[jiejie]

Originally Posted by su_junwei

Perhaps. you forgot to update gradU every time step. Just recalculate gradU every time step. add the following line to time loop
gradU=fvc::grad(U) ;

regards,Junwei

Hi JunWei

I updated gradU every time step as I have gradU in the data set for each time step. However, the there is no gradU stored in the lagrangian folder. As I have the position data and gradU field data, I want to use the particle position information to check the gradU stored in the lagrangain particle is correct.

[su_junwei]

Hi Jie
I cannot quite following you. Do you want to store the gradU at particle centers into the Lagrangian folder or gradU has been stored but the data was not correct.
If they was not stored, just add a new variable for your particle class and add this IO utility for it like other properties in particle class. The IO utility is implemented in the following function

1) constructor function like this

Foam::solidparticle::solidparticle
(
const Cloud<solidparticle>& cloud,
Istream& is,
bool readFields
)
2) readFields
3) writeFields
4) operator<<

If gradU has been written into lagrangian folder but not correct. I think they are not updated. Note that the volTensorField gradU you declare in your createFields.H is not the field of gradU in particle center. The first is the gradU in the patchcenters and cellCenters, whereas the second is the gradu in several discrete points.

regards, Junwei

[jiejie]

Hi JunWei

Sorry about the confusion. I tried to store the gradU at particle centers into the Lagrangian folder, but it only contains the fist particle.

I added the following lines in the constructor:

######################################
Foam::solidParticle::solidParticle
(
const Cloud<solidParticle>& cloud,
Istream& is,
bool readFields
)
:
Particle<solidParticle>(cloud, is, readFields)
{
if (readFields)
{
if (is.format() == IOstream::ASCII)
{
d_ = readScalar(is);
is >> U_;
//- ** velocity gradients
is >> gradU_;
}
else
{
is.read
(
reinterpret_cast<char*>(&d_),
//sizeof(d_) + sizeof(U_)

//- **velocity gradients
sizeof(d_) + sizeof(U_) + sizeof(gradU_)
);
}
}

// Check state of Istream
is.check("solidParticle::solidParticle(Istream&)") ;
}
#####################################

and I asked it to write out the gradU

####################################
void Foam::solidParticle::writeFields(const Cloud<solidParticle>& c)
{
Particle<solidParticle>::writeFields(c);

label np = c.size();

IOField<scalar> d(c.fieldIOobject("d", IOobject::NO_READ), np);
IOField<vector> U(c.fieldIOobject("U", IOobject::NO_READ), np);

//- **velocity gradients
IOField<tensor> gradU(c.fieldIOobject("gradU", IOobject::NO_READ), np);

label i = 0;
forAllConstIter(Cloud<solidParticle>, c, iter)
{
const solidParticle& p = iter();

d[i] = p.d_;
U[i] = p.U_;

//- **velocity gradients
gradU[i] = p.gradU_;

i++;
}

d.write();
U.write();

//- **velocity gradients
gradU.write();
}
####################################

in the lagrangain folder, gradU only contains:

FoamFile
{
version 2.0;
format ascii;
class tensorField;
location "101.3";
object gradU;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

1301{(0 0 0 0 0 0 0 0 0)}

// ************************************************** *********************** //

Instead of having all zeros in 1301{(0 0 0 0 0 0 0 0 0 0)}, it should have the gradU for all 1301 particles. I have no problem with U.

[jiejie]

Hi JunWei

I have attached my solidparticleIO.C here.

It looks like it counts the number of particles in the gradU in lagrangian folder, but all the gradU for each particle is not there.

[su_junwei]

Hi Jie
1301{(0 0 0 0 0 0 0 0 0 0)} means there are 1301 particles with the same value. tensor::zero. The possible reason is that the gradU of the particle location was not updated. The attachment you provided is right I think.
Regards, Junwei

[jiejie]

Hi JunWei

I am still looking at how to update the gradU for each of the particle and I came across couple question as follows:

1. if gradU has 1301{(0 0 0 0 0 0 0 0 0 0)} at each time step, it means all the gradU stored in the particles are zero?

2. In the particle.C:

template<class ParticleType>
00411 void Foam::Particle<ParticleType>::transformPosition(const tensor& T)
00412 {
00413 position_ = transform(T, position_);
00414 }

It says this transforms the physical properties of the particle.
But in parcel.C

void Foam:: parcel::transformProperties(const tensor& T)
00689 {
00690 U_ = transform(T, U_);
00691 }

It says this transforms the position and physical properties of the particle.
How come it uses the velocity U_ to transform the position of the particle and it use position_ to transform the position of the particle?

Thank you very much.

[su_junwei]

1) if gradU has 1301{(0 0 0 0 0 0 0 0 0 0)} at each time step, it means all the gradU stored in the particles are zero?
Yes
2) transform can be used in rotation. T is rotational tensor.

Junwei

[jiejie]

figuring out how to update gradU.

Thanks for your time JunWei

[jiejie]

One quick question: if I need to do an inner product of a tensor "T" and a vector "a", should it be "T & a" instead of "a & T"?

[ehsan]

Dear All,

tauMC is calculated via:

volTensorField tauMC("tauMC",mu*dev2(fvc::grad(U)().T()));

Could someonhe help me to find gard(U) from tauMC (revers of above relation)

the notation gard(U)=tauMC/mu gives error.

Thanks

[francescomarra]

Dear All,

After running dieselFoam, I would be like to observe further properties of the lagrangian field, as the Weber number of each particle, similarly to the way the diameter d or the density rho of each particle is reported in the lagrangian folder at each time step. However I am a newbe with C++ and I do not understand how I can adopt the function We available in the parcelFunctions.C to compute and store this quantity for each particle. Is this a simple task or does it require a long coding effort ?
Any help will be very gratefully appreciated.

Thank you in advance,

Franco

[ycui]

Dear Su Jun Wei,

I was trying to use your way to calculate the velocity gradient tensor.

It works fine with one core. But error comes for parallel computing when try to access fvc::grad().

Do you have any idea?

Best,

ycui

Originally Posted by su_junwei

Oh sorry.

I am confused last night. Just use fvc::grad(U_) and make an interpolation on the particle location like this

volTensorField gradU=fvc::grad(U_);
autoPtr<interpolation<tensor> >gradUInterpolator_ =
interpolation<tensor>::New(interpolationSchemes, gradU);
//interpolationSchemes is a word for assigning interpolation schemes("cell", "cellPoint" or "cellPointFace")

tensor gradUatPoint = gradUInterpolator->interpolate(ball.position(), ball.cell());

If the gradient vary little across the space, interpolation can be neglected, and use the gradient at the cell center to represent the gradient at the particle location. In this situation, just use the following code
tensor gradUatPoint =gradU[ball.cell()];
ball is the particle defined in OpenFOAM.

Hope this helps

regards, Junwei Su