Hints on ANSYS Modeling ANSYS gets much easier to use if you take advantage of the FILE.LOG capability. This file saves all the entries you make in the Graphical User Interface (GUI). From the minute you turn on ANSYS, it records everything in this file. Thus, this file will have what you just did, along with what you did last time and all the times before that until the last time you deleted or erased it!!

cageysyndicateUrban and Civil

Nov 15, 2013 (3 years and 6 months ago)


Hints on ANSYS Modeling

ANSYS gets much easier to use if you take advantage of the FILE.LOG capability. This
file saves all the entries you make in the Graphical User Interface (GUI). From the
minute you turn on ANSYS, it records everything in this fil
e. Thus, this file will have
what you just did, along with what you did last time and all the times before that until the
last time you deleted or erased it!!

So, to use it well…use the SEARCH feature of your operating system to find the
directory (folde
r) where FILE.LOG is saved by ANSYS. Place a shortcut to this folder
on your desktop. Then, with ANSYS not running, delete all the files in this directory
that start with FILE with various extensions such as FILE.DB, FILE.RST, FILE.ERR,
etc…delete them
all. Then, turn on ANSYS. Note that it creates a new FILE.LOG.
Create a shortcut to this FILE.LOG on your desktop. Click on the shortcut at any time to
see what is in it. Note what is in it as soon as you open ANSYS.

Now, Enter some Parameters. Us
e the shortcut to look at the current version of the
FILE.LOG. ANSYS can write into this file, even if you have it open!!. You can save a
copy with a new name such as MYPROJECT_PARAMETERS to indicate the place
where you left off. You could also give it
any name to indicate where you are such as

Now, go to the preproccessor. Add the element type 95 and pick the option of extra
output as Nodal Stress. Now look at the FILE.LOG using the shortcut. Notice that you
can see that eleme
nt type1 is defined as the 20NODE SOLID95. Further, just below, you
see that the 4 options are listed as if you entered commands. The action of the mouse
saved the 4 choices….the extra nodal output is option 5 and choice number 2.

Notice that there are
!* as lines that separate things making it easier to read. ANSYS
ignores anything following the exclamation (!) on any line. Thus

!********************REPRESENTS A COMMENT*************

you can enter comments to help you remember what you did in your l
og files. For
example you can say

!**********Select Solid element type 95 and enter the extra output at nodes********

or later

!***************enter the Young’s Modulus and Poisson’s ratio as E and NU********

Once you understand the log files, you c
an use them to save lots of time because you can
restart using the log files.

To do this, either from a fresh start of ANSYS or after issuing the command
FILE/Clear & Start NEW…..choose FILE/Read Input from… and then select the log
file that you sa
ved with the name MYPROJECT_InProgress.txt. When you issue this
command, all the things you did get repeated and the FILE.LOG gets an entry such as

',, 0 Which basically tells the name
of the input file, i
ts type and location. Mine was in C:
FEM81 …the directory on my C
drive where I store results from ANSYS version 8.1

Once you have meshed your object, you can save the FILE.LOG with a new name such
as MYPROJECT_MESHED. You can then change any paramete
r and then get to this
point easily. You can then add the boundary conditions (the displacement constraints and
the loading forces on the nodes) These are attached to specific nodes by their node
numbers. You can see the numbering method by seeing what
is in the log file. You can
also turn on the node numbers to observe them yourself. Now…consider what happens if
you change the number of elements along a specific side. This will change the number of
nodes along that side so when ANSYS numbers the node
s, the particular nodes that you
locked down for example with the ALL DOF =0 command will no longer be in the same
place or even adjacent to one another. Thus, if you change the number of element
parameters (NL or NH or NT in our examples) you will have t
o reenter the loads and

If you do not change these, then you can continue on to the solution and to the post
processor and then save the FILE.LOG file with yet another
name….MYPROJECT_XSTRESS for example which shows the solution complete
the plotted shape and stresses. Then, you can edit this file to change the value of the
parameter for the height of the beam for example *Set, H,1 is the old value….change it
to *Set, H, 1.5 then you can run the problem again to see the re
sults for a beam of
1.5 inches high. This makes it easy to study geometry effects.

Remember that these solutions are always linear….which means that doubling the load
doubles the stress and doubles the displacement at any point. Remember you can def
the sample by prescribing a displacement that is not zero is you hold some other place
from moving….The same as if you had pulled it to make the nodes move to where you
told them to go.

The maximum and minimum are indicated on every contour plot by MX

and MN and the
maximum displacement is always shown on the side of the output.

You can use symmetry to model only a portion of a beam. If you think about it, all these
beam problems are symmetric about the center both left to right and front to back.

Therefore, if we modeled only half and constrained the nodes at the left edge to only not
move left and right, we could put the load on the left edge pointing down with the full
value and then fix Uy and Uz of the right hand support and get the same answe
r with only
half the model…Further, we could cut this piece that is left the thin way and set Uz=0
along the centerline, change the load to half and still get the same answer. Since we only
have one element thick in out examples, this is not useful but if
we had had multiple
elements in the thickness direction, we could get away with half the number using this
symmetry about the Z=0 plane.

These symmetry considerations are very helpful. You will not have to model both sides
of your bridges…only ¼!

Now fo
r the rivets…

The rivet expands and fills the hole. Thus it is as if the hole is filled with aluminum. In
other words, a tight rivet acts as if the material has no hole. There may still be a stress
concentration if the loads get large enough to cause t
he rivet to separate but we can think
of the rivet as effectively filling the hole. We model this then by applying loads to the
surface nodes in the physical location where the rivet is to be. We can use a “circle”
when picking the nodes to load, then, t
urn the view 90 degress and use a box to
“unselect” the nodes inside the object. This lets us select just the surface nodes. Now
apply a displacement condition so they cannot move in the vertical direction. Apply
force loads on the other end of the riv
eted piece. Use symmetry to model only one half
of the riveted joint and even more symmetry to model only one half of the lower piece,
realizing that on the centerline, the lateral displacements will be zero. See the screen
shot below which shows how to

pick the nodes for the half rivet hole. These nodes will
be constrained in the Uy direction

The nodes on the left hand side will now be set to have Ux =0 while the nodes on the
bottom will be given an Fy=

LOAD/Number of Nodes
on Bottom.

One of the nodes in the center of the rivet is Locked with Ux=Uy=Uz=0

Now we solve and look at the stresses.

Below I show the Y stress. Note that it is very uniform over most of the part and just
below the rivet, shows

tension as a stress concentration. There are some “ARTIFACT”
or FALSE stress concentrations at the edges on the bottom because we put the same load
on every node…the nodes on the edges should have less load to represent the same stress
because the load o
n a node logically represents the force caused by the area surrounding
that node and the area “stops” at the edge. Thus, we really need to apply half the load at
edges and even less at corners…There are better ways to apply loads by applying
stresses. We

could also apply a displacement load at the bottom which would be then
uniform because the uniform displacement is what results from a uniform stress. This is
where we have to realize that FEM is a tool that requires careful attention to all details…

astly, we can look at the total reaction load by choosing “List Result
solution” . As seen in the next screen shot

If we now pick structural forces Fy we will get the following:

Which, after scrolling to the b
ottom shows 53 as the total load on all the fixed points.
This corresponds to the 53 nodes on the bottom of the model because I put one pound on
each node.

After thinking about these results and trying this yourself, this should give you an idea of
e stresses associated with the rivets…

You can try holding up your bridge (those plates we did earlier) by circles of nodes on
their surfaces which represent rivets…we can look at the total shear forces in the rivets
and compare that with the forces we wi
ll measure to shear rivets.

This will show that the rivets will tend to twist the structure a bit. Note that the rivets
here actually caused bending of the lapped part…even though we only did ¼ of it…

Hey…This ANSYS FEM lets us experiment a bit before w
e ever build our first
prototype…that is the whole idea.

Have a good weekend,

Professor Quesnel.