How should I go about doing this parametric study? [COMSOL]

[tugni925]

I need to do a sweep study (parametric study) - to see how the plot (Fig.3) will change when I change the geometric parameters "a_1", "b" and "t". I first do a cell periodicity study to find effective material properties for a set of "a_1", "b" and "t". Based on this data I need to do a linear buckling analysis, which gives me the plot in Fig.3. I have done this succesfully for a given value of "a_1", "b" and "c" - but im not sure how to do this for multiple sets of "a_1", "b" and "c". (Besides doing it manually of course).


Fig.1 This study finds the elasticity matrix for varying geometry defined by "a_1", "b" and "c"

Fig 2: Want to do a sweep over these "k" values for sets of "a_1", "b" and "c".

Fig 3: This is a plot for a given value of "a_1", "b" and "t". Basically I want to see how this plot changes as "a_1", "b" and "t" changes.

Ps. I dont want this plot for all combinations of "a_1", "b" and "t", but for sets of "a_1", "b" and "t".

 

[FEA way]

Try using different type of sweep than the one currently selected (Specified combinations).

 

[tugni925]

I think I might not have explained well; my problem is I want to take all the data from cell periodicity study and run a linear buckling analysis for all this data. I know changing "Specified combinations" to "All combinations" will mix it all, but this is not what I want, I want the specified combinations. I guess my question is more along the lines of: "how can I get plot in Study 2 for all the data I got from my cell periodicity study".

 

[FEA way]

I think I get it now. It’s possible to create plots showing the variation of some solution variable (in this case critical load factor) when selected parameters are changed. To do this you have to select parametric sweep as a study used for the plot. Then it’s just a matter of some adjustments done to the plot.

 

[tugni925]

I am not quite sure how to select parametric sweep as a study - how would it look if we did this? In this model the Bandstructure is defined as 1D plot group, x-axis shows the k-values, and y-axis is the critical load factor. I did try to add "a_1" and "b" to the y-axis but that just gives me 3 straight line plots as seen to the right:

these are the options for the x-axis data:

do I maybe need to change plot type?

 

[FEA way]

If you want to plot critical load factor vs k for each given combination of parameters a_1, b and t then you should select parametric solution as a data set (it seems that you already did that) and leave only critical load factor in expression field for y axis. The rest should be done automatically by Comsol - you should get a plot with multiple curves corresponding to selected parameter combinations.