topic Re: Rigid links in custom directions & degrees of freedom in Robot Structural Analysis Forum

Rigid links in custom directions & degrees of freedom

Anonymous — Tue, 16 May 2017 09:03:56 GMT

Hi,

Please see attached model.

I wish to do away with the bars 15, 3 and 16 and make nodes (304, 21 and 306) the master nodes of (47, 8 and 46) respectively.

The idea is to make these nodes move together in UZ, and also move together in a direction parallel to 14_Edge(5) and 14_Edge(6), but free to move independently in a direction perpendicular to these edges.

Due to the orientation of my building, it appears I am not able to use rigid links to model master / slave relationships in custom directions.

My intended work-around is to connect rigid bars from nodes (304, 21 and 306) to nodes (47, 8 and 46) respectively. The properties of the rigid bars will then be modified to be very stiff in bending but very weak axially.

I would however prefer to use the rigid links option, if there is a way around it.

Thoughts?

Re: Rigid links in custom directions & degrees of freedom

Rafacascudo — Tue, 16 May 2017 12:30:57 GMT

rigid bars from nodes (304, 21 and 306) to nodes (47, 8 and 46) respectively
is not really perpendicular to de panel edges . But if this is what you need , you could model rigid bars between these nodes and release the correspondent uy or uz local directions on these rigid bars accordingly

Re: Rigid links in custom directions & degrees of freedom

Anonymous — Tue, 16 May 2017 12:51:58 GMT

My question really is...

Is there a way use rigid links to model master / slave relationships in custom directions ?

Apart from using rigid bars, is there another way this can be done in RSA?

Cheers.

Re: Rigid links in custom directions & degrees of freedom

Artur.Kosakowski — Tue, 16 May 2017 13:16:32 GMT

Hi @Anonymous

As a rigid links directions are related to the global coordinate system in case you need to follow e.g. a skew beam it is better to replace it with a rigid bar rotated (gamma angle) as per direction of a beam and define releases at its end where the slave node of the rigid link would be. Mind that for other that Skyline and Frontal solver this is done anyway (use of RLink elements to model rigid links).

If you find your post answered press the Accept as Solution button please. This will help other users to find solutions much faster. Thank you.

Re: Rigid links in custom directions & degrees of freedom

Anonymous — Mon, 22 May 2017 15:49:55 GMT

@Artur.Kosakowski

I am really not getting along with RSA today.

Following your message above take a look at the attached. I have tried to create bars to mimic a debonded shear key connected to a slab which sits on a corbel. So that the corbel supports the slab vertically, the shear keys support the slab laterally, and the debond allows the slab to move transversely.

I have tried to arrive at the below such that :

Iz produce the same Ux deflection as a 32mm diameter shear key

Iy to be very stiff to represent vertical support of the corbel

Ax to be as little as possible so that the bar can move freely in Uy; representing the debonded sleeve of the dowel.

I end up with large inconsistencies in the Ux deflections as I increase Iz of the bar...

I also get instability warnings

Fx (kN)	Ax	Iz	Deflection (mm) Ux	Calc messages - instabilities
300	0.0001	1E+48	0	RY type 3
300	0.0001	1E+41	0	RY type 3
300	0.0001	1E+34	0	RY type 3
300	0.0001	1E+29	0.001	RY type 3
300	0.0001	1E+27	0.0616	RY type 3	UY type 2
300	0.0001	1E+24	516423	RY type 3	UY type 2
300	0.0001	1E+23	252	RY type 3
300	0.0001	1E+22	4034	RY type 3
300	0.0001	1E+20	57380	RY type 3
300	0.0001	1E+12	63204	RY type 3

It seems impossible to arrive at a set of properties that

1. Does not give any instability warnings

2. Gives Uy > 50,000mm, Ux <= 150mm, Uz <= 0.1mm

Please comment.

Re: Rigid links in custom directions & degrees of freedom

Rafacascudo — Mon, 22 May 2017 17:36:06 GMT

Due to the fact that the edges (14_Edge(5) and 14_Edge(6)) are not in the same direction and the slab is continuous, the effect you need ,described in your 1st post (free to move independently in a direction perpendicular to these edges), is impossible to achieve.

Any link (with releases or through custom section properties)you create for one of these edges will not "help" or allow the other edge to move freely.

In the direction below it is possible , but that´s not what you really described in your 1st post

Re: Rigid links in custom directions & degrees of freedom

Anonymous — Tue, 23 May 2017 08:01:55 GMT

I think we are missing the point here... and what I am really interested in doing is creating bars that can have custom properties in certain degrees of freedom. That is the question I really want answered. And then I can do with the bars as I see fit.

Lets not focus on my first post because my previous post clarifies my first post.

So.... I referring to the above again... can you please explain if the below properties are achievable or not.

I have tried to arrive at the below such that :

Iz produce the same Ux deflection as a 32mm diameter shear key

Iy to be very stiff to represent vertical support of the corbel

Ax to be as little as possible so that the bar can move freely in Uy; representing the debonded sleeve of the dowel.

I end up with large inconsistencies in the Ux deflections as I increase Iz of the bar...

I also get instability warnings

Fx (kN)	Ax	Iz	Deflection (mm) Ux	Calc messages - instabilities
300	0.0001	1E+48	0	RY type 3
300	0.0001	1E+41	0	RY type 3
300	0.0001	1E+34	0	RY type 3
300	0.0001	1E+29	0.001	RY type 3
300	0.0001	1E+27	0.0616	RY type 3	UY type 2
300	0.0001	1E+24	516423	RY type 3	UY type 2
300	0.0001	1E+23	252	RY type 3
300	0.0001	1E+22	4034	RY type 3
300	0.0001	1E+20	57380	RY type 3
300	0.0001	1E+12	63204	RY type 3

It seems impossible to arrive at a set of properties that

1. Does not give any instability warnings

2. Gives Uy > 50,000mm, Ux <= 150mm, Uz <= 0.1mm

Please comment.

Re: Rigid links in custom directions & degrees of freedom

Anonymous — Tue, 23 May 2017 11:20:10 GMT

@Rafacascudo / @Artur.Kosakowski

I'm simply going to put this down to an error with RSA, because I can't find a reason for this.

After investigating further... take a look at the attached file and screenshot.

Bar 50 and 54 have identical properties, dimensions and Fx forces applied to them. Yet they are showing different Ux deflection values.

I am convinced that the Ux values presented by bar 50 are wrong, because its Ux values do not vary proportionally to a change in its Iz values! (see previous table in post)

For bar 54, which I had to create separately (its not a copy of bar 50) changes to its Ux vary in proportion to changes to its Iz. So this works as it should.

I tried copying bar 50 to create bar 1. And then copied the properties of bar 54 using the copy properties function, in case there was some property that I was unaware of or missed in defining bar 1 and 54 properties. This did not work and bar 1 behaves exactly like bar 50.

Bar 1 and 50 should deflect like bar 54.

I'd appreciate a response addressing this issue directly, in case I am missing something. Please have a look at my file. Don't create yours. This looks to me like an error that needs to be addressed.

Regards.

Re: Rigid links in custom directions & degrees of freedom

Rafacascudo — Tue, 23 May 2017 18:01:25 GMT

Just for the record , I am not official support !!!

Better way to compare displacements of similar models is to look at "nodal displacements" table . If you had checked this table you would have seen that nodes 5 and 323 from bars 1 and 50 have has weird displacements on Y direction for a load totally on direction X.

Then , It happenned that the mistake was with you and it was a pretty basic one. Never trust 2 digits numbers!!!! Always use at least 4 digits for node coordinates!!!!

Due to that , bar 54 is perfectly modelled on Y direction . But bars 1 and 50 are not ,as you can see below.

That was the reason for the enormous and weird displacements on Y direction on bar top nodes of bars 1 and 50. The force applied on X direction was not exactly perpendicular to these 2 bars

Correcting that , you get

That said , The best way to get what you need is to model very rigid bars and assign bar releases for the directions,rotation you want ... released

Re: Rigid links in custom directions & degrees of freedom

Anonymous — Tue, 23 May 2017 19:46:44 GMT

@Rafacascudo

First of all, thanks for looking taking the time to look into this. I think you are the only one who has looked into this properly.

HOWEVER...

I find this VERY strange indeed. What you are telling me is that when copying nodes in RSA one should never trust the copy operation! Note taken.

I copied node 322 by a distance of x=0, y=0.25 to create node 323. This is a disatance of 2 decimal places, not 4!! I needed a 250mm long bar. Why do I need 4 decimal places to define a 250mm long bar in meters?? So then, why on earth should one expect to need to see 4 decimal places to check the accuracy of RSA'scopy operation! Its absolutely unecessary and VERY strange indeed.

I'll be mindful of this in the future, in case it is a bug.

I will contact this forum again if I come across the same problem.

Many thanks.

Re: Rigid links in custom directions & degrees of freedom

Rafacascudo — Tue, 23 May 2017 20:37:26 GMT

How did you copy node 322? By mouse snapping or writing 0 , 0.25 , 0 on the vector field?

Re: Rigid links in custom directions

Anonymous — Wed, 24 May 2017 06:08:31 GMT

By writing 0,0.25 in the vector field.

Re: Rigid links in custom directions

Rafacascudo — Wed, 24 May 2017 11:57:22 GMT

well , in 16 years ,it is the 1st time I see an error like this. These kind of mistake happens when you set the vetor through mouse snaps. It´s very easy to snap to some of these below as they are already activated by default.

I would say that it is nearly impossible to be a RSA bug.

Can you reproduce this error again?

Maybe official support has other ideas of how this happenned

Re: Rigid links in custom directions

Anonymous — Wed, 24 May 2017 12:27:19 GMT

Nearly, impossible. But not absolutely impossible.

I started using RSA only 4 years ago and I have come across all kinds of
very weird errors / bugs. RSA deleting elements or randomly moving them
around after a crash is my most common one - this usually requires to
open the model in repair mode. I can't count the number of times I have
opened RSA in repair mode! This program crashes a lot! For example, in the
last week I have opened my models in repair mode every single day, because
of some error, or crash or randomly omitting / deleting floors from
my model, or whatever.

Lets face it, RSA is very, very powerful, this is why we love it. But it
is also extremely unstable and unpredictable, and this can sometimes be
extremely frustrating.

Thanks for your help.

Re: Rigid links in custom directions

Anonymous — Wed, 24 May 2017 12:29:52 GMT

@Rafacascudo

I will also try and reproduce this error again / look out for if it happens again and let you know.

I'll also be keen to hear from 'official support' as you call it.

You have been very supportive anyway.

Thanks a lot.

Re: Rigid links in custom directions

Artur.Kosakowski — Fri, 26 May 2017 09:50:36 GMT

Hi @Anonymous

As the definition of released d.o.fs in rigid links refers to the global coordinate system in situations when you are working in other planes the best option seems to be replacing them with rigid bars with gamma angle defined in such a way that their local axes correspond to the 'directions' you want to release and define releases in their 'slave' nodes.

For the crashes: have you sent any crash reports?

For the snapping: make sure you switched off the display of the grid when you indicate the translation vector.

I hope this helps.

If you find your post answered press the Accept as Solution button please. This will help other users to find solutions much faster. Thank you.