- Email: [email protected]
Experimental and numerical investigation of lubrication system for reciprocating compressor
International Journal of Refrigeration 108 (2019) 224–233
Contents lists available at ScienceDirect
International Journal of Refrigeration journal homepage: www.elsevier.com/locate/ijrefrig
Experimental and numerical investigation of lubrication system for reciprocating compressor Mustafa Ozsipahi a,∗, Haluk Anil Kose a, Sertac Cadirci a, Husnu Kerpicci b, Hasan Gunes a a b
Department of Mechanical Engineering, Istanbul Technical University (ITU), Gumussuyu, 34437 Istanbul, Turkey Arcelik Research and Development Center, Tuzla, 34950 Istanbul, Turkey
a r t i c l e
i n f o
Article history: Received 22 February 2019 Revised 16 August 2019 Accepted 18 August 2019 Available online 22 August 2019 Keywords: Reciprocating compressor Lubrication system Sliding mesh Moving reference frame Two-phase flow
a b s t r a c t Sufficient lubrication of the moving parts in a variable capacity inverter compressor is vital since it can directly affect the performance and expected lifetime. In this study, the lubrication system of a compact inverter compressor (CIC) is numerically and experimentally investigated. In the numerical modeling, a finite volume-based algorithm is used to model two-phase (air–oil) flow inside the compressor using Volume of Fluid Method (VoF) method. Transient behavior of the oil flow under laminar flow conditions is both simulated by imposing Sliding Mesh (SM) and the Moving Reference Frame (MRF) methods at various crankshaft speeds varying between 1200 and 4500 rpm. The measurements are taken using an experimental setup to compare/validate CFD results obtained from SM and MRF methods. Flow visualizations are performed with a high-speed camera to determine the oil climbing and required time for sufficient lubrication precisely. Moreover, the acceleration of the crankshaft is determined via high-speed camera and employed as a user-defined function to model the start-up period of the compressor. It is shown that with increasing crankshaft speeds, the average oil mass flow rate released from the upper part of the crankshaft is increasing almost linearly. It is also shown numerically that with increasing oil viscosity, the mass flow rate decreases. The comparison of experimental and CFD results reveals that the MRF solutions are in a better agreement with the measurements up to 2800 rpm. The SM method became favorable as a CFD method due to better agreement with measurements between 30 0 0 and 4500 rpm. © 2019 Elsevier Ltd and IIR. All rights reserved.
Étude expérimentale et numérique d’un système de lubrification pour compresseur à piston Mots-clés: Compresseur à piston; Système de lubrification ; Maillage glissant; Cadre de référence mouvant; Écoulement diphasique
1. Introduction The inverter-type compressors are variable capacity compressors since the cooling capacity requirement is adjusted with crankshaft speed. Since the cooling capacity is directly related to the crankshaft speed, the design of the oil management system which provides required amount of the lubricant to the journal bearings and other moving parts are crucial to warrant expected
∗
Corresponding author. E-mail addresses: [email protected] (M. Ozsipahi), [email protected] (H.A. Kose), [email protected] (S. Cadirci), [email protected] (H. Kerpicci), [email protected] (H. Gunes). https://doi.org/10.1016/j.ijrefrig.2019.08.026 0140-7007/© 2019 Elsevier Ltd and IIR. All rights reserved.
lifetime and performance of the compressor. In this context, some of the recent studies in the household refrigerator compressors focus on the improvement of the oil management system (Alves et al., 2011; Ozsipahi et al., 2014). In the last decade increasing usage of the inverter compressors urges manufacturers to modify existing oil management system in hermetic reciprocating compressors. Numerical calculations provide a chance to investigate oil management system design of inverter type reciprocating compressors in more detail. As rapid and costeffective numerical approaches open the way of comparison of the number of various design alternatives with a lesser prototype, peer investigations require higher computation time for the most realistic approach. CFD is a powerful alternative to experiments in the
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Nomenclature
α μ φ ρ σ A dt dx g m p S t u ug V CFD CIC COP cSt fps MRF PISO rpm SM UDF VoF
volume fraction (dimensionless) diffusion coefficient (dimensionless) dynamic viscosity (kg m−1 s−1 ) general scalar (dimensionless) density (kg m−3 ) standard deviation (dimensionless) area (m2 ) time step size (s) minimum cell size (m) gravity (m s−2 ) −1 mass flow rate (kg s ) −2 pressure (N m ) source term (kg m3 s−1 ) time (s) velocity (m s−1 ) mesh velocity (m s−1 ) volume (m3 ) computational fluid dynamics compact inverter compressor coefficient of performance centistokes frame per second moving reference frame pressure-implicit with splitting of operators rotations per minute sliding mesh user defined function volume of fluid
design phase of a product moreover the results of numerical calculations should be consistent with measurements. In the CFD simulations, MRF method is widely preferred to model the motion of the crankshaft that pumps oil from the sump to the moving components of the compressor (Alves et al., 2011; Ozsipahi et al., 2016; 2014; Posch et al., 2018). Important outputs of such CFD calculations are the climbing time of oil and the amount of oil spread from the upper part of the crankshaft at various speeds. Kerpicci et al. (2013) investigated oil flow in the hermetic reciprocating compressor and found oil climbing time by using a high-speed camera and CFD simulations. Ozsipahi et al. (2014) numerically investigated how the amount of oil mass flow rate in constant capacity hermetic compressors was affected by some operating conditions such as crankshaft speed, submersion depth of the crankshaft in the oil sump, oil viscosity and also the shape of the helical oil path on the crankshaft. They showed that the oil mass flow rate was directly proportional to the crankshaft speed, submersion depth and inversely proportional to the oil viscosity. In another study, Ozsipahi et al. (2016) suggested an analytical model for the screw pump used in the crankshaft of the hermetic variable capacity compressors to increase the oil mass flow rate at low speeds. The effect of the screw design on maximizing the flow rate was investigated in the analytical model and found to be consistent with the CFD simulations. Posch et al. (2018) analyzed the oil pump system of a hermetic reciprocating compressor numerically. They have split up the oil pump into its individual pumping parts and developed a tool that can be recommended for use in the design of reciprocating compressors in order to rapidly obtain simulation results. Tada et al. (2014) proposed an uncoupled numerical model for the variable capacity compressors which are employed in domestic applications and investigated the effects of the immersion depth of the crankshaft and the compressor speed both numerically and experimentally. Pizarro-Recabarren and
225
Barbosa (2016) proposed a differential model to calculate the oil film and compressor shell temperature distributions and verified it by experiments. The proposed model can be used to predict the total oil flow rate and the fraction of this flow rate flowing as a film on the shell. Afshari et al. (2017b) pointed out the significance of lubrication on the performance of reciprocating compressors used in air-water heat pumps. The measurements carried out with refrigerant mixtures and several oil samples showed that oil with lowest viscosity provides maximum COP value. It was concluded that oil with less viscosity was suggested for maximum exergetic efficiency in comparison to other samples. In another study, Afshari et al. (2017a) investigated how hydrofluorocarbon (HFC) refrigerant affects compressor operating conditions and system performance by experiments. Experiments have been carried out with refrigerant R404A and appropriate lubricants and tested the effect of ideal gas and real gas assumptions on the energetic analysis. The refrigerant amount was found to be very important in the COP, however the heat exchangers were also studied to increase system performance. Navarro et al. (2012) investigated the effect of oil sump temperature in hermetic compressors for heat pump applications and found relations between the oil sump temperature and several operating conditions such as condensation temperature or design parameters such as stroke and number of cylinders in reciprocating compressor, experimentally. In general, the oil sump temperature increases with increasing condensation temperature and piston stroke. In the CFD simulations where rotating parts are present, there are different techniques to be implemented to approach a more realistic solution. In the literature, dynamic meshes are intensively used in numerical investigation of the screw and rotary compressors. Kovacevic et al. (2003) presented the interface method for the SM method for the screw compressor. It is reported that a good agreement between the predicted and measured performance is a strong indication that CFD is a powerful tool for the design and optimization of the screw compressor. Liu and Hill (20 0 0) suggested Frozen Rotor model, Circumferential Average model, and the transient SM method. Depending on the problem definition such as the existence of non-axisymmetric components or the interaction between the rotating and stationary parts, these models can potentially give different results. Deng and Hu (2016) simulated a rotary two-stage inverter compressor with SM model and confirmed the CFD results with experimental data. Lane et al. (20 0 0) compared the SM and MRF methods for modeling of stirred tanks pointed out that both methods are in good agreement with the measurements and the MRF method is a practical alternative by providing faster solutions. Gullberg and Sengupta (2011) investigated axial fan performance by comparing MRF and SM methods with experiments. They reported that SM is far less sensitive to domain specifications compared to MRF method. Singh et al. (2007) simulated the flow in a continuous flow pump-mixer and compared the results with experimental data obtained on pilot-scale setup. It is noted in the study, in most cases, SM is found to perform better than MRF method but more validations are necessary to endorse the superiority of the modeling approach. Ozsipahi et al. (2018) compared the numerical performance of the SM versus the MRF method in modeling the lubrication system of a CIC. Oil climbing time with infinite acceleration, instantaneous flow fields of the two-phase flow and the mass flow rate predictions obtained by the two methods have been compared in the study. In the present study, CFD simulations of the lubrication system of a CIC are carried out by using the MRF and SM methods and compared to measurements obtained from an in-house built setup. The originality of the study is to see the effects of both numerical approaches on the oil mass flow rate and address the robustness of the numerical approaches by detailed experiments which is not found in the related literature survey. Moreover, the flow
226
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Fig. 1. (a) Photo of a CIC, (b) CAD model of the CIC and (c) Crankshaft of the compressor
Table 1 The details of the compressor model. Model Weight Motor Type Displacement Cooling Capacity COP Working Speeds Oil Charge Height of the crankshaft Dimensions (W × L × H)
Table 2 Mesh independence tests for SM method. VNTZ type 6 kg Brushless DC 11.28 cc 201 W at 3000 rpm 1.9 W/W at 3000 rpm (Standard ASHRAE condition) 1200–4500 rpm 210 cc 9.8 cm 14.5 × 15 × 13.5 cm
visualization test bench is used to validate oil climbing time and flow fields of the oil spreading out from the crankshaft outlet is demonstrated both by experiments and the SM method. The structure of the paper is organized as follows: The details of the handled problem are summarized in Section 2. The numerical models, governing equations and boundary conditions introduced in Section 3. Oil mass flow rate test bench and flow visualizations are addressed in Section 4. Then, the evaluation and comparison of the results are presented in Section 5. Finally, the main findings are summarized in Section 6.
Case ID
Averaged oil mass flow rate [g s−1 ]
CutCell, 790 K CutCell, 1.2 M CutCell, 1.8 M CutCell, 7.9 M Tetrahedral, 2.5 M Tetrahedral, 5 M Tetrahedral, 10 M
3.31 3.29 3.23 3.21 3.31 3.21 3.07
Relative error [%] 0.6 – −1.8 −2.4 0.6 −2.4 −6.7
ous kinematic viscosities depending on the operating conditions (GMBH, 2018). Lubrication oil is filled into the compressor in assembling process and the compressor is expected to self-lubricate during the operation of the refrigerator. The crankshaft which acts like an oil pump delivers oil from the sump to the rotating and moving parts of the compressor. Oil is sucked from the bottom of the crankshaft and transferred with the centrifugal forces caused by the rotation and the viscous effects. Oil climbs through the eccentric hole up to the first journal bearing. It further moves on the double-helical paths which join to each other before the main journal bearing. Finally, oil is spread out from the hole located at the uppermost of the crankshaft.
2. Problem definition In this study, the compressor of interest is a hermetic reciprocating compressor for household appliances which is shown in Fig. 1. The coefficient of the performance of the refrigeration units is directly influenced by the performance of the compressors. Environmental laws, legislation and regulations are forcing the manufacturer to improve their products. With improved manufacturing techniques and novel compression concepts, the variable capacity hermetic compressors are designed more compact for both reducing the production cost and increasing cooling domain in the refrigerator. Moreover, due to the changing energy regulations, the expectancy of the higher isentropic and volumetric efficiencies are still a challenging issue in the design of compact inverter compressors. Depending on the refrigeration capacity inverter type compressors operate at various rotational speeds. Oil management and lubrication mechanism in journal bearings may change drastically between compressor maximum and minimum speeds. Nevertheless, sufficient lubrication should be provided to the bearings and all moving parts to avoid any mechanical damage on the compressor. The technical specifications and dimensions of the compressor used in this study are shown in Table 1. The lubricants which are widely used in hermetic reciprocating compressors are synthetic and mineral based oils with vari-
3. CFD modelling 3.1. Sliding mesh The SM method is a special type of dynamic mesh motion which boundaries and mesh cells move together in a rigid-body motion. As a consequence, SM method is inherently unsteady due to the motion of the mesh with time. The cells in the zones are not deforming and the governing equations of the fluid motion are different than used in MRF method. SM method is defined as the most accurate method for simulating flows in multiple reference frame, but also the most computationally demanding (Fluent, 2013). In the first method of CFD calculations, the numerical domain is simplified. The interface encapsulating the crankshaft is introduced in the numerical model as shown in Fig. 2. Since SM method provides the opportunity to track the motion of the crankshaft in temporal domain, it is possible to obtain instantaneous oil iso-surfaces. The conservation equation in general integral form for any scalar used in the SM model is given in the following equation:
d dt
V
ρφ dV +
∂V
ρφ (u − u g ).dA =
∂V
∇ φ .dA +
V
Sφ dV
(1)
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
227
Table 3 Mesh independence tests for MRF method. Case ID
Averaged oil mass flow rate [g s−1 ]
CutCell, 986 K CutCell, 1.3 M CutCell, 2.7 M
1.61 1.32 1.3
Relative error [%] 22 – −1.5
3.3. Governing equations and boundary conditions
Fig. 2. Simplified domain used in SM method: red part: crankshaft, grey part on the top and bottom of the crankshaft: interface.
In the numerical modeling, the VoF model with implicit scheme was used since there is an interface between the free surface of oil and the refrigerant (in the present study air). Air was preferred in CFD calculations because the experiments were carried out at atmospheric conditions without refrigerant. For the twophase model, the continuity and momentum equations are given in Eqs. (4) and (5), respectively where α represents the volume fraction (Fluent, 2013).
∂ (α ρ ) + ∇ (αl ρl u l ) = 0 ∂t l l ρ
∂ u u ) = ρ g − ∇ p + ∇ μ ∇ u + ∇u T + ∇ (u ∂t
(4)
(5)
The boundary conditions for the MRF method were defined as follows: •
• •
•
Same pressure was imposed at the top of the free surface in the oil sump and crankshaft outlet. Angular velocity was defined on the walls of the crankshaft. All remaining walls and interfaces between domains were stationary. The thermophysical properties of all fluids were constant.
Fig. 3. Computational domain used in MRF method.
Since the time rate of change of the cell volume is zero, V n+1 = V n is valid for the SM model. The first order temporal derivative in Eq. (1), can be rewritten using first-order forward difference formula as given in the following equation:
d dt
(ρφ )n+1 − (ρφ )n V ρφ dV = t V
(2)
In order to satisfy mass conversation, the time derivative of the control volume can be formulated as given in the following equation: nf
dV j = 0 g, j .A = u dt
(3)
j
3.2. Moving Reference Frame In the MRF method, the relative motion of a moving zone with respect to adjacent zones is not taken into account, in other words, the mesh is fixed. This means freezing the motion of the crankshaft of a hermetic compressor in a fixed position and computing the instantaneous flow field with the crankshaft in the corresponding position. In many engineering applications, where a rotating part is present in the flow domain, MRF is widely preferred as the numerical approach. Some examples for this include turbomachinery applications with weak rotor-stator interaction, wind turbines and mixing tanks. Fig. 3 shows the fixed domain used in MRF where the red part represents the crankshaft, the grey part is the outer wall around the crankshaft and the blue part is the oil sump.
In the SM method, the only moving part was the crankshaft and the submersion depth of the crankshaft in the oil sump was defined similarly to the MRF method. The fluid flow in the lubrication system was solved numerically using the finite-volume based ANSYS-Fluent v16 package. PISO algorithm was used for pressure– velocity coupling in the simulations. 3.4. Mesh and time dependency tests Intense mesh independence tests have been carried out for both numerical methods. A reference case at certain operating conditions was selected to obtain the required mesh convergence results. The reference case was chosen at 30 0 0 rpm with a 15 mm immersion depth of using a 5 cSt oil kinematic viscosity. Unsteady calculations were performed with PISO algorithm for pressurevelocity coupling solving second order accurate momentum equations. The time step size was dt = 5 × 10−4 s. Table 2 shows the mesh independence tests for CutCell and tetrahedral meshes. Table 3 indicates the mesh independency tests for MRF method. Relative errors in the averaged oil mass flow rates with respect to the solution from 1.3 million mesh were calculated for the determination of the resulting mesh for that method. Table 4 shows the time sensitivity analysis which were performed for the SM method using 1.2 million number of elements in the mesh. In this analysis the relative errors were calculated with respect to the following time step size, as a result time step size was chosen 5 × 10−5 seconds. Parallel computations have been performed on a 28-core processor computer and average simulation can have a wall-clock time approximately 20 and 60 h for the MRF and SM method, respectively.
228
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Fig. 4. 2D- sketch of the experimental setup (the figure is not proportional).
Fig. 5. (a) Standard deviation of the oil mass flow rates at 30 0 0 rpm, 5 cSt; (b) temperature fluctuations during measurement.
Table 4 Time independence tests.
4. Experimental studies
Time step size [s]
Averaged oil mass flow rate [g s−1 ]
0.002 0.001 5 × 10−4 2.5 × 10−4 1 × 10−4 5 × 10−5 2.5 × 10−5
3.70 3.55 3.29 2.45 1.85 1.61 1.55
Relative error [%] 138.9 124.2 112.6 58.2 19.3 3.8 –
Table 5 The list of the instrumentation. Data acquisition Thermocouples High speed camera Spotlights PID controller Power controller Heater
Keysight 34970A T type (%0.4 accuracy) Photron 512 PCI 32K dedocool COOLH Autonics TX4S Autonics SPC1-35 Sheet type adhesive flexible resistance (1200 W)
An experimental setup was built for oil mass flow rate measurements and flow visualizations to validate the CFD results both quantitatively and qualitatively. The details of the devices used in the experiment are given in Table 5.
4.1. Oil mass flow rate measurements Fig. 4 shows the 2D-sketch of the setup and its components. The temperatures of the heater were controlled via PID feedback control system to maintain the oil viscosity in the operating range. Temperatures were measured with a data acquisition system via thermocouples located in the oil tank and compressor. To monitor the temperatures carefully, seven T-type thermocouples were used, three of them in the oil tank, other three in the oil sump and the last one in the cylinder head of the compressor. Measurements were started after the oil temperatures reached the operating conditions of the CIC. Crankshaft speed was adjusted by a signal generator.
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
229
were repeated at least 15 times. If the crankshaft speed of the compressor was higher than 2400 rpm, the run time was 1 minute and measurements were repeated 30 times. Kinematic viscosity of the oil was 5 cSt at the temperature of 40 ◦ C (Fuchs, 2018). Fig. 5(a) shows the repeatability of the measurements reflected on the standard deviation with 2σ using 30 measurements. The repeatability analysis indicated that the standard deviation in the oil mass flow rates was calculated σ = 0.034 g/s from the mean value of 1.69 g/s. The analysis showed that half of the total measurements were in the bandwidth of 1.67–1.69 g/s, thus the measurements were reliable at the standard operating conditions of the CIC. Fig. 5b shows the temperature fluctuations during sampling. It was seen that there were only negligible deviations in the temperatures changes. The only considerable fluctuation was observed for the thermocouple T2 which was located on the left side of the oil tank since the collected oil in the beaker was returned into left side of the oil tank for a new measurement. Fig. 6. Flow visualization test bench
At steady state conditions, the compressor operates for a minimum of ten minutes and oil was collected in a beaker. The run time for oil collection was about 2 min and the measurements
4.2. Flow visualizations Fig. 6 shows the experimental setup with flow visualization equipment, which consisted of one high-speed camera, two spotlights for illumination, PC for recording and storage. The acceleration of the crankshaft between the initial time t = 0 to 7.8 s
Fig. 7. Instantaneous snapshots of the CIC at 30 0 0 rpm, high speed camera at 10 0 0 fps.
230
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Fig. 8. Instantaneous isosurfaces of the oil volume fraction at 30 0 0 rpm from the top-view of the CIC.
Fig. 9. Oil mass flow rate variation during CIC start-up for 5 cSt oil.
was monitored with a high-speed camera (see Fig. 7). After the crankshaft speed increases to 750 rpm in 1 second, it reaches 30 0 0 rpm with an increase of 100 rpm per second as shown in Fig. 9. Although the oil climbing time is seen at 0.75 s, oil droplets
Fig. 10. Oil mass flow rate variation at various crankshaft speeds with SM method for 5 cSt oil.
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
appear in snapshots up to 6.35 s. Transition from oil droplets to an oil film is captured around 6.8 s, which is called, required time for sufficient lubrication. In this case, the crankshaft speed is around 1200 rpm and CFD analyses and measurements show that below this range it is unsafe to run the compressor. 5. CFD results 5.1. Instantaneous contours Fig. 8 shows the instantaneous isosurfaces of the oil volume fraction which clearly indicated the distribution of the lubricant inside the CIC at various time levels. The instantaneous isosurfaces of
231
the oil volume fraction were obtained from SM method since the sliding mesh enabled to track the crankshaft position. The CutCell mesh had 1.2 million elements and time step was 5 × 10−5 s. To show the flow contours of the spreading oil from the uppermost of the crankshaft more clearly, a plane at the top of the crankshaft outlet was created. Oil contours were illustrated as isosurfaces on the plane. The oil droplets at four different crankshaft positions in one period were visualized from CFD simulations and the steady state conditions for a developed oil film distribution at 30 0 0 rpm with infinite acceleration could be observed first at t = 0.4625 s. Using a time increment of t = 0.0125 s, the proceeding snapshots revealed the oil traveling inside the CIC and revealed how the
Fig. 11. Effect of viscosity on mass flow rates at various crankshaft speeds with SM: (a) 1600 rpm, (b) 3000 rpm, (c) 4500 rpm and with MRF (d) 4500 rpm.
232
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
reciprocating parts of the CIC were lubricated during the operation cycle. 5.2. Effect of Acceleration on Oil Mass Flow Rates To test the effect of the crankshaft acceleration on the oil mass flow rates during start-up, linear acceleration data was derived from the flow visualization snapshots and two linear functions were fitted to the visual data as shown in Fig. 9. These functions were implemented as motor start-up curves in the form of User Defined Function (UDF) into the CFD code and the time step size was set equal to 5x10−4 s to reduce computational time. The oil mass flow rates during the start-up of the crankshaft were evaluated with SM and MRF methods. In the SM method, the inlet was represented by a plane 25 mm above the crankshaft bottom hole, whereas the in the MRF method the inlet was represented by the crankshaft bottom hole itself. The first oil droplets were observed at t=3.8 s and 8 s in the MRF and SM methods, respectively. Fig. 9 clearly demonstrates the oil climbing time for both CFD methods and it also revealed the mass conservation was achieved between the crankshafts bottom and outlet for 10 seconds.
Fig. 12. Comparison of CFD results with experiments for 5 cSt oil.
5.3. Effect of Crankshaft Speed on Oil Mass Flow Rates The CFD calculations using SM method at various crankshaft speeds were summarized graphically in Fig. 10. The transient behavior of the oil mass flow rates and the climbing times with infinite acceleration could be understood well using the SM method. With increasing speed, mass flow rates increase as expected. No oil droplets spread out from the crankshaft outlet at 1200 rpm. The numerical output of the SM method was fluctuating due to the mathematical nature and mesh topology of the method (Fluent, 2013), thus time sampling was used in the quasi-periodic phase after 0.7 seconds. In the time sampling period, the oil mass flow rates were almost identical at 40 0 0 and 4500 rpm. The reason for this was thought to be the constructive constraints in the CIC design. The mass flow rates from CFD analysis were evaluated as averaged values over time to make a compatible comparison with measurements. 5.4. Effect of Viscosity on Oil Mass Flow Rates The forces acting on the fluid were centrifugal and viscous forces. The impact of the viscous forces became obvious with increased viscosity. The effect of the oil viscosity on mass flow rates was demonstrated in Fig. 11 at different crankshaft speeds. Increased viscosity causes more friction forces resulting in decreased mass flow rates. Depending on the crankshaft speeds, the oil climbing time changed, thus low crankshaft speed meant delayed oil climbing time. The mass flow rates overlapped for 3 and 5 cSt oils at high speeds such as 4500 rpm. In the SM method, sharp fluctuations were observed which was not the case for the MRF method at all speeds.
steeper compared to MRF method. In the range between 30 0 0 and 4500 rpm, SM predictions agreed better with the measurements than the MRF predictions. It should be noted that, the increase in the mass flow rates converged to a value indicating that, there was only a slight change after a critical crankshaft speed. It was shown that, after 3300 rpm, the discrepancies between the MRF results and the measurements were considerably high.
6. Conclusion In this study, the lubrication system of a CIC was investigated both numerically and experimentally. The CFD calculations have been performed for the immiscible two-phase flow under laminar flow conditions using various crankshaft speeds between 1200 and 4500 rpm for oils with kinematic viscosities of 3, 5, 10 and 15 cSt. Two numerical methods, namely MRF and SM were proposed in the prediction of the averaged mass oil flow rates from the outlet of the crankshaft. The formation of oil droplets and their development during the operation of the compressor were observed with a high-speed camera and the oil mass flow rate measurements were carried out on an in-house built setup to confirm the CFD results. The concluding remarks are as follows: •
•
5.5. Validation of CFD results with measurements
•
Fig. 12 shows the comparison of the predicted averaged mass flow rates at all crankshaft speeds with the measurements for 5 cSt oil at an immersion depth of 11.5 mm. As Fig. 12 indicates, in the range between 1200 and 2600 rpm, the measurements showed a monotone increase. In this range, all the CFD results were lower than the experimental results. Also, the MRF predictions were relatively closer to the measurements than the SM predictions. No oil was observed from the crankshaft outlet in the SM method at 1200 rpm and the rate of change of the oil mass flow rates was
•
•
•
With increasing crankshaft speed the oil mass flow rates increased. The linear acceleration of the crankshaft was compared versus infinite acceleration and the mass equilibrium was achieved after 10 seconds for both MRF and SM methods. The SM solutions were fluctuating during time sampling which was not the case in the MRF solutions. The measurements were performed in an environment subjected to the same boundary conditions as in simulations revealing that both CFD methods could have different tendencies at various crankshaft speeds. The numerical results obtained from MRF were closer to the measurements up to 2600 rpm. The SM method became favorable as a CFD method due to better agreement with measurements between 30 0 0 and 4500 rpm. The MRF method is cost-effective in terms of simulation time.
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Acknowledgements This study has been conducted within the 2+2 joint project with partners Istanbul Technical University, Arcelik A.S., TU Dresden Bitzer-Chair of Refrigeration, Cryogenics and Compressor Technology and Fuchs Schmierstoffe GmbH. The authors are indebted to the Scientific and Technological Research Council of Turkey (TUBITAK) for financial support through the project number 215M873. References Afshari, F., Comakli, O., Adiguzel, N., Zavaragh, H.G., 2017a. Influence of refrigerant properties and charge amount on performance of reciprocating compressor in air source heat pump. Journal of Energy Engineering 143 (1), 04016025. doi:10. 1061/(ASCE)EY.1943-7897.0 0 0 0377. Afshari, F., Comakli, O., Lesani, A., Karagoz, S., 2017b. Characterization of lubricating oil effects on the performance of reciprocating compressors in airwater heat pumps. International Journal of Refrigeration 74, 505-516. doi:10.1016/j.ijrefrig. 2016.11.017. Alves, M., Barbosa Jr., J., Prata, A., Ribas Jr., F., 2011. Fluid flow in a screw pump oil supply system for reciprocating compressors. International Journal of Refrigeration 34 (1), 74–83. doi:10.1016/j.ijrefrig.2010.08.003. Deng, L., Liang, S. X. J., Hu, Y., 2016. Cfd analysis and experiment study of the rotary two-stage inverter compressor with vapor injection. In: International Compressor Engineering Conference. Fluent, A., 2013. 15.0 Theory Guide. Fuchs, S., 2018. Refrigeration Oils 2018/2019. Gullberg, P., Sengupta, R., 2011. Axial fan performance predictions in cfd, comparison of mrf and sliding mesh with experiments. In: SAE 2011 World Congress & Exhibition. SAE International. doi:10.4271/2011-01-0652. Kerpicci, H., Yagci, A., Onbasioglu, S., 2013. Investigation of oil flow in a hermetic reciprocating compressor. International Journal of Refrigeration 36 (1), 215–221. doi:10.1016/j.ijrefrig.2012.09.002.
233
Kovacevic, A., Stosic, N., Smith, I., 2003. Three dimensional numerical analysis of screw compressor performance. Journal of Computational Methods in Sciences and Engineering 3 (2), 259–284. doi:10.3233/JCM- 2003- 3206. Lane, G., Schwarz, M., Evans, G., 20 0 0. Chapter 34 - comparison of cfd methods for modelling of stirred tanks. In: van den Akker, H., Derksen, J. (Eds.), 10th European Conference on Mixing. Elsevier Science, Amsterdam, pp. 273–280. doi:10.1016/B978-044450476- 0/50035- 2. Liu, Z., Hill, D. L., 20 0 0. Issues surrounding multiple frames of reference models for turbo compressor applications. In: International Compressor Engineering Conference. Navarro, E., Corbern, J., Martnez-Galvan, I., Gonzlvez, J., 2012. Oil sump temperature in hermetic compressors for heat pump applications. International Journal of Refrigeration 35 (2), 397–406. doi:10.1016/j.ijrefrig.2011.10.006. Ozsipahi, M., Cadirci, S., Gunes, H., Sarioglu, K., 2016. A comparison of an analytical and numerical flow model for the lubrication system of a variable speed hermetic compressor. Isi Bilimi Ve Teknigi Dergisi/ Journal of Thermal Science and Technology 36 (2), 41–49. Ozsipahi, M., Cadirci, S., Gunes, H., Sarioglu, K., Kerpicci, H., 2014. A numerical study on the lubrication system for a hermetic reciprocating compressor used in household refrigerators. International Journal of Refrigeration 48, 210–220. doi:10.1016/j.ijrefrig.2014.09.002. Ozsipahi, M., Gunes, H., Kerpicci, H., 2018. On the cfd modeling approach of lubrication system for a hermetic reciprocating compressor: Sliding mesh vs moving reference frame. In: International Compressor Engineering Conference. Pizarro-Recabarren, R.A., Barbosa, J.R., 2016. The effect of the lubricating oil on heat transfer in a hermetic reciprocating compressor. Journal of the Brazilian Society of Mechanical Sciences and Engineering 38 (1), 189–208. doi:10.1007/ s40430- 015- 0346- 0. Posch, S., Hopfgartner, J., Berger, E., Zuber, B., Schllauf, P., Almbauer, R., 2018. Numerical analysis of a hermetic reciprocating compressor oil pump system. International Journal of Refrigeration 85, 135–143. doi:10.1016/j.ijrefrig.2017.09.017. Singh, K., Mahajani, S., Shenoy, K., Patwardhan, A., Ghosh, S., 2007. Cfd modeling of pilot-scale pump-mixer: Single-phase head and power characteristics. Chemical Engineering Science 62 (5), 1308–1322. doi:10.1016/j.ces.2006.10.028. Tada, M., Hoffmann, T., Couto, P., A.L., M., Bortoli, M., 2014. Numerical and experimental examination for oil pump system using a simplified uncoupled simulation model. In: International Compressor Engineering Conference.
Contents lists available at ScienceDirect
International Journal of Refrigeration journal homepage: www.elsevier.com/locate/ijrefrig
Experimental and numerical investigation of lubrication system for reciprocating compressor Mustafa Ozsipahi a,∗, Haluk Anil Kose a, Sertac Cadirci a, Husnu Kerpicci b, Hasan Gunes a a b
Department of Mechanical Engineering, Istanbul Technical University (ITU), Gumussuyu, 34437 Istanbul, Turkey Arcelik Research and Development Center, Tuzla, 34950 Istanbul, Turkey
a r t i c l e
i n f o
Article history: Received 22 February 2019 Revised 16 August 2019 Accepted 18 August 2019 Available online 22 August 2019 Keywords: Reciprocating compressor Lubrication system Sliding mesh Moving reference frame Two-phase flow
a b s t r a c t Sufficient lubrication of the moving parts in a variable capacity inverter compressor is vital since it can directly affect the performance and expected lifetime. In this study, the lubrication system of a compact inverter compressor (CIC) is numerically and experimentally investigated. In the numerical modeling, a finite volume-based algorithm is used to model two-phase (air–oil) flow inside the compressor using Volume of Fluid Method (VoF) method. Transient behavior of the oil flow under laminar flow conditions is both simulated by imposing Sliding Mesh (SM) and the Moving Reference Frame (MRF) methods at various crankshaft speeds varying between 1200 and 4500 rpm. The measurements are taken using an experimental setup to compare/validate CFD results obtained from SM and MRF methods. Flow visualizations are performed with a high-speed camera to determine the oil climbing and required time for sufficient lubrication precisely. Moreover, the acceleration of the crankshaft is determined via high-speed camera and employed as a user-defined function to model the start-up period of the compressor. It is shown that with increasing crankshaft speeds, the average oil mass flow rate released from the upper part of the crankshaft is increasing almost linearly. It is also shown numerically that with increasing oil viscosity, the mass flow rate decreases. The comparison of experimental and CFD results reveals that the MRF solutions are in a better agreement with the measurements up to 2800 rpm. The SM method became favorable as a CFD method due to better agreement with measurements between 30 0 0 and 4500 rpm. © 2019 Elsevier Ltd and IIR. All rights reserved.
Étude expérimentale et numérique d’un système de lubrification pour compresseur à piston Mots-clés: Compresseur à piston; Système de lubrification ; Maillage glissant; Cadre de référence mouvant; Écoulement diphasique
1. Introduction The inverter-type compressors are variable capacity compressors since the cooling capacity requirement is adjusted with crankshaft speed. Since the cooling capacity is directly related to the crankshaft speed, the design of the oil management system which provides required amount of the lubricant to the journal bearings and other moving parts are crucial to warrant expected
∗
Corresponding author. E-mail addresses: [email protected] (M. Ozsipahi), [email protected] (H.A. Kose), [email protected] (S. Cadirci), [email protected] (H. Kerpicci), [email protected] (H. Gunes). https://doi.org/10.1016/j.ijrefrig.2019.08.026 0140-7007/© 2019 Elsevier Ltd and IIR. All rights reserved.
lifetime and performance of the compressor. In this context, some of the recent studies in the household refrigerator compressors focus on the improvement of the oil management system (Alves et al., 2011; Ozsipahi et al., 2014). In the last decade increasing usage of the inverter compressors urges manufacturers to modify existing oil management system in hermetic reciprocating compressors. Numerical calculations provide a chance to investigate oil management system design of inverter type reciprocating compressors in more detail. As rapid and costeffective numerical approaches open the way of comparison of the number of various design alternatives with a lesser prototype, peer investigations require higher computation time for the most realistic approach. CFD is a powerful alternative to experiments in the
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Nomenclature
α μ φ ρ σ A dt dx g m p S t u ug V CFD CIC COP cSt fps MRF PISO rpm SM UDF VoF
volume fraction (dimensionless) diffusion coefficient (dimensionless) dynamic viscosity (kg m−1 s−1 ) general scalar (dimensionless) density (kg m−3 ) standard deviation (dimensionless) area (m2 ) time step size (s) minimum cell size (m) gravity (m s−2 ) −1 mass flow rate (kg s ) −2 pressure (N m ) source term (kg m3 s−1 ) time (s) velocity (m s−1 ) mesh velocity (m s−1 ) volume (m3 ) computational fluid dynamics compact inverter compressor coefficient of performance centistokes frame per second moving reference frame pressure-implicit with splitting of operators rotations per minute sliding mesh user defined function volume of fluid
design phase of a product moreover the results of numerical calculations should be consistent with measurements. In the CFD simulations, MRF method is widely preferred to model the motion of the crankshaft that pumps oil from the sump to the moving components of the compressor (Alves et al., 2011; Ozsipahi et al., 2016; 2014; Posch et al., 2018). Important outputs of such CFD calculations are the climbing time of oil and the amount of oil spread from the upper part of the crankshaft at various speeds. Kerpicci et al. (2013) investigated oil flow in the hermetic reciprocating compressor and found oil climbing time by using a high-speed camera and CFD simulations. Ozsipahi et al. (2014) numerically investigated how the amount of oil mass flow rate in constant capacity hermetic compressors was affected by some operating conditions such as crankshaft speed, submersion depth of the crankshaft in the oil sump, oil viscosity and also the shape of the helical oil path on the crankshaft. They showed that the oil mass flow rate was directly proportional to the crankshaft speed, submersion depth and inversely proportional to the oil viscosity. In another study, Ozsipahi et al. (2016) suggested an analytical model for the screw pump used in the crankshaft of the hermetic variable capacity compressors to increase the oil mass flow rate at low speeds. The effect of the screw design on maximizing the flow rate was investigated in the analytical model and found to be consistent with the CFD simulations. Posch et al. (2018) analyzed the oil pump system of a hermetic reciprocating compressor numerically. They have split up the oil pump into its individual pumping parts and developed a tool that can be recommended for use in the design of reciprocating compressors in order to rapidly obtain simulation results. Tada et al. (2014) proposed an uncoupled numerical model for the variable capacity compressors which are employed in domestic applications and investigated the effects of the immersion depth of the crankshaft and the compressor speed both numerically and experimentally. Pizarro-Recabarren and
225
Barbosa (2016) proposed a differential model to calculate the oil film and compressor shell temperature distributions and verified it by experiments. The proposed model can be used to predict the total oil flow rate and the fraction of this flow rate flowing as a film on the shell. Afshari et al. (2017b) pointed out the significance of lubrication on the performance of reciprocating compressors used in air-water heat pumps. The measurements carried out with refrigerant mixtures and several oil samples showed that oil with lowest viscosity provides maximum COP value. It was concluded that oil with less viscosity was suggested for maximum exergetic efficiency in comparison to other samples. In another study, Afshari et al. (2017a) investigated how hydrofluorocarbon (HFC) refrigerant affects compressor operating conditions and system performance by experiments. Experiments have been carried out with refrigerant R404A and appropriate lubricants and tested the effect of ideal gas and real gas assumptions on the energetic analysis. The refrigerant amount was found to be very important in the COP, however the heat exchangers were also studied to increase system performance. Navarro et al. (2012) investigated the effect of oil sump temperature in hermetic compressors for heat pump applications and found relations between the oil sump temperature and several operating conditions such as condensation temperature or design parameters such as stroke and number of cylinders in reciprocating compressor, experimentally. In general, the oil sump temperature increases with increasing condensation temperature and piston stroke. In the CFD simulations where rotating parts are present, there are different techniques to be implemented to approach a more realistic solution. In the literature, dynamic meshes are intensively used in numerical investigation of the screw and rotary compressors. Kovacevic et al. (2003) presented the interface method for the SM method for the screw compressor. It is reported that a good agreement between the predicted and measured performance is a strong indication that CFD is a powerful tool for the design and optimization of the screw compressor. Liu and Hill (20 0 0) suggested Frozen Rotor model, Circumferential Average model, and the transient SM method. Depending on the problem definition such as the existence of non-axisymmetric components or the interaction between the rotating and stationary parts, these models can potentially give different results. Deng and Hu (2016) simulated a rotary two-stage inverter compressor with SM model and confirmed the CFD results with experimental data. Lane et al. (20 0 0) compared the SM and MRF methods for modeling of stirred tanks pointed out that both methods are in good agreement with the measurements and the MRF method is a practical alternative by providing faster solutions. Gullberg and Sengupta (2011) investigated axial fan performance by comparing MRF and SM methods with experiments. They reported that SM is far less sensitive to domain specifications compared to MRF method. Singh et al. (2007) simulated the flow in a continuous flow pump-mixer and compared the results with experimental data obtained on pilot-scale setup. It is noted in the study, in most cases, SM is found to perform better than MRF method but more validations are necessary to endorse the superiority of the modeling approach. Ozsipahi et al. (2018) compared the numerical performance of the SM versus the MRF method in modeling the lubrication system of a CIC. Oil climbing time with infinite acceleration, instantaneous flow fields of the two-phase flow and the mass flow rate predictions obtained by the two methods have been compared in the study. In the present study, CFD simulations of the lubrication system of a CIC are carried out by using the MRF and SM methods and compared to measurements obtained from an in-house built setup. The originality of the study is to see the effects of both numerical approaches on the oil mass flow rate and address the robustness of the numerical approaches by detailed experiments which is not found in the related literature survey. Moreover, the flow
226
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Fig. 1. (a) Photo of a CIC, (b) CAD model of the CIC and (c) Crankshaft of the compressor
Table 1 The details of the compressor model. Model Weight Motor Type Displacement Cooling Capacity COP Working Speeds Oil Charge Height of the crankshaft Dimensions (W × L × H)
Table 2 Mesh independence tests for SM method. VNTZ type 6 kg Brushless DC 11.28 cc 201 W at 3000 rpm 1.9 W/W at 3000 rpm (Standard ASHRAE condition) 1200–4500 rpm 210 cc 9.8 cm 14.5 × 15 × 13.5 cm
visualization test bench is used to validate oil climbing time and flow fields of the oil spreading out from the crankshaft outlet is demonstrated both by experiments and the SM method. The structure of the paper is organized as follows: The details of the handled problem are summarized in Section 2. The numerical models, governing equations and boundary conditions introduced in Section 3. Oil mass flow rate test bench and flow visualizations are addressed in Section 4. Then, the evaluation and comparison of the results are presented in Section 5. Finally, the main findings are summarized in Section 6.
Case ID
Averaged oil mass flow rate [g s−1 ]
CutCell, 790 K CutCell, 1.2 M CutCell, 1.8 M CutCell, 7.9 M Tetrahedral, 2.5 M Tetrahedral, 5 M Tetrahedral, 10 M
3.31 3.29 3.23 3.21 3.31 3.21 3.07
Relative error [%] 0.6 – −1.8 −2.4 0.6 −2.4 −6.7
ous kinematic viscosities depending on the operating conditions (GMBH, 2018). Lubrication oil is filled into the compressor in assembling process and the compressor is expected to self-lubricate during the operation of the refrigerator. The crankshaft which acts like an oil pump delivers oil from the sump to the rotating and moving parts of the compressor. Oil is sucked from the bottom of the crankshaft and transferred with the centrifugal forces caused by the rotation and the viscous effects. Oil climbs through the eccentric hole up to the first journal bearing. It further moves on the double-helical paths which join to each other before the main journal bearing. Finally, oil is spread out from the hole located at the uppermost of the crankshaft.
2. Problem definition In this study, the compressor of interest is a hermetic reciprocating compressor for household appliances which is shown in Fig. 1. The coefficient of the performance of the refrigeration units is directly influenced by the performance of the compressors. Environmental laws, legislation and regulations are forcing the manufacturer to improve their products. With improved manufacturing techniques and novel compression concepts, the variable capacity hermetic compressors are designed more compact for both reducing the production cost and increasing cooling domain in the refrigerator. Moreover, due to the changing energy regulations, the expectancy of the higher isentropic and volumetric efficiencies are still a challenging issue in the design of compact inverter compressors. Depending on the refrigeration capacity inverter type compressors operate at various rotational speeds. Oil management and lubrication mechanism in journal bearings may change drastically between compressor maximum and minimum speeds. Nevertheless, sufficient lubrication should be provided to the bearings and all moving parts to avoid any mechanical damage on the compressor. The technical specifications and dimensions of the compressor used in this study are shown in Table 1. The lubricants which are widely used in hermetic reciprocating compressors are synthetic and mineral based oils with vari-
3. CFD modelling 3.1. Sliding mesh The SM method is a special type of dynamic mesh motion which boundaries and mesh cells move together in a rigid-body motion. As a consequence, SM method is inherently unsteady due to the motion of the mesh with time. The cells in the zones are not deforming and the governing equations of the fluid motion are different than used in MRF method. SM method is defined as the most accurate method for simulating flows in multiple reference frame, but also the most computationally demanding (Fluent, 2013). In the first method of CFD calculations, the numerical domain is simplified. The interface encapsulating the crankshaft is introduced in the numerical model as shown in Fig. 2. Since SM method provides the opportunity to track the motion of the crankshaft in temporal domain, it is possible to obtain instantaneous oil iso-surfaces. The conservation equation in general integral form for any scalar used in the SM model is given in the following equation:
d dt
V
ρφ dV +
∂V
ρφ (u − u g ).dA =
∂V
∇ φ .dA +
V
Sφ dV
(1)
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
227
Table 3 Mesh independence tests for MRF method. Case ID
Averaged oil mass flow rate [g s−1 ]
CutCell, 986 K CutCell, 1.3 M CutCell, 2.7 M
1.61 1.32 1.3
Relative error [%] 22 – −1.5
3.3. Governing equations and boundary conditions
Fig. 2. Simplified domain used in SM method: red part: crankshaft, grey part on the top and bottom of the crankshaft: interface.
In the numerical modeling, the VoF model with implicit scheme was used since there is an interface between the free surface of oil and the refrigerant (in the present study air). Air was preferred in CFD calculations because the experiments were carried out at atmospheric conditions without refrigerant. For the twophase model, the continuity and momentum equations are given in Eqs. (4) and (5), respectively where α represents the volume fraction (Fluent, 2013).
∂ (α ρ ) + ∇ (αl ρl u l ) = 0 ∂t l l ρ
∂ u u ) = ρ g − ∇ p + ∇ μ ∇ u + ∇u T + ∇ (u ∂t
(4)
(5)
The boundary conditions for the MRF method were defined as follows: •
• •
•
Same pressure was imposed at the top of the free surface in the oil sump and crankshaft outlet. Angular velocity was defined on the walls of the crankshaft. All remaining walls and interfaces between domains were stationary. The thermophysical properties of all fluids were constant.
Fig. 3. Computational domain used in MRF method.
Since the time rate of change of the cell volume is zero, V n+1 = V n is valid for the SM model. The first order temporal derivative in Eq. (1), can be rewritten using first-order forward difference formula as given in the following equation:
d dt
(ρφ )n+1 − (ρφ )n V ρφ dV = t V
(2)
In order to satisfy mass conversation, the time derivative of the control volume can be formulated as given in the following equation: nf
dV j = 0 g, j .A = u dt
(3)
j
3.2. Moving Reference Frame In the MRF method, the relative motion of a moving zone with respect to adjacent zones is not taken into account, in other words, the mesh is fixed. This means freezing the motion of the crankshaft of a hermetic compressor in a fixed position and computing the instantaneous flow field with the crankshaft in the corresponding position. In many engineering applications, where a rotating part is present in the flow domain, MRF is widely preferred as the numerical approach. Some examples for this include turbomachinery applications with weak rotor-stator interaction, wind turbines and mixing tanks. Fig. 3 shows the fixed domain used in MRF where the red part represents the crankshaft, the grey part is the outer wall around the crankshaft and the blue part is the oil sump.
In the SM method, the only moving part was the crankshaft and the submersion depth of the crankshaft in the oil sump was defined similarly to the MRF method. The fluid flow in the lubrication system was solved numerically using the finite-volume based ANSYS-Fluent v16 package. PISO algorithm was used for pressure– velocity coupling in the simulations. 3.4. Mesh and time dependency tests Intense mesh independence tests have been carried out for both numerical methods. A reference case at certain operating conditions was selected to obtain the required mesh convergence results. The reference case was chosen at 30 0 0 rpm with a 15 mm immersion depth of using a 5 cSt oil kinematic viscosity. Unsteady calculations were performed with PISO algorithm for pressurevelocity coupling solving second order accurate momentum equations. The time step size was dt = 5 × 10−4 s. Table 2 shows the mesh independence tests for CutCell and tetrahedral meshes. Table 3 indicates the mesh independency tests for MRF method. Relative errors in the averaged oil mass flow rates with respect to the solution from 1.3 million mesh were calculated for the determination of the resulting mesh for that method. Table 4 shows the time sensitivity analysis which were performed for the SM method using 1.2 million number of elements in the mesh. In this analysis the relative errors were calculated with respect to the following time step size, as a result time step size was chosen 5 × 10−5 seconds. Parallel computations have been performed on a 28-core processor computer and average simulation can have a wall-clock time approximately 20 and 60 h for the MRF and SM method, respectively.
228
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Fig. 4. 2D- sketch of the experimental setup (the figure is not proportional).
Fig. 5. (a) Standard deviation of the oil mass flow rates at 30 0 0 rpm, 5 cSt; (b) temperature fluctuations during measurement.
Table 4 Time independence tests.
4. Experimental studies
Time step size [s]
Averaged oil mass flow rate [g s−1 ]
0.002 0.001 5 × 10−4 2.5 × 10−4 1 × 10−4 5 × 10−5 2.5 × 10−5
3.70 3.55 3.29 2.45 1.85 1.61 1.55
Relative error [%] 138.9 124.2 112.6 58.2 19.3 3.8 –
Table 5 The list of the instrumentation. Data acquisition Thermocouples High speed camera Spotlights PID controller Power controller Heater
Keysight 34970A T type (%0.4 accuracy) Photron 512 PCI 32K dedocool COOLH Autonics TX4S Autonics SPC1-35 Sheet type adhesive flexible resistance (1200 W)
An experimental setup was built for oil mass flow rate measurements and flow visualizations to validate the CFD results both quantitatively and qualitatively. The details of the devices used in the experiment are given in Table 5.
4.1. Oil mass flow rate measurements Fig. 4 shows the 2D-sketch of the setup and its components. The temperatures of the heater were controlled via PID feedback control system to maintain the oil viscosity in the operating range. Temperatures were measured with a data acquisition system via thermocouples located in the oil tank and compressor. To monitor the temperatures carefully, seven T-type thermocouples were used, three of them in the oil tank, other three in the oil sump and the last one in the cylinder head of the compressor. Measurements were started after the oil temperatures reached the operating conditions of the CIC. Crankshaft speed was adjusted by a signal generator.
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
229
were repeated at least 15 times. If the crankshaft speed of the compressor was higher than 2400 rpm, the run time was 1 minute and measurements were repeated 30 times. Kinematic viscosity of the oil was 5 cSt at the temperature of 40 ◦ C (Fuchs, 2018). Fig. 5(a) shows the repeatability of the measurements reflected on the standard deviation with 2σ using 30 measurements. The repeatability analysis indicated that the standard deviation in the oil mass flow rates was calculated σ = 0.034 g/s from the mean value of 1.69 g/s. The analysis showed that half of the total measurements were in the bandwidth of 1.67–1.69 g/s, thus the measurements were reliable at the standard operating conditions of the CIC. Fig. 5b shows the temperature fluctuations during sampling. It was seen that there were only negligible deviations in the temperatures changes. The only considerable fluctuation was observed for the thermocouple T2 which was located on the left side of the oil tank since the collected oil in the beaker was returned into left side of the oil tank for a new measurement. Fig. 6. Flow visualization test bench
At steady state conditions, the compressor operates for a minimum of ten minutes and oil was collected in a beaker. The run time for oil collection was about 2 min and the measurements
4.2. Flow visualizations Fig. 6 shows the experimental setup with flow visualization equipment, which consisted of one high-speed camera, two spotlights for illumination, PC for recording and storage. The acceleration of the crankshaft between the initial time t = 0 to 7.8 s
Fig. 7. Instantaneous snapshots of the CIC at 30 0 0 rpm, high speed camera at 10 0 0 fps.
230
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Fig. 8. Instantaneous isosurfaces of the oil volume fraction at 30 0 0 rpm from the top-view of the CIC.
Fig. 9. Oil mass flow rate variation during CIC start-up for 5 cSt oil.
was monitored with a high-speed camera (see Fig. 7). After the crankshaft speed increases to 750 rpm in 1 second, it reaches 30 0 0 rpm with an increase of 100 rpm per second as shown in Fig. 9. Although the oil climbing time is seen at 0.75 s, oil droplets
Fig. 10. Oil mass flow rate variation at various crankshaft speeds with SM method for 5 cSt oil.
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
appear in snapshots up to 6.35 s. Transition from oil droplets to an oil film is captured around 6.8 s, which is called, required time for sufficient lubrication. In this case, the crankshaft speed is around 1200 rpm and CFD analyses and measurements show that below this range it is unsafe to run the compressor. 5. CFD results 5.1. Instantaneous contours Fig. 8 shows the instantaneous isosurfaces of the oil volume fraction which clearly indicated the distribution of the lubricant inside the CIC at various time levels. The instantaneous isosurfaces of
231
the oil volume fraction were obtained from SM method since the sliding mesh enabled to track the crankshaft position. The CutCell mesh had 1.2 million elements and time step was 5 × 10−5 s. To show the flow contours of the spreading oil from the uppermost of the crankshaft more clearly, a plane at the top of the crankshaft outlet was created. Oil contours were illustrated as isosurfaces on the plane. The oil droplets at four different crankshaft positions in one period were visualized from CFD simulations and the steady state conditions for a developed oil film distribution at 30 0 0 rpm with infinite acceleration could be observed first at t = 0.4625 s. Using a time increment of t = 0.0125 s, the proceeding snapshots revealed the oil traveling inside the CIC and revealed how the
Fig. 11. Effect of viscosity on mass flow rates at various crankshaft speeds with SM: (a) 1600 rpm, (b) 3000 rpm, (c) 4500 rpm and with MRF (d) 4500 rpm.
232
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
reciprocating parts of the CIC were lubricated during the operation cycle. 5.2. Effect of Acceleration on Oil Mass Flow Rates To test the effect of the crankshaft acceleration on the oil mass flow rates during start-up, linear acceleration data was derived from the flow visualization snapshots and two linear functions were fitted to the visual data as shown in Fig. 9. These functions were implemented as motor start-up curves in the form of User Defined Function (UDF) into the CFD code and the time step size was set equal to 5x10−4 s to reduce computational time. The oil mass flow rates during the start-up of the crankshaft were evaluated with SM and MRF methods. In the SM method, the inlet was represented by a plane 25 mm above the crankshaft bottom hole, whereas the in the MRF method the inlet was represented by the crankshaft bottom hole itself. The first oil droplets were observed at t=3.8 s and 8 s in the MRF and SM methods, respectively. Fig. 9 clearly demonstrates the oil climbing time for both CFD methods and it also revealed the mass conservation was achieved between the crankshafts bottom and outlet for 10 seconds.
Fig. 12. Comparison of CFD results with experiments for 5 cSt oil.
5.3. Effect of Crankshaft Speed on Oil Mass Flow Rates The CFD calculations using SM method at various crankshaft speeds were summarized graphically in Fig. 10. The transient behavior of the oil mass flow rates and the climbing times with infinite acceleration could be understood well using the SM method. With increasing speed, mass flow rates increase as expected. No oil droplets spread out from the crankshaft outlet at 1200 rpm. The numerical output of the SM method was fluctuating due to the mathematical nature and mesh topology of the method (Fluent, 2013), thus time sampling was used in the quasi-periodic phase after 0.7 seconds. In the time sampling period, the oil mass flow rates were almost identical at 40 0 0 and 4500 rpm. The reason for this was thought to be the constructive constraints in the CIC design. The mass flow rates from CFD analysis were evaluated as averaged values over time to make a compatible comparison with measurements. 5.4. Effect of Viscosity on Oil Mass Flow Rates The forces acting on the fluid were centrifugal and viscous forces. The impact of the viscous forces became obvious with increased viscosity. The effect of the oil viscosity on mass flow rates was demonstrated in Fig. 11 at different crankshaft speeds. Increased viscosity causes more friction forces resulting in decreased mass flow rates. Depending on the crankshaft speeds, the oil climbing time changed, thus low crankshaft speed meant delayed oil climbing time. The mass flow rates overlapped for 3 and 5 cSt oils at high speeds such as 4500 rpm. In the SM method, sharp fluctuations were observed which was not the case for the MRF method at all speeds.
steeper compared to MRF method. In the range between 30 0 0 and 4500 rpm, SM predictions agreed better with the measurements than the MRF predictions. It should be noted that, the increase in the mass flow rates converged to a value indicating that, there was only a slight change after a critical crankshaft speed. It was shown that, after 3300 rpm, the discrepancies between the MRF results and the measurements were considerably high.
6. Conclusion In this study, the lubrication system of a CIC was investigated both numerically and experimentally. The CFD calculations have been performed for the immiscible two-phase flow under laminar flow conditions using various crankshaft speeds between 1200 and 4500 rpm for oils with kinematic viscosities of 3, 5, 10 and 15 cSt. Two numerical methods, namely MRF and SM were proposed in the prediction of the averaged mass oil flow rates from the outlet of the crankshaft. The formation of oil droplets and their development during the operation of the compressor were observed with a high-speed camera and the oil mass flow rate measurements were carried out on an in-house built setup to confirm the CFD results. The concluding remarks are as follows: •
•
5.5. Validation of CFD results with measurements
•
Fig. 12 shows the comparison of the predicted averaged mass flow rates at all crankshaft speeds with the measurements for 5 cSt oil at an immersion depth of 11.5 mm. As Fig. 12 indicates, in the range between 1200 and 2600 rpm, the measurements showed a monotone increase. In this range, all the CFD results were lower than the experimental results. Also, the MRF predictions were relatively closer to the measurements than the SM predictions. No oil was observed from the crankshaft outlet in the SM method at 1200 rpm and the rate of change of the oil mass flow rates was
•
•
•
With increasing crankshaft speed the oil mass flow rates increased. The linear acceleration of the crankshaft was compared versus infinite acceleration and the mass equilibrium was achieved after 10 seconds for both MRF and SM methods. The SM solutions were fluctuating during time sampling which was not the case in the MRF solutions. The measurements were performed in an environment subjected to the same boundary conditions as in simulations revealing that both CFD methods could have different tendencies at various crankshaft speeds. The numerical results obtained from MRF were closer to the measurements up to 2600 rpm. The SM method became favorable as a CFD method due to better agreement with measurements between 30 0 0 and 4500 rpm. The MRF method is cost-effective in terms of simulation time.
M. Ozsipahi, H.A. Kose and S. Cadirci et al. / International Journal of Refrigeration 108 (2019) 224–233
Acknowledgements This study has been conducted within the 2+2 joint project with partners Istanbul Technical University, Arcelik A.S., TU Dresden Bitzer-Chair of Refrigeration, Cryogenics and Compressor Technology and Fuchs Schmierstoffe GmbH. The authors are indebted to the Scientific and Technological Research Council of Turkey (TUBITAK) for financial support through the project number 215M873. References Afshari, F., Comakli, O., Adiguzel, N., Zavaragh, H.G., 2017a. Influence of refrigerant properties and charge amount on performance of reciprocating compressor in air source heat pump. Journal of Energy Engineering 143 (1), 04016025. doi:10. 1061/(ASCE)EY.1943-7897.0 0 0 0377. Afshari, F., Comakli, O., Lesani, A., Karagoz, S., 2017b. Characterization of lubricating oil effects on the performance of reciprocating compressors in airwater heat pumps. International Journal of Refrigeration 74, 505-516. doi:10.1016/j.ijrefrig. 2016.11.017. Alves, M., Barbosa Jr., J., Prata, A., Ribas Jr., F., 2011. Fluid flow in a screw pump oil supply system for reciprocating compressors. International Journal of Refrigeration 34 (1), 74–83. doi:10.1016/j.ijrefrig.2010.08.003. Deng, L., Liang, S. X. J., Hu, Y., 2016. Cfd analysis and experiment study of the rotary two-stage inverter compressor with vapor injection. In: International Compressor Engineering Conference. Fluent, A., 2013. 15.0 Theory Guide. Fuchs, S., 2018. Refrigeration Oils 2018/2019. Gullberg, P., Sengupta, R., 2011. Axial fan performance predictions in cfd, comparison of mrf and sliding mesh with experiments. In: SAE 2011 World Congress & Exhibition. SAE International. doi:10.4271/2011-01-0652. Kerpicci, H., Yagci, A., Onbasioglu, S., 2013. Investigation of oil flow in a hermetic reciprocating compressor. International Journal of Refrigeration 36 (1), 215–221. doi:10.1016/j.ijrefrig.2012.09.002.
233
Kovacevic, A., Stosic, N., Smith, I., 2003. Three dimensional numerical analysis of screw compressor performance. Journal of Computational Methods in Sciences and Engineering 3 (2), 259–284. doi:10.3233/JCM- 2003- 3206. Lane, G., Schwarz, M., Evans, G., 20 0 0. Chapter 34 - comparison of cfd methods for modelling of stirred tanks. In: van den Akker, H., Derksen, J. (Eds.), 10th European Conference on Mixing. Elsevier Science, Amsterdam, pp. 273–280. doi:10.1016/B978-044450476- 0/50035- 2. Liu, Z., Hill, D. L., 20 0 0. Issues surrounding multiple frames of reference models for turbo compressor applications. In: International Compressor Engineering Conference. Navarro, E., Corbern, J., Martnez-Galvan, I., Gonzlvez, J., 2012. Oil sump temperature in hermetic compressors for heat pump applications. International Journal of Refrigeration 35 (2), 397–406. doi:10.1016/j.ijrefrig.2011.10.006. Ozsipahi, M., Cadirci, S., Gunes, H., Sarioglu, K., 2016. A comparison of an analytical and numerical flow model for the lubrication system of a variable speed hermetic compressor. Isi Bilimi Ve Teknigi Dergisi/ Journal of Thermal Science and Technology 36 (2), 41–49. Ozsipahi, M., Cadirci, S., Gunes, H., Sarioglu, K., Kerpicci, H., 2014. A numerical study on the lubrication system for a hermetic reciprocating compressor used in household refrigerators. International Journal of Refrigeration 48, 210–220. doi:10.1016/j.ijrefrig.2014.09.002. Ozsipahi, M., Gunes, H., Kerpicci, H., 2018. On the cfd modeling approach of lubrication system for a hermetic reciprocating compressor: Sliding mesh vs moving reference frame. In: International Compressor Engineering Conference. Pizarro-Recabarren, R.A., Barbosa, J.R., 2016. The effect of the lubricating oil on heat transfer in a hermetic reciprocating compressor. Journal of the Brazilian Society of Mechanical Sciences and Engineering 38 (1), 189–208. doi:10.1007/ s40430- 015- 0346- 0. Posch, S., Hopfgartner, J., Berger, E., Zuber, B., Schllauf, P., Almbauer, R., 2018. Numerical analysis of a hermetic reciprocating compressor oil pump system. International Journal of Refrigeration 85, 135–143. doi:10.1016/j.ijrefrig.2017.09.017. Singh, K., Mahajani, S., Shenoy, K., Patwardhan, A., Ghosh, S., 2007. Cfd modeling of pilot-scale pump-mixer: Single-phase head and power characteristics. Chemical Engineering Science 62 (5), 1308–1322. doi:10.1016/j.ces.2006.10.028. Tada, M., Hoffmann, T., Couto, P., A.L., M., Bortoli, M., 2014. Numerical and experimental examination for oil pump system using a simplified uncoupled simulation model. In: International Compressor Engineering Conference.