Modeling
About Archicad's design tools, element connections, modeling concepts, etc.

Railing Fundamental Error

hkrna
Enthusiast

I think I figured out why it is almost impossible to model the most basic railing in Archicad, when the railing is turning a tight on a stair or ramp corner.
The problem is that the railing tool uses the posts centre line to define the reference line. The Railings are then offset from this line. It is a basic geometry issue, that due to this offset, a straight portion of the rail would be necessary to generate a level connection. In real life, the reference line is the centre or inner most line of the railing, therefore the turning can be done easily. The balusters and posts are then offset from this reference line.
This is also the logic of the Regulations and Standards in all countries that I know, because the important dimension is the clear distance between the railings. (specifically, not the centre line, but the inner most point of the rail profile)
It would be much appreciated if GS could improve the railing tool to give option to set the reference line to any of the Rails instead of the posts centre line.

hkrna_0-1659189060887.png

 

ArchiCAD since 4.5

Windows 10
8 REPLIES 8
Barry Kelly
Moderator

I don’t have Archicad in front of me at the moment.

But look at the node connections.

You can control how the connect there.

Disconnected, goose neck or direct connections.

I am guessing direct connection is what might suit the best.

note, there are different connections for the top rail, hand rails and other rails, so make sure you are setting the correct one.

 

Barry.

One of the forum moderators.
Versions 6.5 to 27
Dell XPS- i7-6700 @ 3.4Ghz, 16GB ram, GeForce GTX 960 (2GB), Windows 10
Lenovo Thinkpad - i7-1270P 2.20 GHz, 32GB RAM, Nvidia T550, Windows 11

Barry, if the connection controls would solve this issue, I would not have raised it here. 🙂
Typically this problem occurs on every single stair / ramp where the stair / ramp turns 90 deg (or other). The problem is always the inner rail.
If I have space to add a flat portion to the ascending stair (typically an extra going in stairs) the geometry problem is solved.
Here down under it is a regulatory requirement that a railing has to be continuous, going at the same height, with no vertical (goose neck) connection to other rail.
Unfortunately, I am afraid, the only solution would be to allow the reference line to be set by the rail, not the posts.

Just another little annoying thing (not little)
Why can't we lock an associated dimension to the inner point of a rail? Archicad recognises it as a point, yet the dim will not be associated.

hkrna_0-1659193261323.png

 

ArchiCAD since 4.5

Windows 10
Barry Kelly
Moderator

You are correct, only if your railing posts and the rails are centred on the reference line of the railing (the edge of your ramps), will the rails join perfectly at the corner.

By offsetting the post onto the ramp, you are adding half the width of the post to the length of the rails.

And by offsetting the rail to the side of the post, you are adding the width of the rail to the length of the post.

Then the centre axis of the rails will no longer meet at the corner without a flat transition length.

 

BarryKelly_0-1659359458170.pngBarryKelly_1-1659359610754.png

 

So now you need a transition zone on the rails to make the join.

Use the direct connection and put a negative length to one side to create a flat transition.

 

BarryKelly_2-1659360078080.png

 

I just guessed -200mm - you could measure it exactly on an elevation.

But the connection is still not perfect if you look closely as the railing twists.

 

A gooseneck connection give a better result.

If you don't want the vertical portion of the gooseneck it will be a bit of trail and error or measuring in elevation.

 

BarryKelly_3-1659360475494.png

 

But even the goosneck is not correct to my eye either.

There should be a transition around both sides of the corner post.

I don't know how to do this.

 

These are exactly the issues they will have on site.

When ever you are transitioning from rake to flat to rake, there must be a flat transition zone.

 

Barry.

 

One of the forum moderators.
Versions 6.5 to 27
Dell XPS- i7-6700 @ 3.4Ghz, 16GB ram, GeForce GTX 960 (2GB), Windows 10
Lenovo Thinkpad - i7-1270P 2.20 GHz, 32GB RAM, Nvidia T550, Windows 11

I can do the connection easily with beams - just flat as it goes around the post.

BarryKelly_4-1659361683951.png

Just can't get the railing to do this.

 

Barry.

One of the forum moderators.
Versions 6.5 to 27
Dell XPS- i7-6700 @ 3.4Ghz, 16GB ram, GeForce GTX 960 (2GB), Windows 10
Lenovo Thinkpad - i7-1270P 2.20 GHz, 32GB RAM, Nvidia T550, Windows 11

Barry,

I am glad that now you understand the geometry issue, you now have to look at real life: in reality, there will be no flat transitions!

If the rails are built without a flat transmission in the real world, that is how I want to model them.

Unless somebody comes up with a better way, I think, the solution is to move the reference line to the innermost point of the rail and offset the posts from there.

At least, there should be an option to do so.

Thanks anyway to take the time you understand the geometry challenge.

ArchiCAD since 4.5

Windows 10

@hkrna wrote:

Barry,

I am glad that now you understand the geometry issue, you now have to look at real life: in reality, there will be no flat transitions!


As far as I can tell there has to be a flat transition if your rail goes out over the flat landing.

Then only time it doesn't, is if the rail is centred over the edge of the ramp where there is one point at the very corner that is the same height for both ramps and the landing with no flat portions.

 

As soon as the rail is offset out over the flat landing, then if the rail is to be all at the same height, then it must rake up with the ramp, go flat over the landing and then rake up with the next ramp.

It is physically impossible to connect 2 rails parallel with each ram as they just won't intersect.

Take the extreme and let's assume the railing is on the other side of the ramp.

The rails must go flat when they get to the landing.

 

BarryKelly_0-1659405115931.png

Yes there are extra posts here.

On the inside corner it is the same thing, but you wouldn't have 3 post as they would all be to close, so you have one and just flatten the rails as the go around it.

 

If this is not the case in real life, then I suggest that the rails are not perfectly parallel to the ramps.

I'd like to see a photo if you have an example.

 

Barry.

One of the forum moderators.
Versions 6.5 to 27
Dell XPS- i7-6700 @ 3.4Ghz, 16GB ram, GeForce GTX 960 (2GB), Windows 10
Lenovo Thinkpad - i7-1270P 2.20 GHz, 32GB RAM, Nvidia T550, Windows 11

@Barry Kelly wrote:

On the inside corner it is the same thing, but you wouldn't have 3 post as they would all be to close, so you have one and just flatten the rails as the go around it.

That made me try something.

Model the rail with no offset but position it slightly inside the ramp edge.

Add the 3 nodes as it transitions around the flat landing.

 

BarryKelly_1-1659406135484.png

Then turn the 2 outer posts off, leaving just the rails and one corner post.

 

BarryKelly_2-1659406229433.png

The top 2 rails are offset to the side of the posts and now wrap around flat (direct connections with zero extensions).

The bottom rail I left with no offset.

This is what I would expect the rail to look like.

 

Barry.

 

One of the forum moderators.
Versions 6.5 to 27
Dell XPS- i7-6700 @ 3.4Ghz, 16GB ram, GeForce GTX 960 (2GB), Windows 10
Lenovo Thinkpad - i7-1270P 2.20 GHz, 32GB RAM, Nvidia T550, Windows 11

@hkrna wrote:

If I have space to add a flat portion to the ascending stair (typically an extra going in stairs) the geometry problem is solved.


It has to be exactly that amount, if the railing has to follow the stair's slope. If two risers meet at the corner, there would be two riser's height that the rail would have to follow.

This extra going's length can be on the 90° bend of the railing, giving the extra riser height. From this the radius (equal to the rail offset) can be calculated, which is about 19 cm for a typical 30 cm going.

That would be the theory, but I still can't set up the connection to be perfect.

Another workaround is to set an offset for the whole railing, so its reference line will be on the handrail. After deassociating it from the stair the reference line can be moved inwards,. Adding nodes to adjust the heights will be necessary.

Péter Baksa
Software Engineer, Library as a Platform
Graphisoft SE, Budapest