Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.
How to find an average stress?
Posted Jan 28, 2010, 10:08 a.m. EST 1 Reply
Please login with a confirmed email address before reporting spam
Hello!
I would like to find the average stress of a perforated circular membrane, clamped at the edges. I'm using a 3D static simulation, also including an initial stress on the membrane.
Thanks a lot!
Andres
I would like to find the average stress of a perforated circular membrane, clamped at the edges. I'm using a 3D static simulation, also including an initial stress on the membrane.
Thanks a lot!
Andres
1 Reply Last Post Jan 29, 2010, 1:47 a.m. EST
Please login with a confirmed email address before reporting spam
Posted:
1 decade ago
Hi
As I understand you have two separate issues, one "simple" how to calculate an average, the other more complex: which stress to use ?
For the selection of the desired stress (Mises, ...) I leave it to you as this is very model dependent, but for the calculation of an average, its rather simple with the Integration Coupling Variables:
You define your region (line, surface or volume), you define a variable
IntStress = int(my_desired_Comsol_stress_variable,dx,dy,(dz)),
then you get the total area or volume by integrating "1" over the same geometrical entity, and
finally outside of the integration variables you take the ratio to get the average value.
A few things to remember:
the integration is a sum so one must take care that "the sum of a product is not necessarily the product of the individual sums" and the implicit sums (integrations over elements) of COMSOL are not always stricking appearant, so its easy to oversee them.
Then for stress, as often one have non-physical stress concetrations due to meshing and some spiky geometry can easily have unwanted side effects. Therefore, I often evaluate the standard deviation, when I look for an average, this is obtained by integrating the square of your variable, and dividing by the respective area/volume. But you can also decide to limit the area to a given region inside some stress limits, by multiplying the integrand expression by y bolean expression on the maximum stress, or just use the"max" function, the area/volume you use should also then be adapted, for me these are the usual "Engineering Common Sense" you must apply to any model results of the real world.
Good luck
Ivar
As I understand you have two separate issues, one "simple" how to calculate an average, the other more complex: which stress to use ?
For the selection of the desired stress (Mises, ...) I leave it to you as this is very model dependent, but for the calculation of an average, its rather simple with the Integration Coupling Variables:
You define your region (line, surface or volume), you define a variable
IntStress = int(my_desired_Comsol_stress_variable,dx,dy,(dz)),
then you get the total area or volume by integrating "1" over the same geometrical entity, and
finally outside of the integration variables you take the ratio to get the average value.
A few things to remember:
the integration is a sum so one must take care that "the sum of a product is not necessarily the product of the individual sums" and the implicit sums (integrations over elements) of COMSOL are not always stricking appearant, so its easy to oversee them.
Then for stress, as often one have non-physical stress concetrations due to meshing and some spiky geometry can easily have unwanted side effects. Therefore, I often evaluate the standard deviation, when I look for an average, this is obtained by integrating the square of your variable, and dividing by the respective area/volume. But you can also decide to limit the area to a given region inside some stress limits, by multiplying the integrand expression by y bolean expression on the maximum stress, or just use the"max" function, the area/volume you use should also then be adapted, for me these are the usual "Engineering Common Sense" you must apply to any model results of the real world.
Good luck
Ivar