Sol701 rigid target surface definition

[BirleyEngineer]

Dear all,
I'm running a sol701 in FEMAP with NX nastran. I'm modelling the impact of the inclined side of a hollow steel box section structure (modelled in shells)with the floor - to be modelled as rigid and need some help in producing efficient contacts. I have currently successfully run the floor as a heavy steel plate (not rigid)with the edges fully restrainted and it runs ok, but takes very long to solve (about 60hrs) I have used mass scaling and don't wish to explore that feature at the moment in this thread. I also appreciated getting more CPUs and enough RAM, faster drives etc will help, I'm getting another machine but currently use twin Quad core zeons at 2.8GHz with 24Gb of RAM.

Having read the NX nastran advanced non Lin theory and modelling guide I'm still uncertain as to the best way of setting up a rigid contact. My questions are

1, does a rigid target surface need to contain elements or can it just be a defined surface? I ask this as the manual references segments not nodes when referring to the rigid definition.

2, Can somebody give a brief description of setting up a rigid target saying how to use the reference node?

3, what is the best contact type to use?

4, If elements are required on the rigid target, can they be large? Or are there limitations.

Again I'm trying to speed up the run and suspect some time is being wasted in contact searching,

Any help much appreciated.

Regards Jeff.

 

[BlasMolero]

Dear Jeff,
Here you are a few comments regarding Surface-to-Surface Contact Region Parameters for NX NASTRAN Advanced Nonlinear Transient Explicit analysis (SOL701):

1.- For a 2D contact region, a rigid target region could be defined by grid points that are not attached to any elements. For a 2D contact region with underlying 2D elements, the contact side is automatically determined by the program. But if the rigid target algorithm is selected (i.e., TYPE=2 (SOL 601) or XTYPE= 3 (SOL 701) in BCTPARA command), the target region must be attached to shell elements only (i.e., using BSURF or BCPROP) and it is automatically set as rigid.

2.- The master grid point may be used to control the motion of a rigid surface. It is a simply node in the space that you can use to prescribe an enforced motion to the rigid region.

3.- The default "Constraint Function" is the best contact type method.

4.- The problem here is that element size affect the time stepping. Since the time step is determined automatically by the SOL701 solver based on the highest eigenvalue of the smallest element, a single small or excessively distorted element could considerably increase the solution time, even if this element is not relevant to the full model. So, the target for SOL701 explicit analysis is to have ALL ELEMENTS THE SAME ELEMENT SIZE !!.

Best regards,
Blas.


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

IBERISA
48011 BILBAO (SPAIN)
WEB:

http://www.iberisa.com

Blog de FEMAP & NX Nastran:

http://iberisa.wordpress.com/

 

[BirleyEngineer]

Hi Blas,
Many thanks for your reply. Can I please ask you couple of questions-

1.Im running a large model, approx 308k elements, and it's taking a very long time to run -around 60 hrs for 200ms. It's essentially a frame impact with non linear materials. I have 24gb of ram, and have set memory to 5gb, which is well beyond the 2gb required. However, when running my machine is still writing to disk a lot, and the CPUs are only working to an average of 80%. I have set no of CPU to 8 (I only have 8). This seems odd as I expected the CPUs to be running flat out!

2. Also can you advise whether it's more efficient to use a rigid impact region or use a meshed surface with structural properties which is restrained?

3, in principle would it be more efficient to use a very course mesh on the target surface?

4, finally I have used the mass scaling to speed things up a tad, while NX reports how many elements have been updated, it does not report the mass increase. While I can make an estimate it would be good to know exactly how mush mass has been added. Is there a way of getting the mass increase used by mass scaling reported?

Thanks for you help