Model organization approach (layers and automated numbering)

[ANick]

Dear community members,

I am trying to adapt the existing model organization approach from HyperMesh to the FEMAP environment, and I would be glad to hear your feedback and suggestions.

1. General organization. According to the existing model organization approach, model entities (nodes, elements, properties, materials) for each part (CHEXA, CQUAD etc.) and each interface (CBUSH, RBE2 etc.) are allocated in separate layers. I came across many discussions “groups vs layers”. Presented approach simply reflects already existing methodology based on components.

2. Numbering convention. Model numbering (nodes, elements, properties, materials) shall start with part layers and continue with interface layers. Starting ID is defined by the layer ID. E.g. for the layer ID 10,001 each entity is renumbered starting from 10,001. It allows keeping numbering ranges for particular objects/parts in the assembly.

3. Model summary. It is often required to generate model summaries, e.g. materials, properties, elements per layers etc.

Currently 2) and 3) are implemented via HyperWorks API, but I assume corresponding tools may already exist for FEMAP. However, I could not find anything similar yet, so your help is much appreciated.

Questions:
[ol 1]
[li]Considering flat layer hierarchy in FEMAP (no sub-layers or sub-assemblies are allowed) presented approach does not scale for the models with large number of parts/interfaces – list of layers becomes too large and difficult to deal with. Is there any way to solve this limitation? Groups do the work just partially, since their representation in the model tree is anyway flat.[/li]
[li]Are there any API collections available for FEMAP besides FEMAP knowledge base?[/li]
[li]What is your favorite model organization technique when using FEMAP?[/li]
[/ol]

Thank you for your feedback.

 

[rb1957]

if you want to use assemblies, read up on NASTRAN superelements.

another day in paradise, or is paradise one day closer ?

 

[ANick]

rb1957, thanks for the reply.
By assemblies I meant a set of components grouped together in FEMAP itself. Typical example is a model with more than thousand parts which can be represented as physical models or super-elements, e.g. complete assembly of a car, spacecraft or a ship. Flat hierarchy is obviously not sufficient for such models.

 

[rb1957]

i don't know how FeMap works with superelements ... i guess it makes sense (since NASTRAN can analyze superelements) that FeMap can work with them too. I'd expect a modfem of each component and one of the assembly ?

if FeMap is inadequate for your needs, do you have experience with another pre-/post-processor that does what you want ?

i know airplanes (surely large structures) are modelled with NASTRAN superelements, and PATRAN (as "awful" as it is) is used as the pre-/post-processor.

another day in paradise, or is paradise one day closer ?

 

[ANick]

Currently I am using HyperMesh and model organization looks like it is shown in the figure attached. I am trying to come up with a similar approach in FEMAP.

Super-elements is not an issue in fact: plotel elements are imported into the Preprocessor and matrix (*.pch, *.op4) and assembly (*.asm) files are called as includes during the analysis.

 

[rb1957]

you have 5 assemblies ... 5 groups. if you wnat you can separate node numbers and element numbers to suit the group.

not sure what the numbers beside the coloured boxes mean ?


another day in paradise, or is paradise one day closer ?

 

[ANick]

That’s an example model from HyperMesh tutorial In my working model I usually have about 5000 components, about 20 large assemblies with 10-15 sub-assemblies each. A sub-assembly can have a sub-sub-assemblies as well. Think of a spacecraft: structure - system - sub-system - panels/units/interfaces.

 

[rb1957]

I think FeMap can handle superelements, possibly not as seamlessly as HyperMesh, and if this is critical for your designs, then you've got your answer ... stick with HyperMesh, and leave FeMap alone.


another day in paradise, or is paradise one day closer ?

 

[ANick]

well, JPL somehow managed to use FEMAP for their projects (case study). I am pretty sure that complexity level there is very high as well.

Anyway, thanks for the feedback. Can you recommend something regarding the API?

 

[BlasMolero]

Dear ANick,
Well, next release of FEMAP V11.2 due for end of this year (or January 2015) will feature an insteresting capability: sub-levels. This was a query for all of us VARs and resellers of FEMAP in Europe, to support sublevels to include results, geometry, etc.. hopefully this will increasy the easy-of-use and powerfull of the FEMAP graphic user interface, to run geometry & mesh like a CAD system, we will know the details in a few months!!. I am sure the phylosophy sub-levels will expand to all areas of FEMAP.

In current version 11.1.2 we have flat-layers, flat-geometry, flat-properties, flat-groups, etc.. but if you leran to use LAYERS you will relize how powerful is to organize complex models, this is the way I run: during preprocessing both geometry & mesh is organized in layers, easy to create, manage, and hide/display!!. During postprocessing results the use groups is really powerfull, you have a full menu in FEMAP named GROUP to create the groups you like. I love to click in the MODEL INFO TREE on top of PROPERTIES and select GROUP, automatically will create a group for each property.

Best regards,
Blas.

~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director

IBERISA
48004 BILBAO (SPAIN)
WEB:

http://www.iberisa.com

Blog de FEMAP & NX Nastran:

http://iberisa.wordpress.com/

 

[ANick]

Thank you, Blas. I am already looking forward to the next version!
I also use Layers as the basis for the model organization in FEMAP. It is indeed the most intuitive way.
By the way could you recommend any existing solutions for the model renumbering based on Layer' information (e.g. layer ID)?

 

[BlasMolero]

Dear ANick,
Node & Element renumbering is not an issue for me, it dosen´t matter the Node#ID of Element#ID number the model has, this is part of the past, now you can select elements in any moment by property, by layer, by material, etc... then not need to have control of numbering. But if you like to renumbver anything you can do it, use command MODIFY > RENUMBER and you are done!.

Best regards,
Blas.



~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director

IBERISA
48004 BILBAO (SPAIN)
WEB:

http://www.iberisa.com

Blog de FEMAP & NX Nastran:

http://iberisa.wordpress.com/

 

[ANick]

I see your point. The problem is that very often in space industry there is a requirement for the numbering range of the sub-assemblies. Numbering range philosophy also comes in handy when performing calculations based on the FEM results, for example bolt verification for the complete structure (might have more than 1,000 bolted connections) according to specific industry standards. In this case it is required to assess the results not only on the individual bolt level, but also on the interface level. It is just one of several most common use cases. It is very easy to achieve when using the consistent node/element numbering philosophy (e.g. see the first post). I assume that manual renumbering “MODIFY > RENUMBER” is not convenient when working with significant number of layers.

 

[rb1957]

no, i think you can renumber without affecting anything. I'm with you about rigid numbering scemes ... i used to hate them (why limit ourselves ?) but I've been working with someones else's model, built with a numbering scheme, which makes it easier to work with ... element 21600001 is the 1st stringer LH aft for frame 16. but it is a pain working with them ... i find i create elements and spend a bunch of time renumbering nodes and elements ... but it's all pensionable time !?

i thought you were looking to reuse a component (use the same model in different places) ?

personally i hate layers, but like groups.

another day in paradise, or is paradise one day closer ?

 

[BlasMolero]

Hello!,
A trick you can use is to enter a OFFSET value to both node & element numbering, do not accept the "default" suggested by FEMAP.
Yes, you can renumber whatever you like, but if you are ordered you can mesh the model following your rules. In FEMAP you can mesh geometry any time you like, not need to do it at the beginneing or at then end, then why not to make your rules. For instance, if you want to have a range of numbering for such type of elements, then before meshing enter a value for Node#ID and Element#ID, in this case FEMAP will use that number to start asigning numbers to mesh and elements.

In the figure I used OFFSET=1000, then if I list nodes I will see the following: the first node#ID is 1000, as well as first Element#ID 1000

List Nodes
492 Node(s) Selected...
 
     ID    Def CS   Out CS        X1           X2           X3    Color PermBC>   Layer     SEID #Elem #Load  #BC
__________________________________________________________________________________________________________________
    1000        0        0           0.           0.          40.   124 ------        1        0     2     0     0 
    1001        0        0      2.35559           0.      40.0463   124 ------        1        0     4     0     0 
    1002        0        0      4.70755           0.       40.185   124 ------        1        0     4     0     0 
    1003        0        0      7.05224           0.      40.4159   124 ------        1        0     4     0     0

Best regards,
Blas.

~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director

IBERISA
48004 BILBAO (SPAIN)
WEB:

http://www.iberisa.com

Blog de FEMAP & NX Nastran:

http://iberisa.wordpress.com/

 

[ANick]

Of course, explicit assigning of the OFFSET is a rather logical way. It is however a manual operation. This procedure can be improved in the following way: for all visible layers, renumber elements/nodes using layer ID as the OFFSET (as explained in the first post). It can be implemented as a macro based on provided API. In this case layer ID would be first ID in the desired numbering range. It can be further extended by introducing directional renumbering along axes. For example, all interfaces along panel A have a sequential numbering in X direction. But it is already more than required.

This automatic renumbering is usually helpful, when performing import of other models while having significant gaps in the numbering ranges of the main FE model.

Have you seen similar algorithms/scripts for the FEMAP?

 

[BlasMolero]

Dear ANick,
Another trick for you: take a look to command FILE > MERGE.
This command allows geometric or finite element entities from any model currently open in the same instance of FEMAP to be “merged” with the active model. Then you can renumber AUTOMATICALLY any part of your current model. At least two models must be open for this command to be available. To facilitate bringing entities into the active model, a number of overall Renumbering and Duplicates Strategy, Entity Selection, and Model Orientation options are available in the Model Merge Manager dialog box. In addition, the top portion of the dialog box, the Entities to Merge list can be used to choose which entity types to merge into the active model using the check marks in the Entity Type column. Here you are a tutorial:

http://community.plm.automation.sie...ents/siemensplm/Femap-tkb/33/1/File-Merge.zip

Best regards,
Blas.





~~~~~~~~~~~~~~~~~~~~~~
Blas Molero Hidalgo
Ingeniero Industrial
Director

IBERISA
48004 BILBAO (SPAIN)
WEB:

http://www.iberisa.com

Blog de FEMAP & NX Nastran:

http://iberisa.wordpress.com/

 

[ANick]

Thank you, Blas. I think this feature can be helpful. It requires a little bit different logic: instead of renumbering of the model after the import, numbering scheme is ensured during (or to be precise right before) the merge. It however does not solve the problem of re-arranging of the existing numbering ranges (e.g. wrong initial estimation of the offsets), but still it is a quite good solution.

There is just one question left from the original list: are there any API collections available for FEMAP besides FEMAP knowledge base? Sorry for bombarding you with so many questions. Your help is much appreciated.

 

[Naffoussi]

Hi All,

I was following this thread since its start. I work for space industry and I know exactly what ANick is encountering regarding the FEM renumbering.
When we have a requirement from the customer, it should be respected...

Most FE models I made for space domain were large. When creating the mesh I number simultaneously the entities and when a part of the model needs to be updated later it will not cause a problem regarding the renumbering as it is already in a group / layer.

But the problem is when your entire model was randomly numbered and then you receive a requirement with numbering ranges (sub-systems models and IF nodes / elements...)
So I decided to accelerate things by automating them using API. I've created an excel / Femap program which create groups based on numbering ranges (Start ID / End ID)for elements, nodes and RBE received from the customer in an excel format. copy / paste and I am done.

And after reading this thread I have made a macro (yesterday) which renumber checked entities based on its Group IDs and not layer IDs as you asked for.
1/ First select the groups to be used to extract their entities (elements, nodes, prop, material, csys)
2/ check the entities you want to renumber based on their group ID
3/ That's all nothing else to do

You can also look at Group / Operations / Booleans which allow a lot of options.
I tested the macro on a small model with ten groups and it works.
I will look at the possibility of renumbering entities based on layerID.
Please use it and let me know if any issue appears.

Hopes this helps,
Regards.

Seif Eddine Naffoussi, Stress Engineer

http://www.Innovamech.com

33650 Martillac – France

 

[Naffoussi]

step 1 and step 2 are inverted ... first check the entities and then select groups

Regards,

Seif Eddine Naffoussi, Stress Engineer

http://www.Innovamech.com

33650 Martillac – France