J. Masek, 3.1.

2005
Entropy production and diabatic term
for single component uid
Aim of this article is to derive closed set of equations describing single component uid,
starting from basic principles of mechanics and phenomenological thermodynamics.
1 Notations
Shorthand notation for partial derivatives:
,i

i


x
i
Einstein summation convention is used, i.e. twice repeated indexes are summed over
automatically. For example:
u
i,i

3

i=1
u
i,i

ijk
x
j
u
k

3

j=1
3

k=1

ijk
x
j
u
k
Other symbols:
c
p
. . . specic heat at constant pressure
c
v
. . . specic heat at constant volume
d . . . 3D divergence (d d
ii
= u
i,i
= u)
d
ij
. . . deformation rate tensor
_
d
ij
u
(i,j)

1
2
(u
i,j
+ u
j,i
)
_
f . . . external volume force per unit mass
J
heat
. . . heat ux
p . . . pressure
Q . . . diabatic term
_
Q T
ds
dt
_
R . . . gas constant
s . . . specic entropy
T . . . thermodynamic temperature
t . . . time
u . . . velocity
_
u
i

dx
i
dt
_
v . . . specic volume
_
v
1

_
x
i
. . . euclidean coordinates (i = 1, 2, 3)

ij
. . . Kronecker symbol
. . . specic internal energy

ijk
. . . completely antisymmetric symbol
. . . bulk viscosity
. . . shear viscosity
. . . heat conductivity
. . . density

s
. . . volume density of entropy production rate

ij
. . . stress tensor

visc
ij
. . . viscous part of stress tensor
. . . control volume
. . . boundary of control volume
d
dt
. . . time derivative along uid trajectories
_
d
dt


t
+u
_
1
2 Budget equations - integral form
Budget equations in integral form will be formulated for xed control volume placed
inside the uid. Volume does not move and its boundary is opened, so the uid
can cross it freely.
Mass budget:
Change of mass in volume equals to advective ux of mass through the
boundary .

t
_

dV =
_

u
i
dS
i
(1)
Momentum budget:
Change of momentum in volume equals to advective ux of momentum
through the boundary plus net external force acting on uid inside .
Net external force consists of two contributions: net surface force and net
volume force.

t
_

u
i
dV =
_

u
i
u
j
dS
j
+
_

ij
dS
j
+
_

f
i
dV (2)
Angular momentum budget:
Change of angular momentum in volume equals to advective ux of angular
momentum through the boundary plus net torque of external forces
acting on uid inside .

t
_

ijk
x
j
u
k
dV =
_

ijk
x
j
u
k
u
l
dS
l
+
+
_

ijk
x
j

kl
dS
l
+
_

ijk
x
j
f
k
dV (3)
Given formulation of angular momentum budget is equivalent to the assumption that
net torque of internal forces is zero. This assumption holds exactly for monoatomic gas,
but might not be true for more complicated molecules with non-zero dipole momentum.
If there is internal source of angular momentum, uid is called polar.
2
Total energy budget:
Change of total energy in volume equals to advective ux of total energy
through the boundary plus heat ux across the boundary plus work
of external forces acting on uid inside . Work of external forces consists
of two contributions: work of surface forces and work of volume forces.

t
_

_
+
1
2
u
2
_
dV =
_

_
+
1
2
u
2
_
u
i
dS
i

J
heat
i
dS
i
+
+
_

u
i

ij
dS
j
+
_

f
i
u
i
dV (4)
Entropy budget:
Change of entropy in volume equals to advective ux of entropy through
the boundary plus entropy ux due to heat exchange plus entropy
generation caused by irreversible processes.

t
_

s dV =
_

su
i
dS
i

1
T
J
heat
i
dS
i
+
_

s
dV (5)
Second law of thermodynamics requires
s
0. Equality holds for reversible processes.
3 Budget equations - dierential form
Budget equations are valid locally, i.e. they apply to arbitrary control volume . This
enables to convert integral relations (1)(5) into dierential form. Procedure will be
illustrated for mass budget. Interchanging time derivative and integral on LHS of
equation (1) and applying Gauss-Ostrogradski theorem to its RHS gives:
_

t
+ (u)
_
dV = 0
Since control volume can be arbitrary, subintegral term must vanish, leading to usual
forms of continuity equation:

t
+ (u) = 0
d
dt
+ u = 0 (6)
Using rst form of continuity equation (6), auxiliary formula valid for arbitrary
quantity can be obtained:

t
_

dV +
_

u
i
dS
i
=
_

d
dt
dV
3
It is advantageous to use this formula during transformation of remaining budget
equations. Application of outlined procedure is straightforward, leading to dierential
form of budget equations (2)(5):

du
i
dt
=
ij,j
+ f
i
(7)

ijk
d
dt
(x
j
u
k
) =
ijk
(x
j

kl
)
,l
+
ijk
x
j
f
k
(8)

d
dt
_
+
1
2
u
2
_
= J
heat
+ (u
i

ij
)
,j
+ f
i
u
i
(9)

ds
dt
=
_
1
T
J
heat
_
+
s
(10)
Equations (8) and (9) can be further simplied, using momentum equation (7).
Angular momentum equation (8) will be examined rst:

ijk
u
j
u
k
. .
0
+
ijk
x
j
du
k
dt
=
ijk

jl

kl
+
ijk
x
j

kl,l
+
ijk
x
j
f
k

ijk
x
j
_

du
k
dt

kl,l
f
k
_
. .
0
=
ijk

kj

ijk

kj
= 0

ij
=
ji
(11)
Equation (11) means that for non-polar uids stress tensor must be symmetric. It has
therefore only 6 independent components. Using momentum equation (7) and symmetry
of stress tensor (11), total energy equation (9) can be rearranged into internal energy
equation:

d
dt
+ u
i
du
i
dt
= J
heat
+ u
i,j

ij
+ u
i

ij,j
+ f
i
u
i

d
dt
+ u
i
_

du
i
dt

ij,j
f
i
_
. .
0
= J
heat
+ u
i,j

ij

ij
=
ji
= u
i,j

ij
= u
(i,j)

ij
= d
ij

ij

d
dt
= J
heat
+ d
ij

ij
(12)
4 Closing the system - constitutive relations
Budget equations (6), (7), (10)(12) do not form a closed system, since they represent
9 relations between 20 unknowns , u
i
,
ij
, , J
heat
i
, s, T,
s
.
1
In order to close
the system, additional 11 relations must be specied (in fact 12 relations will be
needed, since they will bring up one more unknown pressure p). These are so called
constitutive relations. They must be chosen in such way that they respect fundamental
thermodynamic inequality
s
0 and some other natural assumptions, known as
axioms of constitutive theory. Constitutive relations for single component uid can be
following:
1
External volume force f is supposed to be known.
4
State equation:
= (p, T) (13)
Only 2 state variables are independent. Every state quantity can be expressed as
a function of (p, T).
Expression for internal energy:
= (p, T) (14)
Denition of entropy (Gibbs equation):
T ds = d + p dv (15)
Expression for stress tensor:

ij
= p
ij
+
visc
ij
= p
ij
+ 2
_
d
ij

1
3
d
ij
_
+ d
ij
(16)
= (p, T) shear viscosity
= (p, T) bulk viscosity
Expression for heat ux (Fourier law):
J
heat
= T (17)
= (p, T) heat conductivity
Remarks:
1. Constitutive relations (13)(17) introduce new state variable p. All other functions
are supposed to be known.
2. State equation in the form (13) excludes possibility of phase transitions.
3. Gibbs equation (15) denes specic entropy s = s(, v) for single component
system in local thermodynamic equilibrium. Since s is state variable, ds must
be total dierential. This gives restriction:

v
_
1
T
_

_
p
T
_
v
(18)
4. Stress tensor dened by relation (16) consists of two parts: elastic part p
ij
and
viscous part
visc
ij
. Elastic part of stress tensor is responsible for appearance of
pressure gradient term in momentum equation. Viscous part of stress tensor is a
linear isotropic function of deformation rate tensor d
ij
. This reects isotropy of
the uid.
5. Viscous part of stress tensor
visc
ij
vanishes for d
ij
= 0, i.e. when there is no
deformation of the uid. Deviator d
ij

1
3
d
ij
(traceless part of d
ij
) represents
uid deformation due to pure shear. Divergence d (trace of d
ij
) represents uid
deformation due to volume change. This explains terminology for shear and bulk
viscosity coecients.
6. When radiative transfer is assumed, heat ux J
heat
must include also radiative
energy ux.
7. Assumptions (16) and (17) dene so called newtonian uid.
5
5 Entropy production
Constitutive relations (13)(17) close the system of budget equations (6), (7), (10)(12).
Having closed system, it is possible to obtain diagnostic relation for entropy production
rate
s
. Derivation starts from Gibbs equation (15) and continuity equation (6):
T
ds
dt
=
d
dt
+ p
dv
dt

ds
dt
=
1
T

d
dt
+
p
T

dv
dt

dv
dt
=
d
dt
_
1

_
=
1

d
dt
= u

ds
dt
=
1
T

d
dt
+
p
T
u (19)
Using equations (10) and (12) enables to eliminate time evolution from equation (19).
After few manipulations this gives:

s
=
1
T
2
J
heat
T +
1
T
d
ij

visc
ij
(20)
Using constitutive relations (16) and (17) it can be veried immediately whether
expression (20) respects fundamental thermodynamic inequality
s
0:

s
=

T
2
(T)
2
+
2
T
d
ij
_
d
ij

1
3
d
ij
_
+

T
d
ij
d
ij
=
=

T
2
(T)
2
+
2
T
_
d
ij

1
3
d
ij
_ _
d
ij

1
3
d
ij
_
+

T
d
2
It is clear now that non-negativity of entropy production rate
s
requires 0, 0
and 0.
Sometimes so called diabatic term Q T
ds
dt
is introduced. Employing equations
(10) and (20) it can be expressed in the form:
Q =
1

_
J
heat
+ d
ij

visc
ij
_
(21)
Formula (21) reveals that denoting Q as diabatic term is justied only for inviscid uids,
where
visc
ij
= 0. In such case Q vanishes for adiabatic processes, when there is no heat
exchange. However, this is no more true for viscous uids, where Q can be non-zero
even in the absence of heat ux. This is due to the second term on the RHS of equation
(21), representing viscous dissipation of kinetic energy. To be precise, for viscous uids
condition Q = 0 denes isoentropic, not adiabatic proces.
6
6 Summary
Using the results from previous sections it is now possible to formulate closed system of
equations for 5 prognostic variables p, T, u
i
. It can be derived from momentum equation
(7), continuity equation (6), internal energy equation (12), state equation (13), internal
energy expression (14) and newtonian uid relations (16), (17):
du
i
dt
=
1

ij,j
+ f
i
(22)

p
dp
dt
+

T
dT
dt
= u (23)

p
dp
dt
+

T
dT
dt
=
1

_
J
heat
+ d
ij

ij
_
(24)
= (p, T) (25)
= (p, T) (26)

ij
= p
ij
+ 2
_
d
ij

1
3
d
ij
_
+ d
ij
(27)
J
heat
= T (28)
d
ij
= u
(i,j)
=
1
2
(u
i,j
+ u
j,i
) d = d
ii
= u
i,i
= u
= (p, T) 0 = (p, T) 0 = (p, T) 0
Equations (23), (24) can be combined in order to get tendencies
dp
dt
,
dT
dt
. This can
be illustrated on perfect gas, for which functions (25) and (26) have the form:
=
p
RT
(29)
= c
v
T (30)
R, c
v
= const c
p
= c
v
+ R
It can be checked easily that relations (29), (30) full constraint (18). Inserting them into
equations (23), (24) and doing necessary manipulations leads to prognostic equations
for p and T:
dp
dt
=
c
p
c
v
p u +
p
T
Q
c
v
(31)
dT
dt
=
R
c
v
T u +
Q
c
v
(32)
Finally, it is usefull to list all assumptions needed to get closed system (22)(28).
Most of them were mentioned explicitly, but some are still hidden:
Continuum hypothesis. In reality, uids are composed of atoms and molecules.
Theory using continuum hypothesis cannot be applied on scales comparable with
inter-particle distances. Thermodynamic quantities like p, , T, , s cannot be
dened for the point, they have reasonable meaning only for uid elements which
contain many particles.
Smoothness of the elds. It was assumed that elds are suciently smooth in
space and time, so that all necessary derivatives can be evaluated. This excludes
study of processes with discontinuities like propagation of shock waves, cavitation
or free boundary problems.
7
Single component system. Fluid can be a mixture, but its composition must not
vary in space and time.
Conservation of mass, Newton equations of motion, rst and second law of
thermodynamics. These laws were used to formulate budget equations.
Assumption of newtonian uid. It implies non-polarity of the uid as well as
uid isotropy. Non-negativity of entropy production rate requires non-negative
transport coecients , , .
State equation in the form = (p, T). Fluid density is fully determined
by pressure and temperature. There is no possibility of phase transitions
(evaporation, condensation). Every state variable can be expressed as the function
of (p, T).
Gibbs denition of entropy s. It implies uid in local thermodynamic equilibrium
and imposes integrability constraint (18) on thermodynamic variables.
Remark:
Problem of initial-boundary conditions was not examined here. It is well known
experimental fact that velocity of viscous uid relative to material boundary is zero.
Unfortunately, system (22)(28) is so complicated that general formulation of well-posed
problem is not known. Main complications rise from non-linearity of the system and
possibility of opened boundaries.
8