Re: [deal.II] Re-using outputdata from a previous analysis into a new one

[Jean-Paul Pelteret]

Hi,

That’s quite an interesting coupled problem that you’re solving!

If I understand correctly, you’re wanting your one problem to solve to completion, and you then want to save that state and load it in another, new simulation? If that’s the case then we have save/load functionality built into most (if not all) linear algebra vectors (see here in the documentation of the Vector class, for example). Of course, you’ll need to access that data with a matching DoFHandler (paired with the correct FE / FESystem), but all of those are also serialisable too. Although I don’t think that we have too many tutorials that demonstrate serialisation (I cannot think of one off the top of my head), we do have a collection of tests that can illustrate how its done.

Was that the information that you were looking for? 

Best,

Jean-Paul

On 30. May 2021, at 04:03, SAR <salra...@gmail.com> wrote:

Dear all

I am working on a study to investigate the role of elastic energy on hydride evolution in Zr claddings in NPP. 

Through dealii, I have developed a program that solves the coupled system system of equations :

1. Laplace equation (steady thermal conduction)

2. Mechanical equilibrium 

3. Cahn- Hilliard equation

4. Allen-cahn equation

The program runs well and no issues there. A typical simulation takes large number of time steps to solve and generate a micro-structure. 

However, I want to investigate how will the hydride phase will re-orient depending on a particular mechanical boundary load. 

Although that I can implement a time dependent boundary condition that will allow me to achieve this goal, I dont want to do that. 

I want to use the results from the last time step in my original analysis and use that as a starting point in a new analysis, that includes all of the system information from the previous analysis. 

Is there a way to do that through dealii?

Thank you

--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en
---
You received this message because you are subscribed to the Google Groups "deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dealii+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/1010a77e-c14e-4871-bacf-51e5dbf11cd7n%40googlegroups.com.

[SAR]

Sir, thank you for your reply, 

In my problem, I use a nucleation algorithm to place hydrides into the domain (using the Allen-Cahn equation). Since this function is of random nature, the exact structure will never not be reproduced in an another analysis. 

I want to generate a particular micro-structure and then, subject that one to multiple mechanical boundary conditions (each condition will be a separate analysis case). The equations, objects will all be be exact, I just want to start the new analysis case using the mesh, and solution vectors from the last problem.

I am very thankful for you time and also for the library in general. 

Thank you

[Wolfgang Bangerth]

On 6/1/21 12:54 AM, SAR wrote:
>
> In my problem, I use a nucleation algorithm to place hydrides into the domain
> (using the Allen-Cahn equation). Since this function is of random nature, the
> exact structure will never not be reproduced in an another analysis.
>
> I want to generate a particular micro-structure and then, subject that one to
> multiple mechanical boundary conditions (each condition will be a separate
> analysis case). The equations, objects will all be be exact, I just want to
> start the new analysis case using the mesh, and solution vectors from the last
> problem.

That sounds like you want to "serialize" the state of the program and the
"restart" multiple times from this state. Or maybe only checkpoint the mesh
and solution vectors, and the reload these in a different program that solves
something different with the solution of the first.

Regardless, I think that the keywords you are looking for are
"checkpoint/restart". If you search the archives of this forum, you will find
numerous mentions of this topic.

Best
W.


--
------------------------------------------------------------------------
Wolfgang Bangerth email: bang...@colostate.edu
www: http://www.math.colostate.edu/~bangerth/

[S A]

Sir, I am very thankful for your response It is what I am looking for. 

Moreover, I am deeply thankful for you and rest of the developers on this great library. 

Thank you. 

--
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en
---

You received this message because you are subscribed to a topic in the Google Groups "deal.II User Group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/dealii/v0HmuVVyALw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to dealii+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/603a65e8-3cca-1219-45e0-0628dde97e1d%40colostate.edu.