arXiv:1012.3091v1 [physics.

class-ph] 14 Dec

Report no. pi-other-204

An exercise in thermo-dynamics
PierGianLuca Porta Mana
Perimeter Institute for Theoretical Physics, Canada <lmana@perimeterinstitute.ca> 12 December 2010 (rst drafted 29 October 2010)

Abstract: An exercise in classical thermodynamics requiring an explicit treatment of the time variable is presented, solved, and further explored; a particular solution is found to be supported by experimental evidence. The exercise is discussed to argue in favour of an explicit treatment of time in classical thermodynamics as usually taught in undergraduate and graduate courses.
PACS numbers: 05.70.-a,64.10.+h,01.40.gb MSC numbers: 80-01,97M50,34-04

to Laura
i mean if you dont change in 2 years or even in a day, then whats the point? L. Pasichnyk

1 An exercise in thermo-dynamics
A vessel that can be idealized as a hollow tube of cross-section A is lled with a mass M of an ideal gas as in g. 1. The tube is closed at one end by a piston of mass m that can move freely without friction. A constant force f acts on the piston from the outside (the force can e.g. consist in atmospheric pressure, and also in the weight of the piston itself if the tube is in a vertical position). Until time t0 the piston is at a distance L from the other base of the tube and 1
This document is optimized for on-screen reading and 2-pages-on-1-sheet printing on A4 or letter paper

Porta Mana

An exercise in thermo-dynamics

the gas is in equilibrium. Starting at time t0 heating is suddenly provided thereafter at a rate (t). How long does it take for the piston to increase its distance by an amount ? How do pressure and temperature change during that same time? The gas is characterized by a massic gas constant R (i.e. R = Ri / with Ri the ideal-gas constant and the molar mass of the gas), a linear relation between internal energy and temperature with massic heat capacity c, and it is homogeneous at all times. Assume, approximately, L = 20 m, A = 10
4

(1a)
2

m ,
3

(1b) kg,
4

M = 2 10 f = 9 N,

(1c) (1e) (1f) (1g) (1h)

corresponding to a pressure f / A = 9 10 Pa, (1d)

2

m = 102 kg, R = 3 10 J/(kg K), c = 7 10 J/(kg K),

2

= 5 10 (t) =
with

m,

and

f ( c + R) 2cL m m 2 (8c + 6 R) ( t t0 )3 + ( t t0 )2 + , (1i) 243 R 6 9 R 3 9 R 3 (1j)

= 1 s.

(If the value of L appears impractically large, consider that this is an idealization and the real set-up can be as in the dotted rectangle of g. 3.)

Figure 1: Idealized experimental set-up for the exercise (for the real set-up see g. 3)

Porta Mana

An exercise in thermo-dynamics

2 Solution
Let us denote the distance of the piston from the tubes base at time t by h(t), and the gas pressure, temperature, and internal energy by p(t), T (t), U (t). The equations relevant to the problem are: Momentum balance for the piston: (t) = 0, Ap(t) f mh where a dot indicates derivation with respect to time. Energy balance for the gas: (t), (t) = (t) p(t) Ah U (3) (2)

since the volume is obviously given by Ah(t) and A is constant. Note (t), where Q(t) is the heat received by the gas from that (t) = Q some arbitrarily specied time up to time t. Constitutive law for the pressure of an ideal gas: p(t) = MRT (t) . Ah(t) (4)

And nally constitutive equation for the gas internal energy, as specied in the problem: U (t) = McT (t). (5) (If some of these equations look unfamiliar to you, take a look at 4.1.) The equations above constitute a system of coupled ordinary differential equations for the unknown h, T , p, U . The system can be easily simplied. Substituting the expression for the pressure (4) in the momentum balance (2) we can express T in terms of h: T (t) = 1 (t)], h(t)[ f + mh MR (6a)

from which, substituting in (4) and (5), we also obtain an expression for the pressure: 1 (t)], p(t) = [ f + mh (6b) A and the internal energy: U (t) = c (t)]. h(t)[ f + mh R 3 (6c)

Porta Mana

An exercise in thermo-dynamics

From the latter two equations and the energy-balance (3) we nally obtain a differential equation of third order for h:

(t) = 1 +

c (t), (t)] + cm h(t)h h(t)[ f + mh R R

(7)

where is given. This differential equation must be supplemented by appropriate , and h . The exercise says that until t0 the initial conditions for h, h piston has been at rest at a distance L and the gas in equilibrium. This means h ( t0 ) = L , (t0 ) = 0, h (t0 ) = 0. h (8)

The differential equation above looks complicated, and it does not seem to be solvable analytically. Seeing however that it is of third order, and that (t) in eq. (1i) is a polynomial of third order in (t t0 ), we can try a polynomial solution of the form h(t) =

i =0

a i ( t t0 ) i .

(9)

The initial conditions (8) impose a0 = L, a1 = 0, a2 = 0. Substituting the above expression for h and that for (1i) in the differential equation (7) we obtain a non-linear system of equations (that I let the readers write for themselves) for the remaining coefcients a3 , . . . , a6 . The system seems to be over-determined since it has ten equations for four unknowns, but miraculously we nd that it can be solved with a3 = /(3 )3 and the remaining coefcients equal to zero. Thus the solution is ( t t0 )3 . (10) h(t) = L + 27 3 The solutions for pressure and temperature are f 2 m + ( t t0 ) , (11a) A 9 A 3 f 2 Lm fL ( t t0 ) + ( t t0 )3 + + T (t) = MR 9 MR 3 27 MR 3 2 2 m ( t t0 )4 ; 243 MR 6 (11b) p(t) = 4

Porta Mana

An exercise in thermo-dynamics

The answer to the rst question of the exercise is that the piston increases its distance by in a time 3 after t0 . With the numerical values (1), h(t0 + t) h(t0 ) = 0.5 m at t = 3 s. (12) Also, f 2 m = 90 033 Pa, + A 3 A 2 ( L + )(2 m + 3 f 2 ) T ( t 0 + 3 ) = T ( t 0 + 3 s ) = = 308 K; 3 MR 2 p ( t 0 + 3 ) = p ( t 0 + 3 s ) = compare with the initial values p(t0 ) = f / A = 90 000 Pa, T ( t0 ) = fL = 300 K. MR (13c) (13d) (13a) (13b)

These solutions are shown in g. 2 for the numerical values (1). With those values, is dominated by its quadratic term in time, T by its constant and cubic terms.

3 Exploring the exercise

Let us examine the equilibrium conditions of the gas for t t0 , before the heating is provided. From eqs (6) we nd the expected equilibrium values for pressure, temperature, and energy: p(t) = f , A T (t) = Lf , MR U (t) = cfL , R t t0 . (14a)

These values are in fact valid at any time t in which the piston is (t) = 0, instantaneously in uniform motion, i.e. not accelerating, h even if it has a non-zero velocity. And when the piston is accelerating? If you think that the term in eqs (6) represent a deviation from equilibrium, consider a situmh ation in which the piston is at rest in relation to the tube, but the whole system is subject to a constant acceleration a in the direction of decreasing h. An inertial force ma acts on the piston in addition to f , and the pressure of the gas has to compensate the sum of the two forces. The inertial force also acts on the gas, but it is negligible if Ma/( pA) is small just as the force of gravity on a gas is usually 5

Porta Mana

An exercise in thermo-dynamics

Q W 15

10

t 0.5 h dm 205 204 203 202 201 t 0.5 p Pa 90 030 90 025 90 020 90 015 90 010 90 005 t 0.5 T K 1.0 1.5 2.0 2.5 3.0 s 1.0 1.5 2.0 2.5 3.0 s 1.0 1.5 2.0 2.5 3.0 s

t0

t0

t0

306

304

302 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0

Figure 2: Solution for h(t) = L + ues (1) 6

( t t0 ) 3 27 3

and the numerical val-

Porta Mana

An exercise in thermo-dynamics

neglected in exercises similar to the present one. From this point of does not represent a deviation from instantaneous view, the term mh equilibrium. I shall comment later on the term equilibrium. Let us explore the behaviour of the system for other forms of the heating and for other initial conditions. The particular expression for (t) given in the exercise was obviously chosen so that we could arrive at the solution through an educated guess. In general the differential equation (7) for h given would be much more difcult to solve. I did not nd an analytic solution for h given a generic , and I suspect there is no solution in closed form. But numerical solutions are easily obtained when , the initial conditions, and the physical constants of the problem are numerically given. We can also turn the question topsy-turvy and consider h as assigned and as the unknown. is then easily calculated with eq. (7). In the following formulae we always intend t > t0 unless stated otherwise. In the solution of the problem h is monotonically increasing. Let us now consider an oscillatory motion instead, like h(t) = L + sin 2( t t0 ) , (15)

with amplitude = 0.5 m and period 3 s. Note that this ( t 0 ) = 2 / , motion corresponds to the initial conditions h(t0 ) = L, h (t0 ) = 0. From eq. (7) we nd that such a motion would be obtained h by means of a heating

(t) = 2

f (c + R) 2 4cLm 2( t t0 ) cos 3 R 2c + R 4( t t0 ) , 4 2 m sin R 3 (16a)

which is also having an oscillatory behaviour with amplitude around 30 W. Pressure and temperature f 2( t t0 ) 4 m sin , (16b) A A 2 Lf 4 2 m f 2 4 Lm 2( t t0 ) 2( t t0 ) + T (t) = + sin sin2 2 2 MR MR MR (16c) p(t) = 7

Porta Mana

An exercise in thermo-dynamics

oscillate with amplitudes around 200 Pa and 7 K. These solutions are shown in g. 4 for the numerical values (1). is dominated by its cosine term, T by its constant and sine terms. Now let us not cheat anymore: let us choose some reasonable heating functions and initial conditions for the position h and nd, by numerical integration of (7), how the latter changes in time. We shall examine ve cases.

3.1 First case: no heating, initial equilibrium

When (t) = 0 it is reassuring to nd numerically that if the piston is at rest and the gas in equilibrium until t0 , they also remain so thereafter, as expected.

3.2 Second case: constant heating

The heating is constant and equal to (t) = 5 W; the piston is at rest until t0 . Numerical integration of eq. (7) gives the solution of g. 5: the distance h increases at an average rate of around 1.7 dm/s. The rate is not strictly constant: we observe a superimposed oscillatory motion of about 0.8 s period and 1 cm amplitude; the net result is periodically vanishes, without becoming negative, so that h inthat h creases in steps. The temperature has an analogous behaviour, with an average increase rate of 2.5 K/s. This stepwise increase is reected in the oscillatory behaviour of the pressure around a seemingly constant value.

3.3 Third case: heating for a short time

We now heat the gas in a continuous way for = 1 s, with a heating of the form

(t) =

10 W sin 0 W,

( t t0 ) ,

t0 t

+ t0 , t0 t + t0 .
t

(17)

The solution is shown in g. 6. While the heating is nonzero the distance h increases and then begins to decrease, but maintains an oscillatory motion around a new value after the heating has vanished. The same behaviour can be observed in the temperature and pressure. It is interesting to note that the period of the oscillations 8

Porta Mana

An exercise in thermo-dynamics

o is apparently the same as in the previous case, around 0.8 s. This must be an intrinsic parameter of the system. Indeed we show in 3.6 that this period is approximately given, for times near t0 , by

0.784 s. (If the remaining undamped oscillatory behaviour surprises you or seems unphysical, remember that the piston is assumed to move with no friction and the gas is inviscid.)

o = 2

cmL ( c+ R) f

3.4 Fourth case: heating for the duration of one intrinsic period
The solution of the previous case suggest that if we heat the gas for a duration equal to the observed intrinsic oscillation period o , we may make the distance h increase and then remain stationary at a new value without residual oscillations. So for our fourth case let us choose 20 W, t0 t t0 + o , (t) = (18) 0 W, t t 0 t o , with o given by (22). We almost obtain the guessed behaviour indeed, as shown in g. 7.

3.5 Fifth case: no heating, initial acceleration

What happens if at t0 the piston is at rest but has some acceleration? Now we assume that h has an initial negative acceleration of (t0 ) = 10 m/s2 (as in free fall), and vanishing initial velocity h (t0 ) = 0. This would be the case if the tube was placed vertically h and we suddenly dropped the piston in it at time t0 . We also assume an adiabatic process, (t) = 0. In these conditions we observe an oscillatory behaviour of all quantities, again with the same intrinsic period, as shown in g. 8. One can show numerically that changing the initial acceleration only changes the amplitudes of the oscillations but not their period. This last case can be experimentally observed with an AssmannRchardt apparatus [1; 2] as modied by Schue [3], shown in g. 3. With that apparatus, the oscillatory behaviour we found is used to calculate the ratio of specic heats of the gas, which in our case is given by = 1 + R/c, by measuring the period o approximately given by eq. (22). The values (1) are indeed approximately taken from 9

Porta Mana

An exercise in thermo-dynamics

Schues article, and our solution gives with good approximation the period of oscillation observed in his experiments.

3.6 Approximations
As already said, the differential equation governing our problem, c (t), (t)] + cm h(t)h h(t)[ f + mh R R

(t) = 1 +

(7)r

does not seem to be amenable to a closed-form solution. But if the (t) is small in comparison to the force f , and if the inertial term mh variations of h(t) are small in comparison to its initial value L, i.e. (t) mh h(t) L and small, f L the above equation can approximated by (19)

(t) = 1 +

c (t). (t) + cmL h fh R R

(20)

Integrating between t0 and t and dening (t) := h(t) h(t0 ) cmL h ( t0 ) h c+R we obtain Q(t) = 1 + (21a)

c (t) + cmL h (t), fh (21b) R R where Q(t) is the total heat received by the gas between t0 and t. The last equation describes a driven harmonic motion and can be solved exactly in terms of integrals of Q. The intrinsic period is

o = 2

cmL 0.784 s. ( c + R) f

(22)

The solutions previously obtained numerically do reasonably satisfy the conditions (19) and one can compare the numerical results with those obtained using eqs (21). For t0 t 3 it is found that 4 the difference does not exceed 2 10 L, viz 4 mm, for all cases except the fth one, 3.5, for which we have around 103 L. It should be remarked that for a smaller volume of gas, i.e. smaller L, or bigger piston mass m, the approximations (19) do not hold and eq. (7) can present even more interesting behaviours to be studied, maybe, in a future memoir. 10

Porta Mana

An exercise in thermo-dynamics

Figure 3: Assmann-Rchardt-Schue apparatus, with which the solution of the fth case, shown in g. 8, can be experimentally observed. The part of the apparatus inside the dotted rectangle is idealized in g. 1; notice the piston in the left tube. The distance L of the exercise is an effective distance given by L = Lr + Vv / A, where Lr is the distance of the piston from the base of the real tube and Vv is the volume of the vessel connected to the latter; the same holds for h. The real tube can be between one and two metres long. (Reprinted and adapted from Schue [3] with the permission of the American Chemical Society)

11

Porta Mana

An exercise in thermo-dynamics

Q W 30 20 10 t 0.5 10 20 30 h dm 204 202 200 198 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0 1.0 1.5 2.0 2.5 3.0 s t0

p Pa 90 200

90 100

90 000

89 900 t 0.5 T K 306 304 302 300 298 296 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0 1.0 1.5 2.0 2.5 3.0 s t0

Figure 4: Solution for h(t) = L + sin values (1) 12

2 ( t t0 )

with the numerical

Porta Mana

An exercise in thermo-dynamics

Q W 10 8 6 4 2 t 0.5 h dm 205 204 203 202 201 t 0.5 p Pa 1.0 1.5 2.0 2.5 3.0 s t0 1.0 1.5 2.0 2.5 3.0 s t0

90 100 90 050 90 000 89 950 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0

T K 307 306 305 304 303 302 301 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0

Figure 5: Solution for (t) = 5 W, piston at rest until t0 , and the numerical values (1) 13

Porta Mana

An exercise in thermo-dynamics

Q W 10 8 6 4 2 t 0.5 h dm 1.0 1.5 2.0 2.5 3.0 s t0

202.0 201.5 201.0 200.5 t 0.5 p Pa 1.0 1.5 2.0 2.5 3.0 s t0

90 100

90 000

89 900 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0

T K 303.0 302.5 302.0 301.5 301.0 300.5 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0

Figure 6: Solution for (t) = 10 W sin 0 in t [t0 , + t0 ], 0 W elsewhere, piston at rest until t0 , and the numerical values (1) 14

( t t )

Porta Mana

An exercise in thermo-dynamics

Q W 20

15

10

5 t 0.5 h dm 205 204 203 202 201 t 0.5 p Pa 1.0 1.5 2.0 2.5 3.0 s t0 1.0 1.5 2.0 2.5 3.0 s t0

90 400 90 200 90 000 89 800 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0

T K 308

306

304

302 t 0.5 1.0 1.5 2.0 2.5 3.0 s t0

Figure 7: Solution for (t) = 20 W in t [0, o ], 0 W elsewhere, piston at rest until t0 , and the numerical values (1) and (22) 15

Porta Mana

An exercise in thermo-dynamics

Q W 1.0

0.5 t 0.5 0.5 1.0 1.5 2.0 2.5 3.0 s t0

1.0

h dm 200.0 199.5 199.0 198.5 198.0 197.5 t 0.5 p Pa 91 000 1.0 1.5 2.0 2.5 3.0 s t0

90 500

90 000

89 500 t 0.5 T K 298.5 1.0 1.5 2.0 2.5 3.0 s t0

298.0

297.5 t 0.5 296.5 1.0 1.5 2.0 2.5 3.0 s t0

Figure 8: Solution for (t) = 0 W, an initial negative acceleration of (t0 ) = 10 m/s2 , and the numerical values (1) h 16

Porta Mana

An exercise in thermo-dynamics

4 Arguing for the time variable in classical thermodynamics

During my undergraduate studies I have never encountered an exercise similar in spirit to the one presented here, in courses of thermodynamics or statistical mechanics. More generally, no thermodynamic exercise that posed a question like in how much time. . . ?, even though that kind of question abounded in courses of mechanics and electromagnetism. My experience is not untypical, I think, and it is conrmed by a look at basically any thermodynamics or generalphysics textbook for physicists. The only undergraduate textbook I found in which the thermodynamic quantities are presented with an explicit time-dependence is Sandlers [4, e.g. chs 3, 4] (written for chemical engineers).

4.1 Classical thermodynamics with time

An explicit presence of the time variable in classical thermodynamics amounts simply to this, as was shown in the solution of 2: The independent variables describing the system, say volume V and temperature T , depend on time: V (t), T (t); just like the position of a mass-point depends on time in mechanics. The dependent variables, say pressure p, internal energy U , entropy S, also depend on time through their dependence on the independent variables: p[V (t), T (t)], U [V (t), T (t)], etc. The differentials usually appearing in the rst and second laws are recognized to be differentials with respect to time, and therefore replaced by time derivatives. So the rst and second law assume the forms = P, S . (23) U T are the rates of change of internal energy and S In these equations, U and entropy, is the heating, or rate at which heat is received by the system, and P is the working, or rate at which work is done by the system. The latter is usually specied in terms of the independent and dependent variables of the system and their rates of change, . (The term power is usually reserved for the sum e.g. P = pV of working and increase of kinetic energy, the latter being usually negligible in thermodynamics)

17

Porta Mana

An exercise in thermo-dynamics

This formulation was discussed by Menger [5] and is presented in Truesdell [6, Lecture 26; 7; 8, Lecture 1 and Appendix1A], Truesdell and Bharatha [9], Owen [10], Samohl [11, ch. II], and the already mentioned Sandler [4].

4.2 Advantages
With the solution and analysis of our little exercise of 1 I hope to have shown that time has, instead, its appropriate place in classical thermodynamics and thermodynamic courses. Its explicit presence has many advantages: First, it avoids any mention of inexact differentials and therefore does not require notions of the mathematics of differential forms. Indeed it shows, by eqs (23), why there cannot be differentials related to heating and working: because they are not the rates of change of anything. This consequently allows, as a second advantage, an easier study of problems in which thermal and mechanical phenomena appear together. This kind of problems is often avoided because of the clash between the mathematics typically used in thermodynamics (differentials, no time) and that used in mechanics (explicit time, derivatives), which leads to ugly mathematical gymnics. An example is given in Schues article [3]: he must derive the equation of a harmonic oscillator in a set-up similar to that of our exercise. He uses differentials in deriving the necessary relations between thermodynamic quantities, but then he must combine these with mechanical ones, arriving at an expression with the differential of a force. At that point he has to concede that the force is related to the double timederivative of position, revealing that all preceding differentials were nothing else than time-derivatives in disguise. Moreover, the use of differentials also force him to make approximations to the behaviour of the system. Our solution, 2, was obtained in a mathematically seamless way instead, owing to the presence of time in the basic thermodynamic equations. Third, it makes it possible to study more interesting thermodynamic systems for which the pressure or other quantities depend not only on volume and temperature but also on their rates of change, say ), as it happens e.g. in synovial uids [12; 13] or in pyroelecp(V, T, V tric materials [14]. The mathematics of differentials can barely handle

18

Porta Mana

An exercise in thermo-dynamics

such systems. On this generalization I recommend the very lucid and simply written work of Samohl [11], especially ch. II. Fourth, it is full of physical and mathematical pedagogic potentials, as the interesting third-order ordinary differential equation (7) shows: it leads to a rich class of mathematical problems and physical behaviours (like the driven harmonic oscillator, which is only a special case), providing also pedagogic material for laboratory classes. Cases for the explicit appearance of time in thermodynamics were already made by Eckart [15], by Bridgman [16, see e.g. pp. 2426], extensively and intensively by Menger [5], and very forcefully, with telling historical, physical, and mathematical arguments, by Truesdell and his school of rational thermodynamics (see e.g. [611; 17 22]).

4.3 Common, weak objections

In conversations on this topic with colleagues two main kinds of objection appear against a presentation of thermodynamics with an explicit presence of time. I nd both very weak: The rst objection is that thermodynamics would only be concerned about equilibrium and relations between equilibria and would have nothing to say about changes and rate of changes. I should rather call such a theory thermostatics (cf. Eckart [15]). But why should we limit ourselves to thermostatics when we do not limit ourselves to statics within mechanics? Actually the objection is that there is no proper science of the change of thermal systems thermodynamics can only be thermostatics. I do not give much weight to this view because it is simply contrary to experience, experimental evidence, and practice, as we have seen e.g. in Schues experiment above. According to this view all theories about non-equilibrium thermodynamics, routinely taught and used by engineers and taught in higher physics courses, would be nonsense. That is preposterous. (Some people even say temperature is dened only at equilibrium, contradicting themselves each time they check a weather thermometer). The second objection is that an explicit time dependence and the consequent appearance of differential equations is unwarranted for high-school and undergraduate students in their rst years. But would not classical mechanics present exactly the same problem? Yet, many mechanical problems, like the harmonic oscillator or bal19

Porta Mana

An exercise in thermo-dynamics

listic problems, can be and are routinely approached and solved without the need to study the full theory of differential equations. The same could surely be done with thermodynamics. Moreover, is the use of non-intuitive and typically poorly explained inexact differentials, without explicitly introducing the theory of differential forms, really easier than the use of rates of change?

4.4 Inadequacy of the standard terminology of classical thermodynamics

Part of the inertia against the time variable in thermodynamics comes from the terminology taught in thermodynamic courses. Indeed, even those who are keen on allowing the presence of time in classical thermodynamics do so only after some verbal exorcisms. They say, e.g., that a process is quasi-static or that the thermodynamic system passes through a sequence of equilibrium states, or other similar phraseologies. I nd most of these phraseologies misleading, counter-intuitive, and argue that we should change them. Their inadequacy can be seen from different points of view: First: In mechanics, electromagnetism, quantum theory we use particular terminologies to indicate that specic equations and laws apply to a problem. When we say that a spring is perfectly, linearly elastic, we mean that Hookes law F = k x can be applied. If a body is called rigid, it means that its motion can be fully described by the . When we speak of a Ohmic resistor, equations F = p and M = J we mean that the law V = IR applies with a constant resistance R. Now, when in thermodynamics we say that a process is quasi-static or that the system passes through a sequence of equilibrium states, we simply mean that an equation like p(t)V (t) = nRT (t) can be applied at all times. But while the terms elastic, rigid, Ohmic give the right ideas, quasi-static is misleading and counter-intuitive: in our exercise we found visible changes of more than 2 500 Pa/s and 5 K/s (see g. 8); do we really want to call them quasi-static? Instead of quasi-static I prefer to say that the system is homogeneous (at all times): this implies that it can be described at all times by a single density value, a single pressure value, a single temperature value, etc., instead of eld quantities as in continuum thermomechanics. And obviously there will be some constitutive equations the ideal-gas law being just an example connecting these values and maybe even their rates of change. Note the analogy between a ho20

Porta Mana

An exercise in thermo-dynamics

mogeneous body and a rigid body: rigidity and homogeneity both imply that the eld quantities describing deformations in the full continuum-thermomechanical treatment can be replaced by a smaller set of mechanical or thermal quantities valid for the whole body. As for saying that a visibly changing system is passing through a sequence of equilibrium states, it sounds to me like saying that a moving object is passing through a sequence of states of rest: it is at best an unnecessary and ugly phraseology, at worst simply nonsense. I think that any mention of equilibrium is misleading when thermodynamic quantities are clearly, visibly changing. Let us keep that term, like rest, for situations in which quantities are not changing in time. Again, we mention instantaneous equilibrium only to mean that the ideal-gas equation can be used at all times. But all we should need to say is this: the ideal-gas law is valid, and the gas is not in equilibrium. That is also what the experiment says. Second: The argument typically adduced for the use of the term quasi-static is that the system can be considered static in equilibrium in comparison to some time scale. But all equations used in physics are valid in particular ranges and scales only. If an equation is used in a physical problem it is implied that it makes sense there. If we again use Hookes law F = k x it is understood that the displacement x is within the range of linear elastic behaviour of the spring, away from non-linear or plastic effects for which the law is 1 mv2 with m constant it not valid. If we write a kinetic energy as K = 2 is clear that v shall not be a Lorentz-relativistic velocity. Analogously, if we use p(t)V (t) = nRT (t) it is understood that the system is homogeneous and the ranges and rates of change of p(t), V (t), T (t) are such that the law is valid. Again, there is no need to mention quasistatic processes. Or maybe in Galileian mechanics we should call all kinematics quasi-static because it is necessary that it be almost at rest in comparison with some velocity, the velocity of light. . . Third: It is true that the construction and use of a thermometer or a pressure gauge involves notions like response time and coupling. But so does the construction of position-gauges, velocity-gauges, and dynamometers. Some thermometers are inappropriate when the temperature changes too rapidly, but the same is true of velocity-gauges too. In mechanics and electromagnetics courses the theory of a physical phenomenon and the complex theory of measurement behind it are usually kept quite separate. Positions, accelerations, forces, charges, currents, elds are supposed to be measurable with some 21

Porta Mana

An exercise in thermo-dynamics

appropriate instrument. Springs exactly obeying Hookes law for arbitrary values of elastic constant and elongation are supposed to exist, for the purpose of studying the mathematical structure of the theory. Thus it ought to be in thermodynamics. Fourth: In mechanics we enjoy the liberty to consider abstract situations leaving unmentioned how they can be realized, in order to concentrate us upon the equations and the phenomenology these exhibit. For example we often say that a particle is in a harmonic potential, without further question about how that is achieved (springs? some kind of eld?). Or that there is a uniform and constant electric eld. This liberty is rarely enjoyed in thermodynamics. Yet we should have the right to say, e.g., that the temperature of a system is constant, without needing to mention heath baths all the time. That the temperature is constant, or is changing in some prescribed way, is all we need to know for our problem. Another example: No body is perfectly rigid; the equations for a rigid body make sense only if the forces and torques do not exceed the internal ones that keep the body together. Also, the action of a force on a part of the body will not propagate instantly to the rest of the body: waves will propagate in its interior, and they can be detected at appropriate space-time scales. We do not usually point out all these facts when examining a rigid-body problem. Analogously, no body is perfectly homogeneous, and its equations make sense only when changes happen on a smaller scale than the velocity of sound in the body; these changes propagate from point to point and could be detected at appropriate space-time scales. We should feel the need to point out these facts neither more nor less than we do in the case of the rigid body. Truesdell makes these points very clearly in his typical sharp and witty style: In mechanics, questions about what force and mass really are and how they are to be measured are important, but they are rarely mentioned in books on the mathematical theory. Persons who wish to study them are referred to works on metaphysics, the philosophicalempirical foundations of physics, or experimental procedures, according to taste. With thermodynamics it is a different matter. Whenever anyone tries to do anything rational there, no matter what branch or application he takes up, he may be sure to be pestered by professionals who rise and demand he explain how to dene temperature and entropy for systems not innitesimally near to equilibrium. The 22

Porta Mana

An exercise in thermo-dynamics

persecutor himself, obviously, regards equilibrium and innitesimally near as primitive concepts [8, Lecture 1, pp. 7879]. For ten years or more, everyone studying materials has seen that both thermodynamics and mechanics must be brought to bear, but these sciences as hitherto taught do not mesh. Thus, if we take up almost any recent book with continuum mechanics or material science in its title, we nd a chapter on thermodynamics, but that chapter presents a curious contrast with the same authors pages, earlier in the book, on pure mechanics. There the reader is faced by mappings, elds of vectors and tensors, Jacobians, differential invariants, perhaps even Christoffel symbols and afne connections; of course he is presumed familiar with calculus as calculus has been taught for the last fty years. He can understand dynamical equations in tensorial form: div T + b = x or k . T km ,m + bk = x

He is informed, in each case, what the dependent and independent variables are, he is presented with explicit differential equations and boundary-value problems, he is shown many special solutions in concrete cases and is directed to grand tomes where he can nd thousands more such solutions not given in the book he is reading, and often he is told about some major problems still unsolved and is challenged to solve them himself. The same reader of the same book then reaches the chapter on thermodynamics, where he is faced with the axiom T dS Q. He is told that dS is a differential, but not of what variables S is a function; that Q is a small quantity not generally a differential; he is expected to believe not only that one differential can be bigger than another, but even that a differential can be bigger than something which is not a differential. He is loaded with an arsenal of words like piston, boiler, condenser, heat bath, reservoir, ideal engine, perfect gas, quasi-static, cyclic, nearly in equilibrium, isolated, universe words indeed familiar in everyday life, doubtless much more familiar than tangent plane and gradient and tensor, which he learned to use accurately and uently in the earlier chapters, but words that never nd a place in the mathematical structure at all, words the poor student of science is expected to learn to hurl for the rest of his life in a rhetoric little sharper than a housewifes in the grocery 23

Porta Mana

An exercise in thermo-dynamics

store. The mathematical structure, in turn, is slight. There are no general equations to be solved, no boundary-value or initial-value problems set, no general theorems characterizing classes of solutions. The examples or exercises require no more than calculating partial derivatives or integrals of given functions or their inverses and plugging numbers into the results. The references cited lead to other books containing just the same material, perhaps otherwise explained and ordered, but no broader or clearer in concept, and equally unmathematical. No problems, in the sense that the word problem has in the theories of mechanics or electromagnetism or optics or heat conduction, are solved. Neither are any open problems stated. The reader must presume that thermodynamics is an exhausted as well as exhausting subject, with nothing left to be done [8, Lecture 1, pp. 6061].

Acknowledgements
Many thanks to Louise, Marianna, Miriam per i loro continui supporto e affetto; to Ingemar Bengtsson and Giacomo Mauro DAriano for discussions; to Lucien Hardy and Lee Smolin for encouragement; to swing music; to R. Cheese; and to the developers and maintainAT X, Emacs, AUCT X, MiKT X, arXiv, Inkscape. Research at ers of L E E E the Perimeter Institute is supported by the Government of Canada through Industry Canada and by the Province of Ontario through the Ministry of Research and Innovation.

References
Note: Of two years separated by a virgule, the rst is that of original publication or composition. [1] C. Assmann: Ueber Erwrmung und Erkaltung von Gasen durch pltzliche Volumnderung. (Poggendorffs) Ann. d. Phys. u. Chem. 85/1 (1852), 136. E. Rchardt: Eine einfache Methode zur Bistimmung von c p /cv . Physik. Zeitschr. 30 (1929), 5859. J. A. Schue: Measuring the ratio of specic heats of a gas: Thermodynamic experiments involving harmonic motion. J. Chem. Educ. 34/2 (1957), 78 80.

[2] [3]

24

Porta Mana

An exercise in thermo-dynamics

[4] [5] [6] [7] [8] [9]

S. I. Sandler: Chemical, Biochemical, and Engineering Thermodynamics, 4th ed. (New York: John Wiley & Sons, 1977/2006). First publ. 1977. K. Menger: The mathematics of elementary thermodynamics. Am. J. Phys. 18/2 (1950), 89103. See also Thomsen [23]; Menger [24]. C. A. Truesdell III: The Elements of Continuum Mechanics Springer-Verlag, 1966). (Berlin:

C. A. Truesdell III: Thermodynamics for beginners. In: Parkus and Sedov [25] (1968), pp. 373389. With discussion. C. A. Truesdell III: Rational Thermodynamics, 2nd ed. (New York: Springer-Verlag, 1969/1984). First publ. 1969. C. A. Truesdell III and S. Bharatha: The Concepts and Logic of Classical Thermodynamics as a Theory of Heat Engines: Rigorously Constructed upon the Foundation Laid by S. Carnot and F. Reech (New York: SpringerVerlag, 1977). D. R. Owen: A First Course in the Mathematical Foundations of Thermodynamics (New York: Springer-Verlag, 1984). I. Samohl: Thermodynamics of Irreversible Processes in Fluid Mixtures (Approached by Rational Thermodynamics) (Leipzig: BSB B. G. Teubner Verlagsgesellschaft, 1987). A. D. Knight and J. R. Levick: Time-dependence of the pressure-volume relationship in the synovial cavity of the rabbit knee. J. Physiol. 335/1 (1983), 139152. J. Hardy, A. L. Bertone, and W. W. Muir: Pressure-volume relationships in equine midcarpal joint. J. Appl. Phys. 78/5 (1995), 19771984. A. G. Chynoweth: Dynamic Method for Measuring the Pyroelectric Effect with Special Reference to Barium Titanate. J. Appl. Phys. 27/1 (1956), 78 84. C. Eckart: The thermodynamics of irreversible processes. I. The simple uid. Phys. Rev. 58/3 (1940), 267269. P. W. Bridgman: The Nature of Thermodynamics (Cambridge, USA: Harvard University Press, 1941/1943). First publ. 1941. C. A. Truesdell III and R. A. Toupin: The Classical Field Theories. In: Flgge [26] (1960), pp. IVII, 226858, 859902. With an appendix on invariants by Jerald LaVerne Ericksen. C. A. Truesdell III and W. Noll: The Non-Linear Field Theories of Mechanics, 3rd ed. (Berlin: Springer-Verlag, 1965/2004). Ed. by Stuart S. Antman; rst publ. 1965. C. A. Truesdell III: The Tragicomical History of Thermodynamics 1822 1854 (New York: Springer-Verlag, 1980).

[10] [11]

[12]

[13] [14]

[15] [16] [17]

[18]

[19]

25

Porta Mana

An exercise in thermo-dynamics

[20] [21] [22] [23] [24] [25]

W. A. Day: A Commentary on Thermodynamics (New York: SpringerVerlag, 1988). J. L. Ericksen: Introduction to the Thermodynamics of Solids, Revised ed. (New York: Springer, 1991/1998). First publ. 1991. M. ilhav: The Mechanics and Thermodynamics of Continuous Media (Berlin: Springer-Verlag, 1997). J. S. Thomsen: [Comment:] The mathematics of elementary thermodynamics. Am. J. Phys. 19/8 (1951), 476. See Menger [5] and also Menger [24]. K. Menger: [Reply:] The mathematics of elementary thermodynamics. Am. J. Phys. 19/8 (1951), 476477. See Menger [5]; Thomsen [23]. H. Parkus and L. I. Sedov, eds.: Irreversible Aspects of Continuum Mechanics and Transfer of Physical Characteristics in Moving Fluids: Symposia Vienna, June 2228, 1966 (Wien: Springer-Verlag, 1968). S. Flgge, ed.: Handbuch der Physik: Band III/1: Prinzipien der klassischen Mechanik und Feldtheorie [Encyclopedia of Physics: Vol. III/1: Principles of Classical Mechanics and Field Theory] (Berlin: Springer-Verlag, 1960).

[26]

26