When deriving Faraday's law of induction, one needs to calculate the time derivative of the flux $\Phi = \int_{S(t)} \vec{B} \cdot \vec{n} \;dS$, where $S(t)$ is the surface over which we define this flux. Now, Jackson and Wikipedia (see first equation) state that it is easy to prove that $$\frac{d}{dt}\int_{S(t)} \vec{B} \cdot \vec{n} \;dS = \int_{S(t)} \frac{\partial \vec{B}}{\partial t} \cdot \vec{n}\;dS + \oint_{\partial S(t)} \vec{B} \cdot \vec{v} \times d\vec{l},$$ where I assume $\vec{v}(t,\vec{x})$ is the vector field describing the velocity of the surface $S(t)$.

I fail to see how the second term is obvious and how I could get it from straightforward differentiation, no drawings.

My first idea would be to use that this is in fact a material derivative, and so the second term should come from $\vec{v} \cdot \vec{\nabla} \vec{B}$ plus some kind of Stoke's theorem. Is that possible?


1 Answer 1


So, let's take some closed loop enclosing a surface $S$ within the fluid and see how the magnetic flux through that loop changes as the loop gets carried along by the fluid. Over a time interval $dt$, $S$ moves a distance $\vec{v} \cdot dt$ and ends up as $S'$. The change in the flux is

$$d\Phi=\int_{S'} \vec{B}(t+dt) \cdot d\vec{a} \; - \int_{S} \vec{B}(t) \cdot d\vec{a} \;$$

Now, since $\vec{\nabla} \cdot \vec{B} = 0$ always, we can integrate $\vec B(t+dt)$ over the closed surface formed by the three surfaces $-S$, $S'$ and $R$ (the surface between $S$ and $S'$) taken together, then use the divergence theorem (this might be the "kind of Stoke's theorem" you were thinking of) to obtain

$$\int_{V} \vec{\nabla} \cdot \vec{B}d^3\vec{r} \; = \int_{S'} \vec{B}(t+dt) \cdot d\vec{a} \; + \int_{R} \vec{B}(t+dt) \cdot d\vec{a} \; - \int_{S} \vec{B}(t+dt) \cdot d\vec{a} = 0\;$$

($V$ is the volume enclosed by the three surfaces). Then, our preceding equation becomes

$$d\Phi=\int_{S} \vec{B}(t+dt) \cdot d\vec{a} \; - \int_{S} \vec{B}(t) \cdot d\vec{a} \; - \int_{R} \vec{B}(t+dt) \cdot d\vec{a} \; = dt\int_{S} \frac{\partial \vec{B}}{\partial t} \cdot d\vec{a} \; - \int_{R} \vec{B}(t+dt) \cdot d\vec{a} \; $$

The differential area $d\vec{a}$ is a parallelogram bounded by a line element $d\vec{l}$ around the loop $\partial S$ and the distance vector $\vec{v} \cdot dt$, so

$$d\vec{a}=d\vec{l} \times \vec{v} \cdot dt$$

Meaning that integrating over $d\vec{a}$ is equivalent to adding up these area elements as we move around the loop. That is, our second term becomes

$$ \int_{R} \vec{B}(t+dt) \cdot d\vec{a} \; = \oint_{\partial S} \vec{B} (t+dt) \cdot (d\vec{l} \times \vec{v} \cdot dt) = dt \oint_{\partial S} \vec{B} (t) \cdot (d\vec{l} \times \vec{v})$$

Where I have taken a first order approximation in the last step. Finally, after some arrangement, we recover the expression you were asking for:

$$\frac{d}{dt}\Phi=\int_{S} \frac{\partial \vec{B}}{\partial t} \cdot d\vec{a}\; + \oint_{\partial S} \vec{B} (t) \cdot (\vec{v} \times d\vec{l})$$

Interestingly enough, and since you asked, Alfvén's theorem can be interpreted in terms of material curves, and the ideal induction equation can be written as a material derivative. See this notes for more details (section 2.1: Physical interpretation of MHD). Also, notice the resemblance between Alfvén's and Kevin's circulation theorem in inviscid fluid dynamics.

  • 1
    $\begingroup$ A nice answer; same approach was taken by Purcell in his book. $\endgroup$
    – user36790
    Commented Nov 7, 2016 at 16:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.