I tried to apply nonlinear temperature gradient using the beam interface of COMSOL, however, COMSOL always considers the temperatures of the top and bottom surfaces to calculate linear gradient over the member thickness, regardless of the defined nonlinear or multi-linear temperature gradient formula. Any ideas?
Hi Sallal. I don't have much experience with COMSOL, so please forgive me if this is off base but I can highlight some general issues with beam elements. Your problem may be in the way that the beam elements are implemented in COMSOL. Generally for CAE programs, beams are based off of analytical models and as such will not allow much variation along the cross section. Perhaps consider using isoparametric elements or if you feel comfortable with writing your own program then look at continuum based beam elements. These will both allow for you to apply the temperature gradient along the cross section. Hope this helps, sorry if it doesn't! Please let me know if you need me to elaborate on any of this more.
Thanks Dean,
Actually i'm not about writing my own program, this needs more skills and more time (for me). COMSOL introduces a simplified one-dimensional beam model to calculate the longitudinal stresses and deformations. Using this model, one can define the cross-section geometry and go on. However, the only applicable temperature gradient is the linear one between the top and bottom surfaces. In my case, the required is to apply various types of nonlinear temperature gradients.
Thanks again
Sorry I couldn't be of more help. Beams are always attractive because of there simplicity but it also means that they become limited in what they can be used for. My only advice would be to try different element types in COMSOL (maybe 3-D elements and model them like a beam?). Hopefully you can find a way to do it with the beam elements though. Good Luck!
Yes Dean,
Thats exactly what i did, i made my 3D model and have got the required results. However, the model was huge, because i have a concrete box-girder bridge, hence, the executed calculations were enormous and as a result, the solution time was too long (several days with a very good computer), which prevented me from making parametric studies within the time i have. Therefore, i decided to go back and try the 1D modeling again.
Thanks
- Badges
- Science topic