Professional Documents
Culture Documents
By Doug Oatis
Rigid Surfaces in Workbench In order to move your rigid part, you need
One of the biggest additions to Workbench to apply a remote displacement. This will
v11 was the joint utility and rigid body serve as a pilot node for the top surface
solver. With that inclusion, we now have (refer to ANSYS Classic for pilot node
the ability to perform rigid-to-flexible body information). Make sure the rest of your
contact. For those unaware, this capability model is properly constrained, and that you
exists within ANSYS Classic, and allows use automatic time stepping to ease conver-
you to assume parts as rigid bodies (e.g.. gence.
metal forming/drawing). You can do this You can then generate your mesh and see
by modeling the rigid part as just a surface that only the surface of your ‘Rigid’ body
(for 3D applications) or a line (for 2D appli- has been meshed. You can apply whatever
cation). Workbench currently only sup- mesh controls to obtain an adequate mesh.
ports rigid to flexible for a 3D analysis (not You can see from the figure that I applied a
plane stress/strain or axisymmetric). If you mapped mesh control to the top surface of
plan on doing this analysis, you should have my rigid plate.
‘Beta Options’ turned on (Tools > Options
> Common Settings > User Interface).
To perform this analysis in v11, you first
need to model all of your interacting parts. After the model solves, you may note some
Once you import the assembly into Simula- oddities in plotting the displacement results
tion, you can then set the ‘Stiffness Behav- for the rigid surface. This happens because
ior’ in the geometry branch to be “Rigid.” of internal contact trimming performed by
(Cont. on pg. 2)
www.padtinc.com 1 1-800-293-PADT
August 17, 2007 The Focus Issue 59
(Rigid, cont...)
Simulation. Simulation uses contact trim- You can change this behavior by setting solved for. The results with the modified
ming to reduce the number of contact ele- ‘Contact Trimming’ to “Off” in the details contact settings are shown below.
ments in a model. window for your contact pair.
This analysis was performed without the
ANSYS Kinematics or Rigid Dynamics
licenses. If I wanted to specify any type of
joint between bodies (revolute, spherical,
slot, etc.), I would need those licenses to
perform the analysis. You can do this anal-
ysis as both a static structural or transient,
just make sure your time stepping controls
are adequate enough to capture the system
You can see the effects of this trimming by response for the transient environment.
writing out an ANSYS input file and view-
ing the model in ANSYS Classic. Many problems can be simplified with
rigid contact. Take some time to read the
ANSYS manual on rigid-flexible contact
When you do this, the contact region will to learn more
not be trimmed and all the elements will be
(CMS, cont...)
rial nonlinearity, an explicit dynamics code CMS run took 139 seconds. That is an on the promise of Aerospace transient dy-
for this type of problem. Now we can use order of magnitude different! The differ- namic simulation. If you get the opportuni-
ANSYS to solve those linear transient dy- ence will go up and down depending on ty to use this powerful and time saving
namics problems right in ANSYS. how large your model is, but should still be feature, give it a try and let us know how it
a huge factor. goes. You can learn a lot more about CMS
The process is pretty much the same as for
in the ANSYS manual.
a modal CMS run: 1) build your CMS sub- The results between the two methods were
structures, 2) solve your problem using the identical. Fig-
substructures, 3) expand the parts you want ure 2 shows a
at the times you want. The big difference is graph with the
that in step 2, for the use pass, you do an deflection of
ANTYPE, TRANS instead of MODAL, the head and
and specify the proper time dependent load- tail as well as
ing for your part. In many cases this is a the input accel-
maneuver load for an aircraft or missile. eration load.
All values are
Figure 1 shows our silly test case model.
for the X direc-
All of the macros can be downloaded from
tion, from tail
here <LINK>. The non-CMS simulation
towards head.
used RUNFULL.INP and the CMS solve
used GENCMS.INP, USECMS.INP and This simple test
EXPCMS.INP. The CMS model is broken case shows that
into 4 pieces: the body, head, tail and legs. ANSYS, Inc.
Comparing run times shows the Full Run has stepped up
taking 2,208 wall clock seconds, and the and delivered
Fig. 2: Input Acceleration and Response Displacement
www.padtinc.com 2 1-800-293-PADT
August 17, 2007 The Focus Issue 59
www.padtinc.com 3 1-800-293-PADT
August 17, 2007 The Focus Issue 59
Super Simulation
Snippets RPSD Curves from a PSD Run
As a variation on our usual Awesome APDL feature, we want to mix /gropt,logx,on
in some APDL scripts that you can use to expand the capabilities of /gropt,logy,on
/xrange,default
Workbench Simulation. And since we have an unnatural urge to /yrange,default,,1
always give things silly names: we will call this feature Super Simula- ! Plot it (vars 2 and 3)
tion Snippets. /show,png
/gfile,600
We will start with one that we get a lot of requests for. In V11 of plvar,2,3
Simulation, you can do a PSD run very easily. And much of the /show,close
typical output is readily available. Moreover, if you want a response ! Write to a file for external plotting
*get,n_data,vari,1,nsets
or RPSD curve, you can’t get that automatically. *dim,psd_data,array,n_data,5,1
vget,psd_data(1,1,1),1 !frequency data goes here
Enter APDL and code snippets to once again save the day. This vget,psd_data(1,2,1),2 !mass1 data here
snippet, provided by our intrepid snippeteer Doug Oatis, is a good vget,psd_data(1,5,1),5 !base node data here
example of how to include POST26 post processing into a Simulation
*cfopen,'d_test','txt'
model.
*vwrite,'Freq','mass_n','','','d_base'
! Enter Post26 and store PSD info
%14C %14C %14C %14C %14C
/post26
store,psd,5 *vwrite,psd_data(1,1,1),psd_data(1,2,1),psd_data(1,3,1),
! Grab mass node and base node psd_data(1,4,1),psd_data(1,5,1)
! You need to define d_bass as a named selection %14.5G %14.5G %14.5G %14.5G %14.5G
esel,s,ename,,21
nsle *cfclose
*get,mass_n,node,0,num,max allsel,all
allsel,all /post1
cmsel,s,d_base
*get,base_n,node,0,num,max
allsel,all
! Get response info
nsol,2,mass_n,u,y
nsol,3,base_n,u,y
rpsd,6,2,,3,1
rpsd,7,3,,3,1
! Multiply to get the right units
prod,2,6,,,mass_node,,,1/386.4**2,1,1
prod,5,7,,,mass_node,,,1/386.4**2,1,1
! Set up your graph options
/axlab,x,Frequency (Hz)
/axlab,y,PSD (G**2/Hz)
/gthk,axis,2
/grtyp,0
/gropt,ascal,on
The Focus is a periodic publication of Phoenix Analysis & Design Technologies (PADT). Its goal is to educate and entertain the worldwide AN-
SYS user community. More information on this publication can be found at: http://www.padtinc.com/epubs/focus/about
www.padtinc.com 4 1-800-293-PADT
August 17, 2007 The Focus Issue 59
Kiva.org is a non-profit
revolutionizing the fight
against global poverty
by enabling people to
connect with and make
personal loans to low-
income entrepreneurs in
the developing world.
www.padtinc.com 5 1-800-293-PADT