# Cooling of a reciprocating compressor through oil atomization in the cylinder.

An analysis of the influence o foil atomization in the cylinder of
a reciprocating compressor is presented in this paper. During
compression, oil atomization enhances heat removal from the refrigerant
gas. This cooling effect, which eventually results in a global
temperature decrease of the compressor parts, aims primarily at reducing
the refrigerant superheating in the suction system and inside the
cylinder, which is largely responsible for overall energy losses and a
decrease of the volumetric efficiency. A prototype was constructed and
tested with R-134a in a hot gas-cycle calorimeter. A significant
reduction of the compressor thermal profile has been achieved, with the
largest variations of 30.9[degrees]C (55.6[degrees]F) and 23.6[degrees]C
(42.5[degrees]F) in the discharge chamber and cylinder wall,
respectively. A major dependence of the compressor efficiency parameters
on the refrigerant solubility in the oil has been observed. A simulation
model using an integral control volume formulation for mass and energy
conservation in the cylinder and in other compressor components is
proposed. Based on this model the effect of oil atomization on the
compressor performance is presented and discussed in terms of oil
injection parameters, such as nozzle position, oil temperature and flow
rate.

Introduction

In household refrigeration compressors, a significant portion of the energy losses is associated with refrigerant superheating along the suction path inside the compressor and during gas compression inside the cylinder. Broadly speaking, the energy losses in a compressor are divided into (i) electrical, (ii) mechanical, (iii) thermodynamic, and (iv) cycle losses (Possamai and Todescat 2004). Thermodynamic losses involve the refrigerant flow inside the compressor. They are mostly associated with valve flows (viscous friction and backflow), gas leakage through the piston-cylinder gap and refrigerant superheating along the suction process and during compression. According to Ribas (2007), for R-134a household refrigeration compressors at the ASHRAE low back pressure (LBP) condition, around 50% of the thermodynamic losses are due to suction gas superheating. This superheating causes a reduction of the volumetric efficiency and an increase of the compression work per unit mass (Gosney 1982).

The reversible work input associated with an internally reversible steady-flow device is given by:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

where v is the specific volume of the working fluid, p is the pressure and 1 and 2 refer to the initial and final states. Even in the presence of internal irreversibilities (e.g., friction), where a departure from the ideal input work takes place, it becomes clear from Equation 1 that the specific volume must be as low as possible in order to minimize the work input. As the specific volume of a gas is directly proportional to the temperature, gas cooling during the compression process can be an alternative for reducing the compressor work and improving the energy efficiency of refrigeration systems. Basic calculations (Kremer et al. 2007) showed that, for a single stage refrigeration system operating with R134a between evaporating and condensing pressures of -27[degrees]C (80.6[degrees]F) and 42[degrees]C (107.6[degrees]F), isothermal compression can be as high as 16% more efficient than isentropic compression.

Many aspects of compressor cooling have been addressed in the open literature. Dutta et al. (2001) investigated theoretically and experimentally the effect of liquid refrigerant injection in scroll compressors using R-22. A decrease in the vapor temperature during compression was observed. However, it was reported that the extra work required to compress the vaporized refrigerant gave rise to a decrease in the compressor performance. The experimental and numerical studies of Coney et al. (2002) quantified the decrease in power consumption (approximately 28%) associated with the atomization of water in air reciprocating compressors. Ooi (2005) evaluated numerically the injection of liquid refrigerant in a rotary compressor, taking into account the influence of the nozzle diameter and its position in the compression chamber. He concluded that it is more advantageous to inject the liquid shortly before the discharge of the gas to benefit from the latent heat transfer without the penalty of compressing a volume of gas that is not used for generating cooling capacity. Meunier (2006) conceived a refrigeration cycle with a coefficient of performance (COP) equal to that of the Carnot refrigerator in which superheated vapor is compressed isothermally. However, no attempt to model the compression process has been proposed. Bonjour and Bejan (2006) demonstrated the existence of an optimal configuration for the distribution of cooling water in a multi-stage ammonia compressor that minimizes the compressor power. Wang et al. (2008) investigated theoretically two performance improvements for reducing the compressor power for several refrigerants. The first option was to cool the motor by external means other than using the suction gas. The second was to combine the processes of isothermal and isentropic compression. Their analysis demonstrated that the first option is more advantageous for LBP rather than high back pressure (HBP) applications for all refrigerants investigated. The second option showed that the compression power can be reduced by up to about 16% depending on operating conditions and working fluid. Toublanc (2009) investigated the atomization of lubricant oil during gas compression as a means of achieving near-isothermal compression in C[O.sub.2] refrigeration systems. He developed a mathematical model and carried out an extensive analysis of the effect of the droplet diameter, oil injection flow rate, compressor type (reciprocating or scroll) and compressor speed. For a reciprocating compressor of 8.4 kW cooling capacity, he reported a theoretical improvement of the COP of more than 30%, when the injected oil flow rate was three times larger than the gas flow rate. However, the volumetric capacity saw a reduction of 14%. Shuaihui et al. (2010) developed a mathematical model based on energy and mass balances to quantify the efficiency improvement associated with water cooling of scroll compressors. They showed that the isentropic efficiency can be improved by approximately 7%, and the gas discharge temperature can be reduced by 23[degrees]C (~41[degrees]F) when the cooling water is at 30[degrees]C (86[degrees]F). Little improvement was observed in the volumetric efficiency as a result of compressor cooling.

In the present paper, atomization of lubricant oil (ISO 10) during compression of R-134a in a household reciprocating compressor is investigated both experimentally and theoretically. In addition to lubrication, oil is used for sealing and cooling, and is heat transfer effective in compressors where leakage rates are high and where sequential injection can be applied. It can also be used in situations where extensive increases of temperature are present. Due to the small size of droplets, oil atomization promotes a heat transfer surface enhancement which favors heat removal from the vapor. The resulting cooling effect also contributes to lowering the overall thermal profile of the compressor parts and, consequently, the initial compression temperature. In quantitative terms, for R-134a, atomization of oil at 45[degrees]C (113[degrees]F) with an average flow rate of 0.916 kg/h (2.02 lbm/h) gave rise to a maximum temperature reduction of 25[degrees]C (77[degrees]F) at the top dead center (TDC). The initial compression temperature decreased from 58.8 [degrees]C (137.8[degrees]F) to 44.6[degrees]C (112.3[degrees]F), illustrating the large potential for reducing losses due to refrigerant superheating. Despite the encouraging temperature results, the compressor efficiency parameters, such as the compressor power, the cooling capacity and the COP have proven to be very sensitive to the solubility of the refrigerant-oil pair. The compressor performance deteriorates when an oil with a larger solubility was used. A comprehensive mathematical model has been proposed, which shows, in general, a satisfactory level of agreement with the experimental data.

Materials and Methods

Compressor

The compressor prototype shown in Figure 1 has a split crankcase sealed with fiat face bolt flanges and a silicone rubber gasket. A commercial hollowcone nozzle (Lechler 212.004) was mounted flush on the wall of the compression chamber. Due to the nozzle and housing dimensions and space restrictions, the distance between the nozzle orifice and the valve plate is approximately 13 mm (0.51 in.) (see Figure 1c). The nozzle is fed by a pressurized 2-mm (0.079 in.) I.D. oil line which is connected to the oil separator of the calorimeter loop (see below).

The temperatures of the compressor components were measured with 13 T-type thermocouples, according to the schematic diagram of Figure 2. The temperature at which the oil enters the cylinder is assumed equal to that of the nozzle housing. The temperature of the gas entering the cylinder is assumed equal to the one measured at the suction chamber. Due to space restrictions, the temperatures of the cylinder wall and of the main bearings are measured by inserting the thermocouples into holes drilled in the compressor block, so that the tip of the sensor is located at approximately 1 mm (0.04 in.) from the surfaces of the cylinder and bearings, respectively. As far as the measurement of the crankcase gas temperature is concerned, no attempt has been made to account for the influence of the oil adhering to the tip of the thermocouple. It was assumed that this influence is negligible because of the small thickness (i.e., negligible thermal resistance and capacity) of the oil film formed on the solid surfaces inside the compressor. It should be mentioned that this is the only point at which the gas temperature is measured inside the compressor. All of the remaining temperature measurements (excluding the above mentioned cylinder wall and bearings measurements) were carried out with thermocouples attached to solid surfaces that were exposed to the low-pressure crankcase internal environment. Therefore, since oil atomization takes place in the cylinder and the atomized oil leaves the compressor through the high-pressure discharge system, it is safe to state that the atomized oil does not interfere with the temperature measurements. This is not to say that the oil that is already present in the crankcase does not have any influence on the heat transfer measurements; it is just that this influence is expected to be approximately the same in the cases with and without oil injection in the cylinder.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Although the uncertainty reported by the thermocouple manufacturer is [+ or -] 0.2[degrees]C ([+ or -] 0.36[degrees]F), the one assumed in this study is [degrees]C (+l.8[degrees]F) due to errors originating from the positioning and attachment of the thermocouples to the surfaces. The instantaneous gas pressure in the cylinder was measured with a Kistler 601A absolute pressure transducer with a sampling frequency of 60 kHz and an associated uncertainty of [+ or -]0.12 bar ([+ or -] 17.4 psi).

The influence of the split crankcase, atomization nozzle and pressure and temperature sensors on the compressor performance has been evaluated by comparing the refrigerant mass flow rate of the modified compressor prototype, but without oil atomization, with that of a compressor without any modification. A reduction of the order of 10% in the mass flow rate has been observed. However, this does not represent a source of error in the present analysis, since the modified prototype (without oil atomization) is the baseline for quantifying the influence of oil atomization on the compressor temperature profile.

Superheated Gas Cycle Calorimeter

The compressor was tested with R-134a in a superheated ("hot") gas cycle calorimeter described schematically in Figure 3. The compressor discharge pressure was regulated via a hand-operated needle valve (DV). An oil separator was installed in the discharge line, and was kept at a temperature above the saturation temperature of the refrigerant at the discharge pressure by means of an electric heater wrapped around its external surface. Downstream of DV, at an intermediate pressure between suction and discharge, an accumulator was installed to dampen pressure oscillations. A Coriolis-effect meter (MicroMotion) is used to measure the mass flow rate and, downstream of it, another hand-regulated needle valve (SV) is operated to expand the gas from the intermediate pressure down to the desired suction line pressure. The experimental error associated with the mass flow measurement is [+ or -] 0.15%, according to the manufacturer. A fixed inlet temperature of 32[degrees]C (89.6[degrees]F) is set by an electric heater. The suction and discharge pressures are measured with HBM P3MB absolute pressure transducers (10 and 50 bar--145.0 and 725.5 psi--full-scale for the suction and discharge pressures, respectively) and the temperatures are measured with T-type thermocouples. The experimental uncertainty of the suction and discharge pressure measurements were estimated at [+ or -] 0.004 bar ([+ or -] 0.058 psi) and [+ or -] 0.1 bar ([+ or -] 1.45 psi), respectively. The compressor power consumption was measured with a Yokogawa WT2 l 0 electric power transducer with an estimated error of [+ or -] 3%. After being separated from the discharge gas, the oil is driven through the oil line via a thermostatic bath for temperature control before being reinjected into the compressor, if the oil valve (OV) is open. The experimental apparatus is fully integrated with a signal conditioning and data acquisition module.

[FIGURE 3 OMITTED]

Experimental Procedure

The experiments were performed with R-134a at evaporating and condensing pressures corresponding to -27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F). The room temperature was maintained at 25[degrees]C (77[degrees]F). Tests were carried out with and without oil atomization and, in the latter, the oil injection temperature was approximately 45[degrees]C (113[degrees]F). The experimental procedure is as follows. Refrigerant is charged into the calorimeter, previously submitted to a vacuum of 0.04 mbar (5.8 x [10.sup.-4] psi) to remove moisture and dissolved gases from the oil. After charging the system and switching on the compressor, up to 6 hours are needed for reaching a fully cyclic condition, due to the thermal inertia of the system. In this period, the needle valves DV and SV are constantly adjusted to maintain the discharge and suction pressures to within [+ or -] 1% of the specified condensing and evaporating pressures. In the tests with oil atomization, the temperature of the thermostatic bath is also continuously adjusted so as to keep the oil atomization temperature within the specified boundaries. The oil flow rate was not measured directly during the experiments. However, based on numerical analyses of the oil flow via computational fluid dynamics (CFD) (Kremer 2006) and on independent tests carried out outside the compressor with a constant nozzle pressure drop, with an oil injection temperature of 60[degrees]C (140[degrees]F), the oil flow rate was estimated at approximately 1 [+ or -] 0.2 kg/h (2.20 [+ or -] 0.44 lbm/h).

The fully cyclic criterion establishes that, over an uninterrupted 45 min interval, the variation of the temperature readings must be less than l[degrees]C (1.8[degrees]F) and the variation of the measured compressor power and calculated cooling capacity (i.e., the product of the measured vapor mass flow rate and latent heat of vaporization at -;27[degrees]C, or -16.6[degrees]F) should be less than 1%. When this condition is met, the temperature, mass flow rate and compressor power data are acquired and averaged for the next 10 min and the test is ended with the acquisition and averaging of 50 pressure-volume cycles.

An error propagation analysis (Kremer 2006) yielded uncertainties of [+ or -] 4 and [+ or -] 5% for the cooling capacity and coefficient of performance, respectively. At least five tests were carried out for each condition and the results reported in this paper consist of arithmetic averages of the data at each specific condition. Repeatability tests confirmed that the dispersion of the data with respect to the normal distribution was small, with deviations less than 3.5% of the average values.

Modelling

In-cylinder gas compression

The control volume formulation of the vapor compression process is based on the work of Ussyk (1984), with the required modifications to predict the effect of oil atomization in the cylinder. The mass and energy conservation equations for the vapor are given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (2)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (3)

where u = u (p, T) is the internal energy of the vapor and,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (4)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (5)

Physical properties were calculated from REFPROP 7.0 (Lemmon et al. 2002). The heat transfer coefficient between the cylinder wall and the vapor was obtained from Annand (1963).

Droplet atomization and heat transfer

The compression cycle is divided into n time steps of size [DELTA]t. The droplets are assumed spherical. The number of droplets injected into the cylinder in a given time step is given by,

N = 3 [[??].sub.oil] [DELTA]t/4[pi][R.sup.3]. (6)

In the following analysis, the subscripts m,n (as in [N.sub.m,n]) denote the group of droplets that were injected into the cylinder at a given instant m (i.e., the mth "family") and still are in the cylinder at an instant n > m. It is assumed that all droplets in each family m are identical with respect to their size, velocity and temperature. If coalescence, break-up, and evaporation are neglected, mass conservation for each family m gives,

d[N.sub.m,n]/dt= - [[??].sub.m,n]/[M.sub.m]. (7)

The mass flow rate of droplets belonging to a family m through the discharge valve is calculated based on their mass fraction in the cylinder, [x.sub.m,n], as follows,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (8)

where

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (9)

and F is a droplet discharge slip factor that accounts for the relative velocity between the vapor and the liquid in the discharge stream. When F = 1 (homogeneous two-phase flow), the dynamic mass fraction of droplets during discharge is equal to their in-cylinder mass fraction. As will be shown below, F was estimated from CFD simulations of the two-phase discharge process. The gas mass flow rate through the discharge valve is given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (10)

Due to the presence of droplets in the cylinder, a correction is applied to the instantaneous cylinder volume to obtain the volume of vapor,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (11)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (12)

where [V.sub.c] and its time derivative are calculated from algebraic relationships for the piston position and velocity as a function of the crankshaft angle (Ussyk 1984).

Due to the small size of the droplets and hence the large surface area per unit volume, it is expected that the largest portion of the heat transfer to the oil occurs while it is in the form of high-speed droplets inside the cylinder. The smaller droplets atomized by the nozzle will, in the long run, reach hydrodynamic and thermal equilibrium with the gas, while the larger ones will impact directly on the cylinder wall or on the top of the piston, forming a thin oil film. The oil film will cool the metal parts, but it will have a small impact on the cooling of the gas.

The velocity of the oil droplets is calculated from Newton's Second Law. Neglecting the effect of body forces, assuming that the velocities of the droplets are much larger than the in-cylinder vapor velocity and that the hydrodynamic interaction between neighbouring droplets is small, one has,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (13)

where [C.sub.D] is the drag coefficient calculated from the Morsi and Alexander (1972) correlation. The average droplet radius, [R.sub.m], was also estimated from the CFD simulations of the atomization process, to be described later. The initial condition for [U.sub.m,n] is calculated from,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (14)

where [C.sub.d] is the nozzle discharge coefficient, assumed equal to 0.78 (Lichtarowicz et al. 1965). In Equation 14, [[rho].sub.oil] is the oil density at the injection temperature. Since the vapor pressure of the oil is small, it is assumed that only sensible heat transfer takes place between the droplets and the gas. Temperature gradients inside the droplets and thermal interaction between droplets are ignored. Thus, the temperature of a droplet in family m at an instant n is calculated from,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (15)

where the oil physical properties are calculated at the mixing-cup temperature of the droplets in the cylinder, [h.sub.m,n] is the heat transfer coefficient between the gas and droplets calculated from the Ranz and Marshall (1952) correlation. The heat transfer rate between all oil droplets and the vapor is given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (16)

It is hypothesized that the heat removal from the cylinder wall by the droplets is small and can be neglected. In other words, the thin liquid film forming on the solid walls reaches thermal equilibrium with the wall instantaneously. The small number of droplets that remain in the clearance volume at the end of a compression cycle are assumed not to contribute, in the next cycle, to the transfer of heat from the gas. This is because they are assumed to reach thermal equilibrium with the gas at the end of the cycle and to remain with the same velocity as the cylinder gas during the subsequent cycle.

Suction and discharge processes

Valve displacement during suction and discharge is calculated via a one-degree-of-freedom model with natural frequency and damping coefficients specific for each valve. The resultant forces on the valves and their respective flow rates are obtained with effective force and effective flow areas derived from numerical simulations (Matos 2002). The total flow rate through the discharge valve is calculated assuming a homogeneous two-phase density based on the discharge mass fraction of each phase (Equations 8-10).

Droplet Atomization Parameters Obtained via CFD

In order to determine approximate droplet atomization parameters such as average droplet radius and discharge slip factor, CFD simulations of the compression and atomization processes were carried out using the commercial code FLUENT 6.2.16 (2005). A 2D moving-mesh domain (2100 volumes) was implemented to enable the simulation of the piston displacement inside the cylinder with a frequency of 60 Hz (see Figures 4 and 5). The suction and discharge valves were modelled as single axissymetric orifices that remain fully open during suction and discharge at the specified evaporating and condensing pressures of-27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F). The orifices are assumed to behave ideally in the sense that there are no friction losses involved, and that they close fully and immediately at pre-determined values of crank angle. Thc fluid flow was modelled using the renormalization (RNG) k-[epsilon] turbulence model (Yakhot and Smith 1992) and a built-in Lagrangean formulation (the Discrete Phase Model) was utilized to calculate the droplet field. The nozzle type and geometry were chosen so as to reproduce the atomization conditions of a commercial hollow-cone nozzle (Lechler 212.004). In the computational mesh, the atomizing orifice was positioned on the symmetry axis at the top of the cylinder. This arrangement is somewhat simpler than the actual cylinder-nozzle system geometry shown in Fig. 1c. However, its simplicity can be justified on the basis that the main purpose of the CFD simulation in the present work is to obtain approximate average droplet parameters for the cycle, and not to effectively solve the in-cylinder compression process.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Figure 4 presents the mean droplet diameter as a function of crank angle for an oil atomizer flow rate of 1.08 kg/h (2.38 1bm/h) at 55[degrees]C (131[degrees]F). The bottom dead center (BDC) corresponds to a crank angle of 0[degrees]C. The mean droplet diameter averaged over one cycle was 69 [micro]m (2.72 x [10.sup.-3] in.). Between 150[degrees] and 180[degrees] the average droplet diameter decreases to approximately 40 [micro]m (1.57 x [10.sup.-3] in.) due to break-up induced by vapor acceleration during discharge. As expansion begins, the average droplet diameter is large because of the low pressure difference between the hypothetical oil reservoir (assumed to remain at the condensing pressure) and the cylinder. In the droplet atomization and heat transfer model described previously, a constant droplet diameter of 69 [micro]m (2.72 x [10.sup.-3] in.) was adopted for the whole cycle.

In Figure 5, the total mass of oil in the cylinder is presented as a function of crank angle. The droplet discharge slip factor, F, is calculated from the maximum and minimum of the curve (i.e., the beginning and ending of the discharge process) together with the mass of refrigerant in the cylinder at those instants. At the present conditions, it was observed that the dynamic mass fraction of oil in the discharge stream was proportional to 65% of its overall mass fraction in the cylinder. Hence, F was assumed equal to 0.65.

Compressor thermal model

In order to provide the in-cylinder vapor compression and droplet heat transfer models with appropriate boundary conditions, the model described in the previous sections was incorporated into an existing lumped heat transfer (overall thermal conductance) model that integrates several compressor components as shown in Figure 6. In the original model without oil atomization (Todescat et al. 1992; Fagotti et al. 1994), the compressor was divided into seven control volumes (the vapor in the muffler and suction chamber, the vapor in the cylinder, the vapor in the discharge chamber, the vapor in the discharge muffler, the electric motor, the compressor shell and the overall compressor balance). Steady-sate energy balance equations were written for each component and solved coupled with the transient in-cylinder vapor compression equations. In the present paper, the compressor thermal model was extended so as to include two additional control volumes (the oil injection line and the discharge line) and to account for the flow of oil through the components downstream of the cylinder.

[FIGURE 6 OMITTED]

The energy balances in the control volumes are represented by the following general equation, where the specific terms in Equation 17 associated with each control volume are shown in Table 1.

The heat transfer rates in Equations 18-26 are modelled in terms of overall thermal conductances between the crankcase gas and the gas inside each corresponding component. The heat transfer rate between the crankcase gas and the suction muffler is given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (27)

where [bar.U[A.sub.s]] is a thermal conductance between the crankcase gas and the gas in the suction muffler. By the same token, the heat transfer rates between the crankcase gas and the oil injection line, the discharge chamber, the discharge muffler, the discharge line and the motor are given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (28)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (29)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (30)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (31)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (32)

The rate of heat transfer rate between the cylinder walls and the gas inside the cylinder is calculated based on the following relationship,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (33)

where Aw is the heat transfer area in the cylinder (walls and top of the piston), [T.sub.w] is the cylinder wall temperature and [bar.T] is the cycle average temperature of the gas inside the cylinder. [[??].sub.w] is the average heat transfer coefficient calculated via the Annand (1963) correlation with the Reynolds number calculated with the mean piston velocity.

In the energy balance associated with the compressor crankcase (i.e., the shell itself), Equation 26, the rate of heat transfer between the shell and the crankcase gas is also modeled in terms of an overall thermal conductance as follows,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (34)

where [T.sub.h] is the average temperature of the compressor crankcase. In Equation 25, [Q.sub.ee] is the rate at which heat is transferred from the outer surface of the compressor crankcase to the external environment. Thus,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (35)

where [T.sub.ee] is the temperature of the external environment and [bar.U[A.sub.ee]] is the corresponding overall thermal conductance.

The unknowns in Equations 18-26 are [T.sub.sc], [T.sub.oinj], [T.sub.w], [T.sub.dc], [T.sub.vb], [T.sub.ld], [T.sub.m] [T.sub.h], and [T.sub.ie]. The system is nonlinear in nature due to the dependence of the specific enthalpies with respect to temperature. The flow rates needed in the energy balances are determined via the models for the processes of suction and discharge and droplet atomization. The two-phase refrigerant-oil flow downstream of the discharge chamber is assumed homogeneous (i.e., no slip between the phases) and in thermal equilibrium. The overall thermal conductances in Equations 27-32 and 34-35 are estimated via experimental data (temperatures, flow rate and electric power) obtained for a reference operating condition of evaporating and condensing pressures corresponding to -27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F), for R-134a, and an oil inlet temperature, Top, of 36.2[degrees]C (97.2[degrees]F). Caution must be exercised when the estimated overall thermal conductances are extrapolated to operating conditions other than the reference case, since the associated variations in mass flow rates and temperatures affect the local heat transfer coefficients that, in turn, change the overall thermal conductances. However, it has been verified against experimental data that, when a new operating condition is within a few degrees from the reference case, the reference overall thermal conductances can be used without an appreciable error.

Solution Method

The compression cycle is divided into [10.sup.3] uniform time (or crank angle) steps and an explicit Euler method is used to integrate the time-dependent mass and energy conservation equations for the vapor and liquid droplet fields. The algebraic equations of the compressor thermal model are solved via a multi-dimensional Newton-Raphson method (Press et al. 1992). The two methods are coupled in such a way that, for each iteration in the compressor thermal simulation, a fixed number of in-cylinder compression cycles (say, from 8 to 12) are performed so that convergence is guaranteed in the solution of the periodic gas compression and droplet atomization equations. The convergence criterion for the compressor thermal simulation is based on a temperature difference tolerance (0.002[degrees]C, 0.0036[degrees]F) between successive iterations for each component temperature. If the criteflon is not met, then another set of in-cylinder cycles is executed, and so on until convergence is guaranteed.

Results

Experimental Results

Atomization of polyol ester oil

Experiments were initially conducted with polyol ester oil (POE ISO 10), for this is the lubricant which is normally used in R-134a systems in household refrigeration compressors. The temperature data for the compressor components are shown in Table 2 for the baseline case (i.e., no oil injection) and for cases where the lubricant oil was atomized in the cylinder at approximately 45[degrees]C (113[degrees]F).

As can be seen, atomization of the POE ISO 10 oil in the cylinder caused a significant reduction of the temperatures inside the compressor, more noticeably in regions which are in direct contact with the gas that is cooled down during compression, such as the discharge chamber and the cylinder wall. In these regions, the temperatures decreased by 27.4[degrees]C (49.3[degrees]F) and 25.3[degrees]C (45.5[degrees]F), respectively. The temperature reduction of approximately 3[degrees]C (5.4[degrees]F) at the suction muffler outlet is much smaller, but still large enough to reduce somewhat the inlet superheating losses. The values associated with the temperature decrease of the oil sump and compressor crankcase are quite similar, possibly due to the way in which the oil circulation takes place inside the compressor (i.e., an oil falling film is formed on the inner wall of the compressor crankcase).

Despite the significant reduction in the compressor thermal profile, the performance of the compressor deteriorated in comparison with the reference case. The indicated power, calculated via a numerical integration of the instantaneous pressure-volume experimental data, increased by 4.6%. It is believed that this increase is partially due to losses in the discharge system, which has not been designed to operate under significant oil flow. However, even when the power losses associated with the valve flows are disregarded (and the valves are assumed to behave as ideal orifices), the power effectively used to compress the gas was 2.1% larger in the atomization case. This was a surprising result, since the compression power was expected to decrease with the reduction of the compressor thermal profile. Nevertheless, this effect was observed consistently and seems to be associated with the solubility of R-134a in the POE oil, as will be explained later in this article. The solubility is defined as the equilibrium mass fraction of refrigerant in the liquid mixture. A more detailed discussion on this matter will be provided below. The compressor electric power consumption increased by 11.1%, which is mainly due to the increase of friction losses since the oil viscosity increases with the temperature reduction in the primary bearings and piston-cylinder gap.

The compressor cooling capacity decreased by 3.6% with respect to the reference case. The reduction of the compressor thermal profile influences the cooling capacity in two different ways. Firstly, as the temperatures of the suction muffler and of the cylinder wall decrease with the in-cylinder oil atomization, one expects that the specific volume of the refrigerant entering the cylinder will decrease and this will lead to a larger refrigerant mass entering the cylinder for a given swept volume, which will increase the cooling capacity. On the other hand, as the temperature of the gas in the cylinder decreases, the volumetric efficiency can also decrease as a result of the delay in the opening of the suction valve due to the combined effect of refrigerant outgassing from the lubricant oil and valve stiction (Khalifa and Liu 1998). This reduction of volumetric efficiency is obviously detrimental, and certainly plays a role, together with the effect of refrigerant solublity in the oil, in the determination of the compressor cooling capacity. The nature of the phenomena of refrigerant outgassing and valve stiction will be better explained below, in the context of their inlluence on the compressor power.

As mentioned above, the effective compression power was expected to decrease as a result of the sensible heat transfer to the oil droplets. However, an opposite trend was observed, which is believed to be related to refrigerant outgassing from the lubricant oil as it is atomized into the cylinder. Since the refrigerant solubility is directly proportional to pressure, the oil in the high-pressure oil separator contains a large amount of dissolved refrigerant that comes out of solution when it is injected into the cylinder, which is at a lower pressure. The refrigerant outgassing from the lubricant oil affects the compressor performance in two ways. During the suction process, while the suction valve is open, the refrigerant evaporated from the atomized droplets causes the pressure in the cylinder to increase. Thus, less mass enters the cylinder via the suction valve, in comparison to the reference case. This is detrimental to the cooling capacity, since the refrigerant released from the droplets would not, in practice, have any cooling effect, for it is confined in the atomization loop (compressor and oil separator). After the suction valve is closed, the refrigerant outgassing contributes to increasing the pressure in the cylinder, and so tile compressor power. In the R-134a/POE case, this supersedes the effect of temperature reduction due to heat transfer to the oil.

It is clear from the above analysis that, in order for the in-cylinder oil atomization to be successful as a means of improving the compressor performance, the refrigerant-oil solubility issue has to be dealt with so as to minimize the refrigerant release from the oil during atomization. In what follows, the potential of the technique has been further explored by means of an experimental analysis with a lubricant oil in which the refrigerant has a lower solubility. As will be seen, the results have improved significantly.

Atomization of mineral oil

To test the influence of the oil solubility on the compressor performance, experiments were carried out with mineral oil (MO ISO 10), which is known to have a much lower solubility in R-134a than the polyol ester oil. The mineral oil has been selected in such a way that its physical properties (density and viscosity) are similar to those of the polyol ester oil. Experimental data on the solubility of R-134a in MO ISO 10 are not widely available in the literature, so specific experiments were carried out with the apparatus of Marcelino Neto and Barbosa (2008) to determine the solubility of this refrigerantoil pair for the conditions of this experiment. For instance, at the pressure and temperature upstream of the injection nozzle, the refrigerant solubility in the R- 134a/MO mixture is around 9%, whereas the solubility corresponding to the R-134a/POE is approximately 23%.

The temperature data for the compressor components for the atomization of MO ISO 10 are also shown in Table 2. The temperature of the oil at the injection nozzle, [T.sub.oinj], is approximately the same as in the POE ISO 10 case. In fact, a very similar temperature profile has been observed with both lubricant oils.

The P-V diagrams for the reference case and for the case with atomization of mineral oil are presented in Figure 7. The pressure behavior during compression and expansion is very similar in both cases, with the main differences appearing in the suction and discharge processes, as can be seen from Figures 8 and 9. Oil atomization increases the energy expenditure to take in and discharge the gas, which raises the compression power and hence the cycle mean effective pressure.

[FIGURE 7 OMITTED]

With oil atomization, the suction losses calculated based on the data of Figure 8 increase by 17% with respect to the baseline. This increase can be attributed to a delay in the suction valve opening which gives rise to the more pronounced troughs seen in the pressure signal during the suction process. The delay itself can be caused by an increase in valve stiction (Khalifa and Liu 1998) due to more oil between the valve and seat, or by changes in the pressure pulsation patterns in the suction muffler. By the same token, the discharge losses in the case with oil atomization increase by 57% with respect to the baseline (Figure 9). The increase in the discharge pressure can be associated with (i) a larger flow restriction during discharge due to more oil between the piston and the valve plate, (ii) a larger mass of refrigerant being discharged due to the lower temperatures along the suction path, and (iii) stiction phenomena in the discharge valve.

[FIGURE 8 OMITTED]

The experimental data presented in the first and second lines of Table 3 for the MO ISO 10 atomization case seem to indicate that the oil solubility influences the compressor performance parameters. For instance, the cooling capacity increased by 1.2%, as opposed to the nearly 4% decrease observed in the POE ISO 10 atomization. The effective compression power is still larger than the baseline, but by 0.4%. As a result, the effective COP, i.e., the ratio of the cooling capacity and the effective compression power increased by 0.8% in comparison with the baseline. The effective compression power is the circumscribed area in the p-V diagram (see Figure 7), without the regions below the suction pressure and above the discharge pressure (horizontal dotted lines). The indicated power and the electric power consumption are still larger than those for the reference case (no oil atomization). However, it is believed that these could be improved via, for example, a modification of the design of the valves to deal specifically with operating conditions involving oil atomization, and by reducing the friction losses in the piston-cylinder gap and bearings in general (using lower viscosity oils). The indicated power is the circumscribed area in the p-V diagram, including the regions below the suction pressure and above the discharge pressure. [COP.sub.pV] and COP are defined as the ratio of the cooling capacity and the indicated power and the electric power, respectively.

[FIGURE 9 OMITTED]

Modeling results

Efficiency parameters and comparison with experimental data

The main discrepancies between the calculated and experimental pressures during the gas compression process appear during suction and discharge. Figure 10 shows the calculated valve displacement during discharge as a function of the crank angle for the baseline and MO ISO 10 atomization cases. The cylinder pressure is also shown for comparison purposes. As can be seen, in both cases, the discharge valve opens at 151.3 degrees from the BDC. However, the pressure in the cylinder reaches the discharge value earlier for the oil atomization case. Thus, by comparing the pressures at which the valve opens in both cases, one concludes that there is a delay of approximately 1 degree (crank angle) in the oil atomization case. The model results for valve displacement seems to corroborate the increase in the valve losses derived from the experimental pressure behavior during discharge.

Figure 11 shows the calculated heat transfer from the vapor as a function of the crank angle for atomization of oil at 45[degrees]C (113[degrees]F) with an average flow rate of 0.916 kg/h (2.02 1bm/h). The heat transfer during compression increases more sharply than the baseline and becomes positive earlier in the cycle showing a strong influence of droplet heat transfer. During expansion, heat transfer is similar in both cases (with and without oil) as a result of the very low velocity of the droplets remaining in the clearance volume after discharge. Figure 12, which shows temperatures as a function of the crank angle, reveals a significant reduction of the vapor temperature (maximum of 25[degrees]C, 45[degrees]1, at the TDC) with oil atomization. The initial compression temperature decreases from 58.8 [degrees]C (137.8[degrees]F) to 44.6[degrees]C (112.3[degrees]F), illustrating the large potential for reducing losses due to refrigerant superheating. The average oil temperature increases in the compression region due to the transfer of heat from the gas, but remains fairly constant during the remainder of the cycle, when the heat transfer driving potential between the gas and the droplets is less significant. The discontinuity in the temperature distribution is related to the discharge process, when the droplets are being expelled from the cylinder and the nozzle is totally covered by the piston (no injection). It should be borne in mind that the mathetical model takes no account of refrigerant dissolution in the oil, so its results would, in principle, represent a best case scenario for the droplet atomization heat transfer.

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

[FIGURE 12 OMITTED]

The fourth and fifth lines of Table 3 present the calculated efficiency parameters for the baseline and MO ISO 10 oil atomization cases. In general terms, although a satisfactory agreement is observed between model and experiments, the increase in the cooling capacity due to oil atomization in the cylinder is overshadowed by the increase in the compressor power. As can be seen, the improvement potential due to the atomization of pure oil (i.e., no dissolved refrigerant) given by the mathematical model is associated with an increase of 3.6% in the effective COP. This improvement is basically due to the prediction of an increase in the cooling capacity and a decrease of the effective compression power (which has not been confirmed in the experiments due to refrigerant outgassing during compression). As a result of the increasing discharge losses, the calculated improvement of [COP.sub.pV] is inferior to that of [COP.sub.eff]. Moreover, as expected from the experimental results, the largest contribution to the increase in compressor power is due to the friction losses associated with the temperature reduction. These losses are responsible for a 6.4% decrease in the COP.

It is worth mentioning that, given the refrigerant temperature decrease from 58.8[degrees]C to 44.6[degrees]C for a saturation pressure of -27[degrees]C, one should expect, from the baseline case, an improvement of cooling capacity of 4.7%, which roughly the same as the increase in gas density. However, the increase in cooling capacity was only 1.5%, according to the result presented in Table 3. This apparent inconsistency can be straightened out by considering the fact that the compression process is modified when the gas is cooled by the oil droplets. Thus, the temperature difference between the baseline case and the oil atomization case at the end of the compression process will be larger than at the beginning. So, the comparatively lower temperature at the end of the compression process for the oil atomization case results in a larger residual mass of gas in the cylinder. This means that less mass will enter the cylinder during the suction process, which decreases the cooling capacity in comparison with a hypothetical case in which there is a gas density increase, but there is no gas cooling during compression. It is worth noting that the influence of the oil on the displaced gas volume (Equation 11) is very small (0.04%). So, the reduction in the cooling capacity is solely due to the modification of the compression process.

Parametric analysis

A parametric analysis has been conducted in order to evaluate the effect of some important variables on the efficiency parameters. The effect of nozzle location inside the cylinder is investigated in Table 4. The nozzle is assumed to be mounted flush on the cylinder wall and its position is defined as the distance between the valve plate and the nozzle oririce. Depending on the nozzle position, the orifice remains covered for a certain region of the cycle (before and after the TDC) and no droplet atomization takes place in this period. The temperature at which the oil enters the compressor is maintained at 45[degrees]C in all cases.

As expected, for smaller distances between the orifice and valve plate, more oil enters the cylinder per cycle which results in a slight but steady decrease in compression power. The cooling capacity reduction with decreasing nozzle distances is mainly due to more oil injection during expansion which causes a delay in the opening of the suction valve. Compared with the baseline case, an increase in cooling capacity is observed for all cases, reflecting the beneficial effect of heat transfer in lowering the compressor thermal profile. A slight variation is seen in the indicated power and COP with respect to the nozzle position. However, the former is lower than in the baseline case due to a reduction in the effective compression power (i.e., disregarding energy dissipation during suction and discharge).

The effect of oil flow rate for a fixed nozzle position (13 mm, 0.51 in.) and oil temperature (45[degrees]C, 113[degrees]F) is seen in Table 5. This was achieved using a multiplying factor for the oil flow rate in Equation 6. Due to the overall temperature reduction, the cooling capacity increases with oil flow rate. However, the rate of increase is not constant due to some negative effects related to oil injection, such as the large fraction of the cylinder volume taken up by the oil during expansion. The COP reaches a maximum at a certain value of oil flow rate. This is due to an increase in indicated power associated with the high oil fraction in the discharge stream. Convergence problems were encountered for oil flow rates larger than 2.753 kg/h (6.07 Ibm/h) reflecting also the large amount of oil in the clearance volume.

The effect of oil temperature entering the injection line for a fixed nozzle position (13 mm, 0.51 in.) and flow rate (0.916 kg/h, 2.02 lbm/h) is depicted in Table 6. The higher the oil temperature, the lower the droplet-vapor heat transfer potential and hence the cooling capacity. Moreover, higher oil temperatures also increase the initial compression temperature of the vapor (due to an overall higher thermal profile), thus degrading the cooling capacity. This also reflects negatively on the indicated power and COP.

Conclusions

This paper presented a combined experimental and numerical analysis of the influence of atomization of lubricant oil in the cylinder of a reciprocating refrigeration compressor with R-134a. A compressor prototype has been prepared and tested with and without the atomization of lubricant oil in the cylinder in a purpose-built calorimeter at evaporating and condensing pressures corresponding to -27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F), respectively. The experimental results showed a significant decrease in the compressor thermal profile due to an enhanced heat removal from the refrigerant gas during compression. For example, with the atomization of polyol ester oil, the temperature reduction in the discharge chamber and the cylinder wall regions were as large as 27.4[degrees]C (49.3[degrees]F) and 25.3[degrees]C (45.5[degrees]F), respectively. The compressor efficiency parameters, however, were seen to be negatively affected by the solubility of the refrigerant in the lubricant oil, that is, the effective compressor power, the losses in the suction and discharge processes and the friction losses increased while the cooling capacity decreased in comparison to the baseline case (no oil atomization). An improvement of the performance parameters was observed when a lubricant oil with a lower solubility was used.

A comprehensive model was presented to evaluate the effect of in-cylinder oil atomization in reciprocating compressors. The model combined overall energy balances for the compressor and its main internal components with a more detailed approach for investigating the gas compression process with the presence of droplets under thermal non-equilibrium with the gas inside the cylinder. Despite the many assumptions regarding the atomization process, the model results were in satisfactory agreement with the experimental data, and the influence of several parameters such as oil injection temperature, nozzle position and oil mass flow rate could be investigated. Although some increase in the compressor cooling capacity was observed, the coefficient of performance decreased as a result of higher compressor power in the case with oil atomization. The effect of refrigerant solubility in the oil has not been considered in the model. One expects that taking this into account in future studies will improve the prediction of the compressor performance parameter, such as the cooling capacity and the [COP.sub.pV]

In order to bring the proposed technology closer to the application in real systems, some design strategies will have to be devised to reduce the thermodynamic losses and make the cooling solution more efficient. These range from simple measures (like adopting a lower viscosity oil to cope with the increase in friction losses) to more complex actions (such as dealing with the solubility/miscibility issue and with the reliability and long-life of the atomization system). Despite the small potential gain observed in the operating condition investigated in this work, the study opens room for further developments that may improve the relevance of the proposed cooling technology for different applications, such as ammonia compressors.

DOI: 10.1080/10789669.2012.646571

Acknowledgements

The material presented in this paper is a result of a long-standing technical-scientific partnership between the Federal University of Santa Catarina (UFSC) and Embraco. The authors are grateful to Mr. Fernando A. Ribas Jr. (Embraco) for encouragement and technical advice. Financial support from CNPq and FINEP is duly acknowledged.

Received August 18, 2011; accepted November 3, 2011

References

Annand, W.D. 1963. Heat transfer in the cylinders of reciprocating internal combustion engines, Proceedings of the Institution of Mechanical Engineers 177:973-96.

Bonjour, J., and A. Bejan. 2006. Optimal distribution of cooling during gas compression. Energy 31:409-24.

Coney, M.W., P. Stephenson, A. Malmgren, C. Linnemann, R.E. Morgan, R.A. Richards, R. Huxley, and H. Abdallah. 2002. Development of a reciprocating compressor using water injection to achieve quasi-isothermal compression, Proceedings of the 16th International Compressor Engineering Conference at Purdue, paper C4-3.

Dutta, A.K., T. Yanagisawa, and M. Fukuta. 2001. An investigation of the performance of a scroll compressor under liquid refrigerant injection. International Journal of Refrigeration 24:577-87.

Fagotti, F., M.L. Todescat, R.T.S. Ferreira, and A.T. Prata. 1994. Heat transfer modeling in a reciprocating compressor. Proceedings of the 12th International Compressor Engineering Conference at Purdue, pp. 605-10.

FLUENT 6.2.16, 2005, User Manual, Fluent Inc. Gosney, W.B. 1982. Principles of Refrigeration. Cambridge, UK: Cambridge University Press.

Khalifa, H.E., and Liu, X. 1998. Analysis of stiction effect on the dynamics of compressor suction valve. Proceedings of the 14th International Compressor Engineering Conference at Purdue, pp. 87-92.

Kremer, R. 2006. Theoretical and experimental analysis of the influence of oil atomization in compression processes, M Eng Diss. Federal University of Santa Catarina (in Portuguese).

Kremer, R., J.R. Barbosa, Jr., and C.J. Deschamps. 2007. Theoretical analysis of the effect of oil atomization in the cylinder of a reciprocating compressor. Proceedings of the International Conference on Compressors and their Systems, London, vol. 1, pp. 85-94.

Lemmon, E.L., M.O. McLinden, and M.L. Huber. 2002. REFPROP 7.0, NIST, USA.

Lichtarowicz, A.K., R.K. Duggins, and E. Markland. 1965. Discharge coefficients for incompressible non-cavitating flow through long orifices. Journal of Mechanical Engineering Science 7(2):2110-19.

Marcelino Neto, M.A., and J.R. Barbosa, Jr. 2008. Solubility, density and viscosity of a mixture of a mixture of R-600a and polyol ester oil. International Journal of Refrigeration 31:34-44.

Matos, EES. 2002. Numerical analysis of the dynamic behavior of reed-type valves in reciprocating compressors. DEng Thesis. Federal University of Santa Catarina (in Portuguese).

Morsi, S.A., and A.J. Alexander. 1972. An investigation of particle trajectories in two-phase flow systems. Journal of Fluid Mechanics 55:193-208.

Meunier, F. 2006. Refrigeration Carnot-type cycle based on isothermal vapour compression. Journal of Refrigeration 29:155-58.

Ooi, K.T. 2005. The effects of liquid injection on the performance of a rotary compressor, Proceedings of the International Conterence on Compressors and their Systems, London, vol. 1, pp. 151-64.

Possamai, F.C., and M.L. Todescat. 2004. A review of household compressor energy performance. Proceedings of the 17th International Compressor Engineering Conference at Purdue, paper C067.

Press, H.W., S.A. Teukolsky, W.T. Vetterling, and B.P. Flannery. 1992. Numerical Recipes in FORTRAN: The art of scientific computing, 2nd Ed., Cambridge, UK: Cambridge University Press.

Ranz, W.E., and W.R. Marshall. 1952. Evaporation from drops--I and II. Chemical Engineering Progress 48:141 and 173.

Ribas Jr., F.A. 2007. Thermal analysis of reciprocating compressors. Proceedings of the International Conference on Compressors and their Systems, London, vol. 1, pp. 277-87.

Shuaihui, S., Z. Yuanyang, L. Liansheng, and S. Pengcheng. 2010. Simulation research on scroll refrigeration compressor with external cooling. International Journal of Refrigeration 33:897-906.

Todescat, M.L., F. Fagotti, A.T. Prata, and R.T.S. Ferreira. 1992. Thermal energy analysis in reciprocating hermetic compressors. Proceedings of the 11th International Compressor Engineering Conferenee at Purdue, pp. 1419-28.

Toublanc, C. 2009. Amelioration du cycle frigorifique transcritique au C[O.sub.2] par une compression refroidie: Evaluations numerique et experimentale. PhD thesis, Le Conservatoire National des Arts et Metiers, Paris.

Ussyk, M.S. 1984. Numerical simulation of hermetic reciprocating compressors. M Eng Diss. Federal University of Santa Catarina (in Portuguese).

Wang, X., Y. Hwang, and R. Radermacher. 2008. Investigation of potential benefits of compressor cooling. Applied Thermal Engineering 28:1791-7.

Yakhot, V. and L.M. Smith. 1992. The renormalization group, the [epsilon]-expansion and derivation of turbulence models. Journal of Scientific Computing 7:35-61.

Rodrigo Kremer, MEng, is Research and Development Engineer. Jader R. Barbosa Jr., PhD, is Professor. Cesar Deschamps, PhD, is Professor.

Rodrigo Kremer, [1] Jader R. Barbosa Jr., [2,*] and Cesar J. Deschamps [2]

[1] Embraeo Compressors, Joinville, Brazil

[2] Mechanical Engineering Department, Federal University of Santa Catarina, Florianopolis, Brazil

* Corresponding author e-mail: jrb@polo.ufsc.br

Introduction

In household refrigeration compressors, a significant portion of the energy losses is associated with refrigerant superheating along the suction path inside the compressor and during gas compression inside the cylinder. Broadly speaking, the energy losses in a compressor are divided into (i) electrical, (ii) mechanical, (iii) thermodynamic, and (iv) cycle losses (Possamai and Todescat 2004). Thermodynamic losses involve the refrigerant flow inside the compressor. They are mostly associated with valve flows (viscous friction and backflow), gas leakage through the piston-cylinder gap and refrigerant superheating along the suction process and during compression. According to Ribas (2007), for R-134a household refrigeration compressors at the ASHRAE low back pressure (LBP) condition, around 50% of the thermodynamic losses are due to suction gas superheating. This superheating causes a reduction of the volumetric efficiency and an increase of the compression work per unit mass (Gosney 1982).

The reversible work input associated with an internally reversible steady-flow device is given by:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.]

where v is the specific volume of the working fluid, p is the pressure and 1 and 2 refer to the initial and final states. Even in the presence of internal irreversibilities (e.g., friction), where a departure from the ideal input work takes place, it becomes clear from Equation 1 that the specific volume must be as low as possible in order to minimize the work input. As the specific volume of a gas is directly proportional to the temperature, gas cooling during the compression process can be an alternative for reducing the compressor work and improving the energy efficiency of refrigeration systems. Basic calculations (Kremer et al. 2007) showed that, for a single stage refrigeration system operating with R134a between evaporating and condensing pressures of -27[degrees]C (80.6[degrees]F) and 42[degrees]C (107.6[degrees]F), isothermal compression can be as high as 16% more efficient than isentropic compression.

Many aspects of compressor cooling have been addressed in the open literature. Dutta et al. (2001) investigated theoretically and experimentally the effect of liquid refrigerant injection in scroll compressors using R-22. A decrease in the vapor temperature during compression was observed. However, it was reported that the extra work required to compress the vaporized refrigerant gave rise to a decrease in the compressor performance. The experimental and numerical studies of Coney et al. (2002) quantified the decrease in power consumption (approximately 28%) associated with the atomization of water in air reciprocating compressors. Ooi (2005) evaluated numerically the injection of liquid refrigerant in a rotary compressor, taking into account the influence of the nozzle diameter and its position in the compression chamber. He concluded that it is more advantageous to inject the liquid shortly before the discharge of the gas to benefit from the latent heat transfer without the penalty of compressing a volume of gas that is not used for generating cooling capacity. Meunier (2006) conceived a refrigeration cycle with a coefficient of performance (COP) equal to that of the Carnot refrigerator in which superheated vapor is compressed isothermally. However, no attempt to model the compression process has been proposed. Bonjour and Bejan (2006) demonstrated the existence of an optimal configuration for the distribution of cooling water in a multi-stage ammonia compressor that minimizes the compressor power. Wang et al. (2008) investigated theoretically two performance improvements for reducing the compressor power for several refrigerants. The first option was to cool the motor by external means other than using the suction gas. The second was to combine the processes of isothermal and isentropic compression. Their analysis demonstrated that the first option is more advantageous for LBP rather than high back pressure (HBP) applications for all refrigerants investigated. The second option showed that the compression power can be reduced by up to about 16% depending on operating conditions and working fluid. Toublanc (2009) investigated the atomization of lubricant oil during gas compression as a means of achieving near-isothermal compression in C[O.sub.2] refrigeration systems. He developed a mathematical model and carried out an extensive analysis of the effect of the droplet diameter, oil injection flow rate, compressor type (reciprocating or scroll) and compressor speed. For a reciprocating compressor of 8.4 kW cooling capacity, he reported a theoretical improvement of the COP of more than 30%, when the injected oil flow rate was three times larger than the gas flow rate. However, the volumetric capacity saw a reduction of 14%. Shuaihui et al. (2010) developed a mathematical model based on energy and mass balances to quantify the efficiency improvement associated with water cooling of scroll compressors. They showed that the isentropic efficiency can be improved by approximately 7%, and the gas discharge temperature can be reduced by 23[degrees]C (~41[degrees]F) when the cooling water is at 30[degrees]C (86[degrees]F). Little improvement was observed in the volumetric efficiency as a result of compressor cooling.

In the present paper, atomization of lubricant oil (ISO 10) during compression of R-134a in a household reciprocating compressor is investigated both experimentally and theoretically. In addition to lubrication, oil is used for sealing and cooling, and is heat transfer effective in compressors where leakage rates are high and where sequential injection can be applied. It can also be used in situations where extensive increases of temperature are present. Due to the small size of droplets, oil atomization promotes a heat transfer surface enhancement which favors heat removal from the vapor. The resulting cooling effect also contributes to lowering the overall thermal profile of the compressor parts and, consequently, the initial compression temperature. In quantitative terms, for R-134a, atomization of oil at 45[degrees]C (113[degrees]F) with an average flow rate of 0.916 kg/h (2.02 lbm/h) gave rise to a maximum temperature reduction of 25[degrees]C (77[degrees]F) at the top dead center (TDC). The initial compression temperature decreased from 58.8 [degrees]C (137.8[degrees]F) to 44.6[degrees]C (112.3[degrees]F), illustrating the large potential for reducing losses due to refrigerant superheating. Despite the encouraging temperature results, the compressor efficiency parameters, such as the compressor power, the cooling capacity and the COP have proven to be very sensitive to the solubility of the refrigerant-oil pair. The compressor performance deteriorates when an oil with a larger solubility was used. A comprehensive mathematical model has been proposed, which shows, in general, a satisfactory level of agreement with the experimental data.

Materials and Methods

Compressor

The compressor prototype shown in Figure 1 has a split crankcase sealed with fiat face bolt flanges and a silicone rubber gasket. A commercial hollowcone nozzle (Lechler 212.004) was mounted flush on the wall of the compression chamber. Due to the nozzle and housing dimensions and space restrictions, the distance between the nozzle orifice and the valve plate is approximately 13 mm (0.51 in.) (see Figure 1c). The nozzle is fed by a pressurized 2-mm (0.079 in.) I.D. oil line which is connected to the oil separator of the calorimeter loop (see below).

The temperatures of the compressor components were measured with 13 T-type thermocouples, according to the schematic diagram of Figure 2. The temperature at which the oil enters the cylinder is assumed equal to that of the nozzle housing. The temperature of the gas entering the cylinder is assumed equal to the one measured at the suction chamber. Due to space restrictions, the temperatures of the cylinder wall and of the main bearings are measured by inserting the thermocouples into holes drilled in the compressor block, so that the tip of the sensor is located at approximately 1 mm (0.04 in.) from the surfaces of the cylinder and bearings, respectively. As far as the measurement of the crankcase gas temperature is concerned, no attempt has been made to account for the influence of the oil adhering to the tip of the thermocouple. It was assumed that this influence is negligible because of the small thickness (i.e., negligible thermal resistance and capacity) of the oil film formed on the solid surfaces inside the compressor. It should be mentioned that this is the only point at which the gas temperature is measured inside the compressor. All of the remaining temperature measurements (excluding the above mentioned cylinder wall and bearings measurements) were carried out with thermocouples attached to solid surfaces that were exposed to the low-pressure crankcase internal environment. Therefore, since oil atomization takes place in the cylinder and the atomized oil leaves the compressor through the high-pressure discharge system, it is safe to state that the atomized oil does not interfere with the temperature measurements. This is not to say that the oil that is already present in the crankcase does not have any influence on the heat transfer measurements; it is just that this influence is expected to be approximately the same in the cases with and without oil injection in the cylinder.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

Although the uncertainty reported by the thermocouple manufacturer is [+ or -] 0.2[degrees]C ([+ or -] 0.36[degrees]F), the one assumed in this study is [degrees]C (+l.8[degrees]F) due to errors originating from the positioning and attachment of the thermocouples to the surfaces. The instantaneous gas pressure in the cylinder was measured with a Kistler 601A absolute pressure transducer with a sampling frequency of 60 kHz and an associated uncertainty of [+ or -]0.12 bar ([+ or -] 17.4 psi).

The influence of the split crankcase, atomization nozzle and pressure and temperature sensors on the compressor performance has been evaluated by comparing the refrigerant mass flow rate of the modified compressor prototype, but without oil atomization, with that of a compressor without any modification. A reduction of the order of 10% in the mass flow rate has been observed. However, this does not represent a source of error in the present analysis, since the modified prototype (without oil atomization) is the baseline for quantifying the influence of oil atomization on the compressor temperature profile.

Superheated Gas Cycle Calorimeter

The compressor was tested with R-134a in a superheated ("hot") gas cycle calorimeter described schematically in Figure 3. The compressor discharge pressure was regulated via a hand-operated needle valve (DV). An oil separator was installed in the discharge line, and was kept at a temperature above the saturation temperature of the refrigerant at the discharge pressure by means of an electric heater wrapped around its external surface. Downstream of DV, at an intermediate pressure between suction and discharge, an accumulator was installed to dampen pressure oscillations. A Coriolis-effect meter (MicroMotion) is used to measure the mass flow rate and, downstream of it, another hand-regulated needle valve (SV) is operated to expand the gas from the intermediate pressure down to the desired suction line pressure. The experimental error associated with the mass flow measurement is [+ or -] 0.15%, according to the manufacturer. A fixed inlet temperature of 32[degrees]C (89.6[degrees]F) is set by an electric heater. The suction and discharge pressures are measured with HBM P3MB absolute pressure transducers (10 and 50 bar--145.0 and 725.5 psi--full-scale for the suction and discharge pressures, respectively) and the temperatures are measured with T-type thermocouples. The experimental uncertainty of the suction and discharge pressure measurements were estimated at [+ or -] 0.004 bar ([+ or -] 0.058 psi) and [+ or -] 0.1 bar ([+ or -] 1.45 psi), respectively. The compressor power consumption was measured with a Yokogawa WT2 l 0 electric power transducer with an estimated error of [+ or -] 3%. After being separated from the discharge gas, the oil is driven through the oil line via a thermostatic bath for temperature control before being reinjected into the compressor, if the oil valve (OV) is open. The experimental apparatus is fully integrated with a signal conditioning and data acquisition module.

[FIGURE 3 OMITTED]

Experimental Procedure

The experiments were performed with R-134a at evaporating and condensing pressures corresponding to -27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F). The room temperature was maintained at 25[degrees]C (77[degrees]F). Tests were carried out with and without oil atomization and, in the latter, the oil injection temperature was approximately 45[degrees]C (113[degrees]F). The experimental procedure is as follows. Refrigerant is charged into the calorimeter, previously submitted to a vacuum of 0.04 mbar (5.8 x [10.sup.-4] psi) to remove moisture and dissolved gases from the oil. After charging the system and switching on the compressor, up to 6 hours are needed for reaching a fully cyclic condition, due to the thermal inertia of the system. In this period, the needle valves DV and SV are constantly adjusted to maintain the discharge and suction pressures to within [+ or -] 1% of the specified condensing and evaporating pressures. In the tests with oil atomization, the temperature of the thermostatic bath is also continuously adjusted so as to keep the oil atomization temperature within the specified boundaries. The oil flow rate was not measured directly during the experiments. However, based on numerical analyses of the oil flow via computational fluid dynamics (CFD) (Kremer 2006) and on independent tests carried out outside the compressor with a constant nozzle pressure drop, with an oil injection temperature of 60[degrees]C (140[degrees]F), the oil flow rate was estimated at approximately 1 [+ or -] 0.2 kg/h (2.20 [+ or -] 0.44 lbm/h).

The fully cyclic criterion establishes that, over an uninterrupted 45 min interval, the variation of the temperature readings must be less than l[degrees]C (1.8[degrees]F) and the variation of the measured compressor power and calculated cooling capacity (i.e., the product of the measured vapor mass flow rate and latent heat of vaporization at -;27[degrees]C, or -16.6[degrees]F) should be less than 1%. When this condition is met, the temperature, mass flow rate and compressor power data are acquired and averaged for the next 10 min and the test is ended with the acquisition and averaging of 50 pressure-volume cycles.

An error propagation analysis (Kremer 2006) yielded uncertainties of [+ or -] 4 and [+ or -] 5% for the cooling capacity and coefficient of performance, respectively. At least five tests were carried out for each condition and the results reported in this paper consist of arithmetic averages of the data at each specific condition. Repeatability tests confirmed that the dispersion of the data with respect to the normal distribution was small, with deviations less than 3.5% of the average values.

Modelling

In-cylinder gas compression

The control volume formulation of the vapor compression process is based on the work of Ussyk (1984), with the required modifications to predict the effect of oil atomization in the cylinder. The mass and energy conservation equations for the vapor are given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (2)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (3)

where u = u (p, T) is the internal energy of the vapor and,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (4)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (5)

Physical properties were calculated from REFPROP 7.0 (Lemmon et al. 2002). The heat transfer coefficient between the cylinder wall and the vapor was obtained from Annand (1963).

Droplet atomization and heat transfer

The compression cycle is divided into n time steps of size [DELTA]t. The droplets are assumed spherical. The number of droplets injected into the cylinder in a given time step is given by,

N = 3 [[??].sub.oil] [DELTA]t/4[pi][R.sup.3]. (6)

In the following analysis, the subscripts m,n (as in [N.sub.m,n]) denote the group of droplets that were injected into the cylinder at a given instant m (i.e., the mth "family") and still are in the cylinder at an instant n > m. It is assumed that all droplets in each family m are identical with respect to their size, velocity and temperature. If coalescence, break-up, and evaporation are neglected, mass conservation for each family m gives,

d[N.sub.m,n]/dt= - [[??].sub.m,n]/[M.sub.m]. (7)

The mass flow rate of droplets belonging to a family m through the discharge valve is calculated based on their mass fraction in the cylinder, [x.sub.m,n], as follows,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (8)

where

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (9)

and F is a droplet discharge slip factor that accounts for the relative velocity between the vapor and the liquid in the discharge stream. When F = 1 (homogeneous two-phase flow), the dynamic mass fraction of droplets during discharge is equal to their in-cylinder mass fraction. As will be shown below, F was estimated from CFD simulations of the two-phase discharge process. The gas mass flow rate through the discharge valve is given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (10)

Due to the presence of droplets in the cylinder, a correction is applied to the instantaneous cylinder volume to obtain the volume of vapor,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (11)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (12)

where [V.sub.c] and its time derivative are calculated from algebraic relationships for the piston position and velocity as a function of the crankshaft angle (Ussyk 1984).

Due to the small size of the droplets and hence the large surface area per unit volume, it is expected that the largest portion of the heat transfer to the oil occurs while it is in the form of high-speed droplets inside the cylinder. The smaller droplets atomized by the nozzle will, in the long run, reach hydrodynamic and thermal equilibrium with the gas, while the larger ones will impact directly on the cylinder wall or on the top of the piston, forming a thin oil film. The oil film will cool the metal parts, but it will have a small impact on the cooling of the gas.

The velocity of the oil droplets is calculated from Newton's Second Law. Neglecting the effect of body forces, assuming that the velocities of the droplets are much larger than the in-cylinder vapor velocity and that the hydrodynamic interaction between neighbouring droplets is small, one has,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (13)

where [C.sub.D] is the drag coefficient calculated from the Morsi and Alexander (1972) correlation. The average droplet radius, [R.sub.m], was also estimated from the CFD simulations of the atomization process, to be described later. The initial condition for [U.sub.m,n] is calculated from,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (14)

where [C.sub.d] is the nozzle discharge coefficient, assumed equal to 0.78 (Lichtarowicz et al. 1965). In Equation 14, [[rho].sub.oil] is the oil density at the injection temperature. Since the vapor pressure of the oil is small, it is assumed that only sensible heat transfer takes place between the droplets and the gas. Temperature gradients inside the droplets and thermal interaction between droplets are ignored. Thus, the temperature of a droplet in family m at an instant n is calculated from,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (15)

where the oil physical properties are calculated at the mixing-cup temperature of the droplets in the cylinder, [h.sub.m,n] is the heat transfer coefficient between the gas and droplets calculated from the Ranz and Marshall (1952) correlation. The heat transfer rate between all oil droplets and the vapor is given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (16)

It is hypothesized that the heat removal from the cylinder wall by the droplets is small and can be neglected. In other words, the thin liquid film forming on the solid walls reaches thermal equilibrium with the wall instantaneously. The small number of droplets that remain in the clearance volume at the end of a compression cycle are assumed not to contribute, in the next cycle, to the transfer of heat from the gas. This is because they are assumed to reach thermal equilibrium with the gas at the end of the cycle and to remain with the same velocity as the cylinder gas during the subsequent cycle.

Suction and discharge processes

Valve displacement during suction and discharge is calculated via a one-degree-of-freedom model with natural frequency and damping coefficients specific for each valve. The resultant forces on the valves and their respective flow rates are obtained with effective force and effective flow areas derived from numerical simulations (Matos 2002). The total flow rate through the discharge valve is calculated assuming a homogeneous two-phase density based on the discharge mass fraction of each phase (Equations 8-10).

Droplet Atomization Parameters Obtained via CFD

In order to determine approximate droplet atomization parameters such as average droplet radius and discharge slip factor, CFD simulations of the compression and atomization processes were carried out using the commercial code FLUENT 6.2.16 (2005). A 2D moving-mesh domain (2100 volumes) was implemented to enable the simulation of the piston displacement inside the cylinder with a frequency of 60 Hz (see Figures 4 and 5). The suction and discharge valves were modelled as single axissymetric orifices that remain fully open during suction and discharge at the specified evaporating and condensing pressures of-27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F). The orifices are assumed to behave ideally in the sense that there are no friction losses involved, and that they close fully and immediately at pre-determined values of crank angle. Thc fluid flow was modelled using the renormalization (RNG) k-[epsilon] turbulence model (Yakhot and Smith 1992) and a built-in Lagrangean formulation (the Discrete Phase Model) was utilized to calculate the droplet field. The nozzle type and geometry were chosen so as to reproduce the atomization conditions of a commercial hollow-cone nozzle (Lechler 212.004). In the computational mesh, the atomizing orifice was positioned on the symmetry axis at the top of the cylinder. This arrangement is somewhat simpler than the actual cylinder-nozzle system geometry shown in Fig. 1c. However, its simplicity can be justified on the basis that the main purpose of the CFD simulation in the present work is to obtain approximate average droplet parameters for the cycle, and not to effectively solve the in-cylinder compression process.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

Figure 4 presents the mean droplet diameter as a function of crank angle for an oil atomizer flow rate of 1.08 kg/h (2.38 1bm/h) at 55[degrees]C (131[degrees]F). The bottom dead center (BDC) corresponds to a crank angle of 0[degrees]C. The mean droplet diameter averaged over one cycle was 69 [micro]m (2.72 x [10.sup.-3] in.). Between 150[degrees] and 180[degrees] the average droplet diameter decreases to approximately 40 [micro]m (1.57 x [10.sup.-3] in.) due to break-up induced by vapor acceleration during discharge. As expansion begins, the average droplet diameter is large because of the low pressure difference between the hypothetical oil reservoir (assumed to remain at the condensing pressure) and the cylinder. In the droplet atomization and heat transfer model described previously, a constant droplet diameter of 69 [micro]m (2.72 x [10.sup.-3] in.) was adopted for the whole cycle.

In Figure 5, the total mass of oil in the cylinder is presented as a function of crank angle. The droplet discharge slip factor, F, is calculated from the maximum and minimum of the curve (i.e., the beginning and ending of the discharge process) together with the mass of refrigerant in the cylinder at those instants. At the present conditions, it was observed that the dynamic mass fraction of oil in the discharge stream was proportional to 65% of its overall mass fraction in the cylinder. Hence, F was assumed equal to 0.65.

Compressor thermal model

In order to provide the in-cylinder vapor compression and droplet heat transfer models with appropriate boundary conditions, the model described in the previous sections was incorporated into an existing lumped heat transfer (overall thermal conductance) model that integrates several compressor components as shown in Figure 6. In the original model without oil atomization (Todescat et al. 1992; Fagotti et al. 1994), the compressor was divided into seven control volumes (the vapor in the muffler and suction chamber, the vapor in the cylinder, the vapor in the discharge chamber, the vapor in the discharge muffler, the electric motor, the compressor shell and the overall compressor balance). Steady-sate energy balance equations were written for each component and solved coupled with the transient in-cylinder vapor compression equations. In the present paper, the compressor thermal model was extended so as to include two additional control volumes (the oil injection line and the discharge line) and to account for the flow of oil through the components downstream of the cylinder.

[FIGURE 6 OMITTED]

The energy balances in the control volumes are represented by the following general equation, where the specific terms in Equation 17 associated with each control volume are shown in Table 1.

The heat transfer rates in Equations 18-26 are modelled in terms of overall thermal conductances between the crankcase gas and the gas inside each corresponding component. The heat transfer rate between the crankcase gas and the suction muffler is given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (27)

where [bar.U[A.sub.s]] is a thermal conductance between the crankcase gas and the gas in the suction muffler. By the same token, the heat transfer rates between the crankcase gas and the oil injection line, the discharge chamber, the discharge muffler, the discharge line and the motor are given by,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (28)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (29)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (30)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (31)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (32)

The rate of heat transfer rate between the cylinder walls and the gas inside the cylinder is calculated based on the following relationship,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (33)

where Aw is the heat transfer area in the cylinder (walls and top of the piston), [T.sub.w] is the cylinder wall temperature and [bar.T] is the cycle average temperature of the gas inside the cylinder. [[??].sub.w] is the average heat transfer coefficient calculated via the Annand (1963) correlation with the Reynolds number calculated with the mean piston velocity.

In the energy balance associated with the compressor crankcase (i.e., the shell itself), Equation 26, the rate of heat transfer between the shell and the crankcase gas is also modeled in terms of an overall thermal conductance as follows,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (34)

where [T.sub.h] is the average temperature of the compressor crankcase. In Equation 25, [Q.sub.ee] is the rate at which heat is transferred from the outer surface of the compressor crankcase to the external environment. Thus,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII.] (35)

where [T.sub.ee] is the temperature of the external environment and [bar.U[A.sub.ee]] is the corresponding overall thermal conductance.

The unknowns in Equations 18-26 are [T.sub.sc], [T.sub.oinj], [T.sub.w], [T.sub.dc], [T.sub.vb], [T.sub.ld], [T.sub.m] [T.sub.h], and [T.sub.ie]. The system is nonlinear in nature due to the dependence of the specific enthalpies with respect to temperature. The flow rates needed in the energy balances are determined via the models for the processes of suction and discharge and droplet atomization. The two-phase refrigerant-oil flow downstream of the discharge chamber is assumed homogeneous (i.e., no slip between the phases) and in thermal equilibrium. The overall thermal conductances in Equations 27-32 and 34-35 are estimated via experimental data (temperatures, flow rate and electric power) obtained for a reference operating condition of evaporating and condensing pressures corresponding to -27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F), for R-134a, and an oil inlet temperature, Top, of 36.2[degrees]C (97.2[degrees]F). Caution must be exercised when the estimated overall thermal conductances are extrapolated to operating conditions other than the reference case, since the associated variations in mass flow rates and temperatures affect the local heat transfer coefficients that, in turn, change the overall thermal conductances. However, it has been verified against experimental data that, when a new operating condition is within a few degrees from the reference case, the reference overall thermal conductances can be used without an appreciable error.

Solution Method

The compression cycle is divided into [10.sup.3] uniform time (or crank angle) steps and an explicit Euler method is used to integrate the time-dependent mass and energy conservation equations for the vapor and liquid droplet fields. The algebraic equations of the compressor thermal model are solved via a multi-dimensional Newton-Raphson method (Press et al. 1992). The two methods are coupled in such a way that, for each iteration in the compressor thermal simulation, a fixed number of in-cylinder compression cycles (say, from 8 to 12) are performed so that convergence is guaranteed in the solution of the periodic gas compression and droplet atomization equations. The convergence criterion for the compressor thermal simulation is based on a temperature difference tolerance (0.002[degrees]C, 0.0036[degrees]F) between successive iterations for each component temperature. If the criteflon is not met, then another set of in-cylinder cycles is executed, and so on until convergence is guaranteed.

Results

Experimental Results

Atomization of polyol ester oil

Experiments were initially conducted with polyol ester oil (POE ISO 10), for this is the lubricant which is normally used in R-134a systems in household refrigeration compressors. The temperature data for the compressor components are shown in Table 2 for the baseline case (i.e., no oil injection) and for cases where the lubricant oil was atomized in the cylinder at approximately 45[degrees]C (113[degrees]F).

As can be seen, atomization of the POE ISO 10 oil in the cylinder caused a significant reduction of the temperatures inside the compressor, more noticeably in regions which are in direct contact with the gas that is cooled down during compression, such as the discharge chamber and the cylinder wall. In these regions, the temperatures decreased by 27.4[degrees]C (49.3[degrees]F) and 25.3[degrees]C (45.5[degrees]F), respectively. The temperature reduction of approximately 3[degrees]C (5.4[degrees]F) at the suction muffler outlet is much smaller, but still large enough to reduce somewhat the inlet superheating losses. The values associated with the temperature decrease of the oil sump and compressor crankcase are quite similar, possibly due to the way in which the oil circulation takes place inside the compressor (i.e., an oil falling film is formed on the inner wall of the compressor crankcase).

Despite the significant reduction in the compressor thermal profile, the performance of the compressor deteriorated in comparison with the reference case. The indicated power, calculated via a numerical integration of the instantaneous pressure-volume experimental data, increased by 4.6%. It is believed that this increase is partially due to losses in the discharge system, which has not been designed to operate under significant oil flow. However, even when the power losses associated with the valve flows are disregarded (and the valves are assumed to behave as ideal orifices), the power effectively used to compress the gas was 2.1% larger in the atomization case. This was a surprising result, since the compression power was expected to decrease with the reduction of the compressor thermal profile. Nevertheless, this effect was observed consistently and seems to be associated with the solubility of R-134a in the POE oil, as will be explained later in this article. The solubility is defined as the equilibrium mass fraction of refrigerant in the liquid mixture. A more detailed discussion on this matter will be provided below. The compressor electric power consumption increased by 11.1%, which is mainly due to the increase of friction losses since the oil viscosity increases with the temperature reduction in the primary bearings and piston-cylinder gap.

The compressor cooling capacity decreased by 3.6% with respect to the reference case. The reduction of the compressor thermal profile influences the cooling capacity in two different ways. Firstly, as the temperatures of the suction muffler and of the cylinder wall decrease with the in-cylinder oil atomization, one expects that the specific volume of the refrigerant entering the cylinder will decrease and this will lead to a larger refrigerant mass entering the cylinder for a given swept volume, which will increase the cooling capacity. On the other hand, as the temperature of the gas in the cylinder decreases, the volumetric efficiency can also decrease as a result of the delay in the opening of the suction valve due to the combined effect of refrigerant outgassing from the lubricant oil and valve stiction (Khalifa and Liu 1998). This reduction of volumetric efficiency is obviously detrimental, and certainly plays a role, together with the effect of refrigerant solublity in the oil, in the determination of the compressor cooling capacity. The nature of the phenomena of refrigerant outgassing and valve stiction will be better explained below, in the context of their inlluence on the compressor power.

As mentioned above, the effective compression power was expected to decrease as a result of the sensible heat transfer to the oil droplets. However, an opposite trend was observed, which is believed to be related to refrigerant outgassing from the lubricant oil as it is atomized into the cylinder. Since the refrigerant solubility is directly proportional to pressure, the oil in the high-pressure oil separator contains a large amount of dissolved refrigerant that comes out of solution when it is injected into the cylinder, which is at a lower pressure. The refrigerant outgassing from the lubricant oil affects the compressor performance in two ways. During the suction process, while the suction valve is open, the refrigerant evaporated from the atomized droplets causes the pressure in the cylinder to increase. Thus, less mass enters the cylinder via the suction valve, in comparison to the reference case. This is detrimental to the cooling capacity, since the refrigerant released from the droplets would not, in practice, have any cooling effect, for it is confined in the atomization loop (compressor and oil separator). After the suction valve is closed, the refrigerant outgassing contributes to increasing the pressure in the cylinder, and so tile compressor power. In the R-134a/POE case, this supersedes the effect of temperature reduction due to heat transfer to the oil.

It is clear from the above analysis that, in order for the in-cylinder oil atomization to be successful as a means of improving the compressor performance, the refrigerant-oil solubility issue has to be dealt with so as to minimize the refrigerant release from the oil during atomization. In what follows, the potential of the technique has been further explored by means of an experimental analysis with a lubricant oil in which the refrigerant has a lower solubility. As will be seen, the results have improved significantly.

Atomization of mineral oil

To test the influence of the oil solubility on the compressor performance, experiments were carried out with mineral oil (MO ISO 10), which is known to have a much lower solubility in R-134a than the polyol ester oil. The mineral oil has been selected in such a way that its physical properties (density and viscosity) are similar to those of the polyol ester oil. Experimental data on the solubility of R-134a in MO ISO 10 are not widely available in the literature, so specific experiments were carried out with the apparatus of Marcelino Neto and Barbosa (2008) to determine the solubility of this refrigerantoil pair for the conditions of this experiment. For instance, at the pressure and temperature upstream of the injection nozzle, the refrigerant solubility in the R- 134a/MO mixture is around 9%, whereas the solubility corresponding to the R-134a/POE is approximately 23%.

The temperature data for the compressor components for the atomization of MO ISO 10 are also shown in Table 2. The temperature of the oil at the injection nozzle, [T.sub.oinj], is approximately the same as in the POE ISO 10 case. In fact, a very similar temperature profile has been observed with both lubricant oils.

The P-V diagrams for the reference case and for the case with atomization of mineral oil are presented in Figure 7. The pressure behavior during compression and expansion is very similar in both cases, with the main differences appearing in the suction and discharge processes, as can be seen from Figures 8 and 9. Oil atomization increases the energy expenditure to take in and discharge the gas, which raises the compression power and hence the cycle mean effective pressure.

[FIGURE 7 OMITTED]

With oil atomization, the suction losses calculated based on the data of Figure 8 increase by 17% with respect to the baseline. This increase can be attributed to a delay in the suction valve opening which gives rise to the more pronounced troughs seen in the pressure signal during the suction process. The delay itself can be caused by an increase in valve stiction (Khalifa and Liu 1998) due to more oil between the valve and seat, or by changes in the pressure pulsation patterns in the suction muffler. By the same token, the discharge losses in the case with oil atomization increase by 57% with respect to the baseline (Figure 9). The increase in the discharge pressure can be associated with (i) a larger flow restriction during discharge due to more oil between the piston and the valve plate, (ii) a larger mass of refrigerant being discharged due to the lower temperatures along the suction path, and (iii) stiction phenomena in the discharge valve.

[FIGURE 8 OMITTED]

The experimental data presented in the first and second lines of Table 3 for the MO ISO 10 atomization case seem to indicate that the oil solubility influences the compressor performance parameters. For instance, the cooling capacity increased by 1.2%, as opposed to the nearly 4% decrease observed in the POE ISO 10 atomization. The effective compression power is still larger than the baseline, but by 0.4%. As a result, the effective COP, i.e., the ratio of the cooling capacity and the effective compression power increased by 0.8% in comparison with the baseline. The effective compression power is the circumscribed area in the p-V diagram (see Figure 7), without the regions below the suction pressure and above the discharge pressure (horizontal dotted lines). The indicated power and the electric power consumption are still larger than those for the reference case (no oil atomization). However, it is believed that these could be improved via, for example, a modification of the design of the valves to deal specifically with operating conditions involving oil atomization, and by reducing the friction losses in the piston-cylinder gap and bearings in general (using lower viscosity oils). The indicated power is the circumscribed area in the p-V diagram, including the regions below the suction pressure and above the discharge pressure. [COP.sub.pV] and COP are defined as the ratio of the cooling capacity and the indicated power and the electric power, respectively.

[FIGURE 9 OMITTED]

Modeling results

Efficiency parameters and comparison with experimental data

The main discrepancies between the calculated and experimental pressures during the gas compression process appear during suction and discharge. Figure 10 shows the calculated valve displacement during discharge as a function of the crank angle for the baseline and MO ISO 10 atomization cases. The cylinder pressure is also shown for comparison purposes. As can be seen, in both cases, the discharge valve opens at 151.3 degrees from the BDC. However, the pressure in the cylinder reaches the discharge value earlier for the oil atomization case. Thus, by comparing the pressures at which the valve opens in both cases, one concludes that there is a delay of approximately 1 degree (crank angle) in the oil atomization case. The model results for valve displacement seems to corroborate the increase in the valve losses derived from the experimental pressure behavior during discharge.

Figure 11 shows the calculated heat transfer from the vapor as a function of the crank angle for atomization of oil at 45[degrees]C (113[degrees]F) with an average flow rate of 0.916 kg/h (2.02 1bm/h). The heat transfer during compression increases more sharply than the baseline and becomes positive earlier in the cycle showing a strong influence of droplet heat transfer. During expansion, heat transfer is similar in both cases (with and without oil) as a result of the very low velocity of the droplets remaining in the clearance volume after discharge. Figure 12, which shows temperatures as a function of the crank angle, reveals a significant reduction of the vapor temperature (maximum of 25[degrees]C, 45[degrees]1, at the TDC) with oil atomization. The initial compression temperature decreases from 58.8 [degrees]C (137.8[degrees]F) to 44.6[degrees]C (112.3[degrees]F), illustrating the large potential for reducing losses due to refrigerant superheating. The average oil temperature increases in the compression region due to the transfer of heat from the gas, but remains fairly constant during the remainder of the cycle, when the heat transfer driving potential between the gas and the droplets is less significant. The discontinuity in the temperature distribution is related to the discharge process, when the droplets are being expelled from the cylinder and the nozzle is totally covered by the piston (no injection). It should be borne in mind that the mathetical model takes no account of refrigerant dissolution in the oil, so its results would, in principle, represent a best case scenario for the droplet atomization heat transfer.

[FIGURE 10 OMITTED]

[FIGURE 11 OMITTED]

[FIGURE 12 OMITTED]

The fourth and fifth lines of Table 3 present the calculated efficiency parameters for the baseline and MO ISO 10 oil atomization cases. In general terms, although a satisfactory agreement is observed between model and experiments, the increase in the cooling capacity due to oil atomization in the cylinder is overshadowed by the increase in the compressor power. As can be seen, the improvement potential due to the atomization of pure oil (i.e., no dissolved refrigerant) given by the mathematical model is associated with an increase of 3.6% in the effective COP. This improvement is basically due to the prediction of an increase in the cooling capacity and a decrease of the effective compression power (which has not been confirmed in the experiments due to refrigerant outgassing during compression). As a result of the increasing discharge losses, the calculated improvement of [COP.sub.pV] is inferior to that of [COP.sub.eff]. Moreover, as expected from the experimental results, the largest contribution to the increase in compressor power is due to the friction losses associated with the temperature reduction. These losses are responsible for a 6.4% decrease in the COP.

It is worth mentioning that, given the refrigerant temperature decrease from 58.8[degrees]C to 44.6[degrees]C for a saturation pressure of -27[degrees]C, one should expect, from the baseline case, an improvement of cooling capacity of 4.7%, which roughly the same as the increase in gas density. However, the increase in cooling capacity was only 1.5%, according to the result presented in Table 3. This apparent inconsistency can be straightened out by considering the fact that the compression process is modified when the gas is cooled by the oil droplets. Thus, the temperature difference between the baseline case and the oil atomization case at the end of the compression process will be larger than at the beginning. So, the comparatively lower temperature at the end of the compression process for the oil atomization case results in a larger residual mass of gas in the cylinder. This means that less mass will enter the cylinder during the suction process, which decreases the cooling capacity in comparison with a hypothetical case in which there is a gas density increase, but there is no gas cooling during compression. It is worth noting that the influence of the oil on the displaced gas volume (Equation 11) is very small (0.04%). So, the reduction in the cooling capacity is solely due to the modification of the compression process.

Parametric analysis

A parametric analysis has been conducted in order to evaluate the effect of some important variables on the efficiency parameters. The effect of nozzle location inside the cylinder is investigated in Table 4. The nozzle is assumed to be mounted flush on the cylinder wall and its position is defined as the distance between the valve plate and the nozzle oririce. Depending on the nozzle position, the orifice remains covered for a certain region of the cycle (before and after the TDC) and no droplet atomization takes place in this period. The temperature at which the oil enters the compressor is maintained at 45[degrees]C in all cases.

As expected, for smaller distances between the orifice and valve plate, more oil enters the cylinder per cycle which results in a slight but steady decrease in compression power. The cooling capacity reduction with decreasing nozzle distances is mainly due to more oil injection during expansion which causes a delay in the opening of the suction valve. Compared with the baseline case, an increase in cooling capacity is observed for all cases, reflecting the beneficial effect of heat transfer in lowering the compressor thermal profile. A slight variation is seen in the indicated power and COP with respect to the nozzle position. However, the former is lower than in the baseline case due to a reduction in the effective compression power (i.e., disregarding energy dissipation during suction and discharge).

The effect of oil flow rate for a fixed nozzle position (13 mm, 0.51 in.) and oil temperature (45[degrees]C, 113[degrees]F) is seen in Table 5. This was achieved using a multiplying factor for the oil flow rate in Equation 6. Due to the overall temperature reduction, the cooling capacity increases with oil flow rate. However, the rate of increase is not constant due to some negative effects related to oil injection, such as the large fraction of the cylinder volume taken up by the oil during expansion. The COP reaches a maximum at a certain value of oil flow rate. This is due to an increase in indicated power associated with the high oil fraction in the discharge stream. Convergence problems were encountered for oil flow rates larger than 2.753 kg/h (6.07 Ibm/h) reflecting also the large amount of oil in the clearance volume.

The effect of oil temperature entering the injection line for a fixed nozzle position (13 mm, 0.51 in.) and flow rate (0.916 kg/h, 2.02 lbm/h) is depicted in Table 6. The higher the oil temperature, the lower the droplet-vapor heat transfer potential and hence the cooling capacity. Moreover, higher oil temperatures also increase the initial compression temperature of the vapor (due to an overall higher thermal profile), thus degrading the cooling capacity. This also reflects negatively on the indicated power and COP.

Conclusions

This paper presented a combined experimental and numerical analysis of the influence of atomization of lubricant oil in the cylinder of a reciprocating refrigeration compressor with R-134a. A compressor prototype has been prepared and tested with and without the atomization of lubricant oil in the cylinder in a purpose-built calorimeter at evaporating and condensing pressures corresponding to -27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F), respectively. The experimental results showed a significant decrease in the compressor thermal profile due to an enhanced heat removal from the refrigerant gas during compression. For example, with the atomization of polyol ester oil, the temperature reduction in the discharge chamber and the cylinder wall regions were as large as 27.4[degrees]C (49.3[degrees]F) and 25.3[degrees]C (45.5[degrees]F), respectively. The compressor efficiency parameters, however, were seen to be negatively affected by the solubility of the refrigerant in the lubricant oil, that is, the effective compressor power, the losses in the suction and discharge processes and the friction losses increased while the cooling capacity decreased in comparison to the baseline case (no oil atomization). An improvement of the performance parameters was observed when a lubricant oil with a lower solubility was used.

A comprehensive model was presented to evaluate the effect of in-cylinder oil atomization in reciprocating compressors. The model combined overall energy balances for the compressor and its main internal components with a more detailed approach for investigating the gas compression process with the presence of droplets under thermal non-equilibrium with the gas inside the cylinder. Despite the many assumptions regarding the atomization process, the model results were in satisfactory agreement with the experimental data, and the influence of several parameters such as oil injection temperature, nozzle position and oil mass flow rate could be investigated. Although some increase in the compressor cooling capacity was observed, the coefficient of performance decreased as a result of higher compressor power in the case with oil atomization. The effect of refrigerant solubility in the oil has not been considered in the model. One expects that taking this into account in future studies will improve the prediction of the compressor performance parameter, such as the cooling capacity and the [COP.sub.pV]

In order to bring the proposed technology closer to the application in real systems, some design strategies will have to be devised to reduce the thermodynamic losses and make the cooling solution more efficient. These range from simple measures (like adopting a lower viscosity oil to cope with the increase in friction losses) to more complex actions (such as dealing with the solubility/miscibility issue and with the reliability and long-life of the atomization system). Despite the small potential gain observed in the operating condition investigated in this work, the study opens room for further developments that may improve the relevance of the proposed cooling technology for different applications, such as ammonia compressors.

Nomenclature Roman A = area [c.sub.p] = specific heat capacity at constant pressure [C.sub.d] = nozzle discharge coefficient [C.sub.D] = droplet drag coefficient COP = coefficient of performance h = specific enthalpy of the gas [??] = heat transfer coefficient m = mass of gas in the cylinder M = mass of a single droplet [??] = refrigerant mass flow rate [[??].sub.t] = total mass flow rate (droplets and refrigerant) through the discharge valvef [??] = mass flow rate of droplets N = number of atomized droplets per time step P = gas pressure [p.sub.c] = condensing pressure [[??].sub.e] = cooling capacity [[??].sub.o] = heat transfer rate to the oil R = droplet radius [??] = gas constant t = time T = temperature [bar.T] = average temperature of the gas during a cycle u = specific internal energy of the gas U = droplet velocity [bar.UA] = average thermal conductance [??] = volume occupied by the gas [[??].sub.c] = cylinder volume [??] = volume flow rate [[??].sub.eff] = effective compression power [[??].sub.ele] = electric power input [[??].sub.i] = indicated power [[??].sub.m] = power dissipated as heat in the bearings and piston cylinder gap w = specific work (per unit mass) x = mass fraction of liquid droplets in the gasliquid mixture Z = compressibility factor Greek [rho] = density Subscripts c = cylinder C = condensation d = discharge dc = discharge chamber ee = external environment epc = gas flow entering the cylinder via the piston-cylinder clearance h = compressor crankcase (shell) ie = gas in the internal environment l = leakage ld = discharge line m = electric motor m,n = time instants within the compression cycle od = oil at the discharge orifice oil = oil oinj = oil injected in the cylinder op = oil injection line r = backflow rev = reversible s = suction sc = suction muffler spc = gas flow exiting the cylinder via the pistoncylinder clearance vb = discharge muffler

DOI: 10.1080/10789669.2012.646571

Acknowledgements

The material presented in this paper is a result of a long-standing technical-scientific partnership between the Federal University of Santa Catarina (UFSC) and Embraco. The authors are grateful to Mr. Fernando A. Ribas Jr. (Embraco) for encouragement and technical advice. Financial support from CNPq and FINEP is duly acknowledged.

Received August 18, 2011; accepted November 3, 2011

References

Annand, W.D. 1963. Heat transfer in the cylinders of reciprocating internal combustion engines, Proceedings of the Institution of Mechanical Engineers 177:973-96.

Bonjour, J., and A. Bejan. 2006. Optimal distribution of cooling during gas compression. Energy 31:409-24.

Coney, M.W., P. Stephenson, A. Malmgren, C. Linnemann, R.E. Morgan, R.A. Richards, R. Huxley, and H. Abdallah. 2002. Development of a reciprocating compressor using water injection to achieve quasi-isothermal compression, Proceedings of the 16th International Compressor Engineering Conference at Purdue, paper C4-3.

Dutta, A.K., T. Yanagisawa, and M. Fukuta. 2001. An investigation of the performance of a scroll compressor under liquid refrigerant injection. International Journal of Refrigeration 24:577-87.

Fagotti, F., M.L. Todescat, R.T.S. Ferreira, and A.T. Prata. 1994. Heat transfer modeling in a reciprocating compressor. Proceedings of the 12th International Compressor Engineering Conference at Purdue, pp. 605-10.

FLUENT 6.2.16, 2005, User Manual, Fluent Inc. Gosney, W.B. 1982. Principles of Refrigeration. Cambridge, UK: Cambridge University Press.

Khalifa, H.E., and Liu, X. 1998. Analysis of stiction effect on the dynamics of compressor suction valve. Proceedings of the 14th International Compressor Engineering Conference at Purdue, pp. 87-92.

Kremer, R. 2006. Theoretical and experimental analysis of the influence of oil atomization in compression processes, M Eng Diss. Federal University of Santa Catarina (in Portuguese).

Kremer, R., J.R. Barbosa, Jr., and C.J. Deschamps. 2007. Theoretical analysis of the effect of oil atomization in the cylinder of a reciprocating compressor. Proceedings of the International Conference on Compressors and their Systems, London, vol. 1, pp. 85-94.

Lemmon, E.L., M.O. McLinden, and M.L. Huber. 2002. REFPROP 7.0, NIST, USA.

Lichtarowicz, A.K., R.K. Duggins, and E. Markland. 1965. Discharge coefficients for incompressible non-cavitating flow through long orifices. Journal of Mechanical Engineering Science 7(2):2110-19.

Marcelino Neto, M.A., and J.R. Barbosa, Jr. 2008. Solubility, density and viscosity of a mixture of a mixture of R-600a and polyol ester oil. International Journal of Refrigeration 31:34-44.

Matos, EES. 2002. Numerical analysis of the dynamic behavior of reed-type valves in reciprocating compressors. DEng Thesis. Federal University of Santa Catarina (in Portuguese).

Morsi, S.A., and A.J. Alexander. 1972. An investigation of particle trajectories in two-phase flow systems. Journal of Fluid Mechanics 55:193-208.

Meunier, F. 2006. Refrigeration Carnot-type cycle based on isothermal vapour compression. Journal of Refrigeration 29:155-58.

Ooi, K.T. 2005. The effects of liquid injection on the performance of a rotary compressor, Proceedings of the International Conterence on Compressors and their Systems, London, vol. 1, pp. 151-64.

Possamai, F.C., and M.L. Todescat. 2004. A review of household compressor energy performance. Proceedings of the 17th International Compressor Engineering Conference at Purdue, paper C067.

Press, H.W., S.A. Teukolsky, W.T. Vetterling, and B.P. Flannery. 1992. Numerical Recipes in FORTRAN: The art of scientific computing, 2nd Ed., Cambridge, UK: Cambridge University Press.

Ranz, W.E., and W.R. Marshall. 1952. Evaporation from drops--I and II. Chemical Engineering Progress 48:141 and 173.

Ribas Jr., F.A. 2007. Thermal analysis of reciprocating compressors. Proceedings of the International Conference on Compressors and their Systems, London, vol. 1, pp. 277-87.

Shuaihui, S., Z. Yuanyang, L. Liansheng, and S. Pengcheng. 2010. Simulation research on scroll refrigeration compressor with external cooling. International Journal of Refrigeration 33:897-906.

Todescat, M.L., F. Fagotti, A.T. Prata, and R.T.S. Ferreira. 1992. Thermal energy analysis in reciprocating hermetic compressors. Proceedings of the 11th International Compressor Engineering Conferenee at Purdue, pp. 1419-28.

Toublanc, C. 2009. Amelioration du cycle frigorifique transcritique au C[O.sub.2] par une compression refroidie: Evaluations numerique et experimentale. PhD thesis, Le Conservatoire National des Arts et Metiers, Paris.

Ussyk, M.S. 1984. Numerical simulation of hermetic reciprocating compressors. M Eng Diss. Federal University of Santa Catarina (in Portuguese).

Wang, X., Y. Hwang, and R. Radermacher. 2008. Investigation of potential benefits of compressor cooling. Applied Thermal Engineering 28:1791-7.

Yakhot, V. and L.M. Smith. 1992. The renormalization group, the [epsilon]-expansion and derivation of turbulence models. Journal of Scientific Computing 7:35-61.

Rodrigo Kremer, MEng, is Research and Development Engineer. Jader R. Barbosa Jr., PhD, is Professor. Cesar Deschamps, PhD, is Professor.

Rodrigo Kremer, [1] Jader R. Barbosa Jr., [2,*] and Cesar J. Deschamps [2]

[1] Embraeo Compressors, Joinville, Brazil

[2] Mechanical Engineering Department, Federal University of Santa Catarina, Florianopolis, Brazil

* Corresponding author e-mail: jrb@polo.ufsc.br

Table 1. Definition of the terms associated with each compressor component according to Equation 17 [SIGMA] [Q.sub.in] [SIGMA] [O.sub.mit] Suction muffler [Q.sub.sc] 0 Oil injection line 0 [Q.sub.op] Compression chamber 0 [Q.sub.w] Discharge chamber 0 [Q.sub.dc] Discharge muffler 0 [O.sub.vb] Discharge line 0 [Q.sub.ld] Electric motor 0 [Q.sub.m] Compressor (overall) 0 [Q.sub.ee] Compressor crankcase [Q.sub.ie] [Q.sub.ee] (shell) [SIGMA] [W.sub.in] [SIGMA] [W.sub.out] Suction muffler 0 0 Oil injection line 0 0 Compression chamber [W.sub.i] + [W.sub.m] 0 Discharge chamber 0 0 Discharge muffler 0 0 Discharge line 0 0 Electric motor [W.sub.el] [W.sub.m] + [W.sub.i] Compressor (overall) [W.sub.el] 0 Compressor crankcase 0 0 (shell) [SIGMA] [m.sub.in][h.sub.in] Suction muffler [m.sub.c][h.sub.s] + [m.sub.spc][h.sub.ie] + [m.sub.sr][h.sub.i] Oil injection line [m.sub.oil][C.sub.P,oil] [T.sub.op] Compression chamber [m.sub.s] [h.sub.sc] + [m.sub.dr] [h.sub.dc] + [m.sub.epc][h.sub.ie] +[m.sub.oil]C [P.sub.oil] [T.sub.oinj] Discharge chamber [m.sub.d]h + [m.sub.oil] [c .sub.P,oil] [T.sub.od] Discharge muffler [m.sub.vb][h.sub.dc] + [m.sub.oil] [C.sub.P,oi1][T.sub.dc] Discharge line [m.sub.c][h.sub.vb] + [m.sub.oil][CP.sub.oil] [T.sub.vb] Electric motor 0 [m.sub.c][h.sub.s] + Compressor (overall) [m.sub.oil][C.sub.P,oil] [T.sub.op] Compressor crankcase 0 (shell) [SIGMA][m.sub.out][h.sub.out] Eq. No. Suction muffler [m.sub.s][h.sub.sc] + (18) [m.sub.epc][h.sub.sc] Oil injection line [m.sub.oil][c.sub.P,oil] (19) [T.sub.oinj] Compression chamber [m.sub.d]h + [m.sub.sr] h + (20) [m.sub.spc] h +[m.sub.oil]C [P.sub.oil] [T.sub.od] Discharge chamber [m,sub.dr][h.sub.dc] + (21) [m.sub.c][h.sub.dc] [m.sub.oil] [c.sub.P,oil][T.sub.dc] Discharge muffler [m.sub.vb][h.sub.vb] (22) Discharge line [m.sub.c][h.sub.ld] + (23) [m.sub.oil [C.sub.P,oil][T.sub.ld] Electric motor 0 (24) Compressor (overall) [m.sub.c][h.sub.ld] + (25) [m.sub.oil] [C.sub.P,oil] [T.sub.ld] Compressor crankcase 0 (26) (shell) Table 2. Experimental data. Temperatures of the compressor components and regions for evaporating and condensing pressures corresponding to-27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F), respectively Temperature, [degrees]C ([degrees]F) Baseline POE ISO 10 Component or region (no atomization) External environment, [T.sub.ee] 24.7 (76.5) 24.6 (76.3) Compressor crankcase 34.7 (94.5) 33.5 (92.3) inlet, [T.sub.ps] Suction muffler inlet, [T.sub.s] 45.6 (114.1) 42.1 (107.8) Suction muffler outlet, [T.sub.sc] 56.0 (132.8) 53.1 (127.6) Cylinder wall, [T.sub.w] 89.7 (193.5) 64.4 (147.9) Discharge chamber, [T.sub.dc], 116.3 (241.3) 88.9 (192.0) Discharge muffler, [T.sub.vb] 95.3 (203.5) 78.2 (172.8) Discharge line, [T.sub.ld] 76.7 (170.1) 66.7 (152.1) Crankcase gas, [T.sub.ie] 72.6 (162.7) 61.6 (142.9) Primary bearings, [T.sub.b] 79.1 (174.4) 63.2 (145.8) Crankcase, [T.sub.h] 51.8 (125.2) 47.7 (117.9) Sump oil, [T.sub.oc] 60.1 (140.2) 56.3 (133.3) Oil shell inlet, [T.sub.op] -- 33.1 (91.6) Oil nozzle, [T.sub.oinj] -- 45.6 (114.1) Temperature, [degrees]C ([degrees]F) MO ISO 10 Component or region External environment, [T.sub.ee] 24.6 (76.3) Compressor crankcase 33.5 (92.3) inlet, [T.sub.ps] Suction muffler inlet, [T.sub.s] 41.7 (107.1) Suction muffler outlet, [T.sub.sc] 51.9 (125.4) Cylinder wall, [T.sub.w] 66.1 (151.0) Discharge chamber, [T.sub.dc], 85.4 (185.7) Discharge muffler, [T.sub.vb] 75.2 (167.4) Discharge line, [T.sub.ld] 67.8 (154.0) Crankcase gas, [T.sub.ie] 59.7 (139.5) Primary bearings, [T.sub.b] 62.5 (144.5) Crankcase, [T.sub.h] 47.4 (117.3) Sump oil, [T.sub.oc] 55.6 (132.1) Oil shell inlet, [T.sub.op] 36.2 (97.2) Oil nozzle, [T.sub.oinj] 44.6 (112.3) Table 3. Experimental data and modeling results. Efficiency parameters associated with the atomization of MO ISO 10. The evaporating and condensing pressures are -27[degrees]C (-16.6[degrees]F) and 42[degrees]C (107.6[degrees]F), respectively. Efficiency parameters [Q.sub.e] W (Btu/h) [W.sub.c] W [W.sub.i], W Baseline 180.87 (617.1) 108.19 76.65 (experimental) Atomization 183.03 (624.5) 118.44 79.07 (experimental) Percentage 1.2 9.5 3.2 difference (atomization baseline) (experimental) Baseline (model) 180.8 (616.2) 104.3 76.56 Atomization 183.6 (626.4) 113.1 75.59 (model) Percentage 1.5 8.5 -1.2 difference (atomization baseline)(model) Efficiency parameters [W.sub.eff], W COP [COP.sub.pV] Baseline 70.83 1.67 2.36 (experimental) Atomization 71.11 1.55 2.32 (experimental) Percentage 0.4 -7.2 -1.7 difference (atomization baseline) (experimental) Baseline (model) 71.66 1.73 2.36 Atomization 70.22 1.62 2.43 (model) Percentage -2.0 -6.4:. 3.0 difference (atomization baseline)(model) Efficiency parameters [COP.sub.eff] Baseline 2.55 (experimental) Atomization 2.57 (experimental) Percentage 0.8 difference (atomization baseline) (experimental) Baseline (model) 2.52 Atomization 2.61 (model) Percentage 3.6 difference (atomization baseline)(model) Table 4. Effect of nozzle position on some calculated efficiency parameters Nozzle position, mm (in.) 2(0.08) 7(0.28) [M.sub.oil], kg/h (lbm/h) 1.727 (3.81) 1.354 (2.99) [Q.sub.e], W (Btu/h) 181.5 (619.3) 182.6 (623.0) [W.sub.i], W 75.3 75.4 [COP.sub.pV] 2.41 2.42 Nozzle position, mm (in.) 13 (0.51) [M.sub.oil], kg/h (lbm/h) 0.916 (2.02) [Q.sub.e], W (Btu/h) 183.6 (626.4) [W.sub.i], W 75.6 [COP.sub.pV] 2.43 Table 5. Effect of oil atomization flow rate on some calculated efficiency parameters Oil flow rate, kgh (lbm/h) 0.0 1.376 (3.03) 2.294 (5.06) [Q.sub.e], W (Btu/h) 180.8 (616.9) 184.1 (617.9) 184.3 (628.8) [W.sub.i], W 76.6 75.3 75.3 [COP.sub.pV]- 2.36 2.44 2.45 Oil flow rate, kgh (lbm/h) 2.753 (6.07) [Q.sub.e], W (Btu/h) 184.3 (628.8) [W.sub.i], W 75.4 [COP.sub.pV]- 2.44 Table 6. Effect of oil temperature on some calculated efficiency parameters Oil temperature in the injection line, [degrees]C ([degrees]F) 32 (89.6) 36 (96.8) 40 (104.0) [Q.sub.e], W (Btu/h) 183.9 (627.5) 183.6 (626.4) 183.4 (625.8) [W.sub.i], W 75.5 75.6 75.7 [COP.sub.Pv] 2.44 2.43 2.42 Oil temperature in the injection line, [degrees]C ([degrees]F) 44 (111.2) 48 (118.4) 52(125.6) [Q.sub.e], W (Btu/h) 182.9 (624.1) 182.6 (623.0) 182.2 (621.7) [W.sub.i], W 75.8 75.9 76.0 [COP.sub.Pv] 2.41 2.41 2.40

Printer friendly Cite/link Email Feedback | |

Author: | Kremer, Rodrigo; Barbosa, Jader R., Jr.; Deschamps, Cesar J. |
---|---|

Publication: | HVAC & R Research |

Article Type: | Report |

Geographic Code: | 3BRAZ |

Date: | Jun 1, 2012 |

Words: | 11032 |

Previous Article: | Modeling of under-floor heating systems: a compromise between accuracy and complexity. |

Next Article: | Quality of inhaled air in displacement ventilation systems assisted by personalized ventilation. |

Topics: |