Questions about oscillation caused by applying gravity

[Evan Yi]

Hello, everyone:

I am trying to simulate gravity in my model, which is a reinforced concrete member subjected to impact loading. It is found that Dynamic Relaxation is time-consuming in my cases, and some references recommended the mass-weighted damping method. So I decided to apply gravity and eliminate oscillation by the mass-weighted damping method.

I applied the gravity by *LOAD_BODY_Y, which is kept constant at 9.8e-3 mm/ms^-2. The damp is applied by *DAMPING_GLOBAL. In order to conduct subsequent impact analysis which should not be subjected to damping, a time-variant curve is defined which is suggested by the relevant reference. I did a rough calculation, the critical damping constant is about  162 rad/ms in my cases, so the damping constant is kept at 300 rad/ms to guarantee the system is overdamped.

The result shows that the oscillation caused by suddenly applied gravity is well-suppressed when the damp exists. However, after the damp withdrawal, the system re-oscillation in a small degree. I cannot figure out why it happens since the oscillation has been eliminated for a while. Could any users with experience in applying gravity provide some suggestions? I am sincerely grateful for any suggestions or recommendations.

BTW, the suggested mass-weighted damping method is proposed in Performance and sensitivity analysis of UHPFRC-strengthened bridge columns subjected to vehicle collisions. My outcomes are presented as followed:

Fig. 1 the damping-history curve

Fig. 2 the resultant force of cross-section

[l...@schwer.net]

I have never used this technique for modeling preload due to gravity.

 

Since the damping forces are proportional to nodal velocity, I suggest fringe plots of nodal velocity after the global damping terminates to locate portions of the model with significant nodal velocities.

 

You might also consider tapering off (ramping) the global damping down to zero rather than suddenly turning it off. I assume the gravity Load Body remains on and constant for the duration of the simulation?

 

You first history plot, indicating something is small or zero, is unlabeled ?

 

                --len

Fig. 1 the damping-history curve

Fig. 2 the resultant force of cross-section

 

 

--
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/a4982d26-cfc5-4982-86e7-9a1ae5b21987n%40googlegroups.com.

[James M. Kennedy]

Dear Evan,

 

The following provides some guidance, notes and necessary input features for modelling the preloading

phase of a bolted structure and the subsequent phase where the structure is impacted by a separate object

at a prescribed initial velocity (for example, an SPH bird-strike). Much of this has been drawn from the

very nice shared notes by Jim Day in the following, regarding several preload approaches:

 

http://ftp.lstc.com/anonymous/outgoing/jday/bolt_preload3.pdf

http://ftp.lstc.com/anonymous/outgoing/support/FAQ_docs/preload.pdf

 

The specific manner in which a preload application via dynamic relaxation is made effects the preload

transition to the transient analysis phase. An internal load need not be sustained to maintain equilibrium

after equilibrium is established in the preload phase; example, *INITIAL_STRESS_SECTION. An

external load needs to be sustained to maintain equilibrium of the preloaded phase; examples, *LOAD_

THERMAL or *LOAD_BODY (gravity).

 

Five different preload options are discussed which use *INITIAL_STRESS_SECTION (the bolt stress

is specified directly in these preload options):

1. Explicit Dynamic Relaxation (ExpDR)

2. Implicit Dynamic Relaxation (ImpDR) - IDRFLG=5/6

3. Transient Explicit with Mass Damping

4. Transient Implicit/Explicit Single Switch

5. Two Separate Analyses - dynain mods in Transient Explicit

 

Two different preload options are discussed which use *LOAD_THERMAL (multiple runs are ne-

cessary to tune the preload to give the desired bolt stress):

1. Explicit Dynamic Relaxation (ExpDR) – temperature load

2. Implicit Dynamic Relaxation (ImpDR) - IDRFLG=5/6 – temperature load

 

Also, some options are offered for *LOAD_BODY (gravity) and presented as a variance from the

*LOAD_THERMAL discussion.

 

Sincerely,

James M. Kennedy

KBS2 Inc.

May 17, 2021

 

From: ls-d...@googlegroups.com [mailto:ls-d...@googlegroups.com] On Behalf Of Evan Yi
Sent: Monday, May 17, 2021 1:12 AM
To: LS-DYNA2 <ls-d...@googlegroups.com>
Subject: [LS-DYNA2] Questions about oscillation caused by applying gravity

 

Hello, everyone:

--

[James M. Kennedy]

3. Transient Explicit with Mass Damping

 

As an alternative to using Explicit Dynamic Relaxation (ExpDR), in some cases

the preload can be established in the early part of the regular transient

simulation.

 

*CONTROL_TERMINATION

$#  endtim    endcyc     dtmin    endeng    endmas

  4.000000         0     0.000     0.000     0.000

 

*INITIAL_STRESS_SECTION

$# *INITIAL_STRESS_SECTION will preload a cross-section of solid elements to

$# a prescribed stress value.

$#   issid      csid      lcid      psid       vid

        13        13         2       213         0

        14        14         2       214         0

 

*DEFINE_CURVE_TITLE

$# The question arises as to whether you can ramp up the preload quasi-

$# statically (using *INITIAL_STRESS_SECTION with SIDR=0) and then hold steady.

$# The key is to define the stress vs. time curve such that you ramp up to the

$# preload stress, and then hold that stress constant for a short period of

$# time, long enough for the dynamics to settle out with the aid of *damping.

$#

$# NOTE: Do “not” define the curve beyond the birth time of the initial velocity.

$# You want to allow the bolt stress to change in response to the dynamic load.

$#

$# The preload stress is just intended to bring the model into a state of pre-

$# load equilibrium. Once that equilibrium is established, it's not necessary

$# to prescribe that stress any longer.

Bolt_Stress

$#    lcid      sidr       sfa       sfo      offa      offo    dattyp

         2         0  1.000000  1.000000     0.000     0.000         0

$#                a1                  o1

               0.000               0.000

            1.000000             218.500

            2.000000             218.500

 

*DAMPING_GLOBAL

$# Use time-dependent mass damping (*DAMPING_GLOBAL) to impose near-critical

$# damping until preload is established.

$#    lcid    valdmp       stx       sty       stz       srx       sry       srz

        10     0.000     0.000     0.000     0.000     0.000     0.000     0.000

 

*DEFINE_CURVE_TITLE

$# Drop damping constant to zero after preload is established and transient

$# loading is ready to be applied.

Mass_Damping

$#    lcid      sidr       sfa       sfo      offa      offo    dattyp

        10         0  1.000000  1.000000     0.000     0.000         0

$#                a1                  o1

            1.000000               0.000

            1.250000            0.100000

            1.750000            0.100000

            2.000000               0.000

 

*INITIAL_VELOCITY_GENERATION

$# Apply transient loads AFTER preload is established. Use nonzero birth time

$# or arrival time for transient loads.

$#nsid/pid      styp     omega        vx        vy        vz     ivatn      icid

       999         3     0.000 175.00000     0.000     0.000         0         4

$#      xc        yc        zc        nx        ny        nz     phase    iridid

     0.000     0.000     0.000     0.000     0.000     0.000         1         0

 

*INITIAL_VELOCITY_GENERATION_START_TIME

$# Use *INITIAL_VELOCITY_GENERATION_START_TIME for problems whose transient

$# response is driven by initial velocity. Delays onset of "initial" velocity.

$#   stime

  2.000000

 

*DEFINE_COORDINATE_NODES_TITLE

Local_Coordinate_System for impacting part assigned initial velocity

$#     cid        n1        n2        n3      flag       dir

         4   5000002   5000001   5000003         0X

Hello, everyone:

--

[Evan Yi]

Dear Len:

Thanks for your answer. According to your suggestions, I checked the nodal velocity fringe plot, and it is found that significant velocity happened on the "weight" part which is purposed for applying extra loads. Also, you mentioned that the damping force is proportional to nodal velocity ( viscous damping type I guess?), the minor portion of kinetic energy or nodal velocity caused by suddenly applied gravity might take a long time to be dissipated since its nodal velocity is small. I asked a colleague yesterday and he said his models usually take 500ms to attain static equilibrium. I personally speculate there is room for optimization since the viscous damping is not very suitable in this case and such a long time to attain static equilibrium is also time-consuming.

I am sorry that I did not notice the first plot is unlabeled. The Abscissa axis is time and the Ordinate axis is damping constant. This is the curve that defines time-varied damp, it ramps up during 4ms from zero to 300 rad/ms, keeping constant for 72 ms, and ramp down to zero for 20ms. Yes, the applied gravity is kept constant during the whole analysis.

Thank you again, I will try to troubleshoot it today. If there is any progress, I will post here.

[Evan Yi]

Dear Kennedy:

Thanks for your replies. I tried the DR method, but it always diverges in my model. I have never seen the energy requirement (1e-3) meet in my model, so I give it up. The example you provide could be very helpful to me, and I will dig into it that whether some useful information I can get.