Seek help with Model output, joints/releases

Hi!

I have a few problems that I'm troubleshooting in my GH/karamba definition which we're using for form finding in structural optimization.

Please view illustration and attachments.

1. Bar forces (M.y) on the boundary of the left hand side (y-axis) of the isometric figure are not displaying correct. The right hand side (x-asis) displays correct. Both x- and y-axis are defined in the same way however this problem occurs when I apply releases on the structure. Joints are applied on the inner elements stretching out of the xy-plane. As can be seen in the GH-definition the joints are applied as rotational springs (inducing simple joints?). I'm not 100% sure if join configuration can be structurally compatible but i'm using it for form-finding, so i want to configure it with a easy logic.

2. I've seen a few topic on the Model output of material and cross-section but it's not any clear to me why they are included in the model. I have the same problem as one extra material and 2 extra cross-sections is applied to the model but I cannot find the input for it. Also if i disable the gravity component there is still a leftover load which is undefined...

- I wonder if it effects the structural analysis in any way and may it be apart of the problem?

I would be overwhelmed if someone could resolve the problem!

Best regards,

Johan

Attachments:

Replies to This Discussion

Permalink Reply by Karamba3D on August 5, 2014 at 11:40am

Hi Johan,

when using joints make sure that your system does not get kinematic. This leads to wrong results. You can check this via an eigen-modes analysis (see attached file).

When you place the analysis components (which can get computationally heavy) before the ModelView you avoid unnecessary recalculations.

In case you want elements without bending you could use trusses (see also the attached file).

The extra material is the default material; the two cross sections are the default beam and shell cross sections.

I could not reproduce the problem with the gravity load. Where/how does it show up?

Best,

Clemens

Attachments:

GHdefinition_CP.gh, 71 KB

Permalink Reply by Johan Reissmuller on August 5, 2014 at 1:49pm

Thanks Clemens!

How do I make sure the structure is static using the results from eigen-modes? Would a kinematic structure have no (null) eigen-modes?

Sure, I could disable the bending stiffness (Mz is crucial around corners) but it doesn’t resolve the real problem:

The problem persists even though I disable the joint component to display them as rigid connections instead. The y-axis boundary structure (both sides) still show inverted beam forces (M.y). I’m verifying the problem with another commercial FE-software and it deviates. It should be similar to the x-axis boundary structure. I thought this may be a result of the beams being not interconnected correctly on that side. I tried also varying the “limit distance input” without luck. Still it may be a lack of programming skills on my part, but I tried loads of configuration with the same effect.

The default cross section and material, can they possibly be removed in any way? If I check the Material box inside ModelView the default material (S235) appear even though I specified S355.

Correct, I cannot either reproduce the gravity load problem with the definition you provided. I may be mistaking.

Best,

Johan

Attachments:

rigid connections.jpg, 226 KB

Permalink Reply by Karamba3D on August 7, 2014 at 12:39am

A kinematic structure has rigid body modes. The corresponding eigenvalues are zero (in theory) or small (due to numerical errors). You can visualize the rigid body modes by enabling the display of the local element coordinate systems - in order to see beam rotations about their longitudinal axis - and change the displacement scale.

Regarding the results of the bending moments: did you try out a reduced system to track the error?

When you want to define a material for specific elements you can either specify the elements via their index or their identifier. In your definition you feed elements into the 'ElemIds'-plug of the material-component. This does not produce an error because the element is cast to a string, but the string does not specify any elements either. When you want to specify a material for the whole structure do not supply a value to 'ElemIds' (see manual).

The default cross sections can not be removed. in which respect do they cause problems?

Best,

Clemens

Permalink Reply by Johan Reissmuller on August 8, 2014 at 7:13am

Everything seems to work well now after being more consistent with the numbering of bars and the local coordinate system. I verified the structure in another FEM software and it's OK! I have not verified the eigen-modes yet though.

OK, the material works fine with the help of your input. // I hope the utilization component will include global buckling of elements in the future. Would be of great benefit when using it as a parameter for optimization.

Default cross section isn't a problem and doesn't affect my workflow, however by just a minor remark, the default cs is being exported when using ExToDstV-component to Robot.

Thanks again! Karamba is awesome!

Best regards,

Johan

Attachments: