Sie sind auf Seite 1von 12

COMPUTER

MODELLING
PERGAMON Mathematical and Computer Modelling 30 (1999) 135-146
www.elsevier.nl/locate/mcm

An Air Cooled Tube-Fin Evaporator Model


For an Expansion Valve Control Law
C. APREA
Department of Mechanical Engineering, University of Salerno
Via Ponte Don Melillo, 84084 Fisciano (Salerno), Italy
C. RENNO
DETEC, Faculty of Engineering, University of Napoli Federico II
P. le Tecchio 80, 80125 Napoli, Italy

(Received December 1998; accepted January 1999)

Abstract-For control purposes, a mathematical model of a tube-fin evaporator of a vapour com-


pression plant running with R22 is analyzed. The refrigerant behavior in an evaporating region is
described by a homogeneous model. The balance equations, together with the constitutive equations,
determine a differential system which makes explicit the mechanism of dynamic behavior. At first,
the numerical solution of the steady state is obtained, both in evaporating and superheated regions.
Moreover, the numerical analysis allows us to evaluate the transition phase and to locate the inter-
face. Subsequently, various analytical aspects are discussed. For the nonlinear two-phase flow, the
dependence of the solution on the boundary data is estimated by means of a qualitative analysis.
Then, a linearized model for the single-phase flow is deduced and solved explicitly. The analytical
solution is compared with the numerical results and the degree of superheating is estimated in terms
of the model parameters. @ 1999 Elsevier Science Ltd. All rights reserved.

Keywords-Evaporator models, Control devices, Heat transfer, Phase transition, Numerical


methods, Nonlinear differential systems.

NOMENCLATURE
A heat transfer area per metre (m2m-) X vapour quality
AT inside cross-section area of tube (m2) Y length in air flow direction (m)
specific heat (kJ kg- K-l) z length in refrigerant flow direction (m)
x friction pressure loss (kPa)
G refrigerant msss flow rate (kgs-l ms2) GREEK SYMBOLS
h enthalpy (kJ kg-l) 0. refrigerant void fraction
ma air msss flow rate (kg s-l) P density (kg me3)
P pressure (kPa) @ frictional multiplier
Qulf heat flux (kW mm2) W air absolute humidity (kg kg-)
T temperature (K)
t time (s) SUBSCRIPTS
21 refrigerant velocity (ms-) air
u heat transfer coefficient (kW me2 K-l) ; refrigerant side
cn mass transfer coefficient i inside the tube

The authors want to thank R. Msstrullo and F. de Rossi for their suggestions and the profitable discussions.

oa95-7177/99/s - see front matter. @ 1999 Elsevier Science Ltd. All rights reserved. ?LP-t by 4v#-W
PII: SO8957177(99)09170-3
136 C. APREA AND C. RENNO

1 liquid phase s saturated condition at wall temperature


0 outside the tube W tube wall
1, vapour phase

1. INTRODUCTION
This paper is related to a research on control models adequate for a vapour compression refrig-
eration plant equipped with an air cooled tube-fin evaporator. In general, in these plants, the
main control concerns the refrigerant mass flow, which by means of an expansion valve, must
keep a prefixed degree of superheating at the evaporator exit, also varying the refrigeration duty.
To achieve the best performance of the evaporator, the degree of superheating should be small.
On the other hand, the refrigerant in the liquid phase must not leave the evaporator because the
compressor could be damaged.
To optimize the control law for an expansion valve, we want to estimate the degree of super-
heating in terms of the parameters related to the specific model of the evaporator.
In recent years, many models of tube-fin evaporators have been analyzed and classified (see
[l-6] for good literature). In the framework of distributed parameter models, two classes must
be pointed out; either the refrigerant is regarded as an homogeneous mixture of gas and liquid,
or these two phases are considered separately. The dynamic behavior of these evaporators has
been discussed mainly by numerical methods, as the analytical models concern free boundary
problems for nonlinear partial differential equations. A further difficult task is related to the heat
transfer coefficient which depends on the unknowns and may be obtained by experiments and
numerical approaches. These questions are typical features of inverse problems in mathematical
modelling (see [7]).
We consider the homogeneous model discussed in [l], as it shows good agreement with ex-
perimental data. The balance equations, together with the constitutive equations, determine
a differential system which makes explicit the mechanism of dynamic behavior (see Section 2).
For control purposes, as it is known, the preliminary analysis of the steady-state is basic, be-
cause it represents the initial state of the dynamical system which is going to be influenced by
time-dependent control parameters. For this, at first the steady problem is solved by numerical
methods, both in evaporating and superheated region (see Section 3). This analysis allows us to
evaluate the transition phase and also to locate the interface. Subsequently, various analytical
aspects of the steady problem are discussed. For the nonlinear two-phase flow, the dependence
of the solution on the boundary data is estimated by means of a qualitative analysis (see Section
4). As for the single phase flow, a linearized model is considered and the solution is explicitly
determined. Further, this solution is compared with the numerical results (see Section 5). The
explicit law of variation of the refrigerant temperature along each pipe of the vapour region allows
us to deduce also the explicit correlation among the degree of superheating, the refrigerant mass
flowrate, the total length of the superheated circuit, and other model parameters (see Section 6).
This estimate could be applied to obtain a control law for an expansion valve.

2. MATHEMATICAL MODEL
The coil circuit of the specific evaporator has a finned surface and is constituted by horizontal
parallel pipes with the same diameter and length (Figure 1). The air exchanges heat power with
the pipes and the finned surface; its flow is descending, as well as the y-axis.
The mathematical model is based on the following hypotheses (see [1,2]).
(a) In the evaporating region, the liquid and the gas phases of the refrigerant are incompress-
ible and in thermal equilibrium. Besides, the two-phase evaporating flow inside the coil
is considered as a homogeneous, one-dimensional flow in the z-direction. The z-versor
follows the flow.
(b) Variations of refrigerant kinetic and potential energy are ignored.
Expansion Valve Control Law 137

(cl Axial heat conduction within each pipe is neglected.


(4 The bends of the coil circuit can be assumed to be adiabatic zones with respect to the
heat exchange with the air.

Figure 1. Coil circuit of the evaporator.

2.1. Basic Equations for the Refrigerant


Let (pe, he) and (pV, h,) be the density and the specific enthalpy of the liquid and the vapour
phase. Besides, denote by x the vapour quality and by (I! the void fraction, i.e., the ratio between
the cross-sectional area filled by the vapour and the total cross-sectional area. In the two-phase
region, let

a= [I+$$ P=W,+(l-cr)pe, j;=crp,h,+(l-a)p~h& (2.1)

If p, Tf, and u are the pressure, temperature, and velocity of the refrigerant, according to the
above hypotheses, the balance equations of the mass, momentum, and energy can be given the
form:

ap+
; -g(a4=0,
j$4+g(b)=-$-fm (2.2b)

g +g (Lu) = &Jf.

In (2.2), fi represents the friction pressure loss (see (2.11)), while &,f is the heat flux at the
wall. If T,,, denotes the wall temperature, one has

&If = 4 2 (T, df),


( >
where Di is the inner pipe diameter and Vi denotes the heat transfer coefficient on the refrigerant
side. Clearly, the basic equations (2.2) are still valid in the vapour region, where CYis equal to
one.

2.2. Heat Balance for the Pipe Wall and the Air
Let h, and + be the specific enthalpy and the specific heat of the air, while M, and c,,, denote
the mass and the specific heat of the tube wall. Further, let Ai and A0 denote the heat transfer
area inside and outside the tube; A, includes the finned surface corrected with its efficiency,
which varies slowly with the air heat transfer coefficient (see [S]).
138 C. APREA AND C. RENNO

As the air and the refrigerant are in contact through the pipe wall, the heat balance leads to
the equation
aTw MJo
cp&, = c,, (ha - hs) - AiUi CC, - Tf>, (2.4)

where U, represents the heat transfer coefficient on the air side. The subscript s denotes the
value of the properties at wall temperature. If this temperature is lower than the dew point
temperature, the properties in saturated conditions are evaluated.
As the heat capacity of the air is small, the mass and energy accumulations at any time can
be neglected. So, the balance equations for the energy and the moisture of the air can be given
(see [1,3]) the form:

a(WJJa)
ay = -A,Um(wa -us),

where m, is the air mass flowrate and U, is the mass transfer coefficient. The negative sign in
(2.5),(2.6) is due to the descending y-axis.

2.3. Constitutive Equations and Coefficients


Letv=P-r,h=ilfi- l. In the two-phase region, one has three constitutive equations such as

h=h(a:,Tf), v=v(z,T), P=P(T.). (2.7)

In the single-phase region, we will consider

h=V,4, P=P(TA (2.8)

where the explicit form of p(Tf, v) for the refrigerant R22 (see [8,9]) is

p= 2 i [oi + bi T -+ dieeb T/c] (v - do)-i ,


1

and where T, is the critical temperature, p is in bar, and ai, bi, di are numerical constants.
The moist air is considered as a perfect gases mixture (see [lo]). For this, a subroutine has
been developed and the properties h,, ws at wall temperature T, are numerically explicit. Then,
in the analytical framework, h, and w, can be specified by linear functions of T,,

h, = a, + b,T,,,, wa = c, + d,T,, (2.10)

which can be deduced by interpolating formulae. These formulae have been applied to h, and w, in
the vapour region, with correlation coefficients included between 0.96 and 0.98. The correlations
used for the heat transfer coefficient Vi in the two regions are specified in Section 3.2.
As for the homogeneous model of the refrigerant two-phase flow, the overall pressure gradient fi
can be determined by
(72 (~qO.25
= p; 2 * 0.079 - (2.11)
e PeDa

where !I$ is the two-phase multiplier and (-g)e represents the frictional pressure gradient, i.e.,
the gradient we would get if only the liquid flows in the pipe with the same total mass flowrate G.
Expansion Valve Control Law 139

3. BOUNDARY CONDITIONS-NUMERICAL SOLUTION


3.1. Statement of the Problem
In the two-phase region, the constitutive equations (2.7) allow us to express V, h, and p in terms
of x and Tf. Further, by (2.10), one deduces h, and ws in terms of T,,,. So, the dynamic model
of the evaporator is characterized by (2.2), (2.4), (2.5), and (2.6), which represent a system of
six equations in the unknowns z, U, Tf, T,, h,, w,. The single-phase region is characterized by
two constitutive equations only. By (2.8), p and h can be expressed in terms of Tf and V; thus,
(2.2b), (2.2c), and (2.4)-(2.6) re present a system of five equations in the unknowns V, Tf , T,,
ha, wa.
In order to estimate the response of the evaporator to variations of independent control pa-
rameters, at first, the steady state is analyzed. For this, we specify the related boundary-value
problem. Let %s be the first air layer, confined by the first two pipes of equations y = 0 and
y = 6 (Figure 2). According to the direction of the refrigerant flow, the abscissaz is always
increasing on each line, so that

i + -----------_--------- --+
I
I*

Figure 2. The first air layer.

We remark that the functions h,, w, are defined in ?&, while the other unknowns are defined
on the boundary of ?%J.So, the analysis must be referred to two separate aspects:
(a) one-dimensional rejkigemnt flow along each pipe;
(b) air flow through the layer !R-,.
Owing to the direction of the air flow, h, and We must be prefixed on the initial line y = 0,
while Tf, G, and z must be specified in PO. These data determine uniquely h, and w, in ?&;
z and Tf are determined along the pipe y = 0 and so in A. As the vertical walls are adiabatic
and each layer is thin, we assumethat the values of 2 and Tf in A and B coincide. At thii point,
one knows all the data on the second pipe y = S, which represents the initial line of the next
layer; so, the analysis can be iterated to the whole coil circuit.
In the analytical framework, this problem is very difficult because both equations and coeffi-
cients Vi, fi are not linear. Moreover, the transition phase problem occurs too. The successive
approximations method could be applied to each of the two regions.

3.2. Numerical Solution


The parameters related to the geometry of the specific evaporator are (Figures 2 and 3)

C= 0.496 m, h = 0.173 m, s = 0.105 m, Di = 0.0085 m, D, = 0.00952 m.


140 C. APREA AND C. RENNO

refrigerant inlet +
Figure 3. Evaporator cross section.

The problem has been solved by finite-difference schemes. Indeed, spectral methods also work
efficiently (see [11,12]). For this, the z-interval has been subdivided into 221 parts and the
y-interval into eight parts, thus obtaining 1768 cells. As in our evaporator the distance among
the fins along each pipe is O.O0434m, the cell dimension in the z direction is in conformity with
the physical situation; as for the y direction, the cell dimension is equal to the distance between
two contiguous pipes.
For each cell, the variables related to the one-dimensional flow in the coil circuit are located in
the centre of the cell, while the unknowns related to the air are placed on cell walls.
The working fluid (R22) is the chlorodifluoromethan (CHCIFs) (see [13]), and for its thermo-
dynamic properties the subroutine EASY has been employed (see [14]). Besides, the heat transfer
coefficient for the two-phase region has been estimated by Kandlikars correlation [15-171, while
the Dittus-Boelter equation has been employed for the single-phase flow. For the air side, the
heat transfer correlation is deduced by the correlation of Turaga, according to whether the tube is
wet or dry (see [18,19]). At last, the mass transfer coefficient is connected with the heat transfer
coefficient by the Lewis correlation [20].
The Newton-Raphson iteration method has been applied and a good level of stability has been
obtained by 20 iterations. The main results are as follows.
- The numerical values related to the pressure difference (Ap,) between the inlet and outlet
of the evaporator are in conformity with the current plant values.
- The phase transition occurs along the penultimate pipe of the coil circuit and the behavior
of the refrigerant temperature is quite realistic.
For the analytical model which we are going to discuss, we emphasize the following aspects
deduced by the numerical analysis.
In all the vapour region, the heat transfer coefficient Vi is almost constant and the constitutive
equations (2.8) can be linearized with a good approximation. For instance, when G = 252 kg
s-l m- 2, Vi is included in (0.394-0.408 kW K-l me2); further, the pressure is variable in (3.34-
3.36 bar) and h in (401-410kJkg-I). At last, in the single-phase region, the friction factor is
given by
fz = W$J, (3.1)
where kc is essentially constant.

4. THE NONLINEAR TWO-PHASE FLOW


Let us consider the two-phase region and refer to the first air layer !XJ (Figure 2). If ho(z) is
the boundary datum prefixed along the initial line of !Xc, let
Expansion Valve Control Law 141

h, (0, z) = hrJ(C- z) ) (4.1)

By (2.5),(4.1), for y = 6-, one deduces

h,(y, z) = emc6ho(l- z) + (1 - e-) h,. (4.2)

Owing to (2.4), the flow along each pipe of the circuit is such that

and by (2.10), (4.2), and (4.3), it is possible to express T, by means of Tf . If one denotes by

c6 -

I&(Z) = b, [ho@. - 4 - as] (4.4)
I

along the pipe y = 6, it is


Ttu - Tf = x(6) [Ho(z) - Tfl * (4.5)
This formula also can be applied to the initial pipe y = 0. It suffices to replace x(6) with x(O).
If one puts

(4.6)

there results

4&x = Diq(l - x). (4.7)

REMARK 4.1. The coefficient x defined in (4.4) depends on Vi and so it represents a nonlinear
function of the unknowns T, and Tf. However, x is always lessthan one and (4.5),(4.7) allow us
to estimate the nonlinear heat flux (2.3) by means of the following linear inequality:

~wf=17(1-x)(Ho-Tf)577(Ho-Tf), (4.8)

as HO - Tf is positive in !Rs; the last member of (4.8) does not depend on Vi.
These formulae imply estimates by means of the data. If Go is the prefixed value of the
refrigerant mass flow-rate G for z = 0, the equation (2.2a), for the steady state flow, implies
G(z) = Go.
Let h = ohz + bhTf + ch be the linear interpolation of the constitutive equation (2.7)i and let

?./=ah 1-x
fz = Gouf,
bh ~7= Gobh

where (T= c(z,Tf,T,) and f = f(z,Tf) (see (2.11)).


Then, by (2.2b),(2.2 c) , one deduces the nonlinear equations

dU
--+fu=-G-l% (4.1Oa)
O dz

-$Tf+rz)+~(Tf+rz)=cWo+rs). (4.10b)

If z(0) = 20, p(O) = pc, Tf(0) = To, and

-.pw~, $@<) = e-Jc=f(r)d+, (4.11)


9hE) = e
142 C. APREA AND C. RENNO

by (4.10), one has

Tf + F = (Tf + ~4, cp(z, 0) +


Jz%
cp(.zY
0
5MC)
WJ
(Ho
+F)4, (4.12a)

P + Gou = (P + Go4, $(z, 0) +


J 0
pea, 4. (4.12b)

Now, let Ag denote the variation of g(z) in the interval (0, Z) and let

17
He = SUP IHol + Irl, fo = sup f, (4.13)
OO=Gobh*
As sup(z) = 1 and p(z) is decreasing, by (4.12), there results

IA (Tf + YZ)I I (ITf + WI, + K) (I- e--OO*)


, (4.14a)
IA (p + Gou)[ 5 (2~0 + GOUO) (1 - eWfo) . (4.14b)

The last members in (4.14) depend on (TO, fe, and the prefixed data. As (4.13)3,(4.14a) show,
the greater Go is, the smaller A(Tf + yz) results. Moreover, (4.14a) allows us to estimate how
much this variation is influenced by the prefixed refrigerant mass flowrate Go. The order of fe in
(4.14b) can be determined by means of the numerical results. Obviously, this qualitative analysis
can be applied to each pipe of the two-phase region.

5. SUPERHEATED REGION
Let y = y and y = y be the equations of the lines which represent the penultimate and the
last pipe of the coil circuit. Further, let !X be the last air layer defined by

Denote by z the abscissa on the last pipe and by z = & - z the abscissa on y = y.
According to the numerical results, the phase transition occurs along y = y and the heat
transfer coefficient Vi is almost constant in all of the vapour region. Moreover, the numerical
analysis permits the evaluation of the air enthalpy &(z) along y = y, and so the boundary
datum on the initial line % is known. Then, the formulae (4.2)-(4.5) can be applied to the
layer R too; it suffices to replace HO(Z) with the function

H(z) = be, [ha(e - z) - a,] ) (5.1)

related to the line y = y.


Soon after the phase transition, the ilow is continuous and the constitutive equations (2.8) can
be linearized with a good approximation (Section 3.2). If the initial state (~10,TO) is closed to the
transition region, there results

ap dv 8Tf dh
-apz+b -, +bh mf (5.2)
iG-- p az bz=ahaZ aZ
with
ap = (5.3)

and similarly for ah, bh.


Thus, owing to (5.2) and (4.5), the problem is reduced to the analysis of equations (2.2 b),(2.2c)
in the unknowns Tf and v. If one puts

,2!! k2(6) =
4&x(S)
-1 (5.4)
bh GoDha
Expansion Valve Control Law 143

the momentum and energy equations (2.2b),(2.2c) can be given the form

--Iclv, (5.5a)
Wf
dz + y; = kz (H - Tf) . (5.5b)
REMARK 5.1. The equation of state (2.9) is explicit in p and characterizes interesting properties
of the refrigerant R22. So, the constants aP, bP defined in (5.3) have quite different order and up is
always negative. For instance, when Gc = 252 kg s-l m- 2, the inversion of Tf occurs at a point P
where Tf = -10.9O C and ps = 14.5 kgm- 3. As a consequence, there results a,, = -2856 bp, and
the order of the parameter /3 defined in (5.4) is dominant; this fact will imply a rigorous estimate.
System (5.5) can be reduced to canonical form
aTf -
a2 co(H - Tf) - YCIV, (5.6a)

-av
az = c2 (H - Tf) + clv, (5.6b)
where
Q)(S) = -pg 2 0, Cl = -&, Q(6) = $g$ (5.7)
The characteristic equation of (5.6) is X2 + (cg - ci)X - (cg + 7cz)ci = 0, and has the two real
roots Xi = -a, Xz = b, with a and b positive and given by

a= ; dz+co-Cl], b= f &co+cl]. (5.8)


O[ O[
Along each pipe of ?X for z = 0, let Tf(0) = TO, v(0) = vc and denote by

Bo = (a + V-l [(a - co) To - YCIVO] , A0 = To - Bo, (5.9)


b+co
ga(z) = a,+b e-a(z-)H(x) ds, ga(z) = bz eb(-)H(z) dz. (5.10)
J0 s0
Then, the solution of the initial-value problem related to (5.6) is
Tf = Agema + BOebZ + ga(z) - gb(z), (5.11a)
v = (ycl)- [(a - ca) Tf - (a + b) (BoebZ - gb)] . (5.11b)
REMARK 5.2. The formulae (5.11) describe the exact behavior of Tf and v along each pipe of ?7$
both for y = y and for y = y. What distinguishes the two cases is the initial state (To, vo) and
the dependence of the constants a, b on S = y - y when y = y.
The exact solution (5.11) can be simplified as the orders of its terms are very different. For
instance, when Go is 252 and 6 = 0 (initial pipe), one has: a = 0.273 2 ~0, b = 0.0034, and
b(a--4 2 2.7. 10-6, bfQ,1 (5.12)
a+b b+a-
As a consequence, the term gb in (5.10) is negligible with respect to ga and by (5.11a), one
deduces %
Tf = (TO - Bo) emaz + BoebZ + a e-@-)H(z) dz. (5.13)
J0
Then, except for terms which influence the fifth or sixth digit only, this formula determines
explicitly the law of variation of Tf along each pipe of the vapour region. Moreover, the above
equations make explicit all the other unknowns. For instance, by (5.11b), one obtains the explicit
behavior of v and by (4.5),(4.3), one deduces T, and h,. In order to compare analytical and
numerical results, the explicit formulae (5.13),(5.11b) have been applied to the last pipe of % and
the behavior of Tf and v is sketched in Figures 4 and 5.
The related boundary-data are tif = 0.015 kg s- ', va = 250 m3/h, T, = 23O C, w, = 0.0105,
T evapor, = -10 C. The condenser outlet temperature and the condenser outlet pressure are 30 C
and 14 bar.
144 C. APREA AND C. RENNO

am-
=%mJ
,~
,~ i .<
.<
,I,I .:
Figure 4. Refrigerant temperature.

SPECIFIC VOLUME
0.071-

0.073s -

0.073 -

0.0725 -

0.0715 -

nn7,
. ..I.. - . , . I. I I I, a I. I. I.. . . I

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.r.s 0.5 0.56 0.6 0.85 0.7 0.75 0.8 0.85 0.9 0.9s 1
z

Figure 5. Specific volume.

6. CONTROL ASPECTS
Consider now the refrigerant temperature excursion in the single-phaseregion. Owing to the
order of p (Remark 5.1), there results cu% ~0, b E cl, with cl near to zero. As a consequence,the
initial constants Ao, Bs defined in (5.9) have different orders and Bo is negligible with respect
to AL,. Further, in (5.13), H can be approximated by a mean value HM; for instance, when
Gs = 252, H is included between 5.2 and 6.3. Then, the prevailing term in (5.13) is

Tf g Toema + HM (1 - ema*) , (6.1)


Expansion Valve Control Law 145

and this approximation can be applied both to y = y and y = y, as S is very small and
x(6) = x(O).
According to the numerical results, let P(ys, zs) be the point of the penultimate pipe where
the inversion of Tf occurs and let

t,=e-Zo, To = Tj (Y, ~0) , TI = Tf (y, 0) . (6.2)

By (6.1), one has

Tf (y, JJ) = Tgewaeo + HM (I- eeaeo) , (6.3a)


Tf (Y, 1) = TIeeat + HIM (1 - ceae) . (6.3b)

As Tl = Tf(y,O) = Tf(y,L), by (6.3), one obtains the value of Tf at the exit of the evaporator
in terms of To,
Tf (y, l) = Toe -a(e+eo) + HM 1 _ ,-a(e+eo) .
(6.4)
( >
Then, if we put
AT, = Tf ( y, l) - To, x = lo + c, (6.5)

the dominant term of the degree of superheating is given by

AT, = (HM - TO) (1 - epaA) , (6.6)

with
(6.7)

So, the formulae (6.6),(6.7) reveal the explicit connection among AT,, Go, Di, and the total
length X of the superheated circuit. Further, in (6.7), the term in square brackets depends on the
refrigerant properties, while x(S) is related to 6 and to other model parameters (see (4.4)). At
last, HM depends on the air temperature and the moist air properties.

7. CONCLUSIONS
We have discussed a first numerical-analytical approach to the study of a control problem for
an air-cooled evaporator mounted in a climatic chamber, where high variations of air temperature
and humidity occur. The mathematical model takes into account both the refrigerant pressure
drops inside the tube and the moisture condensation on the air side. The properties of the steady-
state, which represents the initial state of the dynamical system, have been analyzed by coupling
numerical and analytical methods. The numerical solution has shown that the behavior of the
model is congruent with the current plant values. Moreover, the numerical results have allowed us
to locate the interface and to verify the validity of the linearized model for the vapour region. In
the analytical framework, both the two-phase and the single-phase regions have been examined.
The qualitative analysis of the nonlinear two-phase flow has achieved the estimates (4.14), which
express the dependence of the solution on the data. As for the vapour region, the solution of
the linear model has implied the formula (5.13), which gives the explicit law of variation of
the refrigerant temperature along each pipe of the superheated circuit. As a consequence, the
explicit dependence (6.6) of AT, on the mass flow-rate and the other model parameters has been
obtained.
The next developments of this analysis will concern the study of the transient response to
variations of independent control parameters.
146 C. APREA AND C. RENNO

REFERENCES
1. X. Jia, C.P. Tso, P.K. Chia and P. Jolly, A distributed model for prediction of the transient response of an
evaporator, Znt. J. of Refrigeration 18 (5), 336-342 (1995).
2. J. Nyers and G. Stoyan, A dynamical model adequate for controlling the evaporator of heat pump, Int. J.
of Refrigeration 17 (2), 101-108 (1994).
3. H. Wang and S. Touber, Distributed and non-steady-state modelling of an air cooler, Znt. J. of Refigemtion
14, Q&111 (March 1991).
4. J.S. Van der Meer, Simulation of a refrigerant evaporator, Ph.D. Thesis, Delft University of Technology
(October 1987).
5. V.P. Carey, Liquid- Vapour Phase-Change Phenomena, Hemisphere, New York, (1992).
6. G.F. Hewitt, Hemisphere Handbook of Heat Exchanger Design, Hemisphere, New York, (1990).
7. N. Bellomo and L. Preziosi, Modelling Mathematical Methods and Scienlific Computation, CRC Press, Boca
Raton, FL, (1995).
8. R. Mastrullo and P. Mazzei, Introdudone All Anal& Exergetica di Impianti Termici, Giannini, Napoli,
(1986).
9. R. Mastrullo and P. Mazzei, Programma interattivo per il calcolo delle proprieta di refrigeranti e lanalisi
di cicli inversi, Fkddo 1, 61-70 (1988).
10. Ashrae Brochure on Psychrometry, Prepared by Ashrae technical committee on psychrometry (1964-1972).
11. C. Canuto and I. Cravero, A wavelet-based adaptative finite element method for advection-diffusion equa-
tions, Math. Models Meth. Appl. Sci. 7, 265-290 (1997).
12. S.A. Lifits et al., Quasi-aefftz spectral method for Stokes problem, Math. Models Math. Appl. Sci. 8,
1187-1212 (1998).
13. Thermodynamic and physical properties of R22, International Institute of Refrigeration, Paris (1982).
14. F. de Rossi, R. Mastrullo, P. Mazzei and M. Sasso, EASY, Cuen Naples, (1991).
15. S.G. Kandlikar, A general correlation for saturated two-phase flow boiling heat transfer inside horizontal
and vertical tubes, ASME Journal of Heat Transfer 112, 219-228 (1990).
16. S.G. Kandlikar, Flow boiling maps for water, R-22 and R134a in the saturated region, In gth ht. Hent
tinsfer Conference, Jerusalem (August 1990).
17. C. Aprea, V. Betta, A. Greco and R. Msstrullo, Experimental comparison of R22 and R407c heat transfer
coefficients, In Int. Conference on Heat Exchangers for Sustainable Development, Lisbon, (June 1998).
18. M. Turaga, S. Lin and P.P. Fazio, Correlation for heat transfer and pressure drop factors for direct expansion
air cooling and dehumidifying coils, ASHRAE tinsaction 92, 616-629 (1988).
19. M. Turaga, S. Lin and P.P. Fazio, Performance of direct expansion plate finned tube coils for air cooling
and dehumidifying coils, Int. J. of Refrigeration 11, 78-86 (1988).
20. Fundamental Handbook, ASHRAE, (1993).
21. F.M. White, Fluid Mechanics, McGraw-Hill, (1994).
22. C.F. Gerald and P.O. Wheatley, Applied Numerical Analysis, Addison-Wesley, (1994).

Das könnte Ihnen auch gefallen