Sig Turbomachinery MRF Library

From OpenFOAMWiki
Revision as of 20:27, 8 March 2010 by Boroli (Talk | contribs)

1 Continuity Equation

1.1 Mathematical Relations

The total time derivation of any variable  \Phi can be split off into a local term and a convective term. In the absolute frame of reference it gives with the absolute velocity  \vec{c}

 \frac{d \Phi}{d t} = \frac{\partial \Phi}{\partial t} + \vec{c} \bullet \nabla \Phi  \qquad (3)

And in the relative frame of reference it gives with the relative velocity  \vec{w}

 \frac{d^\prime \Phi}{d t} = \frac{\partial^\prime \Phi}{\partial t} + \vec{w} \bullet \nabla^\prime \Phi  \qquad (4)

As for a scalar variable it is valid

 \frac{d \Phi}{d t} = \frac{d^\prime \Phi}{d t}

With eq. 3 and eq. 4 it yields

 \frac{\partial \Phi}{\partial t} = \frac{\partial^\prime \Phi}{\partial t} - \left( \vec{c} - \vec{w} \right) \bullet \nabla \Phi

1.2 Transformation in a Relative Frame of Reference

2 Momentum Equation

2.1 Mathematical Relations

The first time derivation of any vector  \vec{b} formulated in the absolute and relative frame of reference, according to [1], where  \vec{\omega} is the angular frequency between these two coordinate systems. Any quantity with an ' is related to the relative frame of reference.

 \frac{d \vec{b}}{d t} = \frac{d^\prime \vec{b}}{d t}+\vec{\omega} \times \vec{b} \qquad (1)

If  \nabla is a vector operator, then it is valid

 \vec{\nabla} \equiv \vec{\nabla}^\prime  \qquad (2)

If we insert eq. 2-4 into the 1, one can write

 \frac{\partial\vec{b}}{\partial t} + \vec{c} \bullet \nabla \vec{b} = \frac{\partial^\prime\vec{b}}{\partial t} + \vec{w} \bullet \nabla \vec{b} + \vec{\omega} \times \vec{b}  \qquad (5)

With the relationship of eq. 14 this yields to

 \frac{\partial\vec{b}}{\partial t} = \frac{\partial^\prime\vec{b}}{\partial t} - \vec{u} \bullet \nabla \vec{b} + \vec{\omega} \times \vec{b}

Additional, the following identities are valid with any vector  \vec{a} and a scalar  \alpha [2]

 (\alpha \vec{a}) \otimes \vec{b} = \vec{a} \otimes (\alpha \vec{b}) = \alpha (\vec{a} \otimes \vec{b})  \qquad (6)

and [3]

 {\nabla} \bullet \left( {\vec{a} \otimes \vec{b}}\right)  = \left( {\nabla}\bullet{\vec{a}} \right) \vec{b} + \vec{a} \bullet {\nabla}{\vec{b}}  \qquad (7)

and [4]

 {\nabla} \bullet (\vec{a} \times \vec{b})=\vec{b} \bullet ({\nabla} \times \vec{a})-\vec{a} \bullet ({\nabla} \times \vec{b})  \qquad (8)

2.2 Derivation

We are now deriving the momentum equation in the relative coordinate system but formulated with the absolute velocity.

If we use eq. 6, then it can be easily shown, that the following statement is valid

 \left( \varrho \vec{c} \right) \otimes \vec{c} =  \vec{c} \otimes \left( \varrho \vec{c} \right)  \qquad (9)

If we use the solution from eq. 9, and apply them to eq. 7, the following can be shown


\begin{array}{rcl}
\nabla \bullet  \left[ \left( \varrho \vec{c} \right) \otimes \vec{c} \right] & = &
\left[\nabla \bullet \left( \varrho \vec{c} \right)\right] \vec{c} + \left( \varrho \vec{c} \right) \bullet \nabla \vec{c} \qquad (10)\\
 =  \nabla \bullet \left[ \vec{c} \otimes \left( \varrho \vec{c} \right) \right] & = &
\left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right) +  \vec{c} \bullet \nabla \left( \varrho \vec{c} \right) \qquad (11)
\end{array}

Furthermore, if we apply  \vec{b} = \varrho \vec{c} in eq. 5, this gives us

\dfrac{\partial \left( \varrho \vec{c} \right)}{\partial t} + \vec{c} \bullet \nabla \left( \varrho \vec{c} \right) = \frac{\partial^\prime\left( \varrho \vec{c} \right)}{\partial t} + \vec{w} \bullet \nabla \left( \varrho \vec{c} \right) + \vec{\omega} \times \left( \varrho \vec{c} \right) \qquad (12)

Now we add the first term on the right hand side of eq. 11  \left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right) to eq. 12 and apply furthermore eq. 11. Then it follows


\begin{array}{rcl}
\dfrac{\partial \left( \varrho \vec{c} \right)}{\partial t} + \left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right) + \vec{c} \bullet \nabla \left( \varrho \vec{c} \right) 
& = & \dfrac{\partial^\prime\left( \varrho \vec{c} \right)}{\partial t} + \vec{w} \bullet \nabla \left( \varrho \vec{c} \right) + \vec{\omega} \times \left( \varrho \vec{c} \right) + \left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right) \\
= \dfrac{\partial \left( \varrho \vec{c} \right)}{\partial t} + \nabla \bullet  \left[ \left( \varrho \vec{c} \right) \otimes \vec{c} \right] & = & \dfrac{\partial^\prime\left( \varrho \vec{c} \right)}{\partial t} + \vec{\omega} \times \left( \varrho \vec{c} \right) + \underbrace{\vec{w} \bullet \nabla \left( \varrho \vec{c} \right) + \left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right)}_{A1}  \\
\end{array} \qquad (13)

Consider, that the origin of the absolute and relative frame of reference or coincident. Then it is valid to say, that the position vector is equal in both systems  \vec{r} =\vec{r}^\prime . But, be aware, the components of these vectors are not equal, because they are defined in different base systems. If we now use  \vec{b} =\vec{r}^\prime in eq. 1, it gives usthe well known relation between the absolute and relative velocity.


\begin{array}{rcl}
\dfrac{d \vec{r}^\prime}{d t} = \dfrac{d \vec{r}}{d t} & = & \dfrac{d^\prime \vec{r}^\prime}{d t} + \vec{\omega} \times \vec{r}^\prime \\
= \vec{c} & = & \vec{w} + \left( \vec{\omega} \times \vec{r}^\prime \right) = \vec{w} + \vec{u} 
\end{array}\qquad (14)

If we now use the solution from eq. 14 and apply this to the first term on the right hand side of eq. 11, the following is valid


\begin{array}{rcl}
\left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right)
& = & \left[\nabla \bullet \left( \vec{w}+\vec{u} \right)  \right] \left( \varrho \vec{c} \right) \\
& = &  \left[\nabla \bullet \vec{w} \right] \left( \varrho \vec{c} \right) + \underbrace{\left[\nabla \bullet \left( \vec{\omega} \times \vec{r}^\prime \right)\right]}_{A2}  \left( \varrho \vec{c} \right)
\end{array}\qquad (15)

If we now apply eq. 8 to the term A2 of eq. 15, we get

 \nabla \bullet \left( \vec{\omega} \times \vec{r}^\prime \right) = \vec{r}^\prime \bullet \underbrace{({\nabla} \times \vec{\omega})}_{0} - \vec{\omega} \bullet \underbrace{({\nabla} \times \vec{r}^\prime)}_{0} = 0 \qquad (16)

If you consider for example \vec{\omega} = (0, 0, \omega)^T and \vec{r}^\prime = (x^\prime, y^\prime, z^\prime)^T in cartesian coordinates and apply the cross product component-by-component, then the result from eq. 16 maybe become a little bit more clearer. With eq. 16, eq. 15 becomes

 \left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right) = \left[\nabla \bullet \vec{w} \right] \left( \varrho \vec{c} \right) \qquad (17)

Now with eq. 17 and eq. 7 term A1 from eq. 13 writes to


\begin{array}{rcl}
\left[\nabla \bullet \vec{c} \right] \left( \varrho \vec{c} \right) + \vec{w} \bullet \nabla \left( \varrho \vec{c} \right) & = & \left[\nabla \bullet \vec{w} \right] \left( \varrho \vec{c} \right) + \vec{w} \bullet \nabla \left( \varrho \vec{c} \right) \\
& = & \nabla \bullet  \left[ \vec{w} \otimes \left( \varrho \vec{c} \right) \right]
\end{array}\qquad (18)

If we now insert eq. 18 into eq. 13, we get


\dfrac{\partial \left( \varrho \vec{c} \right)}{\partial t} + \nabla \bullet  \left[ \left( \varrho \vec{c} \right) \otimes \vec{c} \right] = \dfrac{\partial^\prime\left( \varrho \vec{c} \right)}{\partial t} + \nabla \bullet  \left[ \vec{w}  \otimes \left( \varrho \vec{c} \right) \right] +
\vec{\omega} \times \left( \varrho \vec{c} \right)\qquad (19)

The momentum equation of an inertial coordinate system is equal to


\dfrac{\partial \varrho \vec{c}}{\partial t} + \nabla \bullet \left( \varrho \vec{c} \otimes \vec{c}\right)
= -\nabla p + \nabla \bullet \tau + \varrho \vec{k}\qquad (20)

If we now apply eq. 19 to eq. 20 we get the momentum equation in a relative coordinate system, see for example [1]


\dfrac{\partial^\prime \left( \varrho \vec{c}\right)}{\partial t} + \nabla \bullet \left( \varrho \vec{w} \otimes \vec{c} \right) + \vec{\omega} \times \left( \varrho \vec{c} \right)
= -\nabla p + \nabla \bullet \tau + \varrho \vec{k}\qquad (21)

Furthermore it can be shown that the tensor 2. rank from the dyadic product of \vec{c} \otimes \vec{w} is not symmetric, if you write it down component-by-component in cartesian coordinates. So that it must be valid to write

 \vec{c} \otimes \vec{w} \boldsymbol{\neq} \vec{w} \otimes \vec{c}

3 Energy Equation

TODO.


4 Shear Stress Tensor

4.1 Mathematical Relations

The product rule is also valid with the Nabla-Operator and arbitrary products \oplus and \boxdot. The arrows above indicates the appliance of the Nabla-Operator.


\nabla \oplus ( \overset{\downarrow}{ \Phi } \boxdot \overset{\downarrow}{ \Psi } ) = \nabla \oplus ( \overset{\downarrow}{ \Phi } \boxdot \Psi ) + \nabla \oplus ( \Phi \boxdot \overset{\downarrow}{ \Psi } ) \qquad (30)

For a cross product of two vectors it is valid


\vec{b} \times \vec{c} = -\vec{c} \times \vec{b} \qquad (31)

In general one writes for a tensor \mathbb{T} = \vec{b} \otimes \vec{c} :


\left( \vec{b} \otimes \vec{c}\right)^T = \vec{c} \otimes \vec{b} \qquad (32)

And for a tensor \mathbb{S} one can also write


\left( \mathbb{S}^T \right)^T = \mathbb{S} \qquad (33)

Furthermore for a tensor \mathbb{T} = \vec{b} \otimes \vec{c} it is valid [2]:


\left( \vec{b} \otimes \vec{c}\right) \times \vec{a} = \vec{b} \otimes \left( \vec{c} \times \vec{a}\right) \qquad (34)

and also


\vec{a} \times \left( \vec{b} \otimes \vec{c}\right)  = \left( \vec{a} \times \vec{b} \right) \otimes \vec{c} \qquad (35)

Therefore the following indentities are valid, starting with eq. 34:


\begin{alignat}{2}
\left( \vec{b} \otimes \vec{c}\right) \times \vec{a} & = & \vec{b} \otimes \left( \vec{c} \times \vec{a}\right) \qquad  \qquad \\
\text{with eq. 31 follows: } & = & - \vec{b} \otimes \left( \vec{a} \times \vec{c}\right)  \qquad (36) \\
\text{with eq. 32 follows: } & = & - \left[ \left( \vec{a} \times \vec{c}\right) \otimes \vec{b} \right]^T  \qquad (37)\\
\text{with eq. 35 follows: } & = & - \left[ \vec{a} \times \left( \vec{c} \otimes \vec{b} \right) \right]^T  \qquad (38)\\
\text{with eq. 32 follows: } & = & - \left[ \vec{a} \times \left( \vec{b} \otimes \vec{c} \right)^T \right]^T \qquad (39)
\end{alignat}

With eq. 39 for the tensor \mathbb{T} = \vec{b} \otimes \vec{c} the following is valid:


\mathbb{T} \times \vec{a} = - \left[ \vec{a} \times \mathbb{T}^T \right]^T \qquad (40)

If eq. 40 gets transposed and then eq. 33 is applied to the result, one yields:


\left( \mathbb{T} \times \vec{a} \right)^T = - \left( \left[ \vec{a} \times \mathbb{T}^T \right]^T \right)^T  = - \left( \vec{a} \times \mathbb{T}^T \right) \qquad (41)

4.2 Transformation in a Relative Frame of Reference

For a newtonian fluid the shear stress tensor  \tau can be expressed in the inertial frame of reference with  \lambda = \mu^\prime - \frac{2}{n_D} \mu as


\tau = \mu \left[ \nabla \vec{c} + \left( \nabla \vec{c} \right)^T \right] + \lambda \nabla \bullet \vec{c} \qquad (50)

Where in three-dimensional the number of dimensions  n_D = 3 . Furthermore the volume viscosity  \mu^\prime vanishes after Stokes' hypotheses.

As already expressed in eq. 16, the last term on the right hand side is equal to


\nabla \bullet \vec{c} = \nabla \bullet \vec{w} \qquad (51)

The same procedure is also applied to the following expression


\nabla \vec{c} + \left( \nabla \vec{c} \right)^T = \nabla \left\lbrace \vec{w} + \vec{u} \right\rbrace  + \left( \nabla \left\lbrace \vec{w} + \vec{u} \right\rbrace \right)^T = \nabla \vec{w} + \left( \nabla \vec{w} \right)^T + \nabla \vec{u} + \left( \nabla \vec{u} \right)^T \qquad (52)

For the last two terms on the right hand side, one can write:


\nabla \vec{u} + \left( \nabla \vec{u} \right)^T = \nabla \left( \vec{\omega} \times \vec{r}^\prime \right) + \left( \nabla \left( \vec{\omega} \times \vec{r}^\prime \right) \right)^T \qquad (53)

If one now applies the product rule from eq. 30, one yields the following:


\begin{alignat}{2}
\nabla \left( \vec{\omega} \times \vec{r}^\prime \right)
+ \left( \nabla \left( \vec{\omega} \times \vec{r}^\prime \right) \right)^T
& = \underbrace{\nabla \overset{\downarrow}{\vec{\omega}} }_0  \times \vec{r}^\prime
+ \nabla ( \vec{\omega} \times \overset{\downarrow}{ \vec{r}^\prime } )
+ ( \underbrace{\nabla \overset{\downarrow}{\vec{\omega}} }_0  \times \vec{r}^\prime )^T
+ ( \nabla ( \vec{\omega} \times \overset{\downarrow}{ \vec{r}^\prime } ) )^T \\
& = \nabla ( \vec{\omega} \times \overset{\downarrow}{ \vec{r}^\prime } )
+ ( \nabla ( \vec{\omega} \times \overset{\downarrow}{ \vec{r}^\prime } ) )^T  \qquad (54)
\end{alignat}

As \vec{\omega}  is a single vector and not a vector field, the gradient of a single vector is always zero. If one now apply eq. 36 and eq. 34 at each term of the right hand side of eq. 54, one yields with the metric tensor \mathbb{G}:


\nabla \vec{u} + \left( \nabla \vec{u} \right)^T
 =  - ( \underbrace{ \nabla \vec{r}^\prime }_{\mathbb{G}}  ) \times \vec{\omega} 
- ( ( \underbrace{ \nabla \vec{r}^\prime }_{\mathbb{G}}  ) \times \vec{\omega} )^T \qquad (55)

With the help of eq. 41 it follows:


\nabla \vec{u} + \left( \nabla \vec{u} \right)^T
= - \mathbb{G} \times \vec{\omega} + \vec{\omega} \times \mathbb{G}^T \qquad (56)

As the metric tensor \mathbb{G} is symmetric, one can write \mathbb{G} = \mathbb{G}^T . Thus from eq. 56 it follows


\nabla \vec{u} + \left( \nabla \vec{u} \right)^T
= - \mathbb{G} \times \vec{\omega} + \vec{\omega} \times \mathbb{G} = 0 \qquad (57)

With the help of eq. 57 it follows for eq. 52:


\nabla \vec{c} + \left( \nabla \vec{c} \right)^T = \nabla \vec{w} + \left( \nabla \vec{w} \right)^T \qquad (58)

Finally with eq. 51 and eq. 58 the shear stress tensor for a newtonian fluid in the relative frame of reference can be expressed as


\tau = \mu \left[ \nabla \vec{w} + \left( \nabla \vec{w} \right)^T \right] + \lambda \nabla \bullet \vec{w} \qquad (59)

5 References

  1. FLUENT 6.3 User's Guide; "Equations for a Rotating Reference Frame"; Chap. 10.2.2
  2. Michael H. Vavra, "Aero-thermodynamics and flow in turbomachines", 1960; Appendix B1