Learning Forum

[Peter Yip]

Hello Forum,

I am running a full SALE calculation, but would like to be able to just somehow get an ASCII output of the pressure history of the contacting element without needing to store large D3PLOTS and probe the element pressure. Is there an easier way to do this for ALE elements? I'm trying to use DATABASE_FSI_SENSOR, but I'm not sure how to set that up correctly... 

Thanks,

Peter

[Ian Do]

Hello Peter,

To define the sensor:

- define a node ==> 

- define a segment (this could be a dummy LAG segment) that this node belongs to so the code can establish a normal vector ==> 

- define the offset distance along that normal => a sensor will be located at the tip of that vector

a "sensor" will be attached to a Lagrangian segment/node and moves with it.
the p inside the ALE elm which is passed by by this sensor is saved.  The
p history result may plotted from either the binary file or ASCII file output (user can choose)

$===============================================================================
*DATABASE_FSI_SENSOR
$       DT    BINARY
      0.01
$ DBFSI_ID       NID SEGMENTID    OFFSET       ND1       ND2       ND3
        11        72        37       0.5
        12        59       261       0.5
$===============================================================================
Hope this helps.

Regards,
Ian Do, PhD

 

[Peter Yip]

 

 

Hi Ian,

Thank you for your help. If the SALE simulation is in 3D (all hexahedral elements), how do I define a node and/or segment? Do I just generate a hex element using the ShapeMesher Tool where I want the sensor? Or is it via DATABASE_HISTORY_NODE and/or DATABASE_HISTORY_NODESET? Or SET_SEGMENT?

Thanks,

Peter

 

 

[Peter Yip]

Hi Ian,

I think I may have answered my own question... I think I can create a node via EleTol -> NodEdit and create a Lagrangian shell element via EleTol -> EleEdit. However, now that I've introduced Lagrangian entities, do I need to do some sort of CONSTRAINED_LAGRANGE_IN_SOLID or ALE_STRUCTURED_FSI cards to make sure the elements are properly referenced with each other?

Peter

[Ian Do]

Hello Peter,

No you do not need to couple the dummy LAG elms to the ALE mat, as there is no real physical interaction beween them that we want. They are there as sensor locators only. They have no physical meaning. If the code complains, you may define *PART, *SECITON_SHELL, *MAT_RIGID (or something) for these dummy shells.

Ian

[Peter Yip]

So I have a domain that encompasses a sphere impacting a surface which I have made using ALE_MESH_CONTROL_POINTS. I then created both the sphere and the target using ALE_STRUCTURED_MESH_VOLUME_FILLING where the sphere was made using ELLIPSOID geometry and the target was made using a BOXCOR geometry. My question is that I can't select the ALE element in PrePost (says 'no node/element picked') when I want to make a segment from the coordinates of the ALE nodes. Is there a way to select the ALE nodes so then I can base my Lagrangian nodes and elements from those?

 

 

[Peter Yip]

Also, on top of that, is there a way to visualize the volume filling? It's somewhat frustrating that I don't know what the sphere and target look like until I run the simulation...

 

[Ian Do]

[Q1] select node set or other sets from S-ALE mesh

[A1] I am sorry this is being developed. Currently we do not have an easy way to do this selection. If you really need this, I will write out a procedure to get it, but it is tedious!

[Q2] Visualize vol-filling on the fly

[A2] This is also another developmental request I have submitted to our LSPP team. I hope that they will have time to implement this very useful feature also. Currently, the way to check is to run the deck with ncycle=1 on the command line ==> then postprocess d3plot. SELPAR ==> NONE==> FLUID ==> select each AMMG to see the AMMG filling.

[Peter Yip]

So if I can't select nodes or nodesets from the S-ALE mesh, what would you advise is the best way to make a lagrangian node / segment that fills is where I want to get the pressure of an element over time (lagrangian sensor). Do I just get an lagrangian shell element somewhere within the element I want the sensor?

[Ian Do]

Hi Peter,

You theoretically can select and define any node/segment/solid sets of the S-ALE mesh based on the control point indices via set_node/segment/solid_general. 

What I am saying is that sometimes is not convenient or easy. Therefore I am asking for developmental features that allows easier approaches to

[Q] Do I just get an lagrangian shell element somewhere within the element I want the sensor?

[A] For the sensor - I described the way to define it above ==> 

- define a dummy node ==> 

- define a segment (this could be a dummy LAG segment) that this node belongs to so the code can establish a normal vector ==> 

do the same thing via command KW or GUI.

 ...

Is this procedure above not clear?

Ian