Mechanical Engineering Journal

Bulletin of the JSME Mechanical Engineering Journal Vol.3, No.6, 2016 Experimental study on the behavior of the two phase flow shock waves occurring in the ejector refrigeration cycle Haruyuki NISHIJIMA*, Kyohei TSUCHII* and Masafumi NAKAGAWA* * Toyohashi University of Technology 1-1 Hibarigaoka, Tempaku-cho, Toyohashi, Aichi, 441-8580, Japan E-mail: nishijima@nak.me.tut.ac.jp Received 20 April 2016 Abstract Conservation of energy is becoming increasingly important for the protection of the environment. Improving the efficiency of a refrigeration cycle is a critical factor to achieve this goal. Recently, an ejector system was developed that reduces the energy requirements of the compressor in the refrigeration cycle. Two-phase-flow shock waves appear in the ejector under certain operating conditions and increase the pressure difference between suction inlet and outlet. Such shock waves play an important role in the ejector s compression mechanism and thus merit a thorough investigation. In this work, we visualize the structure of a two-phase-flow shock wave in an ejector nozzle using a high-speed camera to monitor an optical beam transmitted through a refrigerant (hot water). As the pressure rises in the ejector outlet, the shock wave moves from the outlet to the nozzle throat and changes from an oblique shock wave to a normal shock wave. The shape of the output nozzle may modify the structure of the shock waves. Key words : Multiphase flow, Supersonic flow, Shock waves, Visualization, Ejector 1 Introduction As the problem of global warming becomes increasingly severe, energy conservation has become a global challenge. Improving the efficiency of a refrigeration cycle is a critical factor to achieve this goal. Thus, the compressor used in refrigeration and air-conditioning equipment must be optimized to maximize the efficiency of the refrigeration cycle. Recently, the use of a two-phase flow ejector to obtain a highly efficient refrigeration cycle has drawn increasing attention. A conventional vapor-compression refrigeration cycle loses energy during decompression expansion in the expansion valve, as shown in Fig. 1. This loss is due to kinetic-energy consumption by swirl flow in the expansion valve. Figure 2 summarizes the ejector refrigeration cycle, wherein the expansion energy drives the ejector. Recovering this expansion energy increases the compressor suction pressure. Accordingly, the ejector cycle can reduce compressor energy consumption by recovering the expansion energy lost in the conventional vapor-compression refrigeration cycle (Takeuchi et al., 2004). Although the ejector refrigeration cycle based on the two-phase flow ejector has been commercialized by Takeuchi et al. (2003), further improvements in the ejector efficiency are needed to meet the current demands for energy conservation. The two-phase flow ejector comprises a nozzle, a suction nozzle, a mixing section, and a diffuser, as shown in Fig. 3. The refrigerant accelerates to supersonic speeds in the nozzle and slows down to subsonic speeds in the mixing section and diffuser. As the speed of sound in two-phase flow is lower than that in gas, a two-phase-flow shock wave easily forms in the nozzle and mixing section. Such shock waves increase the pressure difference between the suction inlet and the ejector outlet and play an important role in the ejector s compression mechanism. Thus, the characteristics of these shock waves merit investigation. In previous work, we found that pseudo shock waves (dispersive shock waves) occur when the refrigerant has a large (small) density ratio for the vapor of the liquid and explained these shock waves based on a theory involving momentum relaxation (Nakagawa and Harada, 2010, Harada and Nakagawa 2011). In addition, by monitoring an optical beam Paper No.16-00255 J-STAGE Advance Publication date: 9 December, 2016 1

reflected in the nozzle-outlet expansion chamber, we found that an oblique shock wave forms when there is no upward pressure (Nakagawa et al., 2005). Furthermore, the position of the shock wave in the nozzle depends on the outlet pressure of the nozzle and the nozzle-inlet quality. However, the structure of the two-phase-flow shock wave that occurs in the ejector remains to be confirmed (Yamanaka and Nakagawa, 2012). The structure of these shock waves defines the complex flow in and the performance of the two-phase flow ejector. In general, the energy lost by the shock wave depends on its structure: the energy lost by a normal shock wave is greater than that lost by an oblique shock wave. Therefore, to maximize ejector efficiency, we must understand the structure of two-phase-flow shock waves in the ejector. Experimental visualization of flow is an important method to study shock wave structure in a two-phase flow ejector. Fabri and Siestrunk (1958) used Schlieren methods to visualize the different flow patterns in a supersonic air ejector. Matsuo et al. (1981) also used such methods to analyze the performance of a supersonic air ejector. Dvorak and Safarik (2005) used the Schlieren technique to study the transonic instability in the mixing-section inlet of a high-speed ejector. Bouhanguel et al. (2011) used the laser-sheet flow-imaging technique to investigate flow in the ejector. Most of these visualization techniques are intended for supersonic, high-speed air or vapor flow. A current study (unpublished) clarifies the structure of the shock wave in a two-phase flow ejector that occurs when a vapor and fine droplets flow through the ejector at supersonic speed. The present study expands on the knowledge gained from our previous studies (Yamanaka and Nakagawa, 2012). We use hot water as refrigerant and, to confirm the shock-wave structure, visualize two-phase-flow shock waves in the ejector nozzle by monitoring a transmitted optical beam. Fig.1 (a) A conventional vapor-compression refrigeration cycle consists of compressor, condenser, expansion valve, and evaporator. (b) A P-h diagram of a conventional refrigeration cycle. The red dot-dashed line is the isentropic curve. A conventional refrigeration cycle loses kinetic energy because of swirl flow during decompression expansion in the expansion valve. Fig.2 (a) Schematic illustration of ejector refrigeration cycle. (b) P-h diagram of ejector refrigeration cycle. In this cycle, the expansion energy drives the ejector. Recovering this expansion energy via the pressure rise in the ejector therefore leads to an increase in compressor-suction pressure. Accordingly, the ejector cycle requires less compressor energy by recovering the expansion energy lost in the conventional vapor-compression refrigeration cycle. 2

Fig.3 The two-phase flow ejector is composed of a nozzle, a suction nozzle, a mixing section, and a diffuser. The refrigerant flowing through the ejector is accelerated to supersonic, which is more than the two-phase sound speed in the nozzle and slows down to subsonic in the mixing section and the diffuser. Therefore, the twophase shock wave forms easily in the nozzle and the mixing section. 2 Experimental study step The experiment involved three main steps shown in Fig.4.: In the first step, the flow in the ejector has to increase to supersonic speeds to create a two-phase-flow shock wave. By measuring the flow rate through the ejector nozzle under various conditions, we determine the critical conditions required to reach supersonic flow in the nozzle. In step 2, under the critical conditions, we vary the nozzle-inlet and ejector-outlet pressure and measure a pressure distribution in the ejector. Finally, in step 3, the flow is visualized by imaging a transmitted optical beam, which allows us to study the structure of the two-phase-flow shock wave in the ejector nozzle. Fig.4 Experimental study step. 3 Experiment setup 3.1 The summary of the experiment setup Figure 5 shows the experiment setup, which consists of a high-pressure tank, heater, mixer, two-phase flow ejector, condenser, flow meter, a metal halide light, and a high-speed camera. When the temperature inside the tank reaches the steady state at 152 C, saturated steam (liquid) flows out of the upper (lower) side of the tank. The steam mass-flow rate G G is measured with a differential-pressure flow meter, and the mass-flow rate G of the ejector outlet is measured by the Coriolis flow meter downstream from the condenser. The liquid mass-flow rate G L is the difference: G G G. The nozzleinlet quality X n of the ejector is given by X n = (h h G ) (h G h L ) (1) where h = (G G h G + G L h L ) (G G + G L ) (2) is the mean mass enthalpy at the ejector inlet and h L and h G are the enthalpy of liquid and steam, respectively. The 23

pressure and temperature in the liquid and gas lines are measured upstream of the mixer, and then X n is calculated by using REFPROP 8.0 (NIST, 2007). By using the two mass-flow adjustment valves to control the two flow rates (i.e., liquid and gas), we vary the nozzle-inlet quality X n of the ejector. The ejector backpressure P b is measured just outside the outlet tube of the ejector and is adjusted by the valve downstream of the condenser. 3.2 Experimental Ejector Fig. 5 Schematic diagram of experiment setup. In this study, we use a two-dimensional ejector (see Fig. 6) to facilitate visualization via a transmitted optical beam. The experimental ejector comprises an ejector plate, an upper plate, and a lower plate. To measure the pressure at the upper plate, three pressure taps are inserted into the convergent section of the nozzle, four taps are inserted into the divergent section of nozzle, three taps are inserted into the mixing section, and one tap is inserted into the diffuser on the side wall of the ejector. By using either polyetheretherketone or polycarbonate for the upper and lower plates, we can simultaneously measure the pressure in the ejector and visualize the flow through the ejector. The ejector plate is made of 2.0-mm-thick stainless steel. Table 1 gives the specification of the experimental ejector. The basic design of the ejector is given in Yamanaka (2013), and this same design (i.e., not optimized) is used for the present work. The divergent section of the nozzle is about 11 mm long. Fig.6 Schematic diagram of experimental ejector. 24

Table.1 Specifications of experimental ejector. 3.3 Visualization Method Our conventional study (Yamanaka and Nakagawa, 2012) used a digital camera to monitor an optical beam reflected within the experimental nozzle. In this study, however, we visualize the two-phase-flow shock wave by using a highspeed camera to monitor an optical beam (produced by a metal halide light) transmitted through the experimental ejector (see Fig. 7). Table 2 lists the specifications of the visualization devices used in this work. Fig.7 Schematic diagram showing visualization technique. Table.2 Visualization devices. Figure 8 shows how we interpret this visualization. First, we divide the visualization domain into two zones. The flow velocity decreases downstream of the shock wave, so the number density of the droplets should increase in this zone. As shown in Table 3, the refractive index of the saturated steam is almost unity, whereas that of the saturated water is slightly greater than 1.3. Thus, the transmitted light is refracted and reflected out of the transmitted beam by the droplet, 25

leading to less light irradiating the high-speed camera downstream of the shock wave compared with upstream. This allows us to visualize two-phase-flow shock waves by monitoring the transmitted beam. Fig.8 Interpretation of image formed by transmitted optical beam. Table.3 Refractive index of refrigerant. 3.4 Experimental condition Table 4 lists the experimental conditions for this study. The nozzle-inlet temperature is 150 C, which allows twophase flow in the nozzle to accelerate to supersonic speeds for all experimental outlet conditions. In addition, the suction flow rate G s at the ejector is zero to allow us to focus on the shock wave in the nozzle. Under these conditions, the pressure and the enthalpy in the ejector follow the diagram shown in Fig. 9. Because no suction flow occurs, the pressure and enthalpy of the flow at the ejector outlet undergo simple changes: decreasing until the shock wave occurs and increasing afterwards. Figure 10 shows the results of calculations based on the homogeneous equilibrium model (HEM) (Japanese Institution of Mechanical Engineers, 2006), which show the characteristics of flow in the nozzle under these experimental conditions. Initially, the equilibrium speed of sound for the two-phase flow decreases gently with decreasing inlet pressure because the void fraction increases according to the decompression in the nozzle. For a nozzle-inlet quality of 0.3, which corresponds to supersonic flow at about 0.28 MPa, the quality increases gently as the nozzle-inlet pressure drops, and the void fraction increases to ~0.9993. The critical pressure required to reach the speed of sound decreases upon increasing the nozzle-inlet quality from 0.3 to 0.6. Thus, because of decompression in the nozzle inlet, the quality 26

decreases gently. Based on these calculations, we conclude that a two-phase-flow shock wave with these phase conversion can be triggered in our experiment by varying the nozzle-inlet quality. Table.4 Experimental conditions. Fig.9 Pressure and specific enthalpy under given experimental conditions. Fig.10 Characteristics of nozzle flow under given experimental conditions, as calculated by the HEM. 27

4 Results of Experiment 4.1 Flow-Rate Results In two-phase flow, the speed of sound is not easy to estimate because of irreversible processes caused by interphase heat-, mass-, and momentum-transfer phenomena. Because the critical condition cannot be determined based on the speed of sound, we measure the two-phase-flow rates upon varying outlet pressure. This approach clarifies when real choking occurs in the nozzle. To visualize the shock wave, the nozzle flow must be supersonic, so determining when the nozzle flow becomes supersonic is important. Figure 11 shows the nozzle flow rate as measured by the Coriolis mass-flow meter (left side, Fig. 5). Under the critical conditions, the flow rate of the two-phase supersonic nozzle is relatively constant with respect to the upward pressure of the ejector outlet. Therefore, the range indicated by the double dotted line in Fig. 11 gives the range of ejector back pressures corresponding to critical flow in the nozzle. To measure the static pressure and visualize the flow, we set the range of the ejector-outlet pressure to 0.12~0.26 MPa in this experiment. Fig.11 Measured flow rate as a function of ejector back pressure. 4.2 Pressure results Figure 12 shows the static pressure measured at the ejector wall surface under the conditions discussed above as a function of distance from nozzle throat and for the nozzle-inlet qualities X n = 0.3 and 0.6. The solid squares, triangles, and diamonds give the experimental results for P b = 0.26, 0.19, and 0.12 MPa, respectively, These dashed curves are adiabatic expansion curves calculated by the HEM. For both experiment and calculations, the red and blue colors give the results for X n = 0.3 and 0.6, respectively. To calculate the refrigerant properties, we use REFPROT 8.0 (NIST, 2008). As shown in Fig. 12, the slope of the experimental pressure vs distance from the throat of the ejector is less than that of the theoretical curves, which we attribute to the large interphase irreversible process of two-phase flow. The fact that two-phase flow occurs under supercritical conditions is mentioned in section 4.1, and is also confirmed by the decrease in pressure in the divergent section of Fig. 12 (i.e., 0 to 11.07 mm from ejector throat). Furthermore, the increase in ejector-outlet pressure leads to a shock wave peculiar to the two-phase flow in which pressure increases gently 28

in the nozzle at P b = 0.19 and 0.26 MPa, as shown in Fig. 12. The pressure in the nozzle decreases toward the nozzle outlet for P b =0.12 MPa, so we conclude that no shock wave forms in the nozzle under these conditions. Next, we compare the pressure distribution for different values of nozzle-inlet quality (X n = 0.3 and 0.6). Upon increasing the nozzle-inlet quality, the velocity in the nozzle increases. Therefore, the magnitude of decompression in the nozzle is increased by increasing the nozzle-inlet quality. Furthermore, for P b = 0.19 MPa, a clear pressure increase occurs in the nozzle upon increasing the nozzle-inlet quality from 0.3 to 0.6. Therefore, under these experimental conditions, a nozzle-inlet quality X n = 0.6 allows the structure of the shock wave in the nozzle to be easily observed, so it can be monitored as the outlet pressure is varied. In addition, for an outlet pressure P b = 0.19 MPa, the position of the shock wave depends on the nozzle-inlet quality. Fig.12 Static pressure at ejector wall surface as a function of distance from ejector throat. 4.3 Visualization Results 4.3.1 Shock wave structure visualized and compared with pressure measurements With the confirmation that the shock wave occurs in the nozzle under the conditions given above, we visualized the shock wave by using the method described above. Figure 13 compares the visualization and the measured pressure for a nozzle-inlet quality X n = 0.6. These images were acquired at 12 500 frames per second (fps), with a shutter speed of 1/25 000 s, and a resolution of 1012 1012. The top and bottom parts of the images are deleted, leaving the nozzle displayed lengthwise. Because less light irradiates the photodetector downstream of the shock wave, as discussed above, a monochromatic high-speed camera allows us to image the shock wave. The gentle pressure increase peculiar to the two-phase-flow shock wave downstream of the black shadow allows us to confirm the visualization of the two-phase-flow shock wave. For comparison, the results of our conventional visualization method (Yamanaka and Nakagawa, 2012) for P b = 0.26 MPa are shown in the lower image of Fig. 13, which further supports the interpretation of the visualization of the two-phase-flow shock wave by the method proposed herein 29

and highlights the resolution of the proposed technique. A rise in the outlet pressure is known to cause the shock wave to move from the outlet toward the throat of the nozzle (Yamanaka and Nakagawa, 2012). Furthermore, Fig. 13 shows that an increase in outlet pressure causes the shockwave to change from an oblique shock wave to a normal shock wave. Although a normal shock wave appears in Fig. 13, no sudden pressure increase appears across the normal shock wave. This phenomenon is attributed to a weak normal shock wave occurring first in the gas phase and continuing, with the deceleration of the liquid phase (with its larger inertia) contributing to a gradual increase in pressure. Fig.13 Images of shock wave acquired by proposed visualization method (top three images) and by conventional method (bottom image). Bottom panel allows the images to be compared with the pressure measured with a nozzle-inlet quality of Xn = 0.6. 10 2

4.3.2 Dependence of shock wave on nozzle-inlet quality Next, we image the shock wave for several values of nozzle-inlet quality. Figure 14 shows images for an ejector outlet pressure of P b = 0.19 MPa. These images were acquired at 60 000 fps, a shutter speed of 1/66 462 s, and a resolution of 768 272. With a decrease in nozzle-inlet quality X n from 0.6 to 0.3, the oblique shock wave moves toward the nozzle outlet, where it causes the velocity in the nozzle to decrease, before disappearing when the quality falls below 0.475. Fig.14 Images of shock wave for several values of nozzle-inlet quality (Pb = 0.19 MPa). 4.3.3 Structural change of the shock waves in two-phase flow ejector. For all conditions tested, a shock wave occurring in the nozzle-outlet region is an oblique shock wave, whereas a shock wave occurring in the nozzle throat region is a normal shock wave. This structural change in the shock wave might be due to the shape of the nozzle outlet. A shock wave occurring in axial flow becomes a normal shock wave, whereas a shock wave occurring where flow has a transverse component (e.g., through a mixing section), such as in the nozzle outlet, becomes an oblique shock wave, as shown in Fig. 15. 11 2

Fig.15 Possible explanations for structural change of shock waves in two-phase flow ejector. 5. Summary and Conclusion In this study, we use a transmitted optical beam and a high speed camera to visualize the structure of two-phase-flow shock waves in an ejector nozzle. The results lead to the following conclusions: i. A rise in ejector-outlet pressure shifts the shock wave from the outlet toward the throat of the nozzle and causes an oblique shock wave to become a normal shock wave. ii. This oblique shock wave moves toward the nozzle outlet and its speed in the nozzle decreases with a drop in nozzle-inlet quality, until it finally disappears from the nozzle. iii. The shape of the output nozzle may modify the structure of the shock waves. Further visualization experiments covering a large set of conditions should clarify the mechanisms behind this structural change. References Bouhanguel, A., Desevaux, P. and Gavignet, E., Flow visualization in supersonic ejectors using laser tomography techniques (2011), International Journal of Refrigeration, Vol.34, No.7, pp.1633 1640. Dvorak, V., Safarik, P., Transonic instability in entrance part of mixing chamber of high-speed ejector (2005), Journal of Thermal Science, Vol.14, pp.258-264. Fabri, J., Siestrunck, R., Supersonic Air Ejectors (1958), Advances in Applied Mechanics, Vol. 5, Academic Press, New York, pp.1-34. Harada, A., Nakagawa, M., Shock and Expansion Waves in Supersonic Two Phase Jet Flow (2011), JSME Tokai Branch 60th annual meeting, No.113-1 (in Japanese). Matsuo, K., Sasaguchi, K., Tasaki, K., and Mochizuki, H., Investigation of Supersonic Air Ejector (Part1.Performance in the Case of Zero-Secondary Flow) (1981), Bulletin of JSME, Vol.24, No.198, pp.2090-2097 (in Japanese). Nakagawa, M., Hakamada, O., Miyazaki, H., Rarefaction waves and shock waves at the outlet of the supersonic twophase flow nozzle (2005), JSME Tokai Branch 54th annual meeting, No.0531-1 (in Japanese). Nakagawa, M., Harada, A., Two-phase Ejector for refrigeration cycle and shock wave appearing in the supersonic twophase flow nozzle (2010), Maltiphase Flow, Vol.24, No.1, pp.21-28 (in Japanese). NIST Standard Reference Data Program, REFPROP v8.0, Reference fluid thermodynamic and transport properties (2007). 12 2

Takeuchi, H., Nishijima, H., and Ikemoto, T., World's First High Efficiency Refrigeration Cycle with Two-Phase Ejector: EJECTOR CYCLE (2004), SAE Technical Paper 2004-01-0916. The Japanese Society of Mechanical Engineers ed., Handbook of Gas-Liquid Two Phase Flow Technology Second Edition (2006), pp.374-393 (in Japanese). Yamanaka, H., Nakagawa, O., Supersonic Nozzle Flow in the Two-Phase Ejector as Water Refrigeration System by Using Waste Heat (2012), 9th International Conference on Heat Transfer, Fluid Mechanics and Thermodynamics, pp.933-938. Yamanaka, H., Nakagawa, M., Two-phase Flow Ejector as Water Refrigerant by Using Waste Heat (2013), Journal of Physics, Conference Series Vol.433, DOI:10.1088/1742-6596/433/1/012018. 13 2