The MHD formalism

Dimensional equations

Here are the following equations for deriving the MHD model.

  • Maxwell-Thomson equation

(1)\[ \div \vb{B} = 0 \]
  • Induction or Maxwell-Faraday equation

(2)\[ \pdv{B}{t} = - \curl{\vb{E}} \]
  • Maxwell-Ampere without displacement currents

(3)\[ \curl \vb{B} = \mu_0 \vb{j} \]
  • Ohm’s law with resistive, Hall, and hyper-resistive terms

(4)\[ \vb{j} = \sigma \left( \vb{E} + \vb{u} \cross \vb{B} - \frac{1}{ne} \vb{j} \cross \vb{B} + \nu \laplacian \vb{j} \right) \]
  • Mass conservation for the plasma

(5)\[ \pdv{\rho}{t} + \div{\rho \vb{u}} = 0 \]
  • Momentum conservation for the plasma assuming a scalar pressure

(6)\[ \pdv{t}(\rho \vb{u}) + \div(\rho \vb{u} \vb{u}) = -\grad P + \vb{j} \cross \vb{B} \]
  • Energy conservation for the total energy of the plasma \(e_t = e + \flatfrac{\vb{u}^2}{2} \)

(7)\[ \pdv{t}(\rho e_t) + \div(\rho \vb{u} e_t) = - \div(P \vb{u}) + \vb{j} \cdot \vb{E} \]

With some manipulations, and by defining the total pressure \(P* = P + \flatfrac{\vb{B}^2}{2 \mu_0}\), the total energy per unit volume \(E = \rho e_t + \flatfrac{\vb{B}^2}{2 \mu_0}\), and the Poynting vector \(\vb*{\Pi} = \frac{\vb{E} \cross \vb{B}}{\mu_0}\) , we can obtain the following system of equations:

(8)\[\begin{split} \begin{aligned} & \pdv{\rho}{t} = -\div{\rho \vb{u}} \\ & \pdv{t}(\rho \vb{u}) = - \div[ \rho \vb{u}\vb{u} + P^* \vb{I} - \frac{\vb{B} \vb{B}}{\mu_0}] \\ & \pdv{E}{t} = - \div [ ( \rho e_t + P) \vb{u} + \vb*{\Pi}] = - \div[ \left(E + P^* - \frac{\vb{B}^2}{\mu_0} \right) \vb{u} + \vb*{\Pi} ] \\ & \pdv{B}{t} = - \curl \vb{E} %= - \curl[ \eta \vb{j} - \vb{u} \cross \vb{B} ] \end{aligned} \end{split}\]

where the electric field is expressed thanks to Ohm’s law as:

\[ \vb{E} = \eta \vb{j} - \vb{u} \cross \vb{B} + \frac{1}{n e} \vb{j} \cross \vb{B} - \nu \laplacian \vb{j} \]

The system is closed with a polytropic equation of state:

(9)\[ P \rho^{-\gamma} = \mathrm{constant} \implies \rho e = \frac{P}{\gamma - 1} \]

Dimensionless equations

Let us introduce the following reference quantities:

  • A reference magnetic field \(B_0\)

  • A reference plasma density \(n_0\)

  • The mass of the proton \(m_p\)

  • The Alfven velocity \(v_{0}\):

\[ v_{0} = \frac{B_0}{\sqrt{m_p n_0 \mu_0}} \]
  • The proton gyro-frequency \(\omega_0\):

\[ \omega_0 = \frac{e B_0}{m_p} \]

From these reference quantities, we can deduce the other reference quantities

  • the proton inertial length \(\delta_{0}\)

\[ \delta_0 = \frac{v_0}{\omega_0} = \frac{1}{e} \sqrt{\frac{m_p}{n_0 \mu_0}} \]
  • the reference mass density \(\rho_0\):

\[ \rho_0 = m_p n_0 \]
  • the reference pressure \(P_0\)

\[ P_0 = \rho_0 v_0^2 = \frac{B_0^2}{\mu_0} \]
  • the reference electric field \(E_0\)

\[ E_0 = V_0 B_0 = \frac{B_0^2}{\sqrt{m_p n_0 \mu_0}} \]
  • the reference charge current \(j_0\):

\[ j_0 = \frac{B_0}{\delta_0 \mu_0} = e B_0 \sqrt{\frac{n_0 }{m_p \mu_0}} \]
  • the reference resistivity \(\eta_0\)

\[ \eta_0 = \mu_0 v_0 \delta_0 = \frac{B_0 }{e n_0} \]
  • the reference hyper-resistivity \(\nu_0\)

\[ \nu_0 = \mu_0 v_0 \delta_0^3 = \frac{B_0 m_p}{e^3 n_0^2 \mu_0 } \]

By defining non-dimensional fields as \(\tilde{\phi} = \frac{\phi}{\phi_0}\), we can obtain the following equations:

\[\begin{split} \begin{aligned} & \pdv{\tilde{\rho}}{\tilde{t}} = -\tilde{\grad} \cdot \left( \tilde{\rho} \tilde{\vb{u}} \right) \\ & \pdv{\tilde{t}}(\tilde{\rho} \tilde{\vb{u}}) = - \tilde{\grad} \cdot \left[ \tilde{\rho} \tilde{\vb{u}}\tilde{\vb{u}} + \tilde{P}^* \vb{I} - \tilde{\vb{B}} \tilde{\vb{B}} \right] \\ & \pdv{\tilde{E}}{\tilde{t}} = - \tilde{\grad} \cdot \left[ ( \tilde{\rho} \tilde{e}_t + \tilde{P}) \tilde{\vb{u}} + \tilde{\vb*{\Pi}}\right] = - \tilde{\grad} \cdot \left[ \left(\tilde{E} + \tilde{P}^* - \tilde{\vb{B}}^2 \right) \tilde{\vb{u}} + \tilde{\vb*{\Pi}} \right] \\ & \pdv{\tilde{\vb{B}}}{\tilde{t}} = - \tilde{\grad} \cross \vb{\tilde{E}} %= - \curl[ \eta \vb{j} - \tilde{\vb{u}} \cross \tilde{\vb{B}} ] \end{aligned} \end{split}\]

with

\[\begin{split} \begin{aligned} & \tilde{\vb*{\Pi}} = \tilde{\vb{E}} \cross \tilde{\vb{B}} \\ & \tilde{\vb{E}} = \tilde{\eta} \tilde{\vb{j}} - \tilde{\vb{u}} \cross \tilde{\vb{B}} + \frac{1}{\tilde{n}} \tilde{\vb{j}} \cross \tilde{\vb{B}} - \tilde{\nu} \tilde{\nabla}^2 \tilde{\vb{j}} \end{aligned} \end{split}\]

In the following, equations will be considered under their non-dimensional form, with the tildes dropped for readability.

Splitting of the magnetic field

To numerically impose an external magnetic field, the magnetic field is decomposed as

\[ \vb{B} = \vb{B}_0 + \vb{B}_1 \]

where \(\vb{B}_0\) is the imposed external field and \(\vb{B}_1\) is the difference between the total and external magnetic fields.

Inserting this splitting in the definition of the total energy \(E\), one can define the “reduced” total energy \(E_1\):

\[ E = \underbrace{\rho e_t + \frac{\vb{B}_1^2}{2}}_{E_1} + \frac{\vb{B}_0^2}{2} + \vb{B}_0 \cdot \vb{B}_1 \]

The derivative of \(E\) is related to \(E_1\)’s one following

\[\begin{split} \begin{aligned} \pdv{E}{t} & = \pdv{E_1}{t} + \vb{B}_0 \cdot \pdv{\vb{B}_0}{t} + \vb{B}_0 \cdot \pdv{\vb{B}_1}{t} + \pdv{\vb{B}_0}{t} \cdot \vb{B}_1 \\ & = \pdv{E_1}{t} + \vb{B}_0 \cdot \pdv*{\vb{B}}{t} + \pdv{\vb{B}_0}{t} \cdot \vb{B}_1 \end{aligned} \end{split}\]

Moreover the term \(\vb{B}_0 \cdot \pdv*{\vb{B}}{t}\) can be expanded using the induction equation as:

\[\begin{split} \begin{aligned} \vb{B}_0 \cdot \pdv*{\vb{B}}{t} &= - \vb{B}_0 \cdot \left( \curl \vb{E} \right) \\ &= - \div(\vb{E} \cross \vb{B}_0) - \vb{E} \cdot \underbrace{ \left( \curl \vb{B}_0\right)}_{ = \vb{j}_0 = \vb{0}} \end{aligned} \end{split}\]

The rotational of \(\vb{B}_0\) is zero because it is external; the currents generating it are located outside of the computational domain.

Therefore, the equation for energy and magnetic field can be written in terms of \(E_1\) and \(B_1\) as follows

\[\begin{split} \begin{aligned} & \pdv{E_1}{t} = - \grad \cdot \left[ ( \rho e_t + P) \vb{u} + \vb{E} \cross \vb{B}_1 \right] - \pdv{\vb{B}_0}{t} \cdot \vb{B}_1 % = - \grad \cdot \left[ \left(E_1 + P_1^* - \vb{B}_1^2 \right) \vb{u} + \vb{E} \cross \vb{B}_1 \right] \\ & \pdv{\vb{B}_1}{t} = - \grad \cross \vb{E} - \pdv{\vb{B}_0}{t} \end{aligned} \end{split}\]