Re: [sparta-users] Surface temperature distribution

[Steve P. <sj...@gm...>]

ok - that's a different kind of model than I was imagining.

That would need to be a new kind of collision model,
say diffuse/temp that accumulates collision/energy info and stores
it with each element.  A conceptual problem is that
many processors can share ownership of an element
(the element overlaps many grid cells).  So how do you
sync the changing temperature contribution across all
those procs?

This is similar to an issue we are in the process of working
on for surface reaction models, where the element has to
store accumulated chemistry history.

So I would say that in the next couple months we should
have some new infrastructure in the code (for storing and
syncing per-element surface state) that would enable
the model you are asking for.

Steve

On Tue, Feb 4, 2020 at 11:34 AM Marc Schouler <Mar...@on...>
wrote:

> Indeed, I meant surface element.
>
> I would not need to read the temperature in a file, but rather give an
> initialization value (same for all elements).
>
> Then, the temperature would implicitly depend on time through the surface
> heat flux. The final purpose being to update Tw_i of the ith surface
> element as follows:
>
> Tw_i = (etot_i/a+b)^0.25
>
> Where etot_i is the heat flux to the ith element and the other terms are
> known scalar constants.
>
> Marc
>
> Le 04/02/2020 à 17:52, Steve Plimpton a écrit :
>
> >> the possibility to define the wall temperature as a surface variable
> with one value per cell.
>
> The Q is how the user defines the value per element.  I assume you mean
> a specific value for each surf element, not per grid cell.  Do you want to
> read them in from a file?  And this temperature is a constant, not a
> time-dependent variable?
>
> The simple answer is you can't do this now, but I think it should be
> possible.
> E.g. we could extend the format of the input surface file to include
> extra values, like a temp per surf.
>
> The idea of a variable (as SPARTA defines it) is that you could
> write a math function that uses the center point position to
> define a temp.  Which also allow for time-varying.
>
> Steve
>
> On Tue, Feb 4, 2020 at 9:22 AM Marc Schouler <Mar...@on...>
> wrote:
>
>> Dear Steve,
>>
>> Exactly, I am looking for a spatially dependent surface temperature.
>>
>> Since the convective flux is defined for every cell, the idea would be to
>> update Twall from a radiative thermal equilibrium (function of the
>> convective flux) every N iterations.  Ultimately, this should converge
>> toward a temperature distribution with a specific value for each surface
>> cell.
>>
>> I do not think this would require the center point of the triangle for
>> the temperature formula but just the possibility to define the wall
>> temperature as a surface variable with one value per cell.
>>
>> If you think I could implement that easily, I will take a deeper look to
>> the related source files.
>>
>> Thank you very much for your answer,
>>
>> Marc
>>
>>
>>
>> Le 04/02/2020 à 15:15, Steve Plimpton a écrit :
>>
>> The surf_collide diffuse model does allow you to specify
>> temperature as a time-dependent variable.  I think what
>> you are asking is whether it could be a spatially-dependent
>> variable?  Can you explain what kind of functionality you
>> want?  E.g. do you want to be able to use the center point
>> of the triangle in the formula for temperature?  Implementing
>> something like this could be done I think.  Note that you
>> can do something similar to this at a coarse level, by grouping
>> surface elements via the group command.  Then defining
>> one surf_collide diffuse model for each group of surfs.  E.g.
>> you could take a flat surface, divide it into 10 strips, one per group.
>> And assign 10 different temperatures, one per strip.
>>
>> Steve
>>
>>
>> On Mon, Feb 3, 2020 at 5:23 AM Marc Schouler <Mar...@on...>
>> wrote:
>>
>>> Hi everyone,
>>>
>>> I recently switched from the Oct18 to the Jan19 version of SPARTA and I
>>> noticed that many new functionalities were added to the surf_collide
>>> command. I am interested in computing and using a surface temperature
>>> distribution with the diffuse boundary condition but apparently the surf
>>> style is not implemented so it can not be done by defining a temperature as
>>> a surface variable. I was then wondering if there was any way to
>>> compute/converge toward a surface temperature distribution with SPARTA
>>> similarly to what can be done with CFD codes?
>>>
>>> Best regards,
>>>
>>> Marc
>>> _______________________________________________
>>> sparta-users mailing list
>>> spa...@li...
>>> https://lists.sourceforge.net/lists/listinfo/sparta-users
>>>
>>
>>
>