15

CHAPTER 2 LITERATURE REVIEW

2.1

INTRODUCTION Literature review, in the present work, has been conducted from the

thermodynamics and valve dynamics system point of view. A considerable quantity of work has been published in the literature for the refrigerant compressors and rotary compressors. In the published literature some work has been reported on the analysis of refrigeration rotary and reciprocating compressors with simple mathematical model. The work dealing with mathematical modelling and valve dynamics are alone considered in the present work. 2.2 MODEL FOR REFRIGERATION COMPRESSORS Gyberg and Stentoft Nissen (1980) have developed a model for fixed vane refrigerant rotary compressors which is based on a control volume for suction and pressure. The first law of thermodynamics and the law of continuity in dynamic form are used on these control volumes. The thermodynamic properties, mass flow, heat effect and compression power are calculated as a function of time or angle of rotation instead of a static average value. The developed model describes suction mass flow, pressure drop and temperature rise in suction pipe, gas leakage from pressure volume to suction volume, oil leakage from shell and shaft to suction and pressure volume, shaft torque arising from gas forces and pressure, temperature, internal energy and

16 enthalpy from the refrigerant equations. The differential equations for the first law of thermodynamics and the law of continuity are solved numerically by a simple Euler integration. The calculated values of volumetric efficiency are validated with measurements. It was observed that the measured values are closer to the experimental values. Chi and Didion (1982) presented a simulation model of a heat pump. The heat pump utilised a hermetic reciprocating compressor. The model equations were developed using a polytropic approach. Simulation of a 4-ton residential air-to-air heat pump operating in the cooling mode was performed with R-22 as the working fluid. The simulated start-up transients were compared to experimental data. The predicted results correlated reasonably well with the simulated results. Davis et al (1972) have simulated an automotive air conditioning system in conjunction with a vehicle compartment simulation. The model simulated an automotive type reciprocating compressor and predicted the refrigerant discharge state and the shaft work. The refrigerant mass flow rate was defined as an input parameter to the compressor model and was determined by a thermostatic expansion valve model in the system simulation. The compressor was modeled as isentropic using ideal gas relationships. In addition, a refrigerant leakage coefficient, and the mechanical efficiency were designated as input parameters to the model. The compressor model also accounted for work done in the suction and discharge valves by relating the work to the associated pressure drops empirically. The suction valve pressure drop was estimated as a fraction of the total suction pressure and the discharge valve pressure drop was estimated from the ideal gas relation. The model did not account for differences in compressor size or speed. In addition, no experimental verification or simulation results were presented.

17 Davis and Scott (1976) outlined a hermetic compressor model for use in a system simulation. The steady-state compressor model included heat transfer within the compressor shell and pressure drop in the suction and discharge passages. In addition, the model included the electric motor dynamics and allowed for the modeling of different compressor sizes and speeds. The model required that the mechanical and volumetric efficiencies are specified as input parameters. These parameters were to be determined from experimental data. No experimental verification or simulation results were presented. Cecchini and Marchal (1991) developed a general steady-state system simulation model based on experimental data. An attempt was made to develop component models in which the component could be characterized by a small number of parameters estimated from a few experimental data points. The compressor model utilized a polytropic-based expression for predicting the discharge refrigerant state similar in form to the reversible polytropic work of compression. The polytropic exponent was defined as an input parameter to the model. The steady-state refrigerant mass flow rate was estimated using an equation based on the pressure ratio and polytropic exponent with the compressor displacement and the clearance fraction as equation parameters. Although the model accounted for different compressor geometries, none of the model equations contained a reference to the compressor speed. It was unclear what assumptions were made about heat loss from the compressor and how the compressor power was determined. The details concerning the development of the compressor model equations were not provided. The system model was verified through the use of experimental data. For an air-to-air system, the model was able to predict the compressor power within 10 %. Furthermore, when the model was used to simulate an air-to-water heat pump, the compressor power was predicted to within 7 %.

18 Seshaiah et al (2007) have developed a mathematical model of a rotary twin screw compressor. Mathematical analysis was carried out on the basis of the laws of perfect gas and standard thermodynamic relations. Heat transfer coefficient required for computer simulation was experimentally obtained and used in performance prediction. The behaviour of control mechanism has been theoretically investigated by a simulation model of control mechanism of variable displacement swash plate compressor of automotive air conditioning system developed by Changqing Tian et al (2007). Dhar and Sodel (1979) have discussed a vapor-compression system simulation model utilizing a hermetically sealed reciprocating compressor. They assumed that the compression process is polytropic, that the compressor operates at a constant speed, and that the pressure drop in the suction and discharge valves is negligible. Although no equations were given, it was claimed that the refrigerant mass flow rate and the work done by the compressor on the refrigerant could easily be derived as a function of compressor geometry, suction and discharge pressure ratios, refrigerant leakage rate, compressor speed and the specific heat ratio of the refrigerant. The discharge state refrigerant enthalpy was determined by adding the work done by the compressor on the refrigerant to the suction refrigerant enthalpy. Then an overall efficiency was used to determine the rate of energy released from the compressor due to friction and other losses in addition to the work input to the compressor. The model also included calculations of the internal heat transfers within the compressor shell. In addition, a detailed analysis of the oil-refrigerant interaction in the compressor sump was also included. Refrigerant property calculations were performed using curve fits of refrigerant property tables.

19 Domanski and McLinden (1990) outlined a rudimentary steady state compressor model included as a part of a generic system simulation model. The compressor model provided three options: an isentropic compressor, a polytropic compressor and a hermetic polytropic compressor including internal heat transfer and volumetric efficiency. The heat transfer equations for the hermetic compressor were based on simple energy balances with heat transfer coefficients based on Prandtl and Reynolds number correlations. The refrigerant state after compression was determined from the reversible polytropic work of compression and user defined polytropic efficiency. The polytropic efficiency was defined as the ratio of the polytropic work of compression to the energy transferred to the refrigerant during compression. The polytropic exponent was defined using the polytropic efficiency and the isentropic index. The isentropic index was determined from an ideal gas relationship instead of the ratio of specific heat capacities. A polytropic process was chosen rather than an isentropic process for simulations claiming that the polytropic efficiency was more constant than the isentropic efficiency which varies with pressure ratio. The volumetric efficiency was determined from an equation based on the pressure ratio and polytropic exponent. Equation was not provided for predicting the overall power input to the compressor which was composed of power transferred to the refrigerant and heat loss. In addition, the model provided no provisions for different compressor speeds or sizes. No experimental verification was provided, only simulation results comparing the performance of different refrigerants were presented. The refrigeration scroll compressor performance was predicted by a simple mathematical model developed by Baolong Wang et al (2007). Energy and mass conservation equations have been used for modeling with the following assumptions: (i) Refrigerant in the working chamber is homogeneous. (ii) Variations of gravitational and kinetic energy are

20 negligible. (iii) Effects of the oil are neglected. It was suggested that the model could be used to predict the general performance and inner compression process of the injected scroll compressor. The model was validated with the help of experimental results. A simplified analysis has been presented for the estimation of reciprocating hermetic compressor performance by Sung Tai Kim and Tae Sik Nin (1984). The thermodynamic properties are represented by ideal gas equation of state. A comparison was made between the results of simulation and experimental data. It was assumed that the temperature and pressure are the same in the suction chamber and inside the cylinder and similarly in the discharge chamber and inside the cylinder. Lawson and McLaren (1984) have suggested an approach to the development of a simple mathematical model of refrigerating reciprocating compressor. A model was developed with theoretical values of various coefficients which do not require any modifications. The computer model developed allowed detailed calculation of pressure drops through the various flow elements described by the portion of the suction and discharge valves and piston. The dynamics of the reed are modeled assuming the reed to be a cantilever beam. The study was confined to the analysis of the valves fluttering. Only theoretical analysis was carried out. Pattari and Bonne (1984) have developed a simple model for the simulation of refrigerant compressor considering clearance volume, blow-by losses and various polytropic coefficients. The model was used to analyse the physical performance of a compressor, including various losses and physical parameters, which are tailored to a particular, real compressor by using measured data points as published by standard compressor performance maps and determining the unknown parameters by solving the corresponding system of linear and non-linear equations.

21 Yezheng Wu et al (1984) developed a mathematical model consisting of nonlinear ordinary differential and algebraic equations for simulation and optimization. The pressure volume and valve lift diagrams have been obtained by simulation model with different stroke/diameter ratio. An investigation was made by Ruihu Liu and Sicheng Ehou (1984) to determine the temperature distribution on the cylinder wall of a reciprocating refrigeration compressor and the coefficient of heat transfer between the cylinder wall and gaseous refrigerant. The purpose was to study the working process of a refrigerating compressor and to find correlations that can be applied in computer modeling of this type of compressor. The temperature distribution on the cylinder wall was measured at different pressure ratios, different suction temperatures and different speeds. The correlations were obtained by using the least square method. The heat flow rate was calculated based on the differential equation for the first law of thermodynamics, the data of pressure-crankangle diagram and the data of valve displacement using the method of orthogonal polynomial fitting. Investigation has shown that the temperature distribution on the cylinder wall varies with pressure ratio, suction temperature, speed, oil temperature and with location. Ellison et al (1979) presented an overview of the Oak Ridge National Heat Pump Model. This model simulated the steady state performance of a hermetic compressor operating at a constant speed. The model required performance and efficiency parameters as well as heat transfer information obtained from experimental data as input parameters. A comparison was made for a few experimental values with simulation results and found to be correlating well. Hai and Squarer (1974) presented a detailed model for simulating multi-cylinder reciprocating compressors. The governing differential

22 equations were developed from ideal gas relationships and were based on incremental changes in the crankshaft angle. In addition, the model used differential equations to model the valve dynamics throughout the compression cycle. The model equations were capable of predicting the compressor torque in addition to refrigerant mass flow rate. However, there was no prediction of the discharge refrigerant state. The compression cycle was simulated for a constant compressor speed. The cycle averages were compared to three experimental data points. The refrigerant mass flow rate corresponded reasonably well with the experimental data. However, the predicted cycle average torque did not correlate well with the experimental results. James and James (1986) provided a transient model of a heat pump utilizing water cooled hermetic compressor operating at a constant speed. To model the internal heat transfers, the compressor was divided into zones. A mass and energy balance was then used in each zone. However, in actual compressor the cylinder was assumed to be only one zone. Furthermore, many of the internal heat transfers were ignored. The internal heat transfers that were modeled were complicated by the lack of heat transfer coefficients. The refrigerant mass flow rate through the compressor was determined using a constant volumetric efficiency which was a model input parameter. James and James expressed an interest in relating this efficiency to pressure ratio at a later date. The work of compression was determined using integration of the polytropic process. The polytropic exponent was designated as a model input parameter. Since no polytropic efficiency was used, the polytropic equation provided only the reversible work done on the refrigerant. The predicted work of compression does not include frictional and other losses within the compressor. In addition, it was assumed that there was no net heat loss from the compressor. Therefore, a simple energy balance was used to predict the discharge refrigerant state. The only transient equations provided involved the

23 internal heat transfer equations. These transients were due solely to the thermal capacities of the compressor components. No experimental verification of the model's performance was provided. However, a computer code listing of the system model program was included. Singh (1984) has developed a user-oriented simulation program for refrigeration reciprocating compressor accounting inlet and discharge piping interactions. Basic thermodynamic equations and valve dynamics equations have been used. A single degree of freedom, damped mass-spring model was used to describe valve motion. The pressure variations with crankangle and cylinder volume were compared with experimental values. Rajendra Singh (1984) has carried out first law analysis of a compressor to get the data required for second law analysis. Only a theoretical analysis has been carried out. The conclusions from the study were (i) First law and polytropic model match for cylinder pressure and mass of gas in the cylinder but not for cylinder air temperature. (ii) Cylinder air temperature prediction depends strongly on the mass flow rates through inlet and discharge valves, valve deflections and heat transfer rate. MacArthur (1984) modeled a hermetic compressor for use in a heat pump. The transient model was based on an energy balance on the compressor cylinder and included the throttling process of the refrigerant passing through the suction and discharge valves. The model also considered the heat transfer to the cylinder walls and the thermal capacity of the compressor components. The compression process was assumed to be polytropic. The input parameters to the model included compressor displacement, clearance factor, compression efficiency, heat transfer coefficients and thermal masses. Simulation results for the steady state refrigerant mass flow rate in addition to the transient response of the

24 discharge refrigerant temperature and mass flow rate during start-up were presented. However the experimental verification was not provided. A model for hermetic reciprocating compressors of refrigeration applications was presented by Navarro (2007). This model was able to predict compressor efficiency and volumetric efficiency in terms of a certain number of parameters representing the main sources of losses inside the compressor. The model provided helpful information about the way in which the compressor was designed and worked. A statistical fitting procedure based on the Monte Carlo method was developed for its adjustment. The model can predict compressor performance at most points with a maximum deviation of 3%. A possible gas condensation on cold spots inside the cylinder during the last part of the compression stroke was also evaluated. Murphy and Goldschmidt (1985) performed a study of the start-up transients found in a residential air conditioner. Although the objective was to model start-up transients, the constant speed hermetic compressor was modeled using steady-state equations. To account for friction and other losses in the compressor power determination, a modified form of a power equation for an ideal compressor was used. The resulting expression for compressor power was a linear function of only the suction and discharge refrigerant pressures. A least squares analysis was used to determine the equation constants. In a similar manner, the refrigerant mass flow rate was expressed as a function of the refrigerant pressure ratio and the suction refrigerant density. When the compressor power was simulated under system start up, the predicted power was observed to follow the same trends as the experimentally measured values. However, the predicted power was significantly lower than the experimental values although the steady-state values were the same. This indicates that there are other significant transient factors besides the refrigerant suction and discharge pressures which influence the compressor power. Murphy and Goldschmidt offered two explanations for this

25 discrepancy between the predicted and experimental values. First, during start up, the oil could foam and cause increased flow resistance through the compressor. Second, there is an additional superheating of the suction refrigerant in the hermetic compressor shell due to the initially hot components within the shell. This additional superheat would not appear in the experimentally measured suction pressure. This is a likely explanation since the internal heat transfer within the compressor shell was ignored in the model. The authors concluded that steady-state compressor models used for quasi steady-state transient simulations provide good trends during start up, but provide poor qualitative numbers. 2.3 MODEL FOR AIR COMPRESSORS A mathematical model of a rotary compressor used in gas turbine power plants was developed by Venturini (2005) for compressor dynamic simulation. After developing the compressor mathematical model through a physical based approach (laws of conservation and thermal balances), the model was implemented through the MATLAB SIMULINK tool. Activity analysis has been carried out to evaluate the influence of model parameters on the model response. Finally the model was formulated for a small size multistage centrifugal compressor and validated through experimental data taken on the compressor under investigation. The analysis showed that (i) intake duct friction factor mainly affects the inlet compressor pressure and mass flow rate, (ii) exhaust duct friction factor influences the outlet compressor pressure significantly and (iii) an increase in the exhaust duct hydraulic mean diameter allows a greater mass flow rate, while outlet compressor pressure and outlet compressor temperature decrease. Model validation against experimental data showed that the predicted values closely follow the experimental data.

26 A model for the thermodynamic analysis of reciprocating compressors was presented by Stouff (2001). The model was based on main and secondary dimensionless parameters. Expressions for the volumetric effectiveness, the work per unit mass and the indicated efficiency were derived. The model has been used in order to predict the performance of a reciprocating air compressor under various operating conditions. The relative importance of the various losses and the influence of different parameters on the reciprocating compressor behaviour were discussed. Especially the influence of in-cylinder residual mass fraction and the wall to fluid heat transfer on the reciprocating compressor performance were highlighted. 2.4 DYNAMIC BEHAVIOUR OF COMPRESSOR VALVES Piechna (1984) has theoretically investigated the dynamic behaviour of a simple reed valve used in refrigeration compressors. Finite element techniques have been used for the study of dynamic behaviour of a reed valve. Valve deflections have been obtained at different valve thicknesses and different valve stops. It was concluded that (i) the Reed stop with constant height causes a strong bending of reed valve which also causes a pressure pulsation in the cylinder, (ii) the oblique reed stop found to be a much better solution, (iii) the use of oblique stop reduce forces and moments, reduces impact velocity and thereby reducing reed oscillations, (iv) the discharge plenum chamber pulsation can change the valve operation and (v) the parameters of the suction valve has a great influence on the compressor volumetric efficiency. Werner Soedel (1984) has worked on mechanical behaviour of reed valves in refrigerating compressors. He has predicted the effect of varying polytropic index during compression or expansion processes in his study. The fluttering of reed valves both on suction and discharge sides were predicted.

27 Boswirth (1984) has discussed a model accounting for non-steady flow effects in valve channels. Gas mass inertia was found to be the most important effect. Adequate simplifications were introduced to allow for a manageable system of equations. He has obtained the pressure drop across the valve with steady and unsteady flow conditions. Vibrational analysis of compressor ring-valve plates was carried out by Wang Di-seng and Lu Pin (1984). The concept of deformation involved in the motion processes was used. Emphasis of research was centered on flexural warping vibrations during opening and closing. Mechanical modeling and mathematical interpretations of plate motion with variously positioned spring systems were established, enabling theoretical solutions of model shape functions, stresses and strains. The work demonstrated that the dynamic analysis of the valve plate can be accomplished by the hybrid approach of analytical-numerical methods. It was observed from the analysis that (i) non-steady state flexural-torsional vibrations take place during the opening and closing processes of valve plate motion, (ii) dynamic stresses are produced by the flexural-torsional vibrations and cause radial fatigue fractures, (iii) the vibrational characteristics of a multi-small-spring system are closely related to the uniformity of stiffness and the number of small springs and (iv) non-uniformity of spring stiffness or the failure of an individual spring results in the increase of eigen-frequency band width with a loss of stability. 2.5 LEAKAGE LOSSES IN REFRIGERANT COMPRESSORS The total refrigerant loss through the clearance between the piston and cylinder wall in hermetic compressor was studied with the help of a simple mathematical model developed by Rogerio et al (1984). The theoretical prediction was compared with measured values. The analysis was carried out for different clearances with the assumptions (i) the phenomena is

28 a quasi-steady flow, (ii) the boundary condition variations are instantaneously transmitted throughout the fluids, (iii) the oil film has constant thickness, (iv) the flow of both Newtonian fluids is incompressible, (v) the end effects on the flow are disregarded and (vi) inertia forces are negligible compared to viscous and pressure forces. It was observed from the analysis that (i) the relationship between the pressure and viscous forces is important to determine the flow condition in the clearance, (ii) when there is no break in the oil film the leakage of refrigerant is dissolved in the oil and (iii) the experimental results agree with the numerical calculation based on the mathematical model especially for small clearance values. Pandeya (1984) has studied the kinematic differences of different compressors and the effect of these differences on the valve losses. A simple mathematical relationship between the kinematic characteristics of a compressor and the corresponding flow losses was developed. Four different kinematic arrangements of the reciprocating as well as the rotary concepts (common, reverse scotch yoke reciprocating arrangements and rolling piston arrangements) were considered for the analysis. The study showed that (i) suction flow losses in the three reciprocating arrangements are very close to each other, but the discharge flow losses are significantly different, (ii) between the three reciprocating arrangements, the regular reciprocating arrangement has the highest loss and the reverse scotch yoke the lowest and (iii) the rolling piston rotary has the least loss of all the four. 2.6 FAILURE MECHANISM OF VALVES Different reed vale failures in refrigeration systems were analyzed by Glaeser (1999). Failure in each case was by fracture of one or more reeds in the valves. Most of the failures could be diagnosed as impact fatigue in nature. The findings from analysis were (i) the thin section inherent in reed design makes them susceptible to fracture, (ii) the reed must be free of surface

29 defects such as dents and scratches to minimize fatigue fracture, (iii) although the reeds are light weight, they can impact the valve seat with sufficient energy to produce surface damage and fatigue fracture and (iv) bending of the reed during contact produces a fatigue state of stress and open surface cracks in dented zones. Fractographic analysis of broken reeds has shown evidence of fatigue. 2.7 SUMMARY From the literature survey made it has been observed that considerable work has been done on mathematical modelling of rotary and refrigerating compressors. A very few researchers have worked on reciprocating air compressors used for industrial applications. Experimental results from the compressors with the speed range of 500 rpm to 1500 rpm were used in previous work. Therefore the present work was aimed at developing a mathematical model of an automotive reciprocating air compressor and to test four high speed compressors with different configurations for model validation. In Chapter 2 the various works carried out by researchers in the compressor field are explained. The theory of automotive compressor and performance parameters are detailed in Chapter 3. The valve characteristics of high speed compressors are also discussed.