[Magistrane]

Hey Foamers,

I need your help : I am stuck for few days now : I am trying to impose an initial angular speed in my simulation using sidDoFRigidBody. I don't want to modify the physics behind equations.
This variable is representing the angular velocity in sixDoFRigidBodyMotionI.H :

inline Foam::vector Foam::sixDoFRigidBodyMotion::omega() const
{
    return  Q() & (inv(momentOfInertia_) & pi());
}

Q() and momentOfInertia are data of my problem. I think that I can't change them without changing the physics. pi() is not a constant, it is depending of time and acceleration.

Do someone has any idea to add for exemple this vector to omega (0 0 50) ?

Thanks a lot.

Best regards,

Magi

[massive_turbulence]

Originally Posted by Magistrane

Hey Foamers,

I need your help : I am stuck for few days now : I am trying to impose an initial angular speed in my simulation using sidDoFRigidBody. I don't want to modify the physics behind equations.
This variable is representing the angular velocity in sixDoFRigidBodyMotionI.H :

inline Foam::vector Foam::sixDoFRigidBodyMotion::omega() const
{
    return  Q() & (inv(momentOfInertia_) & pi());
}

Q() and momentOfInertia are data of my problem. I think that I can't change them without changing the physics. pi() is not a constant, it is depending of time and acceleration.

Do someone has any idea to add for exemple this vector to omega (0 0 50) ?

Thanks a lot.

Best regards,

Magi

The Q() is a tensor called here as well

inline Foam::tensor& Foam::sixDoFRigidBodyMotion::Q()
{
    return motionState_.Q();
}

https://github.com/OpenFOAM/OpenFOAM...dBodyMotionI.H

I'm not sure what it means actually, I'd have to dig deeper and the problem I'm seeing is how would you add that vector to a tensor even if you just simply tried something like adding them, and also making physical sense. I guess the diagonals of a tensor are the coordinates x,y,z? So that might be it?

momentOfInertia_ is probably a tensor as well.

[Magistrane]

Hello,

Q_ is the orientation matrice. I do not know exactly what it is but openFoam documentation recommand to get it using the surfaceInertia function. So I guess it is a geometry dependant parameter. Changing it is not really an option.

momentOfInertia could be a tensor, but its description is also geometry related...
Isn't it possible to initialize omega too (0 0 50) and then to use this formula without touching it?
Thanks !

[massive_turbulence]

Originally Posted by Magistrane

Hello,

Q_ is the orientation matrice. I do not know exactly what it is but openFoam documentation recommand to get it using the surfaceInertia function. So I guess it is a geometry dependant parameter. Changing it is not really an option.

momentOfInertia could be a tensor, but its description is also geometry related...
Isn't it possible to initialize omega too (0 0 50) and then to use this formula without touching it?
Thanks !

Well... the way I see it is this. You have a default constructor that takes dict files and uses those to initialize your angular velocity (omega).

Here's you constructor initializer list

Foam::sixDoFRigidBodyMotion::sixDoFRigidBodyMotion
(
    const dictionary& dict,
    const dictionary& stateDict
)
:
    motionState_(stateDict),
    motionState0_(),
    restraints_(),
    constraints_(),
    tConstraints_(tensor::I),
    rConstraints_(tensor::I),
    initialCentreOfMass_
    (
        dict.lookupOrDefault
        (
            "initialCentreOfMass",
            vector(dict.lookup("centreOfMass"))
        )
    ),
    initialCentreOfRotation_(initialCentreOfMass_),
    initialQ_
    (
        dict.lookupOrDefault
        (
            "initialOrientation",
            dict.lookupOrDefault("orientation", tensor::I)
        )
    ),
    mass_(readScalar(dict.lookup("mass"))),
    momentOfInertia_(dict.lookup("momentOfInertia")),
    aRelax_(dict.lookupOrDefault<scalar>("accelerationRelaxation", 1.0)),
    aDamp_(dict.lookupOrDefault<scalar>("accelerationDamping", 1.0)),
    report_(dict.lookupOrDefault<Switch>("report", false)),
    solver_(sixDoFSolver::New(dict.subDict("solver"), *this))

from https://github.com/OpenFOAM/OpenFOAM...idBodyMotion.C

In order to get omega(0 0 50) from that perspective is to somehow initialize the variables that are looked up in those dictionary files so that the initial state of omega is that. That's probably the best case scenario without pulling hairs trying to find a needle in a haystack going through lines of code and seeing what each line does to calculate omega and then changing that code flow to initialize omega. Or I could be wrong and there is a solution but I'm not seeing it.


EDIT:

Here's another initializer list

Foam::sixDoFRigidBodyMotionState::sixDoFRigidBodyMotionState
(
    const dictionary& dict
)
:
    centreOfRotation_
    (
        dict.lookupOrDefault
        (
            "centreOfRotation",
            dict.lookupOrDefault("centreOfMass", vector::zero)
        )
    ),
    Q_(dict.lookupOrDefault("orientation", tensor::I)),
    v_(dict.lookupOrDefault("velocity", vector::zero)),
    a_(dict.lookupOrDefault("acceleration", vector::zero)),
    pi_(dict.lookupOrDefault("angularMomentum", vector::zero)),
    tau_(dict.lookupOrDefault("torque", vector::zero))
{}

from sixDoFRigidBodyMotionState.C

I'm not sure where this is (what dictionary file) pi_(dict.lookupOrDefault("angularMomentum", vector::zero)), but you could maybe do

pi_(dict.lookupOrDefault("angularMomentum", omega (0 0 50))) //basically insert the vector there.

[Magistrane]

Well, I really would like to do so, I already changed the moment of inertia to create an huge acceleration at the beginning... But all of these variables are called in other functions not related to omega and it change all the physics properties behind the calcul...
The only change I can do is on Q_ (orientation) but it also has an huge impact on the simulation : this is the rotation matrice. This is physicaly a non-sens to modify this.

Only pi() could be change. I'll look into it but I think that it is time dependant, so potentially has an impact on each simulation steps.

UP : pi() contains pi0()

pi() = rConstraints_ & aDamp_*(pi0() + 0.5*deltaT0*tau());

pi_ is define as angular momentum

dict.add("angularMomentum", pi_);

I could change this value if I had access to it. Because angularMomentum = momentOfInertia * omega
Moment of inertia only depend of the geometry. So omega is proportional to pi_.

[massive_turbulence]

Originally Posted by Magistrane

Well, I really would like to do so, I already changed the moment of inertia to create an huge acceleration at the beginning... But all of these variables are called in other functions not related to omega and it change all the physics properties behind the calcul...
The only change I can do is on Q_ (orientation) but it also has an huge impact on the simulation : this is the rotation matrice. This is physicaly a non-sens to modify this.

Only pi() could be change. I'll look into it but I think that it is time dependant, so potentially has an impact on each simulation steps.

UP : pi() contains pi0()



pi_ is define as angular momentum



I could change this value if I had access to it. Because angularMomentum = momentOfInertia * omega
Moment of inertia only depend of the geometry. So omega is proportional to pi_.

Here's another interesting piece of code you should look at in
sixDoFRigidBodyMotionSolver.C

bool Foam::sixDoFRigidBodyMotionSolver::writeObject
(
    IOstream::streamFormat fmt,
    IOstream::versionNumber ver,
    IOstream::compressionType cmp,
    const bool valid
) const
{
    IOdictionary dict
    (
        IOobject
        (
            "sixDoFRigidBodyMotionState",
            mesh().time().timeName(),
            "uniform",
            mesh(),
            IOobject::NO_READ,
            IOobject::NO_WRITE,
            false
        )
    );

    motion_.state().write(dict);
    return dict.regIOobject::write();
}

"sixDoFRigidBodyMotionState" is the object that looks up the dictionary but I still cannot fathom what the name of that file is. Openfoam is known for terrible documentation.

[Magistrane]

Up.
My Bad. After adding this line in the dynMeshDict:

angularMomentum (0 0 50);

New Angular velocity is compute. I guess i have to play with this parameter.

Thanks a lot