Trying to create a 1 -> 8 pipe manifold.

Creating the outside of the pipe is OK, creating the body of the pipe is also OK. Trying to perform a boolean subtract to remove the inside from the outside is very slow. Am I doing this the right way?

Thanks, Andy

Switching to top view and trying to add a cplane on the origin. As soon as I move the pointer anywhere near the origin MoI freezes, expands to the full screen and becomes unresponsive for several minutes before automatically exiting the cplane mode. Does this every time.

Here is an updated version. In this I have created all eight hollow pipes but the problem is how to deal with the area where they intersect so that is also hollow.

Thanks, Andy

Andy - I don't have a solution as I can't get the 8 tubes to Union into one piece.

I would instead use Rail Revolve to make a tapered, hollow cylinder with a single inlet hole on top. Then attach the 8 outlet tubes to the cylinder, rather than attach to each other. Boolean Union all and add fillets.

Also, if it's possible to arrange the 8 tubes in a circle then you can build just one tube and use Circular Array to get the rest.

Ed Ferguson

Ed - good idea! I will try that. Andy

A quick model. I'm sure something more attractive can be designed :)

I used the Pipe2 script to make the tubes. The script makes it easy to define the exterior diameter as well as the wall thickness.

Is this going to be 3D printed?

Ed Ferguson

Hi Andy,
you can get a result by producing one leg 7.5 rad & the second leg 7.49 rad then boolean union.
Then produce the inner with 5 rad & the second leg 4.99 rad then boolean union.
Boolean subtract inner from outer.
Now boolean subtract a 45 degree segment as shown in the photo.
You can now mirror a few times to get the complete manifold that can be boolean union to give a solid.
Cheers
Barry

Hi Andy, it's difficult for booleans when pieces just barely graze each other. In your case here the top areas of the different pieces are not very precisely aligned with each other, here is what I got after untrimming some of your pieces that had some holes cut out:





So the first step wiill be to rebuild these with a tighter accuracy so all the pieces end on the same plane to start with.

- Michael

Thanks Barry and Michael.

Michael - how do I tell the sweep to make sure the surfaces all end on the plane for the end circle? It's sweep that created those misalignments.

Andy

Hi Andy, so I did a closer examination of these 2 curves here:



When building sweeps for them:



If you zoom in to the top you can see they are not aligned:



The root problem seems to be inaccurate end tangents on one of the curves - instead of having an end tangent going exactly along the z axis direction it is off at a slight angle:






So my first step to clean this up will be to go through and tune up all the curve points to be aligned to axis directions where they should be.

- Michael

Hi,

One big problem with 3D software and boolean operations is on the coincident surface or edges.
So to avoid this problem you can do like Barry with small difference on radius.

My solution is similar as Barry's and I have made The 8 sweeps.
- The first one use two circles with 7.5 and 7.5 mm radius
- the 7 next use two circles 7.5 and 7.45 mm radius to "enter" without be same as biggest.
The boolean union is possible with this construction.

To avoid the problem to do the holes inside, I have moved the top and bottom points of curves 0.1 mm to have non coincident surfaces on main solid and the 8 solids made to have holes (with boolean difference).

Hi Andy, the other thing that's pretty difficult in your original one is even when the curve tangents are tuned up, the way it's set up here has a quite complex intersection between the pieces (here shown during a mesh export done to a very high density using "Divide larger than" to analyze the surface shape interactions):





It would probably help if the rail's curve tangent aligned with the z axis direction a little bit before the end of the curve instead of only right at the very end of the curve. That would probably get the surfaces to be the same shape near the top rather than the different shapes for neighboring surfaces like shown above.

- Michael

Hi Andy,

re:
> Michael - how do I tell the sweep to make sure the surfaces all end on the plane for the end circle?
> It's sweep that created those misalignments.

You've got to make the rail curve to be more accurate - currently some of your rail curves have an end tangent that is out of line with the world z axis (see above screenshot). So that introduces a kind of slant into the sweep.

- Michael

I wrote previously:

> It would probably help if the rail's curve tangent aligned with the z axis direction a little bit before the end
> of the curve instead of only right at the very end of the curve.

This didn't seem to work out very well. It's probably better to hack off a slice off the top of each piece before trimming them against each other, to avoid this complex zone here:





- Michael

Michael - that was my plan, however I went with Barry's approach and that seemed to automatically take care of the problem.

Using a difference in diameters of 0.01mm resulted in a huge improvement in boolean performance.

Andy

Hi Andy, here's the result from tuning up all the curve tangents, then remaking the sweeps, then slicing off the top and trimming surfaces against each other and then patching in the top part separately. Is this going to work for what you need? :



- Michael

Hi Andy,

re:
> Michael - that was my plan, however I went with Barry's approach and that seemed to
> automatically take care of the problem.

That's great, I'm glad you've got it solved! Thanks Barry!

- Michael

Cool result!

Filling 8 beer bottles at a time?