so far:

1) buy a huge scanner, and place the object, along with an object of known size- right next to it. Scan it, and import it into a sketch using sketch tools.

2) Setup the part on a platform square to a good digital camera (DSLR, 10+ megapixel). Same idea for the rest. Seems like the biggest catch here is getting a sharp image with high contrast.

Accuracy isn't huge really- if I could get within .25mm that would do.

I worry about the lens on the camera causing any sort of distortion.

Anybody else encounter this kind of stuff? I really wish I could just spring for an arm but it's totally not in the cards at the moment. I have a big one I've GOT to get done though. The worst part is since we don't have an in house laser or waterjet- prototyping is difficult. :doh:

 

[rollupswx]

I'm not sure that I'd want to sit down and try to measure it all up with calipers.

From your problem description it doesn't sound like you have any choice.
Is this flange completely flat - a sheetmeetal part with an irregular boarder and hole pattern?

I have had some success using idea #2 (and lots of hand measuring/plotting/comparison to part).

I don't think the quality of the camera matters all that much - the important parameter is getting the image in focus with minimum of parallax error. The parallax error will be unavoidable, but if your flange is a flat sheet it should be workable - you just need to accept that you can't take the imported image too seriously - you still have a lot of manual measuring to do.

Can you attach an image of the part here?

 

[ctopher]

By the time you wait/search for info, you could have measured it with calipers. That is usually the route I take.

Chris
SolidWorks 10 SP4.0
ctopher's home
SolidWorks Legion

 

[hititfaster]

I agree - set a datum and get measuring, then print your first effort 1:1 and lay the part over it to compare. HOWEVER in the interest of having another option: will it fit on an A0 sheet of paper? If so:

1. Lay the part on the paper
2. Draw round it as accurately as you can
3. Take the trace to a copy shop, have them scan it at 1:1(we have an A0 scanner so I know this can be done)
4. Have them save it in .pdf format
5. DL and install some software (if you don't already have some) to convert from .pdf to .dxf (for example)
6. Paste the conversion into SW.
7. Draw over the lines.

Robert's ya father's brother.

 

[hititfaster]

EDIT: or... have them save it as .jpg, and drop that straight into a SW sketch? Knocks out a couple of extra steps which should increase accuracy.

 

[peterblais1]

Ya, for simple stuff I do just measure with calipers. When they start putting large or funky radii, stuff like that- is when it either gets really time consuming doing all the trial and error. The bolt pattern is usually the easy part, it's the "pretty" looking flange outline somebody drew up that can be a bastard.

I've scanned some small parts before, and it worked OK, along with a decent amount of measuring to try to determine scale, etc.

I was mostly just curious if I was missing any major tricks haha. At least I did just get a printer big enough to print these new parts out 1:1 to make some xacto prototypes

Thanks guys.

 

[DiegoLGraves]

If you are in the SE Wisconsin area feel free to contact me. We have a Faro arm, and 2d and 3d Virtek's that can help reverse engineer.

Regards, Diego

 

[drawoh]

peterblais1,

You can measure and model by triangulation. Select two features and work out the distance between them. Select any other feature and measure the distances from the first two. Use phantom lines in SolidWorks.

JHG

 

[peterblais1]

Ya, usually I take the min and max readings between two bolt holes using calipers, then average them. It's the best way I've found using a standard caliper. I have some bits that can clamp onto the tips, but have never been able to bring myself to trust their calibration. One of those though, is a cone shape which I could simply drop into the holes and get an accurate center to center reading.

I might dig those tips back out and see if I can get them to work decently well. If I had the bolt pattern laid out to scale, then putting a scan or image on top of it to do the outline would be easy enough to accomplish.

Diego- Thanks for your offer, unfortunately we are in Utah.

 

[ArtL]

Peter One way I have had some luck with is lay down graph paper, tape papers togather if needed, Select a X and Y axis on the paper, align hinge as you want/need, pencil profile and use transfer punches to center mark holes, From there you can make a fairly accurate flat part. Then print out a 1 to 1 print. do it in sections if to big for your printer.Check your print out on part and adjust profile and hole C-lines as needed. I have been able to get really close using these basic tools.