Author |
Message |
Tobias Roeder (tropper)
New member Username: tropper
Post Number: 1 Registered: 04-2008
| Posted on Saturday, April 19, 2008 - 07:47 am: | |
Hi all, I'm new to flexpde (FlexPDE 5.0.21). I want to simulate the melting of silicon by a laser pulse. The laser pulse is absorped from the material thus increasing the temperature. The temperature is transported by heat conduction through the material. My problem is that the resulting temperature profile is quite noisy. I have read about the "smoothing operators" but i'm not sure if i'm alloud to use the del2-operator since there is allready a div(grad()) term in the equation. It also takes longer to run the simulation. Are there other possibilities to smooth the temperature profile without increasing the calculation time by a lot? Thanks Tobias |
Robert G. Nelson (rgnelson)
Moderator Username: rgnelson
Post Number: 1106 Registered: 06-2003
| Posted on Saturday, April 19, 2008 - 01:57 pm: | |
I'm not sure what you intended with the expression uramp(Temp-DeltaTemp-Tmelt, Temp+DeltaTemp-Tmelt) in the source function, but this is the source of the noise. Here is a picture of the values for this expression at t=2e-8:
|
Tobias Roeder (tropper)
New member Username: tropper
Post Number: 2 Registered: 04-2008
| Posted on Sunday, April 20, 2008 - 07:46 am: | |
Dear Mr. Nelson, Thanks for looking at my problem. The uramp function is to simulate the reflectivity jump between liquid and solid silicon. The melting of the silicon surface should start at x0 and than increase in +/- x-direction. Therefore uramp (I used a step function first) should look like a pulse centered at x0 with increasing width during melting and decreasing width during cooling. If there are some spikes in the temperature distribution those spikes reaches the melting point earlier thus creating the noisy uramp. Do I get oszillations do to this term? Is there another way to include the increasing reflectivity of the liquid surface. Thanks a lot for your help Tobias |