Two temperature model for ultrashort ... Log Out | Topics | Search
Moderators | Register | Edit Profile

FlexPDE User's Forum » User Postings » Two temperature model for ultrashort laser ablation « Previous Next »

Author Message
Top of pagePrevious messageNext messageBottom of page Link to this message

douglas
Member
Username: douglas

Post Number: 4
Registered: 08-2010
Posted on Tuesday, August 31, 2010 - 12:55 am:   

I composed two temperature model(TTM) for ultrashort laser ablation.
However, I couldn't solve the problem as followed.
- Invalid floating-point operation
- Folating-point Stack Check

Initially, I thought that student version have a limitation for solve TTM problem.
Therefore I bought professional version.
But I could't solve the problem.

I am seriouly worried about find a solution to error message and TTM.

Could you give me some advice?

Sincerely yours
Dongsig, Shin
application/octet-stream
100831_Q ttm_Journal of Mechanical Science and Technology 21(2007) 1847.pde (4.2 k)
Top of pagePrevious messageNext messageBottom of page Link to this message

rgnelson
Moderator
Username: rgnelson

Post Number: 1400
Registered: 06-2003
Posted on Tuesday, August 31, 2010 - 01:29 pm:   

Clearly you have some terms or parameters wrong in your equation system. Some of your variables try to go instantaneously to infinity.

Plot all your variables and coefficients and look for things that are in error. Use SELECT BUSYMON to force monitors to plot at every trial step, since the first timestep never completes.

See the attached script for an example that does this.

Alternatively, you can simplify the system until you can get one to work, then add the complexity a little at a time until you find the part in error.

application/octet-stream
Q_ttm1.pde (4.6 k)
Top of pagePrevious messageNext messageBottom of page Link to this message

douglas
Member
Username: douglas

Post Number: 5
Registered: 08-2010
Posted on Wednesday, September 01, 2010 - 12:55 am:   

Dear rgnelson

Thank you for sincere advice.
I am really appreciate it.

I changed two parameters as followed.
- Ce=3*Nc*kb ==> Ce=10+3*Nc*kb: Initially, too low value of "Ce" make instantaneouly to infinity
- tau_e=0.5*(1+Nc/(2e27)) ==> tau_e=0.5e-12*(1+Nc/(2e27)) : It was unit error.

Then, I think it might solve many infinity errors.
However, there is another porboem. That is, we need too much time to solve the simulation. (I can speculate it takes several days for finishing the simulation)

Could you recommend for my code to improve simulation speed.


application/octet-stream
Q_ttm1-21095_revision_100901.pde (4.7 k)
Top of pagePrevious messageNext messageBottom of page Link to this message

rgnelson
Moderator
Username: rgnelson

Post Number: 1403
Registered: 06-2003
Posted on Wednesday, September 01, 2010 - 01:51 pm:   

You removed the diffusion term from the Nc equation, so it can do nothing but grow without bound. It needs some way to dispose of the sourced value.
Top of pagePrevious messageNext messageBottom of page Link to this message

douglas
Member
Username: douglas

Post Number: 6
Registered: 08-2010
Posted on Wednesday, September 01, 2010 - 05:28 pm:   

I changed original Nc equation as followed

Nc: dt(Nc)=div(Do*grad(Nc))-gamma*Nc^(3)+alpha*source/(ha*omega)+Nc*(3.6e-10)*exp(-1.5*Eg/(kb*TE))

Howvever, there is weird error such as "Argument of Exp is too large(816.38) Maximum allowed is 690.776"

I don't think arguement of Exp is below 10.
Could you tell me what is problem?

application/octet-stream
Q_ttm1-21095_revision_100902.pde (5.2 k)
Top of pagePrevious messageNext messageBottom of page Link to this message

mgnelson
Moderator
Username: mgnelson

Post Number: 207
Registered: 07-2007
Posted on Friday, September 03, 2010 - 07:06 pm:   

The error report means that the argument of the EXP operator is so large that calculation of this expression would result in a number too large to represent with the computer. (It could also be in the calculation of the derivative of this expression.)

If I run the script with the Nc equation containing only the DIV and source terms,
[ i.e. Nc: dt(Nc)=div(Do*grad(Nc))+alpha*source/(ha*omega) ]
the problem will run, but the value of Nc at the boundary just continues to rise. This suggests to me that something in your system is not allowing Nc to diffuse. Maybe Do is very small?

As suggested earlier, you should start with a simpler model. Get that to work and then add complexity a little at a time.
Top of pagePrevious messageNext messageBottom of page Link to this message

douglas
Member
Username: douglas

Post Number: 7
Registered: 08-2010
Posted on Saturday, September 04, 2010 - 06:01 pm:   

I can see that problem was generated by Nc term using more simple model as attached.

In case of using fixed Nc valu like 1e23, this code was operated very well.
But, DIV and sourse terms of Nc can make sudden increase of TE.

However, according to many references, the Nc term is absolutely correct.
But I realized that I should find alternative Nc equation for eliminate the problems.
(Actually, even if I use fixed Nc value, the simulated temperature was coincident with experimental rusult because some references utilized fixed Nc value. ^^)

Thank you for sincere advice for me

Sincerly yours
Dongsig, Shin

application/octet-stream
No-0-0p-1DTTM-5pulses-Nc.pde (7.0 k)

Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action:

Topics | Last Day | Last Week | Tree View | Search | Help/Instructions | Program Credits Administration