Skip to content

Commit

Permalink
WIP turbulence
Browse files Browse the repository at this point in the history
  • Loading branch information
milancurcic committed Oct 28, 2024
1 parent 27720a0 commit c8a1f68
Showing 1 changed file with 178 additions and 6 deletions.
184 changes: 178 additions & 6 deletions notes.tex
Original file line number Diff line number Diff line change
Expand Up @@ -818,6 +818,7 @@ \subsection*{Further reading}

\newpage
\section{Conservation of mass and momentum}
\label{sec:continuity_momentum}

In this chapter we will derive the fundamental equations for fluid flows:
continuity, momentum, and energy.
Expand Down Expand Up @@ -1136,6 +1137,7 @@ \subsubsection{Incorporating the forces}
\begin{equation}
\frac{\partial \mathbf{u}}{\partial t} + (\mathbf{u} \cdot \nabla) \mathbf{u} =
\frac{1}{\rho} \nabla \cdot \boldsymbol{\sigma} + \frac{\mathbf{F}_b}{\rho}
\label{eq:momentum_cauchy}
\end{equation}
This form of the momentum equation is often called the
\textit{Cauchy momentum equation}\index{Momentum equation!Cauchy}.
Expand Down Expand Up @@ -1220,6 +1222,7 @@ \subsubsection{Pressure gradient}
\index{Momentum equation!Euler}

\subsubsection{Viscous forces}
\label{sec:viscous_forces}

Now, let's look at the shear stress tensor divergence $\nabla \cdot \boldsymbol{\tau}$.
Written out explicitly as a matrix of all its components, $\boldsymbol{\tau}$ is:
Expand Down Expand Up @@ -3722,14 +3725,14 @@ \subsection{Reynolds decomposition}
Insert Eq. (\ref{eq:reynolds_advection_expanded}) into Eq. (\ref{eq:ns_reynolds3}) to get:

\begin{equation}
\frac{\partial (\rho \overline{\mathbf{u}})}{\partial t} +
\frac{\partial \overline{\mathbf{u}}}{\partial t} +
\nabla \cdot (\overline{\mathbf{u}}\, \overline{\mathbf{u}}) =
- \frac{1}{\rho} \nabla \overline{p} +
\nu \nabla^2 \overline{\mathbf{u}} +
\nabla \cdot (\overline{\mathbf{u}' \mathbf{u}'})
\label{eq:ns_reynolds4}
\nu \nabla^2 \overline{\mathbf{u}}
- \nabla \cdot (\overline{\mathbf{u}' \mathbf{u}'})
\label{eq:rans}
\end{equation}
which is the \textit{Reynolds-averaged Navier-Stokes equation}
which is the \textit{Reynolds-Averaged Navier-Stokes (RANS) equation}
\index{Reynolds!averaged Navier-Stokes equation}, the term
$\overline{\mathbf{u}' \mathbf{u}'}$ is called the \textit{Reynolds stress tensor}
\index{Reynolds!stress tensor},
Expand All @@ -3744,14 +3747,183 @@ \subsection{Reynolds decomposition}
\label{eq:continuity_reynolds}
\end{equation}

Between Eqs. (\ref{eq:ns_reynolds4}) and (\ref{eq:continuity_reynolds}) we have
Between Eqs. (\ref{eq:rans}) and (\ref{eq:continuity_reynolds}) we have
two equations with three unknowns: $\overline{\mathbf{u}}$, $\overline{p}$, and
$\overline{\mathbf{u}' \mathbf{u}'}$.
To close the system, we need to find an equation for the Reynolds stress tensor,
which brings us to the closure problem of turbulence.

\subsection{Closure problem}

To illustrate the closure problem of turbulence, let's try to derive the
equation for the evolution of the Reynolds stress $\overline{\mathbf{u}' \mathbf{u}'}$.
Suppose that the Reynolds stress evolves according to the yet to be determined
sources and sinks of the Reynolds stress:

\begin{equation}
\frac{d \left( \mathbf{u}' \mathbf{u}' \right)}{dt} =
\text{sources} - \text{sinks}
\label{eq:reynolds_stress_evolution}
\end{equation}
Expanding the time derivative in a momentum-conservative form and time averaging
yields similar to Eq. (\ref{eq:reynolds_advection_expanded}):

\begin{equation}
\frac{d \left( \overline{\mathbf{u}' \mathbf{u}'} \right)}{dt} =
\frac{\partial \left( \overline{\mathbf{u}' \mathbf{u}'} \right)}{\partial t} +
\nabla \cdot \left( \overline{\mathbf{u}} \overline{\mathbf{u}' \mathbf{u}'} \right) +
\nabla \cdot \left( \overline{\mathbf{u}' \mathbf{u}' \mathbf{u}'} \right)
\end{equation}
See, if we try to seek the equation for the evolution of the Reynolds stress,
we end up with the flux of the flux itself as a new unknown.
Further, if we tried to seek the equation for this new cubic term, we would
end up with an equation that includes a quartic term of $\mathbf{u}'$:


\begin{equation}
\frac{d \left( \overline{\mathbf{u}' \mathbf{u}' \mathbf{u}'} \right)}{dt} =
\frac{\partial \left( \overline{\mathbf{u}' \mathbf{u}' \mathbf{u}'} \right)}{\partial t} +
\nabla \cdot \left( \overline{\mathbf{u}} \overline{\mathbf{u}' \mathbf{u}' \mathbf{u}'} \right) +
\nabla \cdot \left( \overline{\mathbf{u}' \mathbf{u}' \mathbf{u}' \mathbf{u}'} \right)
\end{equation}
The fact that we cannot close the RANS equations unless we somehow approximate
the Reynolds stress tensor is known as the closure problem of turbulence.
On one hand, it's relieving that we don't have to figure out the sources and
sinks for the Reynolds stress tensor in Eq. (\ref{eq:reynolds_stress_evolution}).
On the other hand, we still need to come up with some model or approximation
for the Reynolds stress tensor to solve the RANS equations.

\subsection{Reynolds stress}

Recall from Chapter \ref{sec:continuity_momentum} where we first derived the
Cauchy momentum equation (Eq. \ref{eq:momentum_cauchy}), ignoring the body forces
for brevity:

\begin{equation}
\frac{\partial \mathbf{u}}{\partial t} + (\mathbf{u} \cdot \nabla) \mathbf{u} =
\frac{1}{\rho} \nabla \cdot \boldsymbol{\sigma} + \frac{\mathbf{F}_b}{\rho}
\end{equation}
and the associated stress tensor (Eq. \ref{eq:stress_tensor_decomposition}):

\begin{equation}
\boldsymbol{\sigma} = -p \mathbf{I} + \boldsymbol{\tau}
\end{equation}
where we had described the stress tensor $\boldsymbol{\sigma}$ as a combination
of the normal stresses (pressure) on the diagonal and the deviatoric stresses
off the diagonal:

\begin{equation}
\begin{bmatrix}
-p + \tau_{xx} & \tau_{xy} & \tau_{xz} \\
\tau_{yx} & -p + \tau_{yy} & \tau_{yz} \\
\tau_{zx} & \tau_{zy} & -p + \tau_{zz}
\end{bmatrix}
\end{equation}
Then, in Section \ref{sec:viscous_forces}, we stated that for a Newtonian fluid
the deviatoric stresses can be approximated with the velocity gradients, an
approximation that was established in the laboratory:

\begin{equation}
\nabla \cdot \boldsymbol{\tau} = \nu \nabla^2 \mathbf{u}
\end{equation}
Now, in addition to the viscous stresses, we have the turbulent Reynolds stresses
introduced in Eq. (\ref{eq:rans}).
The turbulent Reynolds stresses arise due to the scale separation between the
large-scale mean flow and the turbulent fluctuations, which we introduced when
we applied the Reynolds decomposition to the velocity field.

Eq. (\ref{eq:rans}) can be rewritten more concisely by applying the divergence
operator to the pressure and Reynolds and viscous stresses as a whole:

\begin{equation}
\frac{\partial \overline{\mathbf{u}}}{\partial t} +
\nabla \cdot (\overline{\mathbf{u}}\, \overline{\mathbf{u}}) =
\frac{1}{\rho} \nabla \cdot \left( \mu \nabla \cdot \overline{\mathbf{u}} - p - \rho \overline{\mathbf{u}' \mathbf{u}'} \right)
\label{eq:rans_expanded}
\end{equation}
If it's not obvious already, notice that $\rho \overline{\mathbf{u}' \mathbf{u}'}$
is the only term that makes Eq. (\ref{eq:rans_expanded}) different from the
original Navier-Stokes equation (Eq. \ref{eq:ns_reynolds3}).
Thus, if we apply a scale separation (\textit{i.e.} the Reynolds decomposition)
to the velocity field such that we distinguish between the mean flow and the
fluctuations, the equation for the mean flow contains an additional term that
quantifies the contribution of the turbulent fluctuations to the mean.
Note that, strictly speaking, $\rho \overline{\mathbf{u}' \mathbf{u}'}$ is a
stress (as in, momentum flux), however it's common to refer to
$\overline{\mathbf{u}' \mathbf{u}'}$ as the Reynolds stress as well, even when
the density is omitted.

Let's look at this Reynolds stress tensor in more detail.
Using our usual notation for the velocity vector to be $\mathbf{u} = (u, v, w)$,
the components of the Reynolds stress tensor are:

\begin{equation}
\overline{u'u'} = \begin{bmatrix}
\overline{u'u'} & \overline{u'v'} & \overline{u'w'} \\
\overline{v'u'} & \overline{v'v'} & \overline{v'w'} \\
\overline{w'u'} & \overline{w'v'} & \overline{w'w'}
\end{bmatrix}
\end{equation}
The diagonal components of this tensor ($\overline{u'u'}$, $\overline{v'v'}$, and
$\overline{w'w'}$) are called the \textit{normal stresses}, and the off-diagonal
components ($\overline{u'v'}$, $\overline{u'w'}$, $\overline{v'w'}$) are called
the \textit{shear stresses}.
The Reynolds stress tensor is symmetric, which means that
$\overline{u'v'} = \overline{v'u'}$, $\overline{u'w'} = \overline{w'u'}$, and
$\overline{v'w'} = \overline{w'v'}$.
It is only the shear stresses that contribute to the turbulent transport of
momentum.
An important property of boundary layer physics, the
\textit{Turbulent Kinetic Energy}\index{Turbulent kinetic energy} (TKE) is half
the sum of the diagonal components of the Reynolds stress tensor:

\begin{equation}
k = \frac{1}{2} \left( \overline{u'u'} + \overline{v'v'} + \overline{w'w'} \right)
\end{equation}
TKE plays an important role in parameterizing the subgrid-scale turbulent
processes in the boundary layer components of weather and ocean prediction
models.
$\overline{u'w'}$ and $\overline{v'w'}$ are also very important quantities in
the study of air-sea interaction, as they govern the momentum exchange between
the atmospheric surface layer, the ocean surface waves, and the upper-ocean
boundary layer.

In numerical models, the vector equations must be written out explicitly in
scalar component form.
It's thus a useful exercise to write out the RANS equation
(Eq. \ref{eq:rans_expanded}) as a system of scalar equations, one for each
component of the mean velocity vector:

\begin{equation}
\frac{\partial \overline{u}}{\partial t} +
\overline{u} \frac{\partial \overline{u}}{\partial x} +
\overline{v} \frac{\partial \overline{u}}{\partial y} +
\overline{w} \frac{\partial \overline{u}}{\partial z} =
- \frac{1}{\rho} \frac{\partial \overline{p}}{\partial x}
- \frac{\partial \overline{u'u'}}{\partial x} - \frac{\partial \overline{v'u'}}{\partial y} - \frac{\partial \overline{w'u'}}{\partial z}
+ \nu \left( \frac{\partial^2 \overline{u}}{\partial x^2} + \frac{\partial^2 \overline{u}}{\partial y^2} + \frac{\partial^2 \overline{u}}{\partial z^2} \right)
\end{equation}

\begin{equation}
\frac{\partial \overline{v}}{\partial t} +
\overline{u} \frac{\partial \overline{v}}{\partial x} +
\overline{v} \frac{\partial \overline{v}}{\partial y} +
\overline{w} \frac{\partial \overline{v}}{\partial z} =
- \frac{1}{\rho} \frac{\partial \overline{p}}{\partial y}
- \frac{\partial \overline{u'v'}}{\partial x} - \frac{\partial \overline{v'v'}}{\partial y} - \frac{\partial \overline{w'v'}}{\partial z}
+ \nu \left( \frac{\partial^2 \overline{u}}{\partial x^2} + \frac{\partial^2 \overline{u}}{\partial y^2} + \frac{\partial^2 \overline{u}}{\partial z^2} \right)
\end{equation}

\begin{equation}
\frac{\partial \overline{w}}{\partial t} +
\overline{u} \frac{\partial \overline{w}}{\partial x} +
\overline{v} \frac{\partial \overline{w}}{\partial y} +
\overline{w} \frac{\partial \overline{w}}{\partial z} =
- \frac{1}{\rho} \frac{\partial \overline{p}}{\partial z}
- \frac{\partial \overline{u'w'}}{\partial x} - \frac{\partial \overline{v'w'}}{\partial y} - \frac{\partial \overline{w'w'}}{\partial z}
+ \nu \left( \frac{\partial^2 \overline{u}}{\partial x^2} + \frac{\partial^2 \overline{u}}{\partial y^2} + \frac{\partial^2 \overline{u}}{\partial z^2} \right)
\end{equation}

\subsection{Turbulent kinetic energy budget}

\subsection{Turbulent cascade}
Expand Down

0 comments on commit c8a1f68

Please sign in to comment.