TUBE twisting

[Bruce WalkerParticipant]

This is a reasonably old topic on Archicad-Talk, but I thought I’d bring it up here.

When the axis of a TUBE path changes, there are twists that occur in the profile. The specific case I have encountered was when scripting for a bent spring (such as wraps around the hose on a commercial kitchen tap). The maths works fine (and the same maths works without issue in SketchUP using Ruby with the FollowMe command), but the TUBE wants to twist. No amount of playing with the angle setting will fix this either.

I have also tested the new TUBE{2} command, but the same issue occurs.

Is it possible to fix this?

The original post: https://archicad-talk.graphisoft.com/viewtopic.php?f=6&t=51395&p=277977

Attachments:

1. 

   spring-1.jpg
   

www.brucepwalker.com
https://www.mindmeister.com/65450406

AC20 6005 AUS Full & AC 21 3009 NZE Full | Win 10 Pro 64 | Intel i7-7700HQ 2.80GHz | 16 Gb RAM
NVIDIA GeForce GTX 1060 8Mb

[Péter BaksaParticipant]

Hi,

The TUBE commands were designed for objects with much less twist, where some auto-correction of the profile rotation comes handy. We don’t want to fix that in the TUBE commands, that would break compatibility. There is a solution however: in a library update in version 21, the “gravity_tube” macro was added.

As you can read in the documentation, the W axis of a cross-section at each particular node of TUBE is tied to the Z axis. With a spatial helix, this makes no sense. Using “gravity_tube”, you can set this W axis freely. In your spatial helix, this gravity vector would be pointing from the path point to the axis of the helix.

Based on your example object, here is a sample code that models a bent helix nicely:

resol 16

! model a helix bent 180
mat = 45		! glass

rWire	= 0.002
rCoil	= 0.011
nCoil	= 18		! number of turns around axis
nTurn	= 36		! number of points in one turn
R	= 0.1		! radius of bending
dFi	= 180 / (nCoil * nTurn)

dim axis[][3]
dim helix[][3]
dim gravity[][3]
n = 0

for fi = 0 to 180 step dFi

	n = n + 1

	! compute points as a helix on a straight cylinder, then bend it
	aCoil = fi / dFi * (360 / nTurn)	! angle around axis
	xCoil = cos(aCoil) * rCoil		! x, y on cylinder
	yCoil = sin(aCoil) * rCoil
	rTurn = R - yCoil			! turn radius of vertical line on cylinder

	! helix axis point
	axis[n][1] = 0
	axis[n][2] = cos(fi) * R
	axis[n][3] = sin(fi) * R

	! helix point
	helix[n][1] = xCoil
	helix[n][2] = cos(fi) * rTurn
	helix[n][3] = sin(fi) * rTurn
	
	! vector from helix to axis
	gravity[n][1] = axis[n][1] - helix[n][1]
	gravity[n][2] = axis[n][2] - helix[n][2]
	gravity[n][3] = axis[n][3] - helix[n][3]

next fi

! show helix
for i = 1 to n - 1

	lin_ helix[i][1], helix[i][2], helix[i][3],
		helix[i + 1][1], helix[i + 1][2], helix[i + 1][3]

next i

! show helix axis and put helix as tube path
for i = 1 to n

	hotspot axis[i][1], axis[i][2], axis[i][3]

	put helix[i][1], helix[i][2], helix[i][3], 0

next i

! try TUBE
TUBE 2, n, 1 + 2 + 16 + 32,
	0, 0, 901,
	rWire, 360, 4001,
	GET(NSP)

! try "gravity_tube"
add 4 * rCoil, 0, 0

! parameters for gravity_tube (derived from TUBE{2}):
!		top_material, bottom_material, cut_material,
!		n, m, mask,
!		u1, w1, s1, mat1,
!		...
!		un, wn, sn, matn,
!		x1, y1, z1, angle1, gravity1_x, gravity1_y, gravity1_z,
!		...
!		xm, ym, zm, anglem, gravitym_x, gravitym_y, gravitym_z

put mat, mat, mat,
	2, n, 1 + 2 + 16 + 32,
	0, 0, 901, mat,
	rWire, 360, 4001, mat
	
for i = 1 to n

	put helix[i][1], helix[i][2], helix[i][3], 0,
		gravity[i][1], gravity[i][2], gravity[i][3]

next i

call "gravity_tube"

del 1

PS. “gravity_tube” might give some warnings with strict error checking options, that will be corrected in the next main release.

Péter Baksa
Software Engineer, Library Team
GRAPHISOFT SE

[Bruce WalkerParticipant]

That’s great Peter – thanks for that. I’ll give it a whirl.

www.brucepwalker.com
https://www.mindmeister.com/65450406

AC20 6005 AUS Full & AC 21 3009 NZE Full | Win 10 Pro 64 | Intel i7-7700HQ 2.80GHz | 16 Gb RAM
NVIDIA GeForce GTX 1060 8Mb

[Bruce WalkerParticipant]

Hi Peter,

Finally got around to trying this out. It works…mostly. There is still one twist near the apex (see attached image). I haven’t delved into the code to figure out why. A promising result, though.

BTW – two corrections to your code: there are two ‘next i’ nestings, and the ‘n’ in the first TUBE command doesn’t work (I replaced with NSP/4).

Attachments:

1. 

   2018-03-30_12-13-32.jpg
   

www.brucepwalker.com
https://www.mindmeister.com/65450406

AC20 6005 AUS Full & AC 21 3009 NZE Full | Win 10 Pro 64 | Intel i7-7700HQ 2.80GHz | 16 Gb RAM
NVIDIA GeForce GTX 1060 8Mb

[Péter BaksaParticipant]

Hi,

The forum engine got rid of [i] even in the code, so we turned off the italic BBCode feature. Please try the code again, now it should work.
If you still see the twist at the apex, please write back the library version you are using. (gravity_tube has been corrected in a library update, it’s possible that your version doesn’t contain that update yet.)

Péter Baksa
Software Engineer, Library Team
GRAPHISOFT SE

[Bruce WalkerParticipant]

Thanks Peter. The code works fine now, but there is still the twist at the apex. My library version (which says it’s up to date) is 4550.

Attachments:

1. ARCHICAD-Library.zip

www.brucepwalker.com
https://www.mindmeister.com/65450406

AC20 6005 AUS Full & AC 21 3009 NZE Full | Win 10 Pro 64 | Intel i7-7700HQ 2.80GHz | 16 Gb RAM
NVIDIA GeForce GTX 1060 8Mb

[Péter BaksaParticipant]

Hi Bruce,

there will be an NZE library update available sometime this week, that should fix this problem too.

Péter Baksa
Software Engineer, Library Team
GRAPHISOFT SE

[Author]

Posts