The piston problem a worked example Consider a system of N moles of an ideal gas with an initial volume V0 and energy

y E0. A piston does work on the system to change the volume to a new value V. Lets see what happens under various conditions, and how this relates to the general statements about the first and second laws. Note that we can take ideal gas equations of state for pressure and energy: p= NRT , V E= NRT ,

where is a constant: 3/2 for a monatomic gas and between 5/2 and 3 for a diatomic gas. The constant volume heat capacity is given by C v = NR. A. Lets first consider reversible processes A1. Adiabatic. This is a classic thermo problem you should know how to do this. Since its reversible and adiabatic, dS=0, and dE= p dV. Substituting the equations of state, we can get a differential equation for the temperature, dV NR dT =NRT V. Note 1: Temperature is not an independent variable in this problem, neither is pressure. They can be derived from equations of state given the thermodynamic variables N, V, E. Solving the equation and inserting the initial conditions, V0, and T0 (N, E0),

V T = 0 T0 V.

Note 2: The statement dS = 0 does not violate postulate 3 (P3 for short) since V is changing and P3 applies to the case of fixed X.
Finally we get an expression for the pressure PV =P0 V 0 , where =11/ and P0=NRT 0 /V 0 is the initial pressure.

Note 3: From any given initial state only a subset of states can be accessed by an adiabatic change of volume. This pressure-volume relation is called an adiabat. A2. Isothermal. This is easier since T (now an independent variable) is fixed. p Since its reversible and isothermal, dE=0, and dS= dV. T Note that the relation dE = 0 is not fundamental it follows from the EOS and dT = 0. Its easy to solve for S: S=S 0NR ln V /V 0 , and S increases during an isothermal expansion.

B. Irreversible adiabatic process. Here is an approximate calculation to illustrate what happens. Let the volume be changing linearly with time: V t =V 0 1rt . It can be expanding (r > 0) or contracting (r < 0). We know that heat is produced during irreversible volume changes of either sign. This suggests that Q(t) ~ r2t, which is in fact correct. The heat is being generated by friction between the moving fluid and the container walls. Lets take the following hypothetical but reasonable form for dQ: dQ=NRT r 2 dt . The reasoning is that Q must also depend linearly on N and we include RT to get an energy. The parameter is a time scale, which is needed to give the whole thing the correct units. When r is small the system will behave reversibly. Note dQ is not following from a thermodynamic state function. The first law is in this case is dE=dQ p ext dV. Lets assume the pressure is roughly balanced so that pext = p. Then we can again get a differential equation for T, in this case as a function of time: NR dT =NRT r 2 dt

r dt , 1rt

where we have used the relation V t =V 0 1rt to convert the dV/V term. Solving and inserting the initial conditions: V V 0 T V =r ln , T0 V0 V0 where we have again used the time dependence of the volume to convert back from t to V. ln

I am not showing all the algebra here: just the main steps. Its important that you work through all the steps yourself. One thing to note is that as r 0 we recover the reversible result. Finally we get the connection between T and V in the irreversible case:

V V V 0 T = 0 exp r . T0 V V0

Note that T is always higher in the irreversible case r0 , regardless of expansion or contraction. If r > 0 then V V0 is positive and vice versa. The final (equilibrium) state along an irreversible path to the same V then has a higher entropy than the initial state (P4) because the energy is higher (at the same volume) and therefore the entropy must be larger too (P3).