Consider a system of $N$ point-masses $m_i$, each travelling at a velocity $\mathbf{v}_i$. Then, the total linear momentum / impulse $\mathbf{p}$ of the system can be calculated as (see Smith, J.O. Physical Audio Signal Processing)

$$\mathbf{p} := \sum_{i=1}^N \mathbf{p}_i = \sum_{i=1}^N m_i \mathbf{v}_i = \sum_{i=1}^N m_i \dot{\mathbf{x}}_i = M \frac{d}{dt} \left(\frac{1}{M}\sum_{i=1}^N m_i \mathbf{x}_i \right) = M \frac{d}{dt} ( \mathbf{x}_c ) = M \mathbf{v}_c$$

where $M = \sum_{i=1}^N m_i$ denotes the total mass and $\mathbf{v}_c$ the velocity of the center of mass.

Similarly, the total linear momentum / impulse $\mathbf{p}$ for a rigid body $K$ can be calculated using points $\mathbf{x}_P$ on the body having the infinitesimal mass $dm$ as

$$\mathbf{p} := \int_{K} d\mathbf{p} = \int_{K} \mathbf{v}_P \: dm = \int_{K} \dot{\mathbf{x}}_P \: dm = M \frac{d}{dt} \left(\frac{1}{M}\int_{K} \mathbf{x}_P \: dm \right) = M \frac{d}{dt} ( \mathbf{x}_c ) = M \mathbf{v}_c$$


Why is it possible to write $$\int_{K} \dot{\mathbf{x}}_P dm = \frac{d}{dt} \left(\int_{K} \mathbf{x}_P dm \right)~?$$
Is it due to the Leibniz Integral Rule or due to the Reynolds Transport Theorem for a constant volume $K$?

  • $\begingroup$ Isn't the latter just a generalization of the former? $\endgroup$ – probably_someone Aug 3 '17 at 8:25
  • $\begingroup$ @probably_someone As I understand it, it is. However, I think that it is not possible to apply Leibniz Rule for each dimension separately, as the integral constants depend on each other, since a body is in all generality not a cuboid. $\endgroup$ – Discbrake Aug 3 '17 at 8:38
  • $\begingroup$ An integral is a linear operator, meaning that $$\frac{\partial}{\partial t} \int f(t,x) {\rm d}x = \int \frac{\partial}{\partial t} f(t,x)\,{\rm d}x$$ $\endgroup$ – ja72 Aug 4 '17 at 21:29
  • $\begingroup$ @ja72 That is only true for constant integration limits, which is not the case for a moving body. $\endgroup$ – Discbrake Aug 5 '17 at 16:11

$$\frac{d}{dt}\int_{K(t)} x_P(t) \: dm = \lim_{dt \: \rightarrow \: 0} \left( \frac{1}{dt}\int_{K(t+dt)} x_P(t+dt) \: dm - \int_{K(t)} x_P(t) \: dm \right)$$

The mass $K(t+dt)$ has changed during $dt$ only at the surface $\partial K$. The contribution is $\int_{\partial K(t)} d\sigma dS$ with the surface element $dS$ and the change of the mass per surface element $d\sigma$. But since no mass does flows out, it holds $K(t+dt) = K(t)$ and the only contribution left in above equation is the differential of $x_P(t)$.

  • $\begingroup$ Could one also use the Reynolds Transport Theory for a 'more formal' derivation? $\endgroup$ – Discbrake Aug 12 '17 at 16:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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