Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

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.

Data Resolution Problem

Please login with a confirmed email address before reporting spam

Hi,

When I plot my results using the 1D Plot Group, Point Graph, I get a jagged curve as opposed to a smooth one. (see attached file).

I have tried changing the time step of the model (making it smaller) and refining the mesh but I still get the same curve.

How can I obtain "smoother" results?


14 Replies Last Post 5 août 2012, 12:29 UTC−4

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13 avr. 2011, 08:33 UTC−4
-
-

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13 avr. 2011, 08:43 UTC−4
Hi

what do you have as time stepping, is it automatic ? or strict, or intermediate ?.

For pluses or periodic shapes you should NOT use automatic as this is tailored for diffusion or asymptotically trend solutions

--
Good luck
Ivar
Hi what do you have as time stepping, is it automatic ? or strict, or intermediate ?. For pluses or periodic shapes you should NOT use automatic as this is tailored for diffusion or asymptotically trend solutions -- Good luck Ivar

Jeff Hiller COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 13 avr. 2011, 09:05 UTC−4
It looks like you may not have updated your plot after running the simulation with the 1e-10 time steps.
It looks like you may not have updated your plot after running the simulation with the 1e-10 time steps.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2012, 00:52 UTC−5
I am having the same problem, and dont know what could possibly be the reason, ive definately redrawn and tried different settings for the 1D plot.
I am having the same problem, and dont know what could possibly be the reason, ive definately redrawn and tried different settings for the 1D plot.


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2012, 01:19 UTC−5
can you attach the model file .. i want to have a try at it ?
can you attach the model file .. i want to have a try at it ?

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2012, 03:30 UTC−5
Thanks,

and if you could also tell me why my source is oscilating so irregulaly, I would be very greatfull!
Thanks, and if you could also tell me why my source is oscilating so irregulaly, I would be very greatfull!


Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 3 févr. 2012, 08:26 UTC−5

Hi

there are several points I can comment:

Trick for the geometry rectangle, in V4 you can add a or several "Layers" and apply it on all boundaries, this reduces the 4 rectangles into a single rectangle call, far less to click and to take care of

Gold, your epslion seems strange, you need at least an "j" or "i" for the imaginary part, then check your sign convention, there are two around COMSOL uses the "-" one

Then check your doc if all these physics solve for sigma = 0, often its better to give al little conduction to air 1S/m or so, and gold is definitively not "0" for sigma

Your time solver, if you check the log file you see that COMSOl solves several steps, then it gets to the time to extract your demanded valaue = "out" then it solves a step again etc, but towards the end it has made a large step and your results out are all interpolations froma few solver steps hence strange behaviour.
You should use a more restricted solver stepping, in the Time Stepping tab of the Transielnt solver node change "free" to intermediate or strickt (Steps taken by the solver:)

And take some tome to read and re-read the doc, its worth it ;)

--
Good luck
Ivar
Hi there are several points I can comment: Trick for the geometry rectangle, in V4 you can add a or several "Layers" and apply it on all boundaries, this reduces the 4 rectangles into a single rectangle call, far less to click and to take care of Gold, your epslion seems strange, you need at least an "j" or "i" for the imaginary part, then check your sign convention, there are two around COMSOL uses the "-" one Then check your doc if all these physics solve for sigma = 0, often its better to give al little conduction to air 1S/m or so, and gold is definitively not "0" for sigma Your time solver, if you check the log file you see that COMSOl solves several steps, then it gets to the time to extract your demanded valaue = "out" then it solves a step again etc, but towards the end it has made a large step and your results out are all interpolations froma few solver steps hence strange behaviour. You should use a more restricted solver stepping, in the Time Stepping tab of the Transielnt solver node change "free" to intermediate or strickt (Steps taken by the solver:) And take some tome to read and re-read the doc, its worth it ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 6 févr. 2012, 00:25 UTC−5
Hi,
thanks for the hints, the layers trick is handy.
and I clearly screwed up my metal.

now do I understand the time stepping settings correctly in, that if I use the strict it will use strickt discrete timesteps and not as is now the case interpolate between vast time regions?


Ive been (re).reading many tutorials, but most I feel focus on very specific tasks while I am often still trying to figure out where top and bottom is. If you have any specific suggestions they would be welcome:)

kr
Hi, thanks for the hints, the layers trick is handy. and I clearly screwed up my metal. now do I understand the time stepping settings correctly in, that if I use the strict it will use strickt discrete timesteps and not as is now the case interpolate between vast time regions? Ive been (re).reading many tutorials, but most I feel focus on very specific tasks while I am often still trying to figure out where top and bottom is. If you have any specific suggestions they would be welcome:) kr

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 6 févr. 2012, 01:30 UTC−5
Hi

I agree it took me also quite some time to get used to the COMSOL approach, hadnt done physics in this way for several years, but gently the puzzle settles and yo get the full image. But it takes some exercices and hours reading (and rereading, I run through all the new doc at least once per year)

--
Good luck
Ivar
Hi I agree it took me also quite some time to get used to the COMSOL approach, hadnt done physics in this way for several years, but gently the puzzle settles and yo get the full image. But it takes some exercices and hours reading (and rereading, I run through all the new doc at least once per year) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 6 févr. 2012, 01:46 UTC−5
anyway it would appear your tips helped, it seems to be working now, so its very much appreciated :)
anyway it would appear your tips helped, it seems to be working now, so its very much appreciated :)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17 févr. 2012, 13:25 UTC−5
I am running a thermomechanical elasto-plastic model with the a focus on residual stress from a transient thermal cycle. I am having a similar "jaggedness" in the spatial domain when extracting stress distributions. The smoothness improves when I increase the number of elements (quadratic) along the line that I'm extracting data from, but even at high density, there is still oscillation in regions of fairly constant values, which is worse in a region of high gradient. I could always make a point graph to hide it, but I'd rather know why it does what it does.

I can't seem to upload a screenshot. Keep getting a "file extension error" message.
I am running a thermomechanical elasto-plastic model with the a focus on residual stress from a transient thermal cycle. I am having a similar "jaggedness" in the spatial domain when extracting stress distributions. The smoothness improves when I increase the number of elements (quadratic) along the line that I'm extracting data from, but even at high density, there is still oscillation in regions of fairly constant values, which is worse in a region of high gradient. I could always make a point graph to hide it, but I'd rather know why it does what it does. I can't seem to upload a screenshot. Keep getting a "file extension error" message.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17 févr. 2012, 14:41 UTC−5
Hi

you are probably getting some numerical oscillations. Thee can come from many resons, the plot smoothing functions (try to turn the plot to wireframe and turn off smoothing (main plot node settings). Then as you did increase the mesh density. You could also try increasing the shape function order. But there are also some cases that create singularitie, such as sharp corners, particularly in EM, but also Solid ...

There is also a good chapter in the doc on the numercal oscillations and the dissusion factors used, in certain physics, it's worth to take a look.

In RF one should not forget that you need at least 5-10 mesh elements per wavelength in each material, and do not forget that the index of refraction can be rather high in some materials, hence the wavelength accordingly shorter, hence this requires a denser mesh (its NOT ONLY the wavelength in vacuum that counts ;)

--
Good luck
Ivar
Hi you are probably getting some numerical oscillations. Thee can come from many resons, the plot smoothing functions (try to turn the plot to wireframe and turn off smoothing (main plot node settings). Then as you did increase the mesh density. You could also try increasing the shape function order. But there are also some cases that create singularitie, such as sharp corners, particularly in EM, but also Solid ... There is also a good chapter in the doc on the numercal oscillations and the dissusion factors used, in certain physics, it's worth to take a look. In RF one should not forget that you need at least 5-10 mesh elements per wavelength in each material, and do not forget that the index of refraction can be rather high in some materials, hence the wavelength accordingly shorter, hence this requires a denser mesh (its NOT ONLY the wavelength in vacuum that counts ;) -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 17 févr. 2012, 14:59 UTC−5
Thanks for the response. I'll look into it.
Thanks for the response. I'll look into it.

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 5 août 2012, 12:29 UTC−4
Hello all,

I am getting sharp peaks instaed of smooth curves .I followed some of your suggestions ,but i coudnt get the desired result.Could you please look into my model.It will be very helpfull for me .Thank you.
Hello all, I am getting sharp peaks instaed of smooth curves .I followed some of your suggestions ,but i coudnt get the desired result.Could you please look into my model.It will be very helpfull for me .Thank you.

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.