Learning Forum

[julius.langenhorst]

Hello,

I'm trying to find the frequency response of a structure that is subjected to accelerations in x-, y-, and z-direction, respectively. Using Full Harmonic gives satisfying results but computational times are too long which is why I want to switch to MSUP. I know from solution #855002 on the Customer Portal that the required required must extend past the range of interest. I'm interested in the first 6 modes and work with the first 200 modes. The only matching result from MSUP is in z-direction. The first image below is from the Full Harmonic and the second from the MSUP analysis, both show displacements.

Below are parts of the APDL code I'm using:

antype,MODAL
modopt,LANB,200,0,0,,OFF
mxpand,200,,,YES,,YES
outres,ERASE
outres,ALL,ALL
outpr,ERASE
outpr,ALL,NONE
resvec,ON
allsel
modcont,ON
thexpand,OFF
acel,89,0,0
allsel
save
solve
lsclear,INER
acel,0,89,0
allsel
save
solve
lsclear,INER
acel,0,0,89
allsel
save
solve
lsclear,INER
finish

The part below is repeated in the same way for the other two directions.

antype,HARMIC
hropt,MSUP,,,YES
hrout,OFF,OFF,ON
harfrq,NINT(first eigenfrequency),NINT(sixth eigenfrequency),,,%freq_array%
nsubst,5000
dmprat,0.0001
kbc,1
resvec,ON
lvscale,1,1
lvscale,0,2
lvscale,0,3
outres,ERASE
outres,ALL,ALL
outpr,ERASE
outpr,ALL,NONE
solve
save
finish

/SOLU
allsel
expass,ON
numexp,ALL
solve
finish

Post-processing is done in /post26. I'm getting the displacement using the NSOL command.

I would like to know why there's no response in x- and y- direction when using MSUP. How must the code be changed to solve this problem?

Thank you in advance!

[Chandra Sekaran]

In the modal analysis you create 3 load vectors. But looks like you have only one load vector created in the Z direction. Can you rerun the modesup harmonic with lvscale applied to only load vector 3. I wonder if load vector 3 even exists. 

[julius.langenhorst]

Thank you for your answer!

The picture in my original post should be the result of load vector 3. The way I structured my analysis is as follows (I probably didn't made that clear enough before):

• Modal analysis and creation of three laod vectors
• First MSUP analysis with subsequent post-processing using first load vector:

lvscale,1,1
lvscale,0,2
lvscale,0,3

• Second MSUP analysis with subsequent post-processing using second load vector:

lvscale,0,1
lvscale,1,2
lvscale,0,3

• Third MSUP analysis with subsequent post-processing using third load vector:

lvscale,0,1
lvscale,0,2
lvscale,1,3

Each MSUP analysis is started by issuing a new /solu command and closed with finish. This way all three load vectors should be used except, of course, they weren't created in the first place. But I used this evaluation scheme on other models and there was a response in x- and y-direction (both for MSUP and Full Harmonic).