AnsweredAssumed Answered

Where to Start In Reverse Engineering a Curved Beam?

Question asked by Wayne Watkins on Jan 7, 2013
Latest reply on Jan 8, 2013 by Mike Lydon

Like • Show 0 Likes0
Comment • 12

We're looking to get a number of similar parts manufactured as soon as possible. We have some old drawings, but they're mostly undimensioned so our vendor is looking for models. No one here has experience with reverse engineering like this and we're not sure of the best approach to take.

We're using Verisurf and a Faro arm, but we're not sure if we should be trying to scan the entire part or perhaps a surface at a time, or how they should be handled in SolidWorks 2013, because so far all I've gotten are really ugly meshes that seem to be of limited use.

Any suggestions or guidance would be appreciated.

Attachments

IMG_20130107_121115.jpg3.2 MB

No one else has this question

Outcomes

12 Replies

Troy Peterson Jan 7, 2013 2:23 PM
Mark Correct
Correct Answer
The old drawings that you have are they in CAD form, I.E. Autocad, or just paper drawings?
Like • Show 0 Likes0
Actions
- Wayne Watkins @ Troy Peterson on Jan 7, 2013 2:25 PM
  Mark Correct
  Correct Answer
  They're old DoD scans of even older paper drawings.
  Like • Show 0 Likes0
  Actions
  - Troy Peterson @ Wayne Watkins on Jan 7, 2013 2:37 PM
    Mark Correct
    Correct Answer
    It looks like a simple enough part. Do you have a QC department? If so they can help you fill in the blanks, radius of the ark, hole locations and so on. If not then you might want to look for an outside company that does dimensional inspection to get you the missing numbers you need to create the part in 3D.
    Like • Show 0 Likes0
    Actions
    - Wayne Watkins @ Troy Peterson on Jan 7, 2013 2:50 PM
      Mark Correct
      Correct Answer
      The engineering department is also, essentially, the inspection department around here. I should have specified that the hole locations aren't important at this point, as we first just need to get the part produced. The curve of the beam is driven by the exterior of the plane we're working on, but I am hoping to avoid going to the loft data if possible.
      Like • Show 0 Likes0
      Actions
      - Troy Peterson @ Wayne Watkins on Jan 7, 2013 2:57 PM
        Mark Correct
        Correct Answer
        Hopefully someone can help you out with your 3D scanner question.
        Like • Show 0 Likes0
        Actions
  - Anna Wood @ Wayne Watkins on Jan 7, 2013 4:36 PM
    Mark Correct
    Correct Answer
    If it is that old it is not real complex geometry. Especially if it is a manual board drawing.
    
    Can you read the dimensions on the original prints?
    
    That looks like it would be very easy to draw from scratch if you can read the old paper prints.
    
    Get a big sheet of paper and do some tracing of the part to paper. Then use some good old fashion descriptive geometry techniques to figure out the radius. Calipers for the hole sizes and spacing. Compare what you come up with against the original drawings. This will require drafting tools such as triangles, a big beam compass and a tape measure.
    
    I think you may be making it harder then it really needs to be.
    
    Cheers,
    
    Anna
    Like • Show 0 Likes0
    Actions
Charles Culp Jan 7, 2013 3:03 PM
Mark Correct
Correct Answer
Since this is, I assume, quite a bit larger than the envelope of your faro arm, I would do this with graph paper and a digital camera. Preferably with an SLR setting the zoom level to 50mm or more. The more zoom the better If you have to use a pocket camera, just zoom in to full optical zoom.

Arrange the graph paper so it forms a full background, taping together multiple sheets. Place the item on the graph paper, and then make sure the camera is on a tripod and plenty far away from the object (you should be 6' or more away). Then use the photos as sketch pictures to model on top.

Do you need more tolerance than this? If you are careful, and you also add in some caliper measurements, etc, you should be able to get within .060 overall.

If you need more tolerance, then the best option is probably to outsource this to someone who has an arm big enough do to this all in one scan. Otherwise you are always going to fight stitching the scans together.
Like • Show 0 Likes0
Actions
Wayne Watkins Jan 7, 2013 3:34 PM
Mark Correct
Correct Answer
We've got a point cloud of scan data from the Faro arm. Granted I have no idea what I'm looking for in a scan like this, but I think it looks pretty good, except for points missing where we had to clamp it down. Is there a trick to using the Scan to 3D tools that won't give me an awful looking mesh, or should I just try something else? Maybe working with one scanned face at a time and then converting from surfaces to solids later on?
Like • Show 0 Likes0
Actions
- Erik Bilello @ Wayne Watkins on Jan 8, 2013 6:00 AM
  Mark Correct
  Correct Answer
  Wayne,
  
  I would tend to agree with Anna and Charles. Even if you have some good scan data, inserting a sketch picture to compare it to can be handy. What format are the scans of the paper drawings you have? I would also try and import those as sketch pictures for reference.
  
  From your picture it appears that the part may have a bit of twist to it and that the rib changes angle, relative to the top of the "T", from one end to the other. Without knowing how the part(s) are used I'm only guessing that the most important face is the "top" of the T. The two long edges of that face are probably most important in establishing the correct shape. They are where I would focus my efforts in getting scan data. If the part does not have a twist you probably only need one edge.
  The profile perpendicular to the "top" face can probably be most easily and accurately determined using conventional methods (calipers, protractors etc.).
  If you can determine the correct shape of the the two "top" edges (And, without seeing an actual part, I think Anna is right that scanning may be overkill for that too.) the rest of the model looks to be fairly easy to make.
  Like • Show 0 Likes0
  Actions
Christopher Figgatt Jan 7, 2013 4:27 PM
Mark Correct
Correct Answer
What software are you using with your laser scanner? This would be a fairly simple part to do in Geomagic or Rapidform. Likewise, it would also be fairly easy to reverse engineer using the point probe with your Faro arm and whatever software you're using. I would start by probing one end of the beam to get a end plane and then measure the various geometry to get the profile of the beam. From there you could measure a spline or polyline to define the curve. Measure the other needed geometry, export to SolidWorks, and use the profile and curve to create a simple sweep.
Like • Show 0 Likes0
Actions
Bjorn Hulman Jan 8, 2013 6:23 AM
Mark Correct
Correct Answer
Looks like a custom weldment with a few cuts and a sketch driven hole pattern to me.
Like • Show 0 Likes0
Actions
Mike Lydon Jan 8, 2013 7:16 AM
Mark Correct
Correct Answer
It looks like some are trying to make this more difficult than it has to be. This looks like a piece of aluminum "I" beam, cut to a "T" and rolled to some radius. Measure the width of the flange and the thickness of the web. Search online. You should be able to find the correct structural section of the base "T". Find the roll radius (lay it on big piece of paper, trace around it, use your high school geometry lesson to find the center point of the arc. Measure it. Open Solidworks, create a matching curve, insert Structural Member (find the correct section in the weldment profiles, or draw from scratch). That should do it. Then add the additional material to the web as shown on the piece, add the holes and you're done.
Like • Show 0 Likes0
Actions

Where to Start In Reverse Engineering a Curved Beam?

Attachments

Outcomes

Related Content

Recommended Content