how to change a volScalarField in bc?
 

I need to add a source term in the time loop, which is calculated in one of my boundary conditions.

First I defined the following field in createFields.H:
It is set to zero by "Hclip/dim_s*0" in the beginning and should be changed/updated in a specific boundary condition.

But how can I change the values of the field in the bc???

I tried the following:
But "sourceH" is still zero...

This type of issue has come up a few times before. The problem is access: the boundary condition only has const access to anything it looks up through the objectRegistry, and private data it creates cannot be accessed at the top level because the solver uses fvPatchField as an interface.

Does your solver need to change it? If not, the solution is fairly simple. Instead of creating the source term at the top level, give it to your custom bc. Then use objectRegistry::lookupObject to find it in your solver.

Thanks David for your explanations!
The solver doesn't have change the scalarField, but just adding it as a source term.
To make sure I got you right:
Did you mean to calculate a scalarField sourceH in my boundary condition and the call it in the solver by something like:
and add Hquelle as the source term?

I'm not really sure about this or in general how to directly call the scalarField from the bc in the solver....

Very close. Instead of objectRegistry::lookupObject... use the name of the objectRegistry sourceH is registered to... you'll probably put it in the mesh, just like nearly everything else:

And you have to make sure that your boundary condition has already created this sourceH object before the solver tries looking it up. So, make sourceH a member of your custom boundary condition, and initialize (to zero) it in all of the constructors.

creating Field
 

I have still problems with creating the volScalarField in my BC:
I tried it like this:

in the header file:
which gives me an error, that the field sourceH has an incomplete type

in the .C file:
giving me the following errors:
I guess I have to include some files. But if I include createTime.H and createMesh.H it still wouldn't work.

"does not have any field named sourceH_"

This is your first problem. It looks like your header file has problems... but you have only quoted a portion of it.

It looks like you want to define what objects it inherits, but left them out. Use either:

or
... or did you misquote your code?

misquote
 

Sorry. You're right! I left something out in my quotation:


It worked fine with all the other variables, like HName_, that's why I don't understand that it has a problem with the volScalarField.

There is probably an error where it complains in the header file that it doesn't know what volScalarField means. This could be an #include problem, or circular dependency.

I tried to to solve this problem by adding
as a second baseClass, like David mentioned. But all I got was an enormous amount of errors....

any other solutions to my initial problem? or is there still a possibility to do it this way?????

thanks in advance!!!!

What are your errors?

"abflussWehrDS.H:85: error: field ‘sourceH_’ has incomplete type"

This is an #include problem. Either you didn't include the correct files that gives you the full definition of volScalarField, or you have circular dependency among your include files. The first option is likely, as volScalarField needs at least two files - the file that gives you the typedef volScalarField, and the one with the full definition.

I strongly dislike these kinds of problems.

I tried it by including
and with:
But the problem is still not solved:
How can I find out which files are missing?

Dear Kathrin, dear David,

The problem of creating the source term / a variable in the boundary condition for use in the main solver, is that if you don't add it correctly to the database, it will be deleted out of it as soon as the fvPatchField is destroyed. This means, as soon as you exit the boundary condition code.

Anyways, the solution is quite simple. Let's say you want to store a scalarList in the database, for use anywhere (in the bc, the solver, etc) you would do:

Checking it out will be done in the usual way:
You could also use const scalarList& straight away as type, but the reason I use an IOList<scalar> is for the next step. Because once you've use your variable and updated it, you can now update it's value in the database (!!! This, although it theoretically is of const type, hence not modifiable :) !!!).
You would do this using the following (perfectly legal) C++ trick:
I hope this helps!

Kind regards,

Francois

Using const_cast is cheating!

"The first rule of the const_cast-club is: 'Don't talk about the const_cast!'"

He he he! :)
Cheating..? Perhaps.. But most importantly: It works! :)

Thanks!!
 

Thank you all for your help. The const_cast works:)

But I have one more question:
Is it also possible to do this for a volScalarField. I could only find IOField<scalar>. Or is it possible to convert a scalarField into a volScalarField, which would also be sufficient.

Well, actually, it has been a while since I implemented this hack, but if I recall properly, the store() function can store any pointer to an IOobject.

So, since your volScalarField is already an IOobject, what you are trying to do should be very simple to do:
And obviously you would do your const_cast with:
Kind regards,

Francois.

From

OpenFOAM-1.6-ext/src/finiteVolume/fields/fvPatchFields/derived/parabolicVelocity

comes:

This is called upcasting in C++: stripping the inherited object to the base class. All bcs are hidden *fields. No need for *_cast magic.

Just so you don't miss it , the key line is , this converts "this" (the class, in this case parabolicBlah...) to a vectorField.

T.