CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   average (

chiara March 20, 2008 12:19

Hi! I would like to know the mean value of the von mises stress calculated on a subdomain. How could I do? I don't know which is the expression for the NUMBER OF (MESH) ELEMENTS of a subdomain and the operation of SUMMATION. My idea is to calculate the sum of the von mises stresses of every element and divide this sum for the number of elements. Is there someone who can help me? Thanx a lot, chiara

rt March 20, 2008 15:18

Re: average
first, what is ur goal?

I assume u used FEM as solution, then what is your FEM space?

von mises stress is a point-wise measue and relating this measue to a sub-doamin is somehow meaninigless, so u should say avarage of von mises stress on a sub-doamin,

in this sense, u should do integration of von mises stress on desired sub-domain and scale its with volume/area of sub-doamin.

note that since this measue is nonlinear, element-wise sum is incorrect, global integration is requited,

so make a something like a macro element in the desird sub-domain and do formal quadrature base integration on its to compute desired quantity.

hope this helps.

chiara March 21, 2008 05:53

Re: average
Hi thanx to have replayed to me. May be my goal was not very clear. I try to explain it better. I applied a load to a subdomain, then I found the numerical solution for the displacement, the stress, etc. I suppose that the von mises stress were calculated for each element of my subdomain. I'm interested in understanding which is the mean value of the von mises stress in the subdomain...for istance if there wer only two elements, one with stress 2e6Pa, the other one with 3e6Pa, the mean value would be (2e6+3e6)/2. The problem is that usually you can write new expressions, but they are applied on each element of the subdomain. On the contrary I need an expression that uses the values of the different elements to calculate th average. I saw there is the possibility to do a temporal integration, but I'm not interested in it, I'm doing a static analysis. Could you be more specific with your suggestion? It's not long time that I'm using comsol and I'm not very practical...!Thank you a lot, chiara

rt March 21, 2008 06:25

Re: average
first i am not user of comsol, so do not have specific comment in this regard.

if u r insisted to do an element-wsie avaraging, u should do weghted avaraging, something like this

A_VMS = sum_over_elements (VMS_e * A_e) / sum_over_elements (A_e)

A_e is element area (2d) or volume (3d)

i do not know usefulness of such treatment !, maybe better is to compute avaraged stress on desired sub-domain and then compute VMS, due to nonlinearity this treatment is more feasible.

P.S.: VMS is usually used to study failure potential (control safety factors) and is quit point-wise, I just give an example, assume L-shaped domain which is loaded from its arm, then we have stress concentration at internal right-angle part which has a large VMS, this structure could fail from this position based on its VMS, while if we compute avarage VMS, we could be in safety margin

chiara March 21, 2008 06:40

Re: average
Thank you a lot, but my problem is doing it in COMSOL! I know how to do an average, my problem is that I don't know (I didn't find it on the user's guide) how to call in comsol the elements and the number of elements. I'm sorry for the misunderstanding, but I thought that this forum was only for comsol users, so I didn't specify I was using this simulation program. Bye, chiara

All times are GMT -4. The time now is 07:53.