Suppose a container separating two gases via a movable massless piston. If the pressure of gas 1 is higher than that of gas 2, the piston will move. In an infinitesimal volume change, the energy lost by gas 1 will be P1.dV and that gained by gas 2 will be P2.dV. As P1 is higher than P2, where is the extra energy going?

Note: no heat exchange is considered. The only change in energy is caused by work.


Let the system $1$ be one gas, the system $2$ be the other gas, and $P$ be the piston. The process is short enough to consider it adiabatic, so the first principle of thermodynamics gives

$$ \textrm{d}E_p + \textrm{d}U_1 + \textrm{d}U_2 = \delta W_{ext} $$

where $E_p$ is the mechanical energy of the piston and $W_{ext}$ the work done by the operator. Indeed, $\textrm{d}U_p = 0$ since the piston is a rigid body and there is no heat transfer, and $\textrm{d}E_1 = \textrm{d}E_2 = 0$ in the first order since the transformation is infinitesimal.

However, the process can be quasi-static, so $\textrm{d}E_p = 0$, so we got

$$ \textrm{d}U_1 + \textrm{d}U_2 = \delta W_{ext}$$

Here, you made the following mistake: you said that $\delta W_{ext} = 0$, but this is physically impossible. Indeed, $P_1 \neq P_2$, so the piston won't evolve smoothly at all if there's no external force applied on it. Thus, if the process is quasi-static, then there is an external force, so

$$ \textrm{d}U_1 + \textrm{d}U_2 \neq 0 $$

But we can also say that the operator doesn't touch the device. Then, $\delta W_{ext} = 0$, however the kinetic energy of the piston have changed during the process, so $\textrm{d}E_p \neq 0$, which gives

$$ \textrm{d}U_1 + \textrm{d}U_2 \neq 0$$

In both cases, at first sight the conservation of energy is violated, however once you've taken the piston into account, there's no problem anymore.

  • $\begingroup$ Thank you for your elaborate answer. But if the piston is massless, then dEp will also be 0. What happens then? $\endgroup$ – Bilal Khalid Dec 20 '16 at 11:16
  • 1
    $\begingroup$ If the piston is massless, and the operator doesn't touch it, then there is a non null force acting on it, so it will have an infinite acceleration: $P_1$ and $P_2$ aren't even defined, and the process isn't infinitesimal! $\endgroup$ – Spirine Dec 20 '16 at 11:30
  • $\begingroup$ A couple of questions. Here's the first: I gather that "evolve smoothly" just means "move," but I don't understand the statement "P 1 ≠P 2 , so the piston won't evolve smoothly at all if there's no external force applied on it. " The difference in pressure gives rise to a net force that will move the piston, just as the OP assumed, does it not? $\endgroup$ – D. Ennis Dec 21 '16 at 11:30
  • 1
    $\begingroup$ You say that "... the kinetic energy of the piston have changed during the process, so dE p ≠0." But how does this answer the question of where "the extra energy" goes if the piston stops moving at the end of the process? $\endgroup$ – D. Ennis Dec 21 '16 at 11:36
  • $\begingroup$ @D. Ennis By "evolves smoothly", I meant "moves in a quasi-static way", so only moves a little bit and really slowly. For your second question: the point is that the piston cannot stand still at the end of the infinitesimal process $\endgroup$ – Spirine Dec 21 '16 at 12:39

The correct answer to your question is simply that the changes in energy are not $P_1dV$ on one side and $P_2dV$ on the other side.   The changes in energy are $-(P_1-P_2)dV$ on the high pressure side and $(P_1-P_2)dV$ on the low pressure side.   There is no "extra energy" to account for.

Explanation:     $dV$ is equal to $Adx$, where $A$ is the area of the piston and $dx$ is its displacement.   The force on the high pressure side of the piston is $P_1A$, and the force on the low pressure side is $-P_2A$.   The net force is the sum of the two, $(P_1-P_2)A$, and the energy transferred, or work done, is the product of this net force and the displacement $dx$.   The work is done by the high pressure chamber on the low pressure chamber, so it is negative for the former and positive for the latter.


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.