topic Re: Silly question about cable in Robot Structural Analysis Forum

Silly question about cable

t.sautierr — Mon, 25 Aug 2014 18:30:41 GMT

the problem :

A cable simply supported, prestressed at 0.99 L, and loaded with uniform load or point load.

case 1 : uniform load along one bar -> OK

case 2 : point load defined relatively at 0.5L of 1 bar -> OK

case 3 : point load on a point with one or two bar -> UNSTABLE ?! why?

case 4 : uniform load on 2 bars -> UNSTABLE?! why ?

I tried to put "cable support" eg : all rotations fixed at the metting point of the bars but nothing better ....

What did I missed?

I'd to be able to make point loads with one bar(cable) without dividing the cable and defining the load on the point and not specifying the location on the bar.

thx a lot.

Re: Silly question about cable

t.sautierr — Mon, 25 Aug 2014 18:31:09 GMT

last file

Re: Silly question about cable

Pawel.Pulak — Tue, 26 Aug 2014 07:50:48 GMT

Hello,

in your model you have divided the cable in 2 elements creating the collinear chain of cables. In such case it is necessary to activate 2nd and 3rd order effects (Non-linear P-delta) to avoid instability and obtain convergence. It is discussed in Help.

I have attached the modification of your model where I have made such change (shown also on the screen capture below). In your original model you have lost the list of loaded elements in load records for cases 2to4. I have not changed it in my modification.

It seems that this model with divided cables was made by you when encountering "Isolated node" warnings for the model before dividing the cable. I have tried to guess your previous steps so I modified your model to contain only one cable element and restored the list of loaded elements in cases 2to4. Bar force load, used in load case 4, uses "Node" setting by default. It results in creating the additional calculation node where the concentrated force is applied. Such setting cannot be used for cable elements because they cannot be divided in smaller calculation elements. That is why "Isolated node" warning is reported in such case - see the screen capture below.

It is necessary to change from "Node" to "No node" to avoid it - such model attached too.

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly!

Regards,

Re: Silly question about cable

t.sautierr — Tue, 26 Aug 2014 09:44:39 GMT

Thanks Pawel!

So if I understand correctly, for point loads :

Can be nodal but :

need to divide the cable at each loaded node + pdelta compulsory

Can be bar loads :

no specific requirement except it exists some remaining calculation nodes along the bar, then set to "no node" in the loading table to avoid the warning.

for uniform loads : nothing to do

for loads coming from FE : i will test 🙂

Is this correct?

I made the test and both model converged, but the results are quite different ...... (see data image) why ? (maybe is it the dimensions of the cable ? 10 mm section for 34 m ....... I will test the same for a more reasonable span. But let's say it is as it is, which model would you say it is the closest to reality (I would say the one with the cable into two parts ? but maybe is it because it gets more deflection so I unconsciously want to be safer).

One last question : do we need to set the mouting case and tick the case (which always the load case with the smallest number I assume?) to be taken into account for other loads cases in the analysis table / structural model ?

Re: Silly question about cable

Pawel.Pulak — Tue, 26 Aug 2014 11:52:08 GMT

@t.sautierr wrote:

Thanks Pawel!

So if I understand correctly, for point loads :

Can be nodal but :

need to divide the cable at each loaded node + pdelta compulsory

Can be bar loads :

no specific requirement except it exists some remaining calculation nodes along the bar, then set to "no node" in the loading table to avoid the warning.

for uniform loads : nothing to do

for loads coming from FE : i will test 🙂

Is this correct?

Yes, it is correct.

What do you mean by loads coming from FE? If you mean loads applied to panels it should work as usual. Cables can be connected to panels - but only in end nodes. If cable will be defined along the edge of some panel or inside it of course it will not connect to internal nodes of FE mesh.

t.sautier wrote:
I made the test and both model converged, but the results are quite different ...... (see data image) why ? (maybe is it the dimensions of the cable ? 10 mm section for 34 m ....... I will test the same for a more reasonable span. But let's say it is as it is, which model would you say it is the closest to reality (I would say the one with the cable into two parts ? but maybe is it because it gets more deflection so I unconsciously want to be safer).

What have you tried to model in these files? The cable was defined with 0.95 relative elongation. It means that the unloaded length of cable was 95% longer that the distance of support nodes.

Such significant length results in huge sag values (about 30 meters for the span of 34 meters) and it is beyond the domain of validity of results for cables in Robot. That is why the differences between modeling cables with and without division were so significant. None of these approaches is correct in such case.

If the cable was defined this way by mistake then it is necessary to correct it. For instance when I tried 0.05 relative elongation the results for cables with and without division were similar.

If such big cable length was used deliberately then none of these modeling methods is correct. In such case the best approach is to model it by the chain of several cables (even 4 can be enough) with zero relative elongation but predicting the deflected shape and obtaining this 95% increase of length by coordinates of nodes. This "predicted" shape does not need to be precise - Robot will find the precise deflected shape during analysis like on the screen capture below:

@t.sautierr wrote:

One last question : do we need to set the mouting case and tick the case (which always the load case with the smallest number I assume?) to be taken into account for other loads cases in the analysis table / structural model ?

Sorry - your question is not clear for me.

In case of structures containing cables the load case with the smallest number is treated as the assembling load case and it is automatically added in other load cases and combinations - it can be noticed checking the sums of forces and reactions in the table of reactions. In case of combinations the assembling load case is added only once with the factor of 1.0 - even when the assembling load case is not specified in the definition of combination. If it is directly specified with the factor different than 1.0 then this value owerwrites the default value of 1.0.

---------------------------------------------
If this post answers your question please click the "Accept as Solution" button. It will help everyone to find answer more quickly!

Regards,

Re: Silly question about cable

t.sautierr — Tue, 26 Aug 2014 12:13:50 GMT

I meant picture 1 meshed loaded pannels supported by cable along its edges, i will make some tests.

What have you tried to model in these files? The cable was defined with 0.95 relative elongation. It means that the unloaded length of cable was 95% longer that the distance of support nodes.

Oups ........ I thought 0.95 = the cable is 0.95L at for assembling case ... sorry (actually the title of the post is relevant then .... this was a very silly question I should read the help ...)

However the method you described below (chains to give a good starting shape for big deformation is a great one.

Sorry - your question is not clear for me.

I meant is it useful to tick this for cable structure but I found the answer in the help : for cable for load case is always taken into account.

Re: Silly question about cable

Pawel.Pulak — Tue, 26 Aug 2014 15:48:24 GMT

t.sautier wrote:
I meant picture 1 meshed loaded pannels supported by cable along its edges, i will make some tests.

If these cables along edges of panels are not divided they will not connect to internal nodes of FE mesh - I have written it in the previous answer.

t.sautier wrote:
I meant is it useful to tick this for cable structure but I found the answer in the help : for cable for load case is always taken into account.

Yes, it is not necessary to activate "Use first case as a starting one..." for models containing cables.

Regards,