<Virindi>
say I have a single tube that I want to split into two tubes. I want the cross sectional area to be constant at every point of flow along the split
<Virindi>
what is the geometry for this?
<Virindi>
obviously, I also want to maintain laminar flow to prevent backpressure
<Virindi>
in the past I modeled this merely as a large diameter chamber with two small diameter tubes coming off the side but clearly this is suboptimal
drfff has joined #openscad
<InPhase>
Virindi: Ah, glad that fixed it. I would have more explicitly suggested an update if I thought it would have impacted your issue. But yeah, the Manifold team is doing some superbly rapid development. (And they have spies *gasp* in here, so say nice things! ;) )
<Virindi>
well I have been running manifold again since april and this is the first time I've had that problem :)
<Virindi>
I had stopped because last year there were too many problems
<InPhase>
As for your tube split, the cross-sectional area is equal if you simply put a divider down the middle of a round tube.
<InPhase>
It's kind of a weird geometry for fluid flow because of the pinched pointy areas, but it's consistent if you break up a circular cross-section into two half circle cross sections.
<Virindi>
hmm. I suppose, if it is infinitely thin, but then you'd need a constant cross section transition from half cylinder to cylinder of smaller diameter
<InPhase>
If you want to avoid the pointy bits, then your next geometry choice is a space-wasting dual oval split.
<Virindi>
how would you model the transition from half a cylinder to a full, smaller diameter cylinder?
<InPhase>
Well putting two cylinders inside of one cylinder gives you 1/4 the area for each.
<InPhase>
The arrangement is kind of obvious. It's two cylinders right up against each other.
<Virindi>
I was planning to convert one 3/8" ID tube to two 1/4" ID tubes, which is very close to equal cross section (yeah I know, "what is an inch???")
<InPhase>
I'm American. Inches don't scare me. But violations of the space time continuum alarm me, and you can't fit two 1/4" tubes into a 3/8" tube. :)
<Virindi>
no.
<Virindi>
it is a coupling. on one side, air comes in through one 3/8" ID tube. on the other side, air leaves through two 1/4" ID tubes
<Virindi>
the goal is to minimize backpressure in the coupling
<InPhase>
Ooooh. You're looking for a coupling that minimally disturbs flow.
<Virindi>
yes.
<InPhase>
That's a totally different problem.
<Virindi>
like I said, I want 1) constant cross section, and 2) laminar flow maintained
<InPhase>
You had phrased it as wanting to split the tube. :)
<InPhase>
Well it's not going go to be perfectly laminar at the junction. But you can make it decent.
<Virindi>
of course an approximation is okay
<Virindi>
it's just that my previous approximations have been poor approximations
<Virindi>
I was hoping to find a better one this time
* InPhase
ponders.
<Virindi>
I'm also american and I hardly know what an "inch" is :P
<Virindi>
I have tape measures which only have metric
<Virindi>
and yes, I do know that flow friction is much greater in two 1/4" tubes than one 3/8" tube as well
<InPhase>
Alright. So your sizing approximation on the split is 12.5% off, which buys some freedom for how good the junction approximation needs to be.
<Virindi>
so by its nature it is still, uh, a "change in impedance"?
<Virindi>
yeah 2x 1/4" should be 0.353553
<InPhase>
It's obvious one can write a numerical solver that will make an optimal transition. But first one should try stupid things and evaluate goodness. Have you already done a hull?
<Virindi>
well, obviously I wanted to model it in openscad, that is why I came here ;)
<Virindi>
what would you hull, the two small tubes coming in at angles and the face of the large tube abruptly ending? that's okay I think
<InPhase>
A hemicircle input and the circular output.
<InPhase>
That's my first take at the simplest OpenSCAD style approach.
<Virindi>
that looks pretty good
<InPhase>
Wrap that into a module with the right smartness on it to make inside and outside and difference for hollow.
<Virindi>
that's no problem
<Virindi>
that also looks like it might print nicely with the two tubes on the bed at -Z
<Virindi>
I've been using openscad over 10 years, I know how to do a module, I have just been having some trouble thinking about the best geometries for this :)
<Virindi>
I like your way though, of just splitting the cylinder in half. I was really overthinking it.
<InPhase>
Yeah, I know you've been around for a while, but can't remember your background. :) There are too many backgrounds to track.
<Virindi>
well thanks :)
<InPhase>
Good luck with your pipes. :)
RichardPotthoff has joined #openscad
snaked has joined #openscad
hyperair has joined #openscad
<J24k63>
Reminder - scADVENT scADVENT is coming - Ü
Paul has joined #openscad
Paul is now known as Guest9753
Guest9753 has quit [Client Quit]
mmu_man has joined #openscad
mmu_man has quit [Ping timeout: 252 seconds]
cart_ has joined #openscad
<Scopeuk>
I have toyed before with making a sort of day by day unlock scifi robot (ideally printable) but really didn't get very far. not really sure it's the best way to go either
JakeSays_ has joined #openscad
JakeSays has quit [Ping timeout: 268 seconds]
ccox_ has quit [Ping timeout: 272 seconds]
ccox has joined #openscad
ccox has quit [Ping timeout: 256 seconds]
ccox_ has joined #openscad
Guest77 has joined #openscad
<Guest77>
Hello. I'm wrestling with "ERROR: The given mesh is not closed! Unable to convert to CGAL_Nef_Polyhedron."
<Guest77>
I've been using OpenSCAD for about 10 years, and this is the first time I have been stumped.
<Scopeuk>
that is most commonly seen when performing operations on a imported stl which is currupt
<Scopeuk>
are you able to share your model on a pastebin?
<Scopeuk>
pastebin?
<Scopeuk>
pastebin!
<Scopeuk>
hmm bots not working
<Guest77>
No, it's a union of two objects generated by a function.
<Guest77>
So, I upgraded. Now I get an error, and no empty shell.
<Guest77>
Hooray, I guess?
<Scopeuk>
my suspicion on this is that this is a floating point error of some description (edge is just either side of some boundary or other)
<Guest77>
Right. I did have it as a very thin element before, at 0.01mm. It was visible in preview though. I've upped it to 0.1mm and compensated the measurements.
<Guest77>
Also, the fascia_outline() used to have the five parts butted up to each other. I edited it to overlap a little, in case the output object had some thin seams within it. No dice.
<Guest77>
Ok. Your STL does look like what I want, thanks. There are a couple of artifacts in Cura, but I could probably ignore them.
<Scopeuk>
it is very strange that my release version and your release version produced differnt results, that is very strange
<Guest77>
Yes. I've tried a whole bunch of things. It's frustrating that I can render either element of the union, but if they are both in the union the rounded part disappears.
<Guest77>
I might try a nightly release. I've generally steered clear of them in case something goes horribly wrong. As I said, until today I was using 2019.05
<Guest77>
Anyway, thank you for your help. I appreciate it. I'm in New Zealand and it's bedtime here, so I'm going to log out. Thanks.
<Scopeuk>
you can have nightly and normal installed at the same time
<Scopeuk>
just put it in a differnt folder when you install
<Scopeuk>
Any way it is very much middle of the night on the other side of the world, feel free to drop back in some other time
JakeSays has joined #openscad
JakeSays_ has quit [Ping timeout: 264 seconds]
hyperair has quit [Remote host closed the connection]
<plasma41>
Is there a way to do a linear extrude along the Z axis that scales the X dimension of the 2D shape but not the Y dimension? The linear_extrude function seems to only support scaling both or neither. Alternately is there a way to apply a separate scale to each of the X and Y dimensions?
<peeps>
plasma41: yes, use scale parameter with an 2 element vector: linear_extrude(..., scale=[x,1])
RichardP_ has joined #openscad
RichardPotthoff has quit [Read error: Connection reset by peer]
muesli has quit [Quit: NO CARRIER]
muesli has joined #openscad
<plasma41>
peeps: Excellent, thanks.
<plasma41>
lol, I got my additive and multiplicative identities mixed up. I need to set my Y scale to 1 not 0.