Solid to SPH issues. Material cutting modeling.

[Pavel Dybskiy]

I'm modeling so called pendulum test where tip of the pendulum, a striker, cuts into the workpiece of interest.

 

Two types of workpiece representation were tried.

1) as an SPH part;

2) as a Solid FE part with Adaptive_Solid_to_SPH  transition.

Known from literature (e.g. Mabrouki, et.al, 2008. “Numerical and experimental study of dry cutting ...)  AL 2024 was used in order to test/verify the modeling approach.

Help is needed to interpret issues encountered.

1) When keyword Define_Adaptive_Solid_to_SPH  was used with SPH coupled to solid FE (iCPL=1) from beginning(iOPT=0), then 

•  (Figure 1)  the Global internal energy (glstat "D") was noticeably  higher than the sum of  partial energies (matsum A,B,C).
• Total Energy was growing (driven by Internal energy growth) with External work = 0

 

Figure 1. Energy

 

2) Workpiece reactions (Fy-horiz,Fz-vert) were significantly different, qualitatively and quantitatively, for SPH (Fig.2) vs FE-to-SPH representation and when iCPL =1 (Fig.3) vs iCPL=0 (Fig.4) were used in FE-to-SPH transition.

 

Figure 2.  Pure SPH. Fy~ 150N

   

Figure 3.  FE-to-SPH, iCPL=1. Fy~ 5000N

  

Figure 4.  FE-to-SPH, iCPL=0. Fy~ +- 20000N

Best Regards

Paul Dybskiy

University of Windsor

[l...@schwer.net]

Have you tried the simulation without coupling, i.e. ICPL=0?

The newly formed chips are similar to debris and thus should be free from the base layer of FE.                         --len

 

ICPL  EQ.0: Failure without coupling (debris simulation),

 

From: ls-d...@googlegroups.com <ls-d...@googlegroups.com> On Behalf Of Pavel Dybskiy
Sent: Wednesday, August 24, 2022 9:45 PM
To: LS-DYNA2 <ls-d...@googlegroups.com>
Subject: [LS-DYNA2] Solid to SPH issues. Material cutting modeling.

 

I'm modeling so called pendulum test where tip of the pendulum, a striker, cuts into the workpiece of interest.

 

 

Two types of workpiece representation were tried.

1) as an SPH part;

2) as a Solid FE part with Adaptive_Solid_to_SPH  transition.

 

Known from literature (e.g. Mabrouki, et.al, 2008. “Numerical and experimental study of dry cutting ...)  AL 2024 was used in order to test/verify the modeling approach.

 

Help is needed to interpret issues encountered.

1) When keyword Define_Adaptive_Solid_to_SPH  was used with SPH coupled to solid FE (iCPL=1) from beginning(iOPT=0), then 

•  (Figure 1)  the Global internal energy (glstat "D") was noticeably  higher than the sum of  partial energies (matsum A,B,C).
• Total Energy was growing (driven by Internal energy growth) with External work = 0

 

 

Figure 1. Energy

 

2) Workpiece reactions (Fy-horiz,Fz-vert) were significantly different, qualitatively and quantitatively, for SPH (Fig.2) vs FE-to-SPH representation and when iCPL =1 (Fig.3) vs iCPL=0 (Fig.4) were used in FE-to-SPH transition.

 

Figure 2.  Pure SPH. Fy~ 150N

 

   

Figure 3.  FE-to-SPH, iCPL=1. Fy~ 5000N

 

 

  

Figure 4.  FE-to-SPH, iCPL=0. Fy~ +- 20000N

 

Best Regards

 

Paul Dybskiy

University of Windsor

 

 

--
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-dyna2/8bf1d68e-2bf4-4cd4-a1e5-1f9e3322c136n%40googlegroups.com.

[James M. Kennedy]

Dear Pavel,

 

The *DEFINE_ADAPTIVE_SOLID_TO_SPH/DES keyword provides a simulation technique considering the effect of the debris since the failed solid elements are replaced with the particles. In this presentation, the results obtained using different options in *DEFINE_ADAPTIVE_SOLID_TO_SPH/DES were compared and the possibility to use this capability for real problems on concrete structures (*MAT_072R3) was discussed:

 

Tokura, S., and Niwa, K., "Evaluation of Debris Modeling Technique on Failure Simulation of Concrete Structures", 11th European LS-DYNA Users Conference, Salzburg, Austria, May, 2017.

 

http://www.dynalook.com/11th-european-ls-dyna-conference/concrete-penetration/

 

Recent LS-DYNA enhancements to air blast loadings:

 

Schwer, L.E., "New Keywords Related to Blast and Penetrations: A Few Simple Applications", 11th German LS-DYNA Forum, Ulm, Germany, October, 2012.

 

https://www.dynamore.de/de/download/papers/ls-dyna-forum-2012/documents/multiphysics-2-2

 

Some further information regarding your issues.

 

This inheriting of history variables would only apply to ICPL=1,IOPT=1 found on the keyword *ADAPTIVE_SOLID_TO_SPH entry. In this case, the failed solid turns to SPH and remains

coupled to remaining solids.

 

The converted SPH particles will couple only with the solid element that give birth to those particles. In this option, when solid elements are converted into SPH particles, the user can either

define the new material model (such as no failure or stronger failure criteria to ensure no further failure happens) or keep the same material model.

With this treatment the SPH particles are visible from the onset of the simulation.  These embedded particles remain inactive and do not participate in contact until the parent solid element is eroded. Generally, contact with the solid elements is treated using one contact definition, while a second contact definition, typically, *CONTACT_AUTOMATIC_NODES_TO _SURFACE, is used to handle contact with the converted SPH  particles.  In the second definition, the slave side is made up of the SPH particles identified with SSTYP=3 in *CONTACT which references IPSPH defined in *DEFINE_ADAPTIVE_SOLID_TO_SPH.

I suggest you overlay time histories of stress for a solid element and the associated SPH particle(s) that take over after the solid fails. Also suggest using a high resolution output from elout for this purpose using *DATABASE_ELOUT, *DATABASE_HISTORY_SOLID, and *DATABASE_HISTORY_SPH. You will also need to write sphout (*DATABASE_SPHOUT), too, since elout does not have the SPH data.

 

This will give you the elout and sphout data which shows that the SPH particles indeed take their history variables from the host solid.

 

Attached (to your personal email) you will find an example wherein the SPH part has been given a different material ID to allow one to make the failure strain much larger than the failure strain of the host solid part.

 

R10 and later includes *MAT_ADD_EROSION criteria in SPH failure and that necessitates that different part IDs without the failure criterion be used for the adapted solid-to-SPH parts when ICPL=1,IOPT=1.

 

Sincerely,

James M. Kennedy

KBS2 Inc.

August 25, 2022

 

From: ls-d...@googlegroups.com [mailto:ls-d...@googlegroups.com] On Behalf Of l...@schwer.net
Sent: Thursday, August 25, 2022 8:33 AM
To: 'Pavel Dybskiy' <pdyb...@gmail.com>; 'LS-DYNA2' <ls-d...@googlegroups.com>
Subject: RE: [LS-DYNA2] Solid to SPH issues. Material cutting modeling.

 

Have you tried the simulation without coupling, i.e. ICPL=0?

The newly formed chips are similar to debris and thus should be free from the base layer of FE.                         --len

 

ICPL  EQ.0: Failure without coupling (debris simulation),

 

From: ls-d...@googlegroups.com <ls-d...@googlegroups.com> On Behalf Of Pavel Dybskiy
Sent: Wednesday, August 24, 2022 9:45 PM
To: LS-DYNA2 <ls-d...@googlegroups.com>
Subject: [LS-DYNA2] Solid to SPH issues. Material cutting modeling.

 

I'm modeling so called pendulum test where tip of the pendulum, a striker, cuts into the workpiece of interest.

 

 

Two types of workpiece representation were tried.

1) as an SPH part;

2) as a Solid FE part with Adaptive_Solid_to_SPH  transition.

 

Known from literature (e.g. Mabrouki, et.al, 2008. “Numerical and experimental study of dry cutting ...)  AL 2024 was used in order to test/verify the modeling approach.

 

Help is needed to interpret issues encountered.

1) When keyword Define_Adaptive_Solid_to_SPH  was used with SPH coupled to solid FE (iCPL=1) from beginning(iOPT=0), then 

•  (Figure 1)  the Global internal energy (glstat "D") was noticeably  higher than the sum of  partial energies (matsum A,B,C).
• Total Energy was growing (driven by Internal energy growth) with External work = 0

 

 

Figure 1. Energy

 

2) Workpiece reactions (Fy-horiz,Fz-vert) were significantly different, qualitatively and quantitatively, for SPH (Fig.2) vs FE-to-SPH representation and when iCPL =1 (Fig.3) vs iCPL=0 (Fig.4) were used in FE-to-SPH transition.

 

Figure 2.  Pure SPH. Fy~ 150N

 

   

Figure 3.  FE-to-SPH, iCPL=1. Fy~ 5000N

 

 

  

Figure 4.  FE-to-SPH, iCPL=0. Fy~ +- 20000N

 

Best Regards

 

Paul Dybskiy

University of Windsor

 

 

--
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-dyna2/8bf1d68e-2bf4-4cd4-a1e5-1f9e3322c136n%40googlegroups.com.

--
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-dyna2/001501d8b887%244376e230%24ca64a690%24%40schwer.net.

[Pavel Dybskiy]

Yes, I tried.

For the cases of pure SPH representation, and  Solid-to-SPH when iCPL=0, energy behaves as it should, total energy decreases if external work is 0 and there is friction present in the system. 

A question here would be how to pinpoint the reason for such a violation in the iCPL=1 case . To my understanding, energy misbehavior is just a side effect of algorithmic mistakes (e.g. incorrect accounting for masses and/or reactions in different materials). Is it correct to draw such a conclusion? 

However, my main concern is not energy by itself . I use it here as a simple universally understood indication that there is something wrong in my ls-dyna simulation. 

My main concern comes from discrepancy observed in reaction forces. Fy ranges from 150N to 10,000N  (see Figures in the first message), and these are my primary objectives. 

My  more specific questions:

1) Is there an efficient way using ls-dyna toolbox to identify which simulation approach is surely incorrect.

2) What can be suggested as the best set of tests to identify and validate correct simulation in this case of material cutting. I guess single element tests (e.g. per https://awg.lstc.com/tiki-index.php?page=AWG+ERIF+Test+Example+15) is not appropriate for SPH elements. Than, what would be the right number of elements for SPH representation. 

3) Maybe I should avoid the use of Solid-to-SPH transition, as the most of literature on material cutting is using SPH.

4) But for some reason ls-dyna has this Solid-to-SPH way of simulation. And my assumption was that Solids are better in stress/strain representation, aren't really they?

Any advice is highly appreciated. Thanks

Paul

You received this message because you are subscribed to a topic in the Google Groups "LS-DYNA2" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ls-dyna2/1Vpp-jk6EfE/unsubscribe.
To unsubscribe from this group and all its topics, send an email to ls-dyna2+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/001501d8b887%244376e230%24ca64a690%24%40schwer.net.

--

Best

-p.d.

[l...@schwer.net]

Paul –

 

1\ I know of no LS-DYNA to tool to discern “correctness” among models.  Recall the useful quote from Professor George Box “All models are wrong, but some are useful.”

 

2\ At a minimum I suggest mesh refinement of the SPH elements, regardless of if you still pursue both approaches. Generally, MANY SPH elements are needed to replace one Lagrange solid, i.e. NQ=3 minimum.

 

3\ All such “mapping” of one solution to another involves error. Save yourself grief and stick to the all SPH approach.

 

4\ Developers get paid for new developments. They are the least capable of judging how effective such developments are for users.

 

               --len

[James M. Kennedy]

Corrected link:

 

https://www.dynalook.com/conferences/11th-european-ls-dyna-conference/concrete-penetration/evaluation-of-debris-modeling-technique-on-failure-simulation-of-concrete-structures

.
To view this discussion on the web visit https://groups.google.com/d/msgid/ls-dyna2/000001d8b89b%2464760f10%242d622d30%24%40kbs2.com.

[Pavel Dybskiy]

Adding the group address ..

---------- Forwarded message ---------
From: Pavel Dybskiy <pdyb...@gmail.com>
Date: Thu, Aug 25, 2022 at 3:35 PM
Subject: Re: [LS-DYNA2] Solid to SPH issues. Material cutting modeling.
To: James M. Kennedy <j...@kbs2.com>

Dear James,

Thank you very much!

Regarding overlaying time-histories of Solids and SPHs I have some earlier observations (see graph below where 1xxxxxx are solids and 9xxxxxx are SPHs), that I didn't have an opportunity to discuss in the group. 

In this example I'm using MAT_TABULATED_JOHNSON_COOK, EOS_TABULATED_COMPACTION, iCPL=1, iOPT=0 in Adaptive_Solid_to_SPH, and  *MAT_ADD_EROSION with erosion controlled by eps_eff=0.25.

My specific questions per figure below are:

1) Why does S_vm jump up in SPH when Solid element is eroded?

2) Why does effective_strain in SPH is 0 before Solid element erosion, whereas effective_plastic_strain in SPH follows exactly what is in Solid element before and after erosion? 

And my more general questions are:

3) Should I expect, in theory, the same reaction of material regardless of effective_strain value which defines Solid erosion. My thought was that, in theory(especially in the case of compression, and when Solids are of the "right" size, and SPH are of the "right" quantity), material reaction should not depend on the value of erosion effective strain, given that SPH inherits stress/strain state from Solid element. 

4) Is it the case that I have the above issues because I have the same failure definitions for Solid and SPH materials?  

 

Best Regards,

Paul Dybskiy

--

Best

-p.d.

--

Best

-p.d.