Fanno processes in dense gases

M. S. Cramer, J. F. Monaco, and B. M. Fabeny

Citation: Physics of Fluids (1994-present) 6, 674 (1994); doi: 10.1063/1.868307

View online: http://dx.doi.org/10.1063/1.868307
View Table of Contents: http://scitation.aip.org/content/aip/journal/pof2/6/2?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

A pilot scale ultrasonic system to enhance extraction processes with dense gases
AIP Conf. Proc. 1433, 358 (2012); 10.1063/1.3703205

Nozzle flows of dense gases

Phys. Fluids A 5, 1246 (1993); 10.1063/1.858610

Relaxation processes and spectra in liquids and dense gases

J. Chem. Phys. 66, 2789 (1977); 10.1063/1.434350

Viscosity of Moderately Dense Gases

J. Chem. Phys. 42, 263 (1965); 10.1063/1.1695686

On the Kinetic Theory of Dense Gases

J. Math. Phys. 4, 183 (1963); 10.1063/1.1703942

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
Copyright by the AIP Publishing. Cramer, M. S.; Monaco, J. F.; Fabeny, B. M., "Fanno processes in dense gases," Phys.
Fluids 6, 674 (1994); http://dx.doi.org/10.1063/1.868307

Fanno processes in dense gases

M. S. Cramer, J. F. Monaco, and B. M. Fabeny
Engineering Science and Mechanics, Virginia Polytechnic Institute and State University, Blacksburg,
Virginia 24061-0219
(Received 9 June 1993; accepted 6 October 1993)
The global behavior of Fanno processes is examined for dense gases. It is shown that three sonic
points corresponding to two local maxima and one local minimum in the entropy can occur if
the fluid is of the Bethe-Zel'dovich-Thompson type. Both analytical and numerical examples of
the nonclassical behavior are provided.

I. INTRODUCTION is the thermodynamic sound speed. The quantities p, p, s,

and T are the fluid pressure, density, entropy, and absolute
One of the simplest models for the frictional heating of temperature.
gases is the Fanno process. This model takes the flow to be Striking qualitative differences between the perfect gas
one-dimensional, steady, and adiabatic and is qiscussed in and dense gas theory were first revealed by Thompson, 16
most texts on gas-dynamics, 1 engineering who demonstrated the critical importance of the funda-
thermodynamics,2 and engineering fluid mechanics. 3 Al- mental derivative
though the most common method of presentation involves
the use of perfect gases, many applications of interest in-
volve large pressures. In such high-pressure applications
r=~+ Jal =.!. J(pa) I (3)
p Jp $ p Jp s
the ideal gas model may no longer give an accurate esti-
mate of the thermodynamic properties and a more complex to virtually all aspects of gas dynamics. With respect to
equation such as that due to van der Waals is required. Fanno flows, Thompson's main contribution was to prove
Examples of high speed flows in which dense gas effects are that any sonic point in a r < 0 fluid will correspond to a
important include the hypersonic and transonic wind tun- local minimum in the flow entropy. In order that the en-
nel designs described by Enkenhus and Parazzoli, 4 tropy increase in Fanno flows, all flows would then be
Simeonides,5,6 Anderson,7,s and Anders;9 the transport of driven away from the sonic, i.e., choking, condition pro-
fuels and other chemicals described by Leung and vided that the r < 0 assumption is satisfied. This contrasts
Epstein,lO and Bober and Chow; II and the cooling of hy- with the well-known perfect gas theory where r> 0 and all
personic aircraft described by Dziedzic et al. 12 Heat trans- flows necessarily become sonic, i.e., choked, provided that
fer and turbomachinery equipmeht in both subcritical and the duct length is sufficiently large. Thompson l6 gives a
supercritical Rankine power cycles also frequently involve definitive picture of the Fanno process provided the fun-
dense gas effects; see, e.g., Jones and Hawkins2 and Rey- damental derivative (3) is either strictly positive or strictly
nolds and Perkins. 13 negative. Tlie main goal of the present study is to extend
The purpose of the present study is to develop a more Thompson's analysis to flows for which r can change sign
complete theory of Fanno flows which is valid for the full within the flow. We also provide specific numerical exam-
range of pressures and temperatures encountered by single- ples of the nonclassical flows predicted by both Thompson
phase fluids. We will also expand the scope of the analysis and the present authors.
to include the full range of fluids of interest in applications Because of the central role played by r in the following
such as chemical and fuel transport, novel wind tunnel study, we first provide a brief review of its variation and the
designs, and Rankine cycle power systems. conditions under which it becomes negative. The first to
Previous investigations of Fanno flows using more gen- give a detailed discussion of the variation and sign of (3)
eral equations of state are reported by Landau and were Bethe l7 and Zel'dovich, IS who demonstrated that r
Lifshitz l4 and Arp et al. 15 The former authors presented a can become negative over a finite range of pressures and
series of local results valid for all gas models. The latter temperatures in fluids with sufficiently large specific heats.
authors· emphasized the importance of the Griineisen pa- For a given fluid with a large specific heat, the range of
rameter, f3a2/cp, on the flow behavior. Here pressures and temperatures corresponding to r < 0 is in the
general neighborhood of the saturated vapor line at pres-
sures on the general order of the value at the thermody-
(1) namic critical point. To provide examples of negative r
fluids and to illustrate the general variation of r we have
is the coefficient of thermal expansivity, cp is the specific plotted a scaled version of r as a function of scaled density
heat at constant pressure and on the critical isotherm in Fig. 1. The fluids chosen are
molecular nitrogen (N2 ), steam (HzO), octane (CSH IS )'
and four commercially available heat transfer fluids: FC-72
(2) (C6F I4 ), PP9 (C l1 F 20 ), PPll (C I4F z4 ), and FC-71
(C 1sF 39N). The physical properties for the first three fluids

674 Phys. Fluids 6 (2), February 1994 1070-6631/94/6(2)/674/10/$6.00 © 1994 American Institute of Physics

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 Fanno flows. Many of the results proven in this section will
be similar to those developed by Landau and Lifshitz 14 and
Thompson,16 although the present authors account for the
global features of the Fanno curve, including sign changes
in r. In Secs. III-V, we provide specific examples of non-
1.0 classical Fanno flows. In particular, we present analytical
solutions for the Fanno curve corresponding to van der
Waals gases in Sec. III and numerical solutions using the
van der Waals gas model and the more accurate Martin-
Hou model in Secs. IV and V. In Sec. VI we provide a
PI'
--0 0.5 short discussion of possible shocked Fanno flows.

II. GENERAL THEORY

The governing equations of a Fanno flow may be de-
0.0
rived by taking the flow to be single-phase, one-
dimensional, adiabatic, free of body forces, and steadily
flowing in a constant area duct. The resultant mass and
energy equations may therefore be written I
vp=v/V=m=const, (4)
h + m2 V212 = const. (5)
Here v is the particle velocity, V == p -1 is the specific vol-
FIG. 1. Variation of the fundamental derivative (3). In each case the ume, m is the mass flux, and h is the enthalpy
temperature is the critical temperature of the fluid of interest.
(6)
where e is the internal energy. Because the flow is adiabatic
were taken from Reid et al. 19 and those for FC-72 and PP9 and involves fluid friction, we naturally assume that the
were taken from the manufacturer's technical publications. entropy increases with distance along the duct. As a result,
The required physical properties of PP 11 and FC-71 were we will impose the following condition as the expression
estimated by Cramer. 20 The equation of state employed for for the second law of thermodynamics:
the generation of Fig. 1 is that due to Martin and HouY ds
Full details of the implementation used may be found in dx>O, (7)
the article by Cramer. 20
The main point to note in Fig. 1 is that the heavier where the x axis will be taken as positive in the direction of
fluids have a local minimum in pr/ a at about one-half to the flow. In fact, if we consider a general fluid flowing
two-thirds the critical density. The heaviest fluids, viz., through a duct with a circular cross section of diameter D,
PPII and FC-71, have a finite region of r <0 at the critical the equation expressing conservation of linear momentum
temperature. It should be noted that both Lambrakis and may be combined with identities derived later in this sec-
Thompson22 and Cramer20 have shown that PP9 also ad- tion to show that
mits a range of pressures and (subcritical) temperatures in ds 4'Tw
its single-phase regime for which r < 0, even though the dx= pTD'
(8)
gas model used here yields r> 0 at the critical tempera-
ture. Further examples of negative r fluids are found in the where 'Tw= 'Tw(x) is the shear stress at the duct wall, de-
work of Thompson and Lambrakis. 23 Summaries of the fined so that 'Tw > 0 if the force generated on the fluid op-
nonclassical dynamics possible in such fluids can be found poses the direction of flow. Thus (7) is consistent with the
in the articles by Thompson, 16 Thompson and expectation that fluid friction opposes the motion. Further-
Lambrakis,23 Menikoff and Plohr,24 Leidner,25 and more, the use of the strict inequality in (7) is consistent
Cramer. 26 with the expectation that the friction remains nonzero.
The previous investigations of negative r fluids indi- Inspection of (5) reveals that the Fanno line is simply
cate that the variation of pr/a on isotherms, isentropes, a downward facing parabola in the h- V plane. In the sub-
shock adiabats, and Fanno lines is qualitatively the same as sequent discussion it will also be useful to determine the
that plotted in Fig. 1. Because of the importance of fluids qualitative features of the isentropes. If we employ Gibbs'
having such a negative r region and the pioneering work relation
done by the authors mentioned above, we will refer to any 1
fluid which possesses such a r < 0 region as a Bethe- dh=T ds+-dp, (9)
Zel'dovich-Thompson (BZT) fluid. p
In the next section we layout the general theory of it may be shown that

Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny 675
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 h
might expect, a local maximum or minimum in the entropy
occurs when the Fanno line is tangent to an isentrope. In
the neighborhood of such a tangency point, the relative
slopes of the Fanno line and isentropes may be inspected in
order to deduce whether the extremum corresponds to a
local maximum or minimum in s.
Before giving a more precise criterion which distin-
guishes between local maxima and minima in the entropy,
we first relate the relative slopes of the Fanno lines, and
therefore the entropy variation, to the local Mach number.
v v If we combine (10) with the differential of (5), we find
that
tal (bl
2
dhl -av
dV
ah I = aV (l-M),
2 (14)
FIG. 2. Sketch of the isentropes in the h- V plane for (a) a fluid having F s
prla>! everywhere and (b) a fluid having prla<t over a finite range
of V. where M=v/a is the local Mach number. Thus, tangency
points are also sonic points, i.e., points where M = 1. Fur-
thermore, (13) may now be rewritten
(10) ds a2 I-M2
dvlF =VT 1+f3a2lcp' (15)
a2
-ahl =T (f3
1+-) >0. (11) from which it is immediately obvious that the entropy in-
as v cp
creases with an increase in V in subsonic regions and de-
From (10), it is clear that the slopes of the isentropes in creases with an increase in V in supersonic regions. This
the h- V plane are also negative. Although the thermal ex- observation is completely consistent with those made in the
pansivity f3 can be negative for liquids, e.g., pure water well-known perfect gas theory.
between 0 and 4 C, we will take f3 > 0 for all vapors con- We now consider the criterion for maxima or minima
sidered in the present study. The nonclassical effects which in the entropy variation along a Fanno line. We note that
arise when f3 < 0 have been summarized in Sec. V of Meni- the slopes of the Fanno line and an isentrope are, by defi-
koff and Plohr. 24 The inequality in (11) is an immediate nition, equal at a tangency point so that the nature of the
consequence of the f3 > 0 assumption once the well-known extremum is necessarily determined by the relative curva-
constraint (cp>cv>O) is imposed; here C v is the specific tures of the Fanno line and the isentrope. If we employ
heat at constant volume. We conclude that the isentropes (12) along with the second derivative of (5) and the def-
do not intersect and that the isentropes corresponding to inition of rh and M, we can show that
the largest entropies lie above those having lower entropies
in the h-V plane. Although it is clear that the Fanno line is
2
ah
av{-
2
d h a
dV2 1 =j12
2
(2a+
pr
M -1
2 )
. (16)
always concave down, the isentropes mayor may not be F
strictly concave up. The precise condition may be seen by Thus, in the neighborhood of the tangency point, denoted
inspection of the following second derivative: by a star, the difference between the enthalpy evaluated on
2 2 the isentrope, hs' and that evaluated on the Fanno line, hF'
a h a ( p r 1)
aV{=j12 2a-- . (12) is

Thus the isentropes are concave up in the h- V plane when- hs-hF-;:::;a*2(pr)1

a- * (V-V*)2
--v;- +0 (V-V*)2
--V;-. (17)
ever pr/ a> 1/2 and concave down whenever pr/ a < 1/2.
Similar conclusions have also been given by Thompson. 1,16 As a result, we conclude that the isentrope lies above the
Isentropes corresponding to a fluid having pr/a> 1/2 ev- Fanno line in the neighborhood of a sonic, i.e., tangency,
erywhere and a fluid having pr/a< 1/2 over a finite range point if r> 0 there and it lies below the Fanno line if r < 0
of volumes are sketched in Fig. 2. at the sonic point. If we employ (11), it is now easily
The entropy variation along a Fanno line can be de- verified that any tangency point corresponds to a local
duced by consideration of the following identity: maximum in entropy if and only if r > 0 there and is a
local minimum if and only if r < 0 there. This conclusion
relating the nature of the extremum to the fundamental
derivative was first given explicitly by Thompson. 16 It is
interesting to note that the key connection between rand
(13) the second derivative of s was also given by Landau and
Lifshitz. 14 Our results are summarized in Figs. 3 and 4,
which was derived by regarding h=h( V,s), by differenti- where (15) has been employed to determine the Mach
ating along the Fanno line, and then by using (11). As one number variation.

676 Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 h
h

Fenno
Line .-

FANNO
LINE

. M<I
.
M>I
..
M<J
..
M>I

v v
(0 ) (b)
v
FIG. 3. Sketch of isentropes and Fanno line for (a) pr/a>~ and (b)
o<pr/a<! at the tangency, i.e., sonic, point.
FIG. 5. Sketch of a Fanno line through a r < 0 regipn. Unmarked lines
are isentropes. The Fanno line is tangent at points a, b, c.

The majority of the conclusions so far have been con-

cerned with local properties. However, if we consider a
fluid for which r > 0 at every point in a given Fanno pro- where, by definition, V c> Vb> Va' However, other config-
cess, then every extremum in the entropy necessarily cor- urations may be constructed for which sa> Sc > Sb' The cor-
responds to a local maximum. Thus, if the process is suf- responding image of the Fanno curve in the s-h plane may
ficiently smooth, only one extremum and therefore only now be detetmined by combining (15) with the derivative
one sonic point can exist because of the fact that multiple of (5) and the definitions of m and M to yield
maxima require the existence of at least one local minima.
We therefore conclude that the only Fanno flows for which dSj 1
M2-1
multiple extrema can exist are those involving BZT fluids. (18)
dh F =TM2 1+f3a2 /cp'
Because the positive r flows will, for the most part, turn
out to be qualitatively similar to those found in the perfect
gas theory, the remainder of the discussion will focus on Thus the slope of the Fanno line in the s-h plane is positive
the behavior of BZT fluids. if the flow is supersonic and it is negative if the flow is
A typical Fanno curve and isentropes for a BZT fluid subsonic. The Fanno curve corresponding to Fig. 5 has
are sketched in Fig. 5. The main assumption in this sketch been sketched in Fig. 6. The arrows in the latter figure
is that r < 0 only over a finite range of volumes V and that indicate the flow direction consistent with (7).
a sonic point does occur in the r < 0 region. If (11) is The pressure variations required to realize these non-
employed, it may be shown that classical flows can be determined through use of Gibbs'
equation (9), the result of which is

ISENTROPES
M>I M< I
'II ....

v h

FIG. 4. Sketch of isentropes and Fanno line for r <0 at the tangency, i.e., FIG. 6. Sketch of the Mollier (s-h) diagram corresponding to the Fanno
sonic, point. line of Fig. 5. Arrows indicate the direction of variation in flow direction.

Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny 677
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 near-sonic flows (M::::: 1) with pr/ a of order one. Under
5 these conditions, (21) may be approximated by

dMI :::::---+0(1),
-
Mpr (22a)
c
dp F P a
a
or, upon integration,
pr
M:::::l_ l P-P*+O(M-l, p-p*), (22b)
a * p* p*
where the * again denotes sonic conditions. We conclude
that the Mach number attains an extremum whenever r
changes sign, at least in the near-sonic limit. A similar
approximation was derived by Cramer and
p
Crickenberger28 in their study of the viscous shock struc-
FIG. 7. Sketch of the p-s diagram for the Fanno line of Fig. 5. ture. We note that (21) and (22) are completely consistent
with the following exact result found in the perfect gas
theory (r>O):
dPI _!... 1+Ml.({3a /cp)
2
(19) dMI y-l M 2)
ds F-V M2_l . dp F =-r;M (. 1+-2- <0.
Equation (19) may be shown to be completely equivalent
to the result given by Landau and Lifshitz. 14 Thus the III. ANALYTICAL SOLUTIONS FOR VAN DER WAALS
pressure increases with s, and therefore the flow direction, GASES
in supersonic flows and decreases as s and x increase when-
ever the flow is subsonic. If (18) and (19) are combined it To provide a partial verification of the general results
is easily demonstrated that derived in Sec. II, we present exact solutions for Fanno
2
flows of van der Waals gases. The van der Waals equation
dPI _ 1 1+M2({3a /cp ) of state reads
(20)
dh F - VM2 1+{3a2/cp >0.
Thus Pa> Ph> Pc in the case depicted in Figs. 5 and 6. The (23)
image of the Fanno line of Figs. 5 and 6 is sketched in Fig.
7 in the p-s plane. Here R is the gas constant, and
We note that the local behavior suggested by (18)-
27 R2T2 RT
(20) is exactly the same as that of the perfect gas theory. a= _ _ _c and b= __c (24)
The main differences between the perfect gas and general 64 Pc 8pc
fluid behavior is the existence of the local minimum in s are material constants correcting the ideal gas relation for
and the global structure. the presence of intermolecular forces and excluded volume,
Finally, we consider the Mach number variations on respectively. The subscript c denotes properties evaluated
the Fanno line. In particular, we note that the Mach num- at the thermodynamic critical point of the fluid. The spe-
ber necessarily attains at least one extremum between the
cific heat at constant volume Cv is a function of temperature
neighboring sonic points seen in Figs. 5-7. Cramer and only for van der Waals gases; however, we will assume
Bese7 have shown that a fairly simple criterion exists for even this variation is negligible in the flows considered in
such extrema provided the flow is isentropic. Because of the present section; i.e., we will take
the entropy variation in Fanno flows, such a simple crite-
rion does not appear possible. The exact expression for the cv=cv(T) =const. (25)
variation of M on the Fanno line can be written
2 Through the use of thermodynamic identities, explicit ex-
dMI M ( pr a aal I_M ) pressions for the enthalpy, thermodynamic sound speed,
dp F =r; -~+T as p 1+{3a /cp •
2 (21)
and entropy of van der Waals gases may be shown to be
Although RT ( t5V) 2a RTref a
h=eref+-g- l+ V _ b --V--t5-+V ' (26)
ref
aal >0
as p V 2a] 1/2'
)2RT(1+8)--V
a= [ ( V-b (27)
in the ideal gas limit, this derivative can change sign in the
dense gas regime, and no simple conclusion regarding the
extrema of the M-p variation appears possible. However, In(T/Tref) (V-b)]
s=R [ t5 +In Vref-b +sref, (28)
some insight can be obtained by restricting our attention to

678 Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 (30)
where Mo and ao are the inlet Mach number and sound
2.0
speed, respectively. The Mach number at any point is ob-
tained by combining the definition M == via with (4) to
yield
M
V Vao
M=m-=Mo - - · (31)
a Vo a
In the following, we fix the inlet conditions by speci-
fying Vo, To, and Mo. The fluid is specified by choosing
the critical properties, the molecular weight, and S. The
20.0 critical properties and molecular weight were then used to
compute the van der Waals parameters (24). The numer-
ical value of the mass flux was computed by combining
(27), evaluated at Vo, To, and (30). Various values of
FIG. 8. Computed Mach number variation for a van der Waals gas. The
inlet conditions were taken to be Vo=O.9Vc. To= 1.015Tc' and the indi-
V> Vo were chosen and substituted in (29). The resultant
cated values of Mo. Circles denote the analytical solution of Sec. III and V, T pairs were then substituted in (26)-(28) and (31) to
the solid line denotes the numerical solution described in Sec. IV. obtain the flow enthalpy, entropy, and Mach number.
As an example, we have computed and plotted the
Mach number and scaled enthalpy versus scaled entropy in
where the subscript ref refers to those properties taken to Figs. 8 and 9 for the heat transfer and Rankine cycle fluid
be at an arbitrary reference state, and o=Rlcv is a measure FC-75 (CgH 160); physical properties were taken from
of the specific heat. Yarrington and Kay.29 The initial volume and temperature
The temperature variation for a flow of fixed mass flux were taken to be Vo=0.9Vc and To= 1.015Tc' Each curve
may now be obtained by combining (26) with (5). After corresponds to a different inlet Mach number, i.e., mass
slight rearrangement, we find flux. The actual points computed are denoted by circles,
whereas the solid lines correspond to the numerical solu-
OVo) 2ao (Vo ) tion described in the following section. The flows corre-
T= [ To ( I+Vo-b +RVo V-I
sponding to Mo=0.18 and 0.20 are the nonclassical cases

+~
om2V6 ( V2)]
1-Ji6 j( SV )
I+V_b' (29)
described in Sec. II. The case Mo=0.16 is on the boundary
between the classical and nonclassical cases. At M o=0.18
and 0.20, the local minimum in the entropy is clearly seen
where the specific volume and temperature at the pipe inlet in each of the figures. In Fig. 8, the three sonic points
are denoted by Vo and To. The mass flux appearing in (29) characteristic of nonclassical cases may also be seen. Al-
is found from (4) to be though no explicit discussion was given in the previous

5-8 0 X 10 2
R
FIG. 9. Computed enthalpy variation for the flows of Fig. 8. The circle and solid lines again denote the analytical and numerical solutions.

Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny 679
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 1.5,,-----~-~---~-~-~--~--.----.

section, the M vs s curves are necessarily looped when

multiple sonic points occur; this looping phenomenon is
also clearly visible in Fig. 8.

IV. NUMERICAL SCHEME

-- --
~--

-
M 1.0
One way to obtain numerical solutions for the Fanno
flow is to solve (4) and (5) as an algebraic system. Here
we take a different approach in that we consider the dif-
ferential form of these equations. We should point out that
there is no particular advantage of one approach over the
other. The choice made here is simply a matter of personal 0.5"--_~_-:...:.;i~~"""","",-~_~__---.~_~_-:"
0.0 1.0 4.0
preference.
For a fixed mass flux, we may consider (5) to be the
algebraic relation between temperature and specific vol-
ume. We may then differentiate (5) with fixed liz to find FIG. 10. Computed Mach number variation for PP II (C1.?24). The inlet

dT V (a
1-{3T
dV Cv+VPT -p:-y--m ,
2
.2) (32)
conditions are Vo=0.75Vc ' To=l.012Tc ' and the indicated values
of Mo.

where y is the ratio of specific heats aQd the Fanno process was found to be in the two-phase re-
gime, the whole curve was discarded as invalid.
PT=:~I v . The fluid used is the commercially available heat trans-
fer fluid PPII. As pointed out in the Introduction, the
In any specific case of interest, the gas model will be pertinent physical properties are summarized by Cramer. 20
known, and all thermodynamic parameters appearing in Results for a range of inlet Mach numbers are plotted
(32) will be known functions of V and T. The constant in Figs. 10--12. The inlet thermodynamic state was taken to
mass flux liz can be determined by the inlet conditions. As be Vo=O.75 Vc and To= l.012Tc, which corresponds to an
in the previous section, we take the inlet conditions to be inlet pressure and temperature of 16.9 atm and 657.95 OK,
Vo, To, and Mo. The approach taken here was to integrate respectively. As in the calculations involving van der
(32) numerically subject to the initial condition T= To at Waals gases, the characteristic looping of the Mach num-
V= Vo. Because h( V,T), s( V,T), and a( V,T) will also be bers and multiple extrema in s are observed.
known (·nce the equation of state is specified, each T, V We have also examined the values of pr/a along the
pair may then be substituted in the above thermodynamic Fanno lines depicted in Figs. 10--12. It was found that the
functions to compute the enthalpy, entropy, and sound extrema in M closely corresponded to the points at which
speed for each V> Vo. The Mach number can be computed r changed sign. This observation is seen to be in complete
from an expression similar to (31). agreement with the conclusion based on the near-sonic ap-
As a check on this computation scheme, we first con- proximation (22).
sidered the van der Waals model introduced in Sec. III.
The results for the same example used in Sec. III are plot- VI. REMARKS ON SHOCKED FLOWS
ted in Figs. 8 and 9. The numerical scheme is in excellent
In the well-known perfect gas theory of Fanno flows,
agreement with the analytical solutions described, in Sec.
shock waves can occur if the inlet flow is supersonic. In the
III. Based on this comparison, as well as others involving
case of Fanno flows of BZT fluids, such shocked flows are
both perfect gas and van der Waals exact solutions, we
likely to be considerably more complicated. In order to
conclude that the numerical scheme provides an accurate
further emphasize the differences between the behavior of
representation .of the exact solutions.
perfect and dense gases, we will simply point out one or
two of the most interesting possibilities.
V. COMPUTATIONS USING THE MARTIN-HOU
In all that follows, the shock waves will be chosen to
EQUATION
satisfy the well-known necessary conditions for an admis-
In this section, we present numerical results for the sible shock. In particular, the entropy inequality will al-
Martin-Hou21 equation of state. The Martin-Hou equa- ways be satisfied and the shocks will be expected to satisfy
tion was chosen" for its accuracy, versatility, and wide- the speed-ordering condition
spread use in applications.
(33)
The basic form of the equation of state, entropy, inter-
nal energy, and specific heat are given in the article by where subscripts and 2 denote conditions immediately
Cramer and Best. 27 Complete details are found in the orig- upstream and downstream of the shock, respectively. Al-
inal article by Martin and HOU. 21 We have also checked though other conditions will be discussed, it is important to
our solutions for pressure and temperature against a stan- note that the present discussion should be regarded as en-
dard vapor pressure model due to Riedel 30 in order to tirely conjectural; a more comprehensive and rigorous pre-
ensure that the flow remained single phase. If any point in sentation will be provided at a later date.

680 Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 -0.05

-0.10 Mo=O.31
h-ho MosO.33
RTc

-0.20

-0.25u-__~~______~____~~ ____~____~____~~____~______~
0.0 1.0 2.0 3.0 4.0

FIG. 11. Computed enthalpy variation for the flow of Fig. 10.

We begin by considering a flow of the type depicted in main point of interest is that an inspection of the global
Figs. 5-7, where the entropy of the high-pressure sonic nature of the Fanno curve of BZT fluids strongly suggests
point is less than that of the low-pressure maxima in en- that the flow may be continued even when a sonic point is
tropy. One of the simplest nonclassical cases is the path encountered. A similar continuation beyond sonic points
marked 0' -su-A-s/ in Fig. 13(a). The pressure distri- was seen to be possible in the nozzle flows studied by
bution corresponding to this flow is sketched in Fig. 13(b). Warner,3! Chandrasekar and Prasad,32 Kluwick,33 and
In this case, the sonic eMsu = 1) expansion shock Su - A can Cramer and Fry.34 A second point of interest is that a
be constructed which implies that the flow may be contin- shock is possible even though the inlet flow is s,ubsonic. In
ued up to the lower sonic point Sf. In this example, the the perfect gas theory of Fanno flows, an initially subsonic

o.o~-----'------'-------'------'------'------.'------r----~

-0.1

-0.2

-0.3

-0.5u-____ ~ ______-L______ ~ __~~______~______~__~~______~

0.0 1.0 2.0 3.0 4.0

FIG. 12. Computed pressure variation for the flow of Fig. 10.

Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny 681

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 the global character of the Fanno curve, and are not re-
p p
0' 0'
stricted to flows having r strictly positive or strictly neg-
ative. The advantage over the well-known perfect gas the-
So ory is that the scope is extended to include a far wider
range of pressures and temperatures than can be described
by the low-pressure theory.
A Although only one sonic point corresponding to a local
maximum in entropy is expected in flows having r> 0, as
many as three sonic points corresponding to two maxima
and one minimum in s may occur in BZT fluids. A more
detailed inspection of the data leading to Figs. 8,9, 10--12
(b)
reveals that the occurrence of r < 0 on the Fanno curve is
(a)
not sufficient to generate the local minimum in the entropy.
Once the Fanno curve enters the r<o region, (22) sug-
FIG, 13, Sketch of shocked Panno flow, (a) p-s diagram, (b) Pox dia- gests that the Mach number will attain an extremum and
gram. The subsonic inlet condition is taken to be at 0'.
the subsequent Mach number variation will be such that
M --+ 1. The third sonic point will be observed if M = 1 be-
flow always remains subsonic and therefore shock free. fore the Fanno line passes out of the r < 0 region. If, on the
The second set of cases involving shocks is illustrated other hand, the Fanno curve encounters the r=o point
in Fig. 14. The inlet state 0" is taken to be on the super- before the Mach number becomes one, (22) suggests that
sonic branch in the general neighborhood of the r < 0 a local maximum or minimum in M will be attained and
sonic point. The first flow follows the path O"-A-B-suc-C-sl the values of M will either decrease or increase away from
which includes the ordinary compression shock A-B and one, respectively. The resultant h-s and p-s diagrams will
the sonic expansion shock su-C. This path is clearly non- then appear classical.
classical in that two shock waves are involved in a single An important consequence of the local maxima and
flow. The second path 0" - A - C' - Sf involves only a minima in the Mach number is that it is no longer conve-
single nonsonic expansion shock A-C'. In Fig. 14(b) we nient to use the Mach number to parameterize the Fanno
have allowed for the fact that the pipe length correspond- flow. At present, it appears that either the density or spe-
ing to the sonic condition Sl is likely to be different if dif- cific volume is a better choice. A similar conclusion can be
ferent numbers of shocks are present. made for isentropic flows.
In concluding this section, we simply note that the The nonclassical global structure of the Fanno curves
nonclassical shape of the Fanno curves may give rise to appears to give rise to new configurations for shocked
shocked flows which differ significantly from those de- flows. A brief discussion was provided in Sec. VI; a more
scribed in the perfect gas theory. Further study is clearly comprehensive theory will be provided in future publica-
required in order to delineate the full range of possibilities tions.
and any limitations.
ACKNOWLEDGMENTS
VII. CONCLUSIONS
This work was supported by the National Science
The present study has provided a general theory of Foundation under Grant No. CTS-8913198.
Fanno flows of single-phase gases. The principal advantage
of this work over that of Thompson 16 is that we delineate
Ip. A. Thompson, Compressible-Fluid Dynamics (McGraw-HilI, New
York, 1972).
2J. B. Jones and G. A. Hawkins, Engineering Thermodynamics, 2nd Ed.
p p (Wiley, New York, 1986).
3p. M. White, Fluid Mechanics, 2nd Ed. (McGraw-HilI, New York,
8 1986).
4K. R. Enkenhus and C. ParazzoIi, "Dense gas phenomena in a free-
piston hypersonic wind tunnel," AIAA J. 8, 60 (1970).
5G. Simeonides, The Aerodynamic Design of Hypersonic Contoured Axi-
symmetric Nozzles Including Real Gas Effects, Von Karman Institute
for Fluid Dynamics Technical Memorandum No. 43 (Von Karman
Institute, Waterloo, 1987).
6G. Simeonides, The VKI Hypersonic Wind Tunnels and Associated Mea-
surement Techniques, Von Karman Institute for Fluid Dynamics Tech-
nical Memorandum No. 46 (Von Karman Institute, Waterloo, 1990).
7W. K. Anderson, "Numerical study of the aerodynamic effects of using
sulfur hexafluoride as a test gas in wind tunnels," NASA Technical
(a) (b) Paper 3086 (1991).
8W. K. Anderson, "Numerical study on using sulfur hexafluoride as a
wind tunnel test gas," AIAA J. 29, 2179 (1991).
PIG. 14. Sketch of shocked Panno flows with the supersonic inlet con- 9J. B. Anders, "Heavy gas wind-tunnel research at Langley Research
dition 0". (a) p-s diagram, (b) pox diagram. Center," ASME Paper 93-FE-S.

682 Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19
 10J. C. Leung and M. Epstein, "A generalized critical flow model for 23p. A. Thompson and K. Lambrakis, "Negative shock waves," J. Fluid
nonideal gases," AIChE J. 34, 1568 (1988). Mech.60,. 187 (1973).
IIW. Bober and W. L. Chow, "Nonideal isentropic gas flow through 24R. Menikolf and B. Plohr, "Riemann problem for fluid flow of real
converging-diverging nozzles," J. Fluids Eng. 112, 455 (1990). materials," Rev. Mod. Phys. 61, 75 (1989).
12W. M. Dziedzic, S. C. Jones, D. C. Gould, and D. H. Petley, "Analyt- 2Sp. Leidner, "Realgaseinflusse in der Gasdynamik," Diplomarbeit, Uni-
ical comparison of convective heat transfer correlations in supercritical versitiit Karlsruhe, 1990.
hydrogen," J. Thermophys. Heat Transfer 7,69 (1993). 26M. S. Cramer, "Nonclassical dynamics of classical gases," in Nonlinear
I3W. C. Reynolds and H. C. Perkins, Engineering Thermodynamics, 2nd Waves in Real Fluids, edited by A. Kluwick (Springer-Verlag, New
Ed. (McGraw-Hill, New York, 1977). York, 1991), p. 91.
14L. D. Landau and E. M. Lifshitz, Fluid Mechanics (Addison-Wesley, 27M. S. Cramer and L. M. Best, "Steady isentropic flows of dense gases,"
Boston, 1959). Phys. Fluids A 3, 219 (1991).
lSV. Arp, J. M. Persichetti, and Chen Guo-bang, "The Griineisen param- 28M. S. Cramer and A. B. Crickenberger, "The dissipative structure of
eter in fluids," J. Fluids Eng. 106, 193 (1984). shock waves indense gases," J. Fluid Mech. 223, 325 (1991).
16p. A. Thompson, "A fundamental derivative in gas dynamics," Phys. 29R. M. Yarrington and W. B. Kay, "Thermodynamic properties of
Fluids 14, 1843 (1971). perfluoro-2-butyitetra-hydrofuran," J. Chern. Eng. Data 5, 24 (1960).
17H. A. Bethe, "The theory of shock waves for an arbitrary equation of 30L. Riedel, "Eine neue Universelle Dampfdruckformel-Untersuchungten
state," Office of Scientific Research and Development, Report No. 545, iiber eine Erweiterung des Theorems der iibereinstimmenden Zustand.
1942. Teil I," Chern. Ing. Tech. 26, 83 (1954).
18Ya. B. Zel'dovich, "On the possibility of rarefaction shock waves," Zh. 31S. M. Warner, "Steady non-isentropic flows of dense gases," Virginia
Eksp. Teor. 4, 353 (1946). Polytechnic Institute and State University, Report No. VPI-E-90-24,
19R. C. Reid, J. M. Prausnitz, and B. E. Poling, The Properties of Gases 1990.
and Liquids, 4th Ed. (Wiley, New York, 1987). 32D. Chandrasekar and Ph. Prasad, "Transonic flow of a fluid with pos-
20M. S. Cramer, "Negative nonlinearity in selected fluorocarbons," Phys. itive and negative nonlinearity through a nozzle," Phys. Fluids A 3, 427
Fluids A 1, 1894 (1989). (1991).
21J. J. Martin and Y. C. Hou, "Development of an equation of state for 33 A. Kluwick, "Transonic nozzle flow of dense gases," J. Fluid Mech.
gases," AIChE J. 1, 142 (1955). 247,661 (1993).
22K. Lambrakis and P. A. Thompson, "Existence of real fluids with a 34M. S. Cramer and R. N. Fry, "Nozzle flows of dense gases," Phys.
negative fundamental derivative r," Phys. Fluids 15, 933 (1972). Fluids A 5, 1246 (1993).

Phys. Fluids, Vol. 6, No.2, February 1994 Cramer, Monaco, and Fabeny 683
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.173.125.76 On: Tue, 08 Apr 2014 15:39:19