[gschaider]

Those who think that groovyBC makes it possible to do stupid things, but think it could be worse, might be interested in avoiding swak4Foam (http://openfoamwiki.net/index.php/Contrib/swak4Foam). It has all the destructive power of this and funkySetFields and allows you to crash simulations, fill disk-space and use more memory than before in many different ways. But if you use it sensible it can be quite useful (but this is boring)

The basic idea behind swak4Foam was to factor out the expression-parsers from groovyBC and funkySetFields and put them into a separate library. That way they can be easily incorporated into other libraries and utilities. As an additional bonus expressions for calculations on cell/face-sets and cell/face-zones are present. Expressions can access values on other entities (patches, zones, etc) the way "old" groovyBC can access other patches

Apart from the core-library the bundle contains:
- the new groovyBC
- the new funkySetFields
- functionObjects for arbitrary calculations (as long as there is a field it can be used) on patches and other stuff
- a new utility to statically set boundary fields to non-uniform values
- and some other little things

It was tested on 1.7, but it compiles on 1.6, too.

Please note that it is very BETA. So if you want a stable version of funkySetFields or groovyBC get the 1.6-versions. They compile on 1.7 without problems. There will never be a separate version of these two (groovy and funky) for 1.7

All documentation is currently in the Wiki-page and the README. Further hints can be found by looking at the examples (and of course the sources).

I have some additional stuff planed, but the basic syntax of the expressions should not change anymore. The API probably will change a bit as I plan to clean it up (but this only affects those who use the core-library for their own stuff)

BTW: swak4Foam stands for "SWiss Army Knife For FOAM". I know this is a bit immodest but I couldn't think of any better name ("expressionParsersForFoam": accurate but boring). And the original name was "Project MacGuyver". But I think that is trademarked.

Have fun with it and report bugs at http://sourceforge.net/apps/mantisbt...e_status_id=90

Bernhard

[alberto]

As usual, thank you!

[robingilbert]

Bernhard,

Thank you so much for conducting the marriage . I was looking for a tool to calculate flow rate through a faceZone.

[Canesin]

Bernhard,

Thanks for all the contributions,

I was looking for an way to insert an equation to eval a variable cp in my code.. I don't want to hardcode the equation, it will be nice to have it written in an dictionary and them parse and eval it. Can I use the swak4Foam parser to do it ?? How ??

In my code cp is very sensible to the magnetic field and temperature.. so I need to include the evaluation of it rigth after the temperature equation solving and do a loop in it, so I have to calculate the volScalarField some times in it...

while ()
	{
		cpOld = cp;
		for (int nonOrth=0; nonOrth<=nNonOrthCorr; nonOrth++)
		    {
		        tmp<fvScalarMatrix> TEqn
		        (
		            fvm::ddt(rho*cp, T)
		          - fvm::laplacian(K, T)
		        );
		        TEqn().relax();
		        TEqn().solve();
		        #include cpEqn.H
		    }
		if ((sum(cpOld)-sum(cp))/sum(cp) < 1e-6)
		{
			delete cpOld;
			break;
		}
	}

So in #include cpEqn.H I should eval the cp from equation as like an normal volScalarField in funkySetFields, so I want to read the dictionary eval the equation (using T, temperature, and H, magnetic field, as variables).

I believe that this is possible using swak4Foam parser, but how is it ?? I will be very glad if you can help me..

Thanks,
Fábio C. Canesin

[gschaider]

Originally Posted by Canesin

Bernhard,

Thanks for all the contributions,

I was looking for an way to insert an equation to eval a variable cp in my code.. I don't want to hardcode the equation, it will be nice to have it written in an dictionary and them parse and eval it. Can I use the swak4Foam parser to do it ?? How ??

In my code cp is very sensible to the magnetic field and temperature.. so I need to include the evaluation of it rigth after the temperature equation solving and do a loop in it, so I have to calculate the volScalarField some times in it...

while ()
    {
        cpOld = cp;
        for (int nonOrth=0; nonOrth<=nNonOrthCorr; nonOrth++)
            {
                tmp<fvScalarMatrix> TEqn
                (
                    fvm::ddt(rho*cp, T)
                  - fvm::laplacian(K, T)
                );
                TEqn().relax();
                TEqn().solve();
                #include cpEqn.H
            }
        if ((sum(cpOld)-sum(cp))/sum(cp) < 1e-6)
        {
            delete cpOld;
            break;
        }
    }

So in #include cpEqn.H I should eval the cp from equation as like an normal volScalarField in funkySetFields, so I want to read the dictionary eval the equation (using T, temperature, and H, magnetic field, as variables).

I believe that this is possible using swak4Foam parser, but how is it ?? I will be very glad if you can help me..

Thanks,
Fábio C. Canesin

so you want to have cp(T,H) and you want to read the function at runtime (if the function is already known hardcoding it into cpEqn.H is definitly the better solution), right?

Have a look at the Examples-Folder: there is a folder called "InterFoamWithSources". The class expressionSource that is used there might be a open to abuse. Basically an expressionSource-object cpExpr is evaluated every time it is called with cpExpr() and it returns a temporay field. So it might even be possible to insert it into the equation like this ddt(rho*cpExpr(),T) (but be aware that it is evaluated every time it is called so from a performance point of view it's better to evaluate it and assign it once)

That example is the only time I have used that thing yet, so you might experience problems

Bernhard

PS: if this is what you're looking for and if it works for you it would be nice if you added a short section on it to the swak-Wikipage

[Canesin]

Originally Posted by gschaider

so you want to have cp(T,H) and you want to read the function at runtime (if the function is already known hardcoding it into cpEqn.H is definitly the better solution), right?

Have a look at the Examples-Folder: there is a folder called "InterFoamWithSources". The class expressionSource that is used there might be a open to abuse. Basically an expressionSource-object cpExpr is evaluated every time it is called with cpExpr() and it returns a temporay field. So it might even be possible to insert it into the equation like this ddt(rho*cpExpr(),T) (but be aware that it is evaluated every time it is called so from a performance point of view it's better to evaluate it and assign it once)

That example is the only time I have used that thing yet, so you might experience problems

Bernhard

PS: if this is what you're looking for and if it works for you it would be nice if you added a short section on it to the swak-Wikipage

Hi Bernhard,

It looks like to be it.. I will adapt and run it as soon as I can. To justify the need for read at run time: my solver is multi-region so every solid region has an cp(T,H) expression. And also I want that the solver is generic enough to run in multiple cases...
I will add the cpExp() after the temperature non-orthogonal loop since evaluating it many times with a bad temperature field looks to be an bad solution physically and in performance terms.

As soon as I get it running I will report here and add in the swak wikipage ..

Best regards,

Fabio C. Canesin

[podallaire]

Hi Bernhard,

thanks for this great contribution !

I'm trying to compile with gcc 4.3 but have some issues with swakExpressionFunctionObject.C / can you confirm that gcc-4.4 or > is needed ?

Regards,

PO

[gschaider]

Originally Posted by podallaire

Hi Bernhard,

thanks for this great contribution !

I'm trying to compile with gcc 4.3 but have some issues with swakExpressionFunctionObject.C / can you confirm that gcc-4.4 or > is needed ?

No. I compile it on a regular basis with 4.3.3 without problems. Just for testing I tried to compile it with 4.1.2 and I get a weird error: it doesn't know dynamicCast. Is that the problem you're having?

Bernhard

[xisto]

Hi Bernhard!
Once more a great tool. Thank you!

I'm trying to model an azimuthal magnetic field in a tube.

And I want set a non uniform intenal field whith the x and y components of the magnetic field, here are the expressions:

Bx=-0.0012/sqrt(x^2+y^20)*sen(arctan(y/x));

By=0.0012/sqrt(x^2+y^20)*cos(arctan(y/x));

Bz=0.

This could be done with funkySetFields?

At the same time I want to impose this components as a non-uniform boundary conditions at the inlet.

For this I can use the groovyBc?

Thanks again for your great work!

Carlos

[gschaider]

Originally Posted by xisto

Hi Bernhard!
Once more a great tool. Thank you!

I'm trying to model an azimuthal magnetic field in a tube.

And I want set a non uniform intenal field whith the x and y components of the magnetic field, here are the expressions:

Bx=-0.0012/sqrt(x^2+y^20)*sen(arctan(y/x));

By=0.0012/sqrt(x^2+y^20)*cos(arctan(y/x));

Bz=0.

This could be done with funkySetFields?

yep.

"vector(-0.0012/sqrt(x^2+y^20)*sen(arctan(y/x)),0.0012/sqrt(x^2+y^20)*cos(arctan(y/x)),0)"

Originally Posted by xisto

At the same time I want to impose this components as a non-uniform boundary conditions at the inlet.

For this I can use the groovyBc?

Thanks again for your great work!

Carlos

Either that, or if it is fixed in time you can use the -keepPatches/-valuePatches options of funky. Or if you are adventurous you can use the new funkySetBoundaryFields-utility

[xisto]

I already use both.

funkySetFields and funkySetBoundaryFilelds

And it works for me.

Excellent tool.

Thanks.

Carlos

[podallaire]

Originally Posted by gschaider

No. I compile it on a regular basis with 4.3.3 without problems. Just for testing I tried to compile it with 4.1.2 and I get a weird error: it doesn't know dynamicCast. Is that the problem you're having?

Bernhard

Hi,

here is what i'm getting :


SOURCE=general/swakExpressionFunctionObject.C ; g++ -m64 -Dlinux64 -DWM_DP -Wall -Wextra -Wno-unused-parameter -Wold-style-cast -Wnon-virtual-dtor -O3 -DNoRepository -ftemplate-depth-40 -I/data-cfd01/projects/OpenFOAM/pod-1.7.x/Libraries/simpleFunctionObjects/lnInclude -I../swak4FoamParsers/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/finiteVolume/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/meshTools/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/sampling/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/triSurface/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/lagrangian/basic/lnInclude -IlnInclude -I. -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/OpenFOAM/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/OSspecific/POSIX/lnInclude -fPIC -c $SOURCE -o Make/linux64GccDPOpt/swakExpressionFunctionObject.o
In file included from general/swakExpressionFunctionObject.C:28:
general/swakExpressionFunctionObject.H:40:36: error: timelineFunctionObject.H: No such file or directory
In file included from general/swakExpressionFunctionObject.C:28:
general/swakExpressionFunctionObject.H:56: error: expected class-name before ‘{’ token
general/swakExpressionFunctionObject.H:56: warning: ‘class Foam::swakExpressionFunctionObject’ has virtual functions and accessible non-virtual destructor
In file included from general/swakExpressionFunctionObject.H:115,
from general/swakExpressionFunctionObject.C:28:
general/swakExpressionFunctionObjectTemplates.C: In member function ‘void Foam::swakExpressionFunctionObject::writeData(Foam ::CommonValueExpressionDriver&)’:
general/swakExpressionFunctionObjectTemplates.C:68: error: there are no arguments to ‘verbose’ that depend on a template parameter, so a declaration of ‘verbose’ must be available
general/swakExpressionFunctionObjectTemplates.C:68: error: (if you use ‘-fpermissive’, G++ will accept your code, but allowing the use of an undeclared name is deprecated)
general/swakExpressionFunctionObjectTemplates.C:76: error: ‘OFstream’ was not declared in this scope
general/swakExpressionFunctionObjectTemplates.C:76: error: ‘o’ was not declared in this scope
general/swakExpressionFunctionObjectTemplates.C:76: error: ‘filePtrs_’ was not declared in this scope
general/swakExpressionFunctionObjectTemplates.C:76: error: no matching function for call to ‘name()’
...

For some reason, file timelineFunctionObject.H is not available.

Thanks

PO

[gschaider]

Originally Posted by podallaire

Hi,

here is what i'm getting :


SOURCE=general/swakExpressionFunctionObject.C ; g++ -m64 -Dlinux64 -DWM_DP -Wall -Wextra -Wno-unused-parameter -Wold-style-cast -Wnon-virtual-dtor -O3 -DNoRepository -ftemplate-depth-40 -I/data-cfd01/projects/OpenFOAM/pod-1.7.x/Libraries/simpleFunctionObjects/lnInclude -I../swak4FoamParsers/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/finiteVolume/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/meshTools/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/sampling/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/triSurface/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/lagrangian/basic/lnInclude -IlnInclude -I. -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/OpenFOAM/lnInclude -I/data-cfd01/software/OpenFOAM/OpenFOAM-1.7.x/src/OSspecific/POSIX/lnInclude -fPIC -c $SOURCE -o Make/linux64GccDPOpt/swakExpressionFunctionObject.o
In file included from general/swakExpressionFunctionObject.C:28:
general/swakExpressionFunctionObject.H:40:36: error: timelineFunctionObject.H: No such file or directory

For some reason, file timelineFunctionObject.H is not available.

Thanks

PO

Because this library needs the simpleFunctionObjects (see Requirement in the README). That's why it is separate from the other functionObjects. It assumes that it finds the sources at $WM_PROJECT_USER_DIR/Libraries. In your case that would be (see your output)

/data-cfd01/projects/OpenFOAM/pod-1.7.x/Libraries/simpleFunctionObjects/

Bernhard

[podallaire]

Originally Posted by gschaider

Because this library needs the simpleFunctionObjects (see Requirement in the README). That's why it is separate from the other functionObjects. It assumes that it finds the sources at $WM_PROJECT_USER_DIR/Libraries. In your case that would be (see your output)

/data-cfd01/projects/OpenFOAM/pod-1.7.x/Libraries/simpleFunctionObjects/

Bernhard

Oops, sorry about that / missed this info on the website.

[alfa_8C]

Hello Bernhard,

I need to create a source and a sink in a model using reacting foam, but with reactions switched off. I have a region acting as source for the components CO2, H2O and CO, and a region acting as a sink for O2.

1.Question: is this pricipally realizable with swak4Foam?

if so

2. As I didn't find any example for swakSourceFields in the net, could you post an example for the synthax to creating a source/sink?

Thanks a lot in advance,
Tony

[gschaider]

Originally Posted by alfa_8C

Hello Bernhard,

I need to create a source and a sink in a model using reacting foam, but with reactions switched off. I have a region acting as source for the components CO2, H2O and CO, and a region acting as a sink for O2.

1.Question: is this pricipally realizable with swak4Foam?

if so

Yes it is. But ... see below

Originally Posted by alfa_8C

2. As I didn't find any example for swakSourceFields in the net, could you post an example for the synthax to creating a source/sink?

There is an example (modified solver and case):
https://openfoam-extend.svn.sourcefo...amWithSources/
in fact it is described in the README (https://openfoam-extend.svn.sourcefo...ak4Foam/README)

Anyway. The problem is in your case that the fields for which the expressions should be applied are only known at runtime so a single source-field won't do, but you'll need a Map of sourceFields where the appropriate one is chosen according to the name of the currently used field (nothing impossible, just an interesting programming problem). Anyway. You'll have to modify the solver anyway.

Bernhard

[eelcovv]

Bernard,

Thanks a lot, I have just downloaded and installed the new groovyBC, funkySetField and the swap4Foam part, and already encountered all kind of very handy new features. Brilliant!

Well, let me take to opportunity to ask two small questions I of features I was looking for for a while. Perhaps you (or anybody) knows a quick answer.

1: is it possible to make an 'if-then-elseif-then-else-' construction with the ternary expression? In gnuplot (also based on c-ternaries) it is possible for instance to do
f(x)= x<0 ? 0 : x>=0 && x<1 ? 2 : 1
but this doesn't seem to work in groovyBC. No I solve it by splitting it up:
f1(x) = x<0?0:0
f2(x) = x>=0 && x<1 ? 2 : 0
f3(x) = x>=1 ? 1:0
f(x)=f1(x)+f2(x)+f3(x)

it works, but expressions tend to become rather complex. Well, perhaps there is a way to do it easier

2: Is there a way to convert cartesian coordinates into polar coordinates of a vector easily ? Or should I just write down the conversion expressions based on x,y, and z ?

Well, again. Thanks! With groovyBC, openFOAM life definitely becomes a lot easier!

Regards,

Eelco

[gschaider]

Originally Posted by eelcovv

Bernard,

Thanks a lot, I have just downloaded and installed the new groovyBC, funkySetField and the swap4Foam part, and already encountered all kind of very handy new features. Brilliant!

Thank you

Originally Posted by eelcovv

Well, let me take to opportunity to ask two small questions I of features I was looking for for a while. Perhaps you (or anybody) knows a quick answer.

1: is it possible to make an 'if-then-elseif-then-else-' construction with the ternary expression? In gnuplot (also based on c-ternaries) it is possible for instance to do
f(x)= x<0 ? 0 : x>=0 && x<1 ? 2 : 1
but this doesn't seem to work in groovyBC. No I solve it by splitting it up:
f1(x) = x<0?0:0
f2(x) = x>=0 && x<1 ? 2 : 0
f3(x) = x>=1 ? 1:0
f(x)=f1(x)+f2(x)+f3(x)

it works, but expressions tend to become rather complex. Well, perhaps there is a way to do it easier

If

x<0 ? 0 : x>=0 && x<1 ? 2 : 1

doesn't behave the way it would in C++ I would consider it a bug (havn'T checked).
Anyway. I'm notoriously bad with remembering operator precedence and therefor believe "brackets are our friend":

x<0 ? 0 : (x>=0 && x<1 ? 2 : 1)

That should behave the way you want it. Presonally I'm a coward and would write it like this

x<0 ? 0 : ((x>=0 && x<1) ? 2 : 1)

(I think it is easier to read, too)

Originally Posted by eelcovv

2: Is there a way to convert cartesian coordinates into polar coordinates of a vector easily ? Or should I just write down the conversion expressions based on x,y, and z ?

I'm sorry. You'll have to do it that way. If I introduce polar coordinates somebody else will want cylindrical coordinates and somebody else will want this with x being "up" and then the parser becomes hard to maintain.

Not sure whether funkySetFields supports "variables" yet (it was on the TODO-list once) but that would make these expressions more readable

Bernhard

[eelcovv]

Hi Bernard,

The way you are suggesting works! Using the nesting brackets around each 'then statement' is the key. Thanks!

Regards

Eelco

[alfa_8C]

Hello Bernhard,

I am currently trying to setup a case using groovyBC for a heatflux BC. The heatflux is adressed to a surface. The area of the surface is fixed but apperently far too small for the given heat release rate - the temperatures close to the surface are exorbitant high. In CFX I would define a subdomain with a volume source term. Is this with your new tool possible?

Best, Tony