Dynamic Simulation and Optimal Matching of A Small-Scale
Dynamic Simulation and Optimal Matching of A Small-Scale
Dynamic Simulation and Optimal Matching of A Small-Scale
This paper describes a principle and method of optimal matching to reduce energy consumption in smallscale refrigeration systems, based on systems analysis. A knowledge of the dynamic characteristics of a refrigeration system is important for predicting the performance of the system. A simulation model of a refrigeration system consisting of a compressor, an evaporator, a condenser and a capillary tube has been established to illustrate optimal matching. For each component a mathematical model has been developed, in which the concept of transient and distributive parameters is introduced. On the basis of dynamic simulation, a method of optimal matching to minimize power consumption is recommended. To test the reliability of the theoretical models, an experiment was carried out on a small-scale refrigeration system. The experimental data were compared with the theoretical results and it is shown that the theory is valid.
Refrigeration systems consisting of a hermetic compressor, a tube-coil evaporator, a capillary tube and a tube-coil condenser are widely used, especially in smallscale refrigerators and air conditioners. As refrigeration systems consume a large amount of energy, knowledge of the operating behaviour of the refrigeration system is important in energy efficiency studies, not only for predicting the system performance, but also for optimizing the combination of system components during the design process. Heat and mass transfer are involved in the refrigeration process. As various parameters such as temperature, pressure, enthalpy and mass flow-rate vary at different times and locations, the refrigeration process is both dynamic and distributed in nature. All the parameters are functions of time and space. To increase the coefficient of performance (COP) of a refrigeration system, the matching characteristics between each component should be thoroughly studied. If the mathematical models of each component and the whole system correctly reflect the actual refrigeration behaviour, it is desirable for system designers to be able to design a refrigeration system on a computer. Many mathematical models have been proposed in the past for modelling refrigeration systems. Research into 0140-7007/91/060329~)7 1991 Butterworth Heinemann Ltd and I I R
simulation models for refrigeration systems falls into two categories: (1) small changes occurring on the refrigerant side, such as an instability phenomenon called 'hunting' or behaviour during an operating mode of capacity control; and (2) large changes occurring on the refrigerant side, such as in the start-up modes. Most research is concerned with the first category. Yasuda et al? investigated a simulation model of a vapour compression refrigeration system, based on conservation of energy in the compressor, condenser and evaporator. By utilizing lumped models for the condenser and distributive models for the evaporator, the transient behaviour of a laboratory-scale refrigeration system was predicted by MacArthur 2. The second type of research is illustrated by the work of Dhar and SoedeP, who carried out a transient analysis of a vapour compression refrigeration system in 1979. Chi and Didion 4 developed a simulation model of a heat p u m p by utilizing a lumped model for each component. The purpose of this paper is to present a simulation model of a small-scale refrigeration system in which the concept of transient and distributive parameters is introduced. Based on the conservation and balance equations for individual components of the system, mathematical models of each component are derived and combined to
329
Nomenclature
A C
C
D E
G g h L 1 m m P Q q T t U
U
V
V
Area (m 2) Flow capacity of valve (m 3 s-~) Specific heat (J kg-J K-1) Diameter of tube (m) Energy transferred by mass flow (k J) Interfacial drag force (N m -3) Mass flux (kg m 2 s-1) Gravitational constant (m s -2) Enthalpy (kJ mol-i) Length of connection rod (m) Pipe length (m) Mass in cylinder (kg) Mass flow-rate (kg s-1) Pressure (N m-2) Heat (J K - i ) Heat flux per unit length Temperature or period (K) Time (s) Internal energy Specific internal energy for compressor model or velocity of refrigerant for heat exchanger model Volume (m 3) Specific volume Power of electric motor (W)
X ~X
P
CO
Distance along the rube (m) Heat transfer coefficient Density (kg m-3) Angular Speed of compressor shaft Rotation angle of compressor shaft
Superscripts
i in o out
Subscripts
Ambient Cylinder Compressor Condenser Capillary Discharge Evaporator Inside Outside Refrigerant Suction Pertaining to the valve Wall
simulate a small-scale refrigeration system. From the detailed results of the simulation, a method of optimizing refrigeration systems is described. This paper also substantiates the theoretical investigation by reporting the results of an experiment. It is very difficult to finely control the variations of a certain parameter in a smallscale refrigeration system with few control devices, such as a capillary tube. However; if auxiliary control devices are installed in such a small system, the data would be difficult to compare with the model, as the auxiliary control devices would affect the experimental data. Therefore, the second type of simulation is utilized, rather than the first.
Description of a simulation model
Four mathematical models of the system components and a mathematical model of the combined system were developed. The four mathematical models are of a hermetic reciprocating compressor, a tube-coil condenser, a capillary tube and a dry expansion evaporator. The actual system used in this study is shown in Figure 1.
Compressor model
butive model for a hermetic compressor. This is because, in such a model, the problem will be three-dimensional and a lot of computation time will be required. Other models are one-dimensional and relatively easy to solve. It would be unsatisfactory for the majority of the time to be spent on the compressor model alone in a simulation process. For this reason, a lumped model was developed for the hermetic compressor. The compression step can be described by a lumped model for an open-type compressor. The model of an open-type compressor was derived based on the following assumptions: (1) the state of refrigerant in the cylinder is homogeneous; (2) the compressor shaft rotates at a constant angular speed; (3) the discharge and suction valve are two-position elements, and the dynamic behaviour of the valves is ignored; (4) there is no pressure vibration in the plenum chamber or suction and discharge lines; and (5) heat transferred from the cylinder wall to the refrigerant in the cylinder or vice versa can instantaneously affect all the refrigerant in the cylinder. Basic equations were derived from the physical laws of mass and energy conservation:
dU=dQ- PdV+dE dm = dms + dmd
(1)
The changes which occur as the refrigerant flows through the hermetic compressor can be divided into three steps: (1) suction heat exchange; (2) compression of the gas; and (3) discharge heat exchange. These are illustrated in Figure 2. Each step influences the other two steps. For instance, the discharge temperature influences the suction heat exchange and then changes the suction temperature, and the suction temperature changes the mass flow-rate of the compressor. It is inadvisable, although more precise, to use a distri330 Int. J. Refrig. 1991 Vo114 November
For the four phases of a compressor cycle, the basic equations were specified as follows: Compression phase: mdu = dQ - Pd V Discharge phase: mdu = dQ - Pd Vd + Pvdm Expansion phase: mdu = dQ - Pd V Suction phase: mdu = dQ - Pd Vs + ( h s - u)dm (2)
(3)
(4) (5)
Evaporator
/ /
//
~.~
\
z~
?
P.t
~Copillary Compressor!
V
Figure 1 Schematic diagram of experimental system Figure l Schema du syst~me expbrimental
dQ/dt =
~clnAcln(Tw- Tr)
(7)
t
Figure 2 Model of hermetic compressor Figure 2 ModUle du compresseur herm~tique
and d m / d t is determined by the equation of the mass flow-rate through the valves:
dm/dt = ~
(8)
J- . . . . .
.-
y __~
dQ
The entire hermetic compressor was considered as an open-type compressor + suction heat exchange + discharge heat exchange. Considering that a typical hermetic compressor consists of an electric motor, a shell, a cylinder and its pertinent parts, and oil, the entire compressor can be divided into two parts from the viewpoint of heat exchange. One part is made up of the cylinder and its pertinent parts, and the other of the electric motor, the shell and the oil. According to the laws of energy conservation: (co V)l,d Tctn/dt = W m - Q 2 - Q3 - Q7 (9)
Figure 3 Simulation model of refrigeration system Figure 3 ModUle de simulation du syst~me frigorifique
for the cylinder and its pertinent parts, and: (c o V)d T/dt = Q3 + Q5 + Q6 + Q1 - Q4
Where d V/dt is determined by the kinetic equation of the piston, as illustrated in Figure 3:
d V/dt=
(lO)
1/ ~2 .I . . . . . [
k2Sin0Cos0 ] (1 - X2Sin2~.)'/2]
(6)
for another part, where W m is the friction work, QI is the heat transferred to the environment, Q2 is the heat exchanged between refrigerant and cylinder while the refrigerant impinges the head of the cylinder, Q3 is the
Rev. Int. Froid 1991 Vol 14 Novembre
331
heat exchanged between the cylinder and the refrigerant out of the cylinder, Q4 is the heat absorbed by the incoming refrigerant, Q5 is the heat transferred from the discharge tube to refrigerant, Q6 is the heat produced by the electric motor, and Q7 is the heat exchanged between the refrigerant inside the cylinder and the cylinder wall.
Modelling o f an evaporator and a condenser
The system of partial differential equations represented by Equations (11), (12) and (14) can be solved numerically by dividing the entire tube length into a large number of sections and applying the finite difference technique. Although this is time-consuming to calculate, it is more precise than lumped models.
Modelling o f a capillary tube
As evaporation and condensation are very complicated, the following assumptions are proposed to simplify the mathematical models: (1) refrigerant flowing in the tube is one-dimensional along the tube axis; (2) in any crosssection of the evaporator and the condenser, the refrigerant is homogeneous and, hence, values of any parameter are identical in any cross-section of the ducts; and (3) energy and mass transfer occur only by convection. The dynamic, one-dimensional forms of the conservation and balance equations for the refrigerant flowing in the duct may be written as: Conservation of mass equation: Op/~t + Q(up)/Ox = 0 (11)
To model the capillary tube, the same assumptions as for the evaporator and the condenser are used, and, hence, the same balance and conservation equations can be employed. In contrast to the entire refrigeration system, the time constant of the capillary tube is much shorter when the flow in the capillary is in a dynamic state. This means that the timewise partial differential terms in Equations (11), (12) and (14) are zero, as O/Ot=O. Hence we have:
d(up)/dx = 0 d(uG + P ) / d x = f + pgcos0 d(uph*)/dx = q + u ( f + pgcos0)
(20)
(21) (22)
Momentum equation: #G/Ot + O(uG + P)/Ox = f + pgcos0 (12) w h e r e f i s the drag force per unit volume exerted on the refrigerant in the direction of flow by the tube wall, and can be stated as: r
;=
The following equations can be derived from Equations (21) and (22): ~v 0h dP (25)
441
Ui'~ pu"
(13)
For horizontal flow cos0=0, and for vertical upflow cos0 = 1. A control volume energy balance in the tube of length dx with the tube wall yields: Energy equation: 8(ph* - P)/c~t + 8(uph*)/~x = q + uOC+ pgcos0) (14)
Gdh/dx- udP/dx = q
The system of the above equations can be written in the following vector form:
AdX/dx = B
(26)
where q denotes the heat transfer per unit volume from the tube wall to the refrigerant, and can be stated as:
q = 4/D~r(Tw - Tr)
(27)
(15)
(16)
Considering matrix A, if the determinant of A, det(A)=0, the equation has no solution, and under this condition, the flow in the duct reaches its critical state, as shown below:
d e t ( A ) = - G ~c~p ! - G- u-v~}
The boundary conditions for the evaporator and the condenser are the enthalpy and the mass flow-rate at the tube inlet and the mass flow-rate at the tube outlet, given by: hb=o=h,,
GI, - o = G~.
3(0v /
_,
= -G ~ (17)
\0v!
-G
(28)
and 2 c~h G[,_L= Gout (19) where h~,, Gin and Gou~are known for the models of the evaporator and condenser, respectively. c~h (29) where s represents an isentropic process, the determinant A becomes:
332
Tomb
~
t
mb
T(*C)
l_
_1
'
Time, t
(30)
If u/c= 1, then det(A)=0. This implies that Equation (26) has no solution if the velocity of flow in the capillary tube reaches its local sonic speed. As this condition is often found at the outlet of the capillary tube, by using this condition, the mass flux of refrigerant in the capillary tube can be determined. The system of two differential equations with the initial condition of the enthalpy and the pressure at the capillary intake was entered into the computer program for numerical solution; the numerical integration was implemented by a Hamming procedure with the RungeKutta procedure for the first three steps. The program automatically searches for that value of mass flux which satisfies the condition specified by the equation det(A) = 0 at the end of the capillary.
functions for calculating the refrigerant properties. For consistency with the experimental validation, a study of performance in the start-up mode was undertaken. It was assumed that the internal refrigerant circuitry was stable and that the pressure and temperature were equal throughout the system. The numerical simulation was arranged such that: . If the nth step had been completed, the initial input values of the compressor for the (n + 1)th step were assumed to the same as the input values of the nth step, that is:
1 7 in 3 (0) __
,~com,,+l - (Ucom)n 2.
in
(31)
The output values of the compressor, the condenser, the capillary and the evaporator were computed in turn and another group of input values was obtained
(U, comJ,+t. i n "t (l) f o r the c o m p r e s s o r f o r the ( n + ] ) t h step, ~
( l f in "~ (0) with ~,~comJn+l, [rr in $ (1) if the difference Comparing ~.'-'com/n+l was not within an acceptable precision
3.
(32)
was substituted for \~rr in tl co m /~ n (0) + 1, where R was a relaxation coefficient. Steps 2 and 3 were repeated until the condition that I f in ~ (1) . , in ~, (0) ~'comJ.+ 1 and -~Ucom]n+ 1 were sufficiently similar was satisfied.
(33)
where W is the input power of the electric motor of the compressor, which can be calculated by the simulation model. The conditions and parameters of optimization vary from problem to problem. As an illustrative example, it
333
Simulation of a small-scale refrigeration system: Z - J . Chen and W.-H. Lin is assumed that a compressor, an evaporator and a condenser have already been selected, and that only the capillary is not determined. The diameter and length of the capillary tube are the parameters of the optimization and the following two inequalities are the conditions of optimization:
dcap > 0
80 a
6O
Z 50
(34) (35)
:5o 2o
0 1.44 2.88 4.:52 5.76
/cap> 0
:5
Mathematically this optimization problem can be represented by the following mathematical model:
Z2
Time (x103 s)
30 b
min[l fT , W(dcap,lcav)dt ]
s.t. dcap > 0 l~p>O
(36)
I0 ~ o -Io -20 -
B(dcap,lcap)=
-- ln(dcap) -- 1n(l~p)
(37)
0 1.44 2.88 4.32 Time (x 103s) 5.76
2
7.2
By introducing:
(38)
and searching for the minimum of Fr(dcap,lcap) until r > 0 and r ~ 0 are satisfied, the solution of the optimal matching problem is obtained.
Results and discussion
To verify the accuracy of the simulation model, the numerical results were compared with experimental data acquired using the equipment with 284 thermocouples and two piezoelectric crystals shown in Figure 1. For a small-scale refrigeration system, it is difficult to undertake experiments such as changing the mass flowrate by changing the position of the valves because additional valves will incur an additional pressure drop, which is undesirable in the system, and will impair the accuracy of the experimental data. It is better to maintain the original refrigeration system, adding as few experimental instruments as possible to minimize the effect of experimental instruments on the system. A method consistent with this requirement is to use the start-up mode and to compare the predicted and experimental values of temperature and pressure versus time. Experiments in the start-up mode were therefore conducted and data were obtained by a data acquisition system and recorded by a computer. The predicted and experimental values of pressure and temperature versus time in the start-up mode are shown in Figures 6 and 7. These figures demonstrate that the predicted behaviour of the refrigeration system in the start-up mode is in overall agreement with the data obtained. The comparatively larger deviation of refrigerant temperature discharged from the compressor shell shown in Figure 7 may be ascribed to the uncertainty of the heat transfer in the compressor shell, as the heat
Figure 6 Variation of temperature with time for a small refrigeration system. (a) Start-up procedure. (1) Compressor discharge tube temperature; (2) compressor shell temperature; (3) inlet temperature of lower condenser; and (4) inlet temperature of main condenser. (b) Regular on off procedure. (l) Refrigerator air temperature; (2) freezer air temperature. (- - - ) Calculated; ( 1 ) experimental Figure 6 Variation de la temperature en fonction du temps pour un systOme frigorifique de petite dimension. (a) procedure de dkmarrage. (1) tempOrature de la conduite de refoulement du compresseur; (2) temperature de l'enveloppe de compresseur; (3) temperature d'entr~e du condenseur inf~rieur; et (4) temperature d'entrke du condenseur principal. (b ) fonctionnement rkgulier par tout ou rien. (1) temperature de I'air du r~frig~rateur; (2) temperature de I'air du congklateur. ( . . . . ) calcuI~es; ( - - ) exp~rimentales
15 12
I
0 1.44
I
2.88
I 4.:52
I 5.76
7.2
Time ( x l O 3 s )
Figure 7 Predicted ( ) and experimental (--) variation of pressure with time during start-up mode. (1) Condenser pressure; (2) evaporator pressure Figure 7 Variation prbvue ( -) et expbrimentale ( - - ) de la pression, en fonction du temps au cours du mode de dkmarrage. (1) pression du condenseur ; (2) pression de l'~vaporateur
transfer in the hermetic compressor is complicated and insufficiently known, especially when oil is involved.
334
Acknowledgements
This research is supported by the N N S F of China and the Doctoral Project Foundation of the State Education Committee.
References
1 2 3 4 Yasuda,H. et al. Simulation model of a vapour compression refrigeration system A S H R A E Trans (1983) MacArthur,J. W. Transient heat pump behaviour, a theoretical investigation Int J Refrig (1984) 7 123-132 Sodel,D. M. Transient analysis of a vapor compression refrigeration system International Congress of Refrigeration, Venice.
Italy, 1979
Chi, J. and Didion, D. A simulation model of the transient performance of a heat pump Int J Refrig (1982) 5 176-184
335