Learning Forum

[valerym.m5]

Good day,

I have encountered many times a situation that is incomprehensible to me, setting the initial speed:

setting the speed direction (-/+) is interpreted in the program UNPREDICTABLY.

in particular, in the attached file the speed is set with the "+" sign - i.e. the direction is up,
but the body = to which the speed is applied - moves down, as I need

maybe I am making mistakes somewhere... or using the program settings incorrectly?

https://drive.google.com/file/d/1UgAJFev6nbMneTF-GcsvzFqwRH2cFDL6/view?usp=sharing

[Pedram Samadian]

Hi Valerym,

I created a simple cube tensile model and applied the same keyword card to define an initial velocity. As can be seen below, the displacements are positive as expected given the positive y-velocity provided. STYP in my model is 2 (i.e part ID) but is 1 in your model (i.e. part set ID). Make sure that your "part set #1" only includes the part you want to apply the initial velocity, not any other parts. 

Thanks,

Pedram

[valerym.m5]

Dear Pedram,

Thank you for your advice.

I checked the correctness of the initial velocity assignment in my model - everything is set correctly there.

It was not by chance that I applied the term "unpredictable" to the assignment of the initial velocity for a rigid body.

I have already had several cases when the model correctly perceives the direction of the initial velocity, but then - in the process of some changes in the model, the specified direction of the initial velocity is perceived by LS-DYNA as the opposite, although I did not redefine the direction of the initial velocity.

*** Unfortunately, I have never been able to "fix" - at what stage of the model change LS-DYNA stops correctly perceiving the specified direction of the initial velocity.

As for your advice - to set the initial velocity for a specific part of the model - I will try to do this now and will write to you about the results.

Regards Valery

[valerym.m5]

 

Dear Pedram,

I followed your advice – I set the initial speed for a specific part.

the model can be obtained from the specified link:

https://drive.google.com/file/d/1qKZq45Eu8MOuKVOMhO2-eEMXLpvbiQLV/view?usp=drive_link 

to my regret, LS-DYNA perceives the set initial speed incorrectly:
initial speed = – 2800 mm per second, direction down.

but as you can see the part moves up.

 

 

 

[Pedram Samadian]

Hi,

Thanks for the details. I am not sure which part the initial velocity was assigned to, but my guess is one part is hitting the other part based on this given velocity, and after the collision, the colliding part changes its moving direction. If that is the case, the reason you do not see the transfer of you part along the expected orientation before the collision might be the low output interval you set. Ensure your output interval in the DATABASE keyword cards is high enough to increase the resolution of the results (more data along the time).

Please keep me posted.

Thanks,

Pedram

[valerym.m5]

the solution step is quite small:
dt 1.72E-07

in addition, if there was an impact/contact with bodies lying below the load, then a wave pattern would be visible in the parts where the material has viscous properties.

[Pedram Samadian]

How long does it take for the part with velocity to reach the stationary part?

I also see the time-integration scheme has changed from implicit to explicit at some point. The implicit time steps are much higher than the explicit ones. What was the time step size for the implicit portion of the simulation?

 

 

[valerym.m5]

the gap between parts = 1.02 mm
initial speed 2800 mm per second
time it takes to start contact = 0.000364 seconds

[valerym.m5]

this is the curve of maximum values ​​of the solution step for the implicit solver

[valerym.m5]

by the way, getting/writing calculation results according to the curve with sequential use of implicit and explicit solver is also a problem for me - which I wrote about here:

https://innovationspace.ansys.com/forum/forums/topic/restart-switch-implicit-gt-explicit/

[Pedram Samadian]

Hi Valery,

Thanks for the details. Unfortunately, I cannot download your model here to figure out what is going on in your model unless you can send it through Ansys Customer Support Space (ACSS).

The contact is established after 0.000364 s, so please ensure that your output interval for d3plot (in *DATABASE_BINARY_D3PLOT) in the implicit portion of your simulation is lower than this. Then plot the velocity history of the part. The velocity sign must be consistent with the direction of the part movement in each output within the period of 0 s to 0.000364 s. 

Thanks,

Pedram 

[valerym.m5]

Dear Pedram,

I will definitely take advantage of your kind offer.

But first I would like to make sure that my assumptions are correct.

 

So:

1. I could not learn to regulate the recording of calculation results at the initial stage of using an explicit solver.
For some unknown reason, the program ignores my instructions and writes the calculation results “arbitrarily” – immediately after switching the solver type – in this model – about 0.02 seconds.
After 0.02 seconds, the results are recorded with a step “close” to the required step, but still not exactly to my requirements specified in the curve.

 

 

 

•  I am interested in the calculation results at the initial moment of time – at the moment of impact of the load.

 

 

 

 

2. Since I couldn’t solve the 1st problem, I tried to use the command:
Velocity_Generation_Start_Time
Where, according to the documentation, you can set a “delay” in time for the initial velocity applied to the “rigid” body.
However, ls-dyna ignores any specified delay values ​​and the initial velocity is activated at the moment of switching from the implicit to the explicit solver.

 

3. Taking into account my unsuccessful attempts 1 and 2.
I decided to achieve the required delay in the time of application of the initial velocity by moving the falling load a distance that I calculated from the specified initial velocity of 1000 mm per second.
However, from the analysis of the solution I found that the initial velocity of the load is not 1000 mm per second, but 2567 mm per second!

0.00805664 sec is the time of motion of the rigid body
20.68639946 mm is the distance the rigid body has moved
2567.62117458 mm/sec is the constant speed of motion

 

If someone wants to get acquainted with my model – see the link below

https://drive.google.com/file/d/1jf3XCB1v87ImHqMO-PfN6SXNVcTPGAUT/view?usp=sharing

 

I will be very grateful for any advice – since in this case I have a stalemate.

———————————-

I tried to solve this problem with other data – trying to understand possible patterns, and I was very surprised, since the initial speed given was different each time?!

Below is another example of the solution

time	Uy
10.0000000	0.0000000
10.0168285	20.1738129
10.0215492	25.8324966
10.0228729	27.4195175
10.0238628	28.6061802
10.0248413	29.7792721
10.0258083	30.9385910
10.0267649	32.0844803
10.0277100	33.2173271
10.0286436	34.3369370
10.0295668	35.4437866
10.0304794	36.5377007
10.0313816	37.6189613
10.0322733	38.6876526
10.0331545	39.7442780
10.0340252	40.7885284
10.0348864	41.8205681
10.0357380	42.8409538
10.0365791	43.8493919
10.0374107	44.8462029
10.0382328	45.8315926
10.0390453	46.8054848
10.0398483	47.7680931
10.0406418	48.7197609
10.0414267	49.6601524
10.0422020	50.5898819
10.0429688	51.5088005
10.0437260	52.4171677
10.0444756	53.3149796
10.0452156	54.2024994
10.0459471	55.0797882
10.0466709	55.9469223
10.0473852	56.8037834
10.0480919	57.6509781
10.0487909	58.4884300
10.0494814	59.3161201
10.0501642	60.1341667
10.0508385	60.9428711
10.0515051	61.7423515
10.0521641	62.5323830
10.0528164	63.3134384
10.0534601	64.0852509

 

Dt	0.0534601
D_Uy	64.0852509
speed	1198.7487

For the last solution, the initial speed was specified -2800 mm per second,

but in reality – according to the calculation – the speed was 1198 mm per second

 

 

 

[valerym.m5]

here is another fact - which I cannot understand:
judging by the graph - the initial velocity of the rigid body increases over a fairly long period of time.

I do not specify this period of time...

ls-dyan somehow determines it itself.....

Nothing is said about this in the documentation... - to be more precise, I did not find anything

hence the question:

and how then can one specify the initial velocity of the rigid body - without the "acceleration time"??

Is it possible to do this at all?

and one more question:

I specify the initial velocity of 1000 mm per second, and here again ls-dyan "acts up"... - the velocity - judging by the graph:

1.1987445068e+03 mm per second

1.0016828537e+01 1.1987445068e+03

 

[valerym.m5]

in my model, the parameter PHASE:=0
EQ.0. Velocities applied immediately

but if I understand the calculation results correctly, then ls-dyna solves the problem with the parameter PHASE:=0 EQ.1. Velocities applied after dynamic relaxation.

and with this ls-dyna somehow independently calculates the period of dynamic relaxation and for some reason it also changes the amplitude/magnitude of the initial velocity that I set.

[Pedram Samadian]

Hi Valery,

Can you please show your specified output interval histroy that includes both the implicit and explicit parts of your model? The image in your 4th comment shows that as the solver changes from implicit to explicit, an output is written in the d3plot. Where is that 0.02 second gap?

The *LSO_TIME_SEQUENCE card may also help you for managing output writing (LS-DYNA Multiphysics Manual).

Using Velocity_Generation_Start_Time, the initial velocity occurs only in the regular transient phase after dynamic relaxation (if used). Neither PHASE value will cause initial velocity to be imposed in the dynamic relaxation phase. It is surprising that the velocity does not start from what you specified, but given the previous explanation, maybe dynamic relation is being applied during that period.

Please also make sure to use the lastest LS-DYNA solver downloadable at LS-DYNA (user=user) Download Page(username: user & password: computer). The double precision type is recommended.

I hope this information helps you.

Pedram

[valerym.m5]

 

Good day Pedra

thank you for your recommendations.

As for the curves concerning the maximum value of the integration step and the step of output of the calculation results – they are in the model.

But as far as I understand, if the model uses switching from an implicit to an explicit solver,
then in this case – ls-dyna does not want to fulfill the exact requirements for output of the calculation results specified in the curve.

it not only writes the results “arbitrarily”, but also changes the “prescribed” value of the initial velocity – as a rule, to a greater one.

maybe I’m wrong about something – since my experience with ls-dyna is very modest.

So I hope to hear the opinion of more qualified engineers.

Reagrds Valery

 

 

[valerym.m5]

 

 

Dear colleagues,

I offer you two models that demonstrate “oddities” that I could not understand.

I will be glad if you write your assumptions… advice…. comments…

====================
Model
STYP-2_Relax.key
Link to the model:
https://drive.google.com/file/d/1P3C9CnTEGN145BxeU_X8ALUofHXnlTHp/view?usp=sharing

The initial speed of the cargo is set: -1000 mm per second

The cargo moves down.

As the calculation results show, the load receives a real initial speed in the calculation: -2594.6645508 mm per second
during the rise time: 0.00009155 sec – see below

Vy.txt
3.6099996948e+02 0.0000000000e+00
3.6100009155e+02 -2.5946645508e+03
3.6100531006e+02 -2.5946645508e+03

======================
Model
No_Relax_1.222e-4-1000mm_0_023sec.key
link to the model :
https://drive.google.com/file/d/1eOWIbv6e2wMkJ7HqZMFYVccMiwD1LyFJ/view?usp=sharing

The initial velocity of the cargo is set: -1000 mm per second

The cargo moves up.

Curveplot
Nodal History
Time
Y-velocity
Node no.
222980 #pts=266
* Minval= -1.0000000000e+03 at time= 0.0000000000
* Maxval= 3.7985620117e+03 at time= 10.0000991821
9.9999990463e+00 0.0000000000e+00
1.0000099182e+01 3.7985620117e+03
1.0001098633e+01 3.7985620117e+03

========================================

List of questions I wanted to would like to get answers:

1. why, when setting one “initial” speed – 1000 mm per second,
in the first model – the cargo moves down, and in the second model – the cargo moves up?

2. why, when setting one “initial” speed – 1000 mm per second,
in the first model – the actual speed of the cargo according to calculations = - (MINUS) 2594.6645508 mm per second, 
and in the second model – the actual speed of the cargo according to calculations = + (PLUS) 3798.5620117  mm per second?

3. Why do the above “initial” speeds reach constant values ​​in different times?

 

 

[valerym.m5]

by the way, here is another model where the initial speed is given with a plus sign:
+1000 mm per second

https://drive.google.com/file/d/1xZ0Byw0l9EbzeD-TFVCRAJDfirI0DbcJ/view?usp=sharing

but in fact, according to calculations - the speed is NEGATIVE = - 1117 mm per second

9.9497890472e+00 0.0000000000e+00
1.0000000000e+01 0.0000000000e+00
1.0008694649e+01 -1.1772938232e+03
1.0009955406e+01 -1.1772938232e+03

i.e. a body with a given positive speed = +1000 mm per second should move up, but for some reason the body moves down?!

[Pedram Samadian]

Hi Valery,

Please remove any relevant mass scaling and dynamic relaxation cards (e.g. control_timestep and control_dynamic_relaxation) from the model and then run it.

Please keep me posted on the outcome.

Thanks,

Pedram

[valerym.m5]

Hi Pedram,

unfortunately, removing the mentioned cards did not produce any positive results.

Regards Valery

[valerym.m5]

 

Hi Pedram,

thank you very much for your advice – I will try to use your advice.

However, I hope to get answers to my questions from more qualified colleagues.

I find it hard to believe that no one has tried to solve such problems before and tried to figure out what is going on?

To my regret, I could not find even a brief description of the "actions" of the program
when switching solvers (implicit <=> explicit) occurs using a curve.

Maybe someone has already found at least some information on this issue?

Regards Valery

 

[Pedram Samadian]

Hi Valery,

I am sorry that the issue is still persisting. Unfortunately, I cannot reproduce the error you are getting in simple models I created. In my last try, I built a simple impact model and used the initial velicity generation card and a combined implicit-explicit time integration scheme (implicit before collision and explicit after that). In the predictions, the part moved in the same direction as that specified in the initial velicity generation card until the collision and could get recordings at the preset time intervals, including the exact time when the explicit solver starts working (you can see it in the image below).

We need to review your model in detail to see what the problem is. As I noted, Ansys employees cannot download or upload files on the Forum platform, but it can be done through Ansys Customer Support Space (ACSS) after users open a ticket. 

Thanks,
Pedram