Noise in dbfsi
Ameen Topa
I am doing mesh refinement analysis on sloshing problem and noticed that with models that has smaller mesh size, the noise levels in dbfsi increases significantly. I decrease the size of both ALE and Lagrange by simply splitting the mesh of the coarse model (so the relative size between the ALE and Lagrange mesh are not changed). The parameters in Constrained Lagrange in Solid are the same for both fine mesh and coarse mesh models.
My question: is there a way to reduce this noise?
Thanks in advance!
Ameen
l...@schwer.net
As the element size decreases, the mesh is able to represent higher frequency content. Perhaps this is what you are observing?
Perhaps the DAMP parameter of *CONSTRAINED_LAGRANGE_IN_SOLID will help?
--len
--
You received this message because you are subscribed to the Google Groups "LS-DYNA2" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ls-dyna2+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/77e49811-8865-48e0-a131-9e709d06f3e7n%40googlegroups.com.
James M. Kennedy
Dear Ameen,
I found this note. See if it is of some help.
[1] The dbfsi P is an averaged value. To get ==> F_dbfsi ~ P_dbfsi * A
[2] The interface-force-file (iff) is more local ==> to get force ==>
F_iff ~ integrate_over_A[ P_iff * dA ]
So in this sense, [2] is more maybe accurate and not exactly the same as [1].
Sincerely,
James M. Kennedy
KBS2 Inc.
January 14, 2021
From: ls-d...@googlegroups.com [mailto:ls-d...@googlegroups.com] On Behalf Of Ameen Topa
Sent: Thursday, January 14, 2021 11:56 AM
To: LS-DYNA2 <ls-d...@googlegroups.com>
Subject: [LS-DYNA2] Noise in dbfsi
Dear all,
--
Ameen Topa
Thank you for your prompt reply. I think it is better to explain with the plots. Here are few things to note:
- First of all, I am running both cases with Murnaghan EOS.
- The force with fine model (20 mm) is represented by the red line. As you can see, at time = 1.35 s there is very high peak in the force, which I want to eliminate.
- Secondly, with the coarse model there is a small peak at t = 0.25 s. I don't mind this. But I wonder why in the finer mesh there is no similar noise at this time.
- I have run the same model with Gruneisen and Linear Polynomial EOS for water. The high oscillation at t = 1.35 s doesn't exist. However, the smaller oscillations of the force similar to point no (3) exist.
l...@schwer.net
Thanks for the additional details.
Your two models seem to have achieved a converged solution – expect for the noted 20mm mesh irregularity at 1.35s.
As a numerical experiment, please try running the model with a 50 or 60mm mesh.
--len
PS – When you again plot dbfsi please select two more dissimilar colors than red and orange.
From: ls-d...@googlegroups.com <ls-d...@googlegroups.com> On Behalf Of Ameen Topa
Sent: Friday, January 15, 2021 6:29 AM
To: LS-DYNA2 <ls-d...@googlegroups.com>
Subject: Re: [LS-DYNA2] Noise in dbfsi
Dear all,
Thank you for your prompt reply. I think it is better to explain with the plots. Here are few things to note:
- First of all, I am running both cases with Murnaghan EOS.
- The force with fine model (20 mm) is represented by the red line. As you can see, at time = 1.35 s there is very high peak in the force, which I want to eliminate.
- Secondly, with the coarse model there is a small peak at t = 0.25 s. I don't mind this. But I wonder why in the finer mesh there is no similar noise at this time.
- I have run the same model with Gruneisen and Linear Polynomial EOS for water. The high oscillation at t = 1.35 s doesn't exist. However, the smaller oscillations of the force similar to point no (3) exist.
Dear Mr. Kennedy,
Thank you for your suggestion. I will look into it.
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/122defdc-2b6a-421a-8d3d-030e47ebd600n%40googlegroups.com.
Ameen Topa
Thank you for the suggestions. I will give it a try.
I forgot to mention, regarding the DAMP parameter in CLIS card, I was using 0.5 in all previous simulations. At the moment, I am running the simulation of the refined model with DAMP = 0.9. Results are not yet.
Sincerely,
Ameen
James M. Kennedy
Dear Ameen,
Two thoughts you might look at: (1) compare single and double precision executables and (2) employ structured (S-ALE) method.
There are instances where double precision is most likely needed: for example, explicit analysis where number of cycles (time steps) is large, e.g., over 200,000 time steps (200,000 is just a ballpark figure... the actual number for any given model could be much higher).
A very nice feature of the S-ALE modeling, especially when making mesh discretization comparison studies, is that one can modify the S-ALE model (wavehitcol.k) below by making a simple parameter change (node numbering) in the three *ALE_STRUCTURED_MESH_CONTROL_POINTS keyword entries.
------------------
ALE (standard) - Intermediate example presentation
https://www.dynaexamples.com/ale/intermediate-example
Intermediate : Wave-Structure Interaction
A container is partially filled with water subjected to gravity. As the wave propagates, it hits a rigid column placed roughly in the middle of the container. This model was developed to make/allow comparisons with an available S-ALE model. To make a direct comparison, using identical cell discretization (0.010 meters) one can modify the S-ALE model (wavehitcol.k) below by making a simple parameter change (node numbering) in the three *ALE_STRUCTURED_MESH_CONTROL_POINTS keyword entries.
Roughly 330,000 elements are in the ALE and S-ALE models with a cell discretization of 0.010 meters (m), while roughly 2,590,000 elements are in the S-ALE model (wavehitcol.k) with a cell discretization of 0.005 meters (m).
The Murnaghan equation of state has become available as a new keyword: *EOS_MURNAGHAN. This equation of state can be used to enforce quasi-static incompressibility while allowing for a reasonable time step in explicit calculations. Any ALE model part is there- fore represented by a *MAT_NULL card, describing the density and viscosity of the material, and an *EOS_MURNAGHAN card, describing the pressure-density relationship.
-----------------
S-ALE - Intermediate example presentation
http://ftp.lstc.com/anonymous/outgoing/hao/sale/models/wavehitcol/
Intermediate : Wave-Structure Interaction
A container is partially filled with water subjected to gravity. As the wave propagates, it hits a rigid column placed roughly in the middle of the container. The total contact forces are recorded for the entire simulation, and compared to experimental data [1].
[1} Gomez-Gesteira, M., Crespo, A.J.C., Rogers, B.D. Dalrymple, R.A., Dominguez, J.M. and Barreiro, A., “SPHysics - Development of a Free-Surface Fluid Solver - Part 2: Efficiency and Test Cases,” Computers & Geosciences, Vol. 48, pp. 300–307, November, 2012.
https://www.sciencedirect.com/science/article/pii/S0098300412000842
[2] Chen, H., "LS-DYNA Structured ALE (S-ALE) Solver", 14th International LS-DYNA Users Conference, Dearborn, Michigan, June, 2016.
[3] Chen, H, and Do, I., "Recent Developments in LS-DYNA S-ALE", 15th International LS-DYNA Users Conference, Dearborn, Michigan, June, 2018.
-------------------
Sincerely,
James M. Kennedy
KBS2 Inc.
January 21, 2021
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/efc4f5d9-aef4-4738-859d-0997d21ace9dn%40googlegroups.com.
Ameen Topa
Thank you for your suggestions.
(1) I am running all my simulations with double precision solver.
(2) S-ALE is very interesting and I am new to it. I will look into the example file and references given.
Sincerely,
Ameen