FEA FOS question...again

I have a question about determining the minimum factor of safety
of a part that probably doesn't have a simple answer I know.

When doing FEA reports on parts it's very important to have a factor of
safety. However, due stress concentrations and various
other factors, how do you fea guru's determine what this fos is?
I can't believe you would follow exactly what the software gives
you and list it to be that...? It seems to me in most cases with
stress concentrations every part would be overdesigned. Is there
a rule of thumb or other factors that I should look for when
determining what the "most accurate" fos to be for a part.

Does anyone know what I'm talking about? I hope so!
Thanks for any light shed on this.
Don

 

Please answer these questions:

1. Loads - Static or repeated?

2. Material - Ductile (most steels, aluminums); Brittle (grey iron,
glass); Visoelastic (plastics)?

3. Geometry - Constant section or stress raisers?\

4. Temperature - Upper and lower limits, average

5. Environmental factors - salt water, plating, corrosives, exposure to
weather, sunlight

6. Geometry - range of thickness due to manufacturing or wear and tear
in critical areas?

7. Standards - is this item covered by existing standards, for example
Mil-specs?

 

Factor of Safety == strength of material (typically yield; sometimes
ultimate) divided by the maximum anticipated stress at any point on the
part. And yes, you do just list whatever the FOS reported by the software
is in your report.

If you have a stress concentration, then that relatively high stress will be
the one used to calculate the FOS. A large part of the reason for doing FEA
analysis is that it allows you to identify and eliminate concentrations

In the optimum case, you add/remove material or change geometry so that the
stress is the same everywhere in the part. This optimum condition, however,
is largely a fantasy. You almost always have some variation in stresses;
your job as an engineer is to minimize it

 

That is a very simplified approach that may overbuild or underbuild the
part. In some cases you will not use the maximum stress reported by the
FEA program. In other cases you will not use the stress at all in
determining the factor of safety. The questions have to be answered to
know the nature of the load and the material.

 

Repeated Loading

Steels and Aluminums

Mostly constant

Factory Atmosphere...

can vary depending on the part.

 

For stress concentrations, linearize stress in the thickness... if you are
not so far from a shell model (L/h>5, R/h>5) see ASME, CODAP (FR)...
use average stress in elements instead of node stresses
meshing has to be larger than thickness for shells
for composite, around a hole, an edge, don't mesh under 4 mm (see average
stress criteria)
A ratio 1 to 1000 between element and max length is a thumb rule
See KI, KII, GI, GII
..... in fact we don't know how to well analyse the stress concentrations! a
standard would have to be done......

 

agreed-- but it's a simplified approach that will (generally) keep you out
of trouble if you're not fully grounded in the subject of FEA--which the OP
clearly is not

 

Well, I went off and had a look at his company's website to see what
they were making. Modular tooling. I suspect that there are two types
of failure mode that he has to watch out for. First will be loading of
a nature that it strains the structure to the breaking point. It must
withstand those loads without failing. For that purpose average stress
on a section, not peak stress in a stress raiser will determine whether
the structure fails. In this case Red is not necessarily bad unless it
covers a large area.

In the second case the tooling may be subjected to repetive loads.
Given the nature of the product the direction and magnitudes of the
loads may not be easy to predict. In this case he may need to make an
influence diagram to see which combination of loads causes the worst
case stresses in his part. There may also be cyclic loads superimposed
on high static loads in clamping hardware. For cyclic loads the stress
concentrations will be the problem. But the allowable stress will not
be yield, it will depend on the magnitude of the stress and the number
of cycles.

Perhaps more important than the FEA at this point is that he knows his
material well. Heat treated properties, fatigue curves and static
stress strain curves.

 

Thanks for your input! This group is always helpful.

 

One of my teachers liked to speak about "factor of ignorance".

It depends on your knowledge of the real conditions. In aerospace they use a
factor close to one ...

 

Hey P.

I really appreciate your help on this. You wouldn't by chance have any
really good examples
of parts/assemblies setup with realistic restraints, loads, surface
contacts etc... that
you can correlate to real world testing? I would like to contact you
through e-mail but
I can't find your full e-mail since they block it now. I would really
like to show you a couple
simple pictures of what I'm trying to do. I want to be sure my setup
is accurate so I don't
have a garbarge-in garbage-out scenario. The Engineer who mainly works
on FEA
here believes that it's garbage out no matter what you put in (yes,
he's gone through training too).
I would really like to understand how others setup simple contraints
seen all the time in machine
design if at all possible. I really would like to try and change his
attitude towards it.
Thanks for all the help.

 

This question has been answered elsewhere on the forum. But I will
recap a bit.

Building Better Products with FEA by Vince Adams and Askenazi.
Roark and Young.

The first is specific to FEA and the second is specific to determining
whether a structure will fail or survive.

You can respond to me directly using Google and clicking on options.