3.3. Initial Conditions

Initial conditions define the state of the system at the start of the analysis. In structural finite element analyses, initial conditions are defined in terms of initial displacements, velocities, and accelerations at all independent degrees of freedom (DOFs).

Because all time-integration schemes (such as the Newmark method and the HHT method) rely on the history of displacements, velocities and accelerations, it is important to define consistent initial conditions. By default, a zero value is assumed for initial displacements, velocities, and accelerations at DOFs that are not otherwise specified (via the IC command).

Inconsistencies in initial conditions introduce errors into the time-integration scheme and lead to excitation of undesired (spurious) modes. Accumulation of these errors over several time increments adversely affects the solution and very often causes the time-integration scheme to fail. Applying numerical damping or other forms of damping can suppress the growth of these errors. However, such additions also affect the solution, especially, when long term transient behavior is being studied in the analysis.

It is not always possible, however, to have complete information about the initial state of a system being modeled for transient analysis. In such situations, it is helpful to run a dummy load step before the actual transient analysis of interest to bring the system into a consistent initial state. The purpose of such a load step is to eliminate the error introduced by inconsistent initial conditions.

3.3.1. Apply Linear Acceleration in a Dummy Transient Analysis

This technique is useful in cases where initial accelerations are non-zero, are known, and are uniform over the entire model. Applying acceleration loading (via the ACEL command) introduces non-zero accelerations into the system. After the analysis has run through one substep, the actual transient analysis can be carried out without the acceleration loading.

Example

Consider a rigid beam of length l rotating in the x-y plane about a pinned end at a constant angular velocity ω. The free end of the beam has a tangential velocity of ωl and a centripetal acceleration of ω2l. The beam is assumed to have all of its mass concentrated at the free end. To perform the analysis, model the rigid beam using the MPC184 element with Lagrange multipliers to enforce the rigid beam constraints. With one end of the rigid beam pinned, apply initial velocity normal to the beam axis at the free end. To introduce centripetal acceleration, use acceleration loading as illustrated in the following input file:

Transient Analysis of a Rigid 3D Beam Rotating About a Fixed Node

/title,Transient analysis of a rigid 3D beam rotating about a fixed node 

/prep7

et,1,mass21
keyopt,1,3,2 !3d mass without rotary inertia
et,2,mpc184
keyopt,2,1,1 !rigid beam
keyopt,2,2,1 !lagrange multiplier

n,1,0.0,0.0  !pinned end (node 1)
n,2,1.0,0.0  !free end (node 2)

type,1
real,1
m = 1.0
r,1,m
en,1,2       !3d mass at free end (node 2)

type,2
real,2
en,2,1,2     !rigid beam

finish

/solu

vel = 6.2831853072       !tangential velocity
ic,2,uy,0.0,vel          !initial condition for velocity
antype,trans
time,1.e-9
acel,0.0,-vel*vel,0.0    !apply centripetal acceleration
kbc,1                    !step loading
nlgeom,on
nsub,1,1,1               !use 1 substep for analysis
trnopt,full, , , , ,HHT  !use HHT time integration
tintp,0.0                !no numerical damping
outres,all,all
solve

d,1,all
ddel,1,rotz
d,2,uz
d,2,rotx
d,2,roty


time,6.0
acel,0.0,0.0,0.0         !remove centripetal acceleration
kbc,1
midtol,on,1e2            !automatic time stepping with MIDTOL
nsub,600,1e7,400
trnopt,full, , , , ,HHT
tintp,0.05               !small numerical damping for HHT
outres,all,all
solve

finish


/post26
/xrange,0.,6.0
nsol,2,2,u,x,ux          !x displacement for node 2
nsol,3,2,u,y,uy          !y displacement for node 2
nsol,4,2,v,x,vx          !x velocity for node 2
nsol,5,2,v,y,vy          !y velocity for node 2
nsol,6,2,a,x,ax          !x acceleration for node 2
nsol,7,2,a,y,ay          !y acceleration for node 2
/axlab,x,Time T
/axlab,y,D/V/A
/gropt,divx,10
/gropt,divy,10
/gthk,curve,2
/title,Transient analysis of a rigid 3D beam rotating about a fixed node 
plvar,ux,uy,vx,vy,ax,ay


finish

3.3.2. Apply Large Numerical Damping Over a Short Interval

This technique is of a more general nature and uses numerical damping to eliminate errors or numerical noise due to inconsistent initial conditions. After the noise has been damped out over several substeps, you can perform the actual transient analysis with smaller numerical damping.

Some potential drawbacks exist in cases where high frequency content of flexible multibody systems is important for analysis. Applying high numerical damping in the dummy analysis can affect the desired high-frequency response. Ansys, Inc. recommends using the HHT method for this technique because the integration scheme shows good dissipation properties with numerical damping.

Example

Consider a rigid-flexible double pendulum made up of a rigid and a flexible beam. One end of the rigid beam is pinned and the other end is hinged to the flexible beam. The other end of the flexible beam is free. The rigid beam is assumed to have all of its mass concentrated at the end that is hinged to the flexible beam. The system is given an initial velocity tangential to the flexible beam axis at its free end, as shown in the following input file:

Transient Analysis of a Rigid-Flexible Double Pendulum

/title,Transient analysis of a rigid-flexible double pendulum

/prep7

et,1,mass21
keyopt,1,3,2   !3d mass without rotary inertia
et,2,mpc184
keyopt,2,1,1   !rigid beam
keyopt,2,2,1   !lagrange multiplier
et,3,mpc184
keyopt,3,1,6   !revolute joint between rigid and flexible beam
et,4,beam188   !flexible beam

n,1,0.0,0.0    !pinned (supported) end of rigid beam
n,2,1.0,0.0    !hinged end of rigid beam (node 2)
n,3,1.0,0.0    !hinged end of flexible beam
n,4,1.25,0.0
n,5,1.5,0.0
n,6,1.75,0.0
n,7,2.0,0.0    !free end of flexible beam (node 7)

type,1
real,1
m = 390
r,1,m
en,1,2         !3d mass at the end of rigid beam

type,2
real,2
en,2,1,2       !rigid beam

local, 11, 0, 0.0, 0.0, 0.0, , , 90
sectype, 3, JOIN, REVO, TESTREVO
secjoin, , 11, 11
type,3
real,3
secnum,3
en,3,3,2       !revolute joint

mp,ex,1,2e11   !material properties for flexible beam
mp,nuxy,1,0.3
mp,density,1,7.8e3

sectype,4,beam,csolid
secdata,1,0.1784124116 !c-s area is 0.1
type,4
real,4
secnum,4
mat,1
en,4,3,4       !flexible beam elements
en,5,4,5
en,6,5,6
en,7,6,7

d,1,all
ddel,1,rotz


finish

/solu

vel = 6.2831853072        !tangential velocity
ic,7,uy,0.0,vel           !initial condition for velocity
antype,trans
time,0.1
kbc,1
nlgeom,on
nsub,50,50,50             !use multiple substeps
trnopt,full, , , , ,HHT   !use HHT time integration
tintp,0.2                 !use high numerical damping
outres,all,all
solve

time,6.0
midtol,on,10              !automatic time stepping with MIDTOL
nsub,100,1e6,100
trnopt,full, , , , ,HHT
tintp,0.05                !small numerical damping for HHT
outres,all,all
solve

finish


/post26
nsol,2,7,u,x,ux      !x displacement for node 7
nsol,3,7,u,y,uy      !y displacement for node 7
nsol,4,2,u,x,ux1     !x displacement for node 2
nsol,5,2,u,y,uy1     !y displacement for node 2
nsol,4,3,v,x,vx      !x velocity for node 7
nsol,5,3,v,y,vy      !y velocity for node 7
nsol,6,7,a,x,ax      !x acceleration for node 7
nsol,7,7,a,y,ay      !y acceleration for node 7
/axlab,x,Time T
/axlab,y,D/V/A
/gropt,divx,10
/gropt,divy,10
/gthk,curve,2
/title,Transient analysis of a rigid-flexible double pendulum
plvar,ux,uy,ux1,uy1,vx,vy,ax,ay


finish