- Email: [email protected]
Influence of driving fluid properties on the performance of liquid-driving ejector
International Journal of Heat and Mass Transfer 101 (2016) 20–26
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Influence of driving fluid properties on the performance of liquid-driving ejector Jianbo Tang a,b, Zhiyi Zhang c, Luyao Li a,b, Junjie Wang a, Jing Liu a, Yuan Zhou a,⇑ a
Key Laboratory of Cryogenics, Technical Institute of Physics and Chemistry, Chinese Academy of Sciences, Beijing 100190, China University of Chinese Academy of Sciences, Beijing 100049, China c Tianjin Key Laboratory of Refrigeration Technology, Tianjin University of Commerce, Tianjin 300134, China b
a r t i c l e
i n f o
Article history: Received 14 September 2015 Received in revised form 7 April 2016 Accepted 10 April 2016 Available online 25 May 2016 Keywords: Ejector Liquid metal Vacuum Bernoulli’s principle
a b s t r a c t This study focuses on the influence of driving fluid properties on the mass transfer process of ejector apparatus. A liquid-actuated ejector system is firstly designed and three experimental cases working with different kinds of driving fluids, i.e., gallium-based liquid metal, aqueous NaI solution and water, are successively investigated. The results show that the ejector system working with liquid metal has high-vacuum ejecting capacity and good temperature stability. With a system power consumption of about 50 W, the liquid metal driving ejector obtains a no-entrainment vacuum pressure of 33 Pa which is three orders of magnitudes lower than that of the water driving case. When the gas-suction pressure exceeds 40 kPa, NaI solution driving case reveals the highest entrainment. The reasons for the differences in the ejector performance are later discussed. Based on the experimental results and theoretical analyses, the method of using high density, low viscosity and low vapor pressure working fluid to improve ejector performance is proposed. Other than conventional methods which focus on optimizing the geometrical design and operating parameters of ejectors, the method discussed here puts forward a new perspective to improve ejector performance. Moreover, the ejector system working with liquid metal demonstrates its advantages such as compact system design, oil-free vacuum pumping as well as high temperature operation. Ó 2016 Published by Elsevier Ltd.
1. Introduction An ejector is a simple-structure device that provides multiple capacities such as vacuum-pumping, compressing, transporting, mixing. It has been widely used in the fields of refrigeration, food industry, as well as chemical, biochemical and medical processes ever since its invention [1–3]. The utilization history of ejector witnesses its continuous evolutions and enormous efforts have been made to improve the performance of different types of ejectors [4]. The performance of the ejector is sensitive to its geometrical design, operating parameters and driving fluid properties. The first two factors are extensively investigated while few studies have been focused on the driving fluid properties [5–8]. As for liquiddriving ejectors, water, as an abundant and easy-handling candidate, is used in most cases [9–11]. Among the few examples that can be found in the open literature, Elgozali et al. [12] studied the effect of viscosity and surface tension on performance of liquid ejector by adding polymeric thickener, sucrose and alcoholic foam ⇑ Corresponding author. E-mail address: [email protected] (Y. Zhou). http://dx.doi.org/10.1016/j.ijheatmasstransfer.2016.04.028 0017-9310/Ó 2016 Published by Elsevier Ltd.
breaker to water. Power et al. [13] and Haak et al. [14] reported experimental studies on high-density mercury vapor jet pumps in 1958 and 1959, respectively. Studies on ejector refrigeration cycles and ejector-absorption cycles encounter a variety of driving fluids such as vaporized halocarbon compound (CFC, HCFCs and HFCs), water, ammonia–water solution, and LiBr–water solution. However, thermodynamic properties (enthalpy, entropy, specific heat, etc.) of the driving fluids which have significant influence on the refrigeration performance are of general interest while others are usually neglected [2,4]. Transport properties (density, viscosity and vapor pressure, etc.) have strong effects on the momentum transfer and mixing process of the ejector and should be considered as prior factors when designing an ejector system or choosing operating parameters. Besides, by simply replacing conventional driving fluids with candidates which offer more desirable transport properties may significantly improve the ejecting performance. Theoretically, driving fluids with high density, low viscosity and low vapor pressure would be ideal candidates. Liquid metal and inorganic salt solution represent two common types of high density, low vapor pressure fluids. Novel functional material GaInSn
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
21
(eutectic gallium–indium–tin) is a nontoxic room-temperature liquid metal which offers many advantages such as ultra-high density and ultra-low vapor pressure while still exhibits good fluidity [15]. Recent years have witnessed its emergence in many unique applications [16–19]. Adding inorganic salt to water is an easy way to create a solution with higher density and lower vapor pressure. In this study, three kinds of driving fluids, namely liquid metal, aqueous solution and water were firstly prepared with their transport properties measured for comparison purpose. The details related to GaInSn handling and oxidation prevention have been described in our previous work [20]. Excellent solubility, good temperature stability and material compatibility are the main reasons for us to choose NaI aqueous solution among all the other inorganic salt solutions that have been tested. The performance of the ejector system working with different driving fluids is later compared in order to study the influence of driving fluid properties on the system performance.
2. Experimental setups and procedures The properties of the driving fluids vary with temperature change. So after the driving fluids are prepared, separate experiments are carried out to measure their saturated pressure and viscosity under different temperatures. The density of each fluid may also change from temperature to temperature. Since temperatureinduced density change is relatively small compared to the density difference between different kinds of fluids, this kind of density change is not considered in this study. As shown in Fig. 1(a), for saturated pressure measurement, a partially filled stainless steel sample fluid container is firstly vacuumed by an auxiliary vacuum pump. The check valve is turned off thereafter so that the container becomes airtight. Then the container is put into a water bath for temperature control. Capillary viscometer is used for viscosity measurement. The setups are illustrated in Fig. 1(b) and the method can be found in Ref. [21]. The system arrangements for ejector performance study are shown in Fig. 2. The driving fluid is firstly pumped to a high pumping pressure before it enters the ejector where it draws in nitrogen from the reservoir. Then the two flows interact with each other inside the ejector where they eventually become a liquid–gas mixture. The mixture is later separated due to gravitational difference in the separator. Nitrogen is expelled to the atmosphere while the driving fluid flows back to the pump to complete the ejector loop. Since the driving fluid properties are temperature dependent,
Fig. 2. Schematic drawing of the test rig (not to scale).
a water coil is installed inside the separator for temperature control. The rotation speed x and electric power input EP of the pump can be controlled with an adjustable controller. The absolute pumping pressure PP of the driving fluid at the ejector entrance is measured by pressure sensor P1 (Model number: PX429S15150A5V, Omega Engineering Inc. Accuracy: ±0.08%). The absolute vacuum pressure inside the reservoir PV and the absolute discharge pressure at the separator PD are measured by pressure senor P2 (Model number: PX429-005A5V, Omega Engineering Inc. Accuracy: ±0.08% and P3 (Model number: PX429-100A5V, Omega Engineering Inc. Accuracy: ±0.08%), respectively. The mass flow rate of the driving fluid G1 is measured by Flowmeter 1 (Model number: DMF-1-3-A, Beijing Sincerity Automatic Equipment Co., Ltd. Accuracy: ±0.2%) and the volume flow rate (standard state) of nitrogen entrainment G2 is measured by Flowmeter 2 (Model number: D0719BM, Beijing Sevenstar Electronics Co., Ltd. Accuracy: ±1%). The temperature of the driving fluid is measured by a sheathed thermocouple (Accuracy: ±0.5 °C) mounted inside the separator. The vacuum pumping capacity, nitrogen entrainment as well as the pump power consumption are the main parameters that are
Fig. 1. Schematic drawing of the experimental setups for (a) saturated pressure measurement; (b) kinematic viscosity measurement. 1-Water bath, 2-sample fluid container, 3-thermal couple, 4-pressure sensor, 5-check valve, 6-to vacuum pump, 7-thermometer, 8-capillary viscometer.
22
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
used to evaluate the system performance. The no-entrainment vacuum pressure is considered as the ultimate vacuum pumping capacity of the ejector under each working condition. To measure the no-entrainment vacuum pressure, the reservoir is precharged by the nitrogen source with check valve V1C open and V2C closed. Check valve V2C is turned off later to cut off nitrogen supply. When the ejector loop gets started, V1C is turned on to expel the nitrogen from the reservoir. After a certain period of time, the reservoir will reach a steady no-entrainment vacuum pressure. The entrainment of the system is measured by another procedure. In this case, V2C remains open throughout the experiment. The pressure of the nitrogen supply at the inlet of Flowmeter 2 is maintained as a constant value by metering valve V2M. Another metering valve V1M is installed between the reservoir and Flowmeter 2 to control the vacuum pressure inside the reservoir. The entrainment at an arbitrary vacuum pressure PV is taken to be equal to the nitrogen flow rate G2 measured by Flowmeter 2. Based on the system arrangement, system parts including the ejector, separator, reservoir as well as their connections are designed using SolidWorks software. Fig. 3(a) shows the assembled
Computer Aided Design (CAD) system prototype. Transparent PMMA (polymethyl methacrylate) is used to construct the ejector parts, separator and the reservoir for flow visualization. The capacity of the reservoir and separator is 800 mL and 900 mL, respectively. The connecting tubes, water coil, wetting part of the valves and sensors are all made of stainless steel since GaInSn alloy is incompatible with materials like aluminum and copper. The ejector is a crucial part of the system and its structure and parameter details are shown in Fig. 3(b). Given the liquid-driving nature of the ejector, a converging-type ejector nozzle is chosen. Also considering the existence of the highest system velocity at the ejector nozzle, an arc-transitional channel is fabricated to reduce flow loss. The nozzle is also designed to ensure that the tangent line at the nozzle exit is horizontal. At the mixing section, a constant area mixing model is used. The mixing section is 6 mm in diameter with respect to the nozzle throat diameter of 2 mm. The mechanical pump used in this study is a volumetric rotary vane pump. Its structure and working mechanism are demonstrated in Fig. 3(c). The wetting parts of the pump are either made of stainless steel or graphite, both materials have good compatibility with all the three kinds of driving fluids.
Fig. 3. (a) CAD system model: 1-pump; 2-controller; 3-ejector; 4-separator; 5-reservoir; 6-nitrogen source; 7-water coil; 8,9-flowmeters; 10,11-check valves, 12,13-metering valves, 14,15,16-pressure sensor; 17-thermometer; (b) structure and parameter details (in millimeter) of the ejector; (c) structure of the pump head.
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
23
Table 1 Comparison on driving fluids properties [15]. Property 3
Density (kg/m ) Melting point (°C) Boiling point (°C)
Ga67In20.5Sn12.5
NaI (aq)
Water
6360 10.5 >1300
1530 <0 >100
1000 0 100
3. Experimental results The comparisons of density and phase-change temperature of the three driving fluids can be found in Table 1 and properties such as saturated pressure and viscosities are presented in Fig. 4. The dash lines represent data of water adapted from the NIST database [22]. The measured data points of water demonstrate good fitting relation with the reference data which indicate the reliability of the measurement. The saturated pressure of both water and NaI solution shows an increase as temperature rises. The difference is that the pressure value as well as the pressure increment of NaI solution is lower than that of water. Since the vapor pressure of the liquid metal is below 106 Pa even though it is over 600 °C [15], it’s safe to neglect its effect in our experiment. Different from the saturated pressure evolution, the viscosities of the fluids show inverse trends as temperature rises. The measured kinematic viscosity v of NaI solution is slightly smaller than that of water. And the liquid metal shows a much smaller kinematic viscosity and notably, it is also found to be more stable when temperature changes. When comparing the dynamic viscosity l which takes fluid density into consideration, the liquid metal shows a much higher value. As can be expected, blending solute in water increases its density and dynamic viscosity, while it reduces the vapor pressure of the solvent. Fig. 5 gives a comparison of the no-entrainment vacuum pressure inside the reservoir under different working conditions. The temperature of the driving fluid for each case is maintained at a constant value of 15 °C. The ejector system working with water reaches a vacuum level of 10 kPa with a power consumption of about 120 W. When we use NaI solution to replace water, lower no-entrainment vacuum pressure is obtained and the vacuum level drops to 1 kPa level while the power consumption rises to 170 W. The liquid metal driving case shows a significant improvement. The vacuum level achieved under liquid metal driving condition is much higher than the other two cases, and better, it consumes the lowest amount of electric power. A vacuum pressure of 33 Pa, which is the lowest vacuum pressure that has ever been reported in the open literature achieved by a single-stage ejector, is detected with about 50 W power input. The no-entrainment vacuum pressure for each case under constant power input and different temperature is also tested and the result is presented in Fig. 6. For water and NaI solution driving cases, when temperature increases form 15 °C to 45 °C, the no-entrainment vacuum pressure rises gradually. The water driving case experiences a 7.3 kPa vacuum pressure increment while for the NaI driving case, a smaller value of 5.8 kPa is detected. The temperature-induced no-entrainment vacuum pressure rise and the pressure–temperature saturation relationship are closely matched when we compare Figs. 6 with 4. So we conclude that the temperature-induced vapor pressure rise of the driving fluid is the main reason for the no-entrainment vacuum pressure rise as the temperature increases. This also explains why liquid metal driving case shows such a small vacuum fluctuation. The ultra-low vapor pressure nature of the liquid metal makes the no-entrainment vacuum pressure not sensitive to temperature variation. The fluctuation detected in the experiment is contributed to other reasons such as flow instability and measurement error. Conclusion can also be made that mercury, a well-known
Fig. 4. Comparison of saturated pressure and viscosities of the different driving fluid under different temperature.
Fig. 5. No-entrainment vacuum pressure of different driving fluids as a function of pump power input. Icons: experimental data points; dash lines: fitting curves.
liquid metal which has higher density (13,600 kg/m3) than GaInSn alloy, may be less competitive in high vacuum ejecting because it has a much higher vapor pressure than GaInSn alloy.
24
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
Fig. 6. The influence of temperature on the non-entrainment vacuum pressure of the ejector.
The nitrogen entrainment of different fluids driving conditions are measured to further study the pumping characteristics of the ejector system. Comparisons are made under the same electric power input of about 60 W. The main working parameters of the ejector system are listed in Table 2. It shows that driving fluid with higher density requires higher pumping pressure and usually has a higher mass flow rate, while driving fluid with lower density obtains a higher volume flow rate and a higher nozzle velocity. Due to its high density, the typical nozzle throat velocity of the liquid metal driving case drops to an incredibly low value of 6.5 m/s. When it comes to the entrainment ratio which also represents an important performance criterion of the ejector, the volume entrainment ratio, rather than mass entrainment ratio, is generally used for liquid–gas ejectors given the great density difference between the driving fluid and the entrained fluid. As shown in Fig. 7, both the nitrogen entrainment and the volume entrainment ratio show a similar trend. In the vacuum pressure range above 40 kPa, the entrainment of NaI solution and water driving cases surpass the liquid metal driving case. But when the vacuum pressure of the reservoir drops below 40 kPa, the entrainment of NaI and water driving cases decrease rapidly to zero. On the contrary, liquid metal driving case shows its advantage in this low vacuum pressure range. The reasons for these different behaviors will be explained in Section 4. A more detailed plot of vacuum pumping performance of the liquid metal driving case against operating time can be found in Fig. 8. As has been shown, by adjusting the electric power input of the pump, different vacuum levels can be obtained. And the ejector system works well as a rough vacuum system with tens of watts power consumption. One major advantage of the liquidring vacuum system is that no lubrication or sealing oil is required so it is desirable for clean vacuum pumping. A commercially available mechanical vacuum pump is also used to compare with the current ejector system. The mechanical vacuum pump can produce
Table 2 Comparison on working parameters under 60 W electric power input. Working parameter
Ga67In20.5Sn12.5
NaI (aq)
Water
Mass flow rate (kg/min) Volume flow rate (L/min) Pumping pressure (Pa) Nozzle throat velocity (m/s)
7.83 1.23 4.87 105 6.5
6.08 3.97 2.59 105 21.1
4.97 4.97 2.52 105 26.4
Fig. 7. Comparison of nitrogen entrainment under the electric power input of about 60 W and driving fluid temperature of 20 °C. Icons: experimental data points of nitrogen entrainment (hollow) and volume entrainment ratio (filled); dash lines: fitting curves.
Fig. 8. Vacuum pressure development of the reservoir as a function of time under different electric power input. Icons: experimental data points; dash lines: fitting curves.
a comparable vacuum pressure of about 50 Pa only with a much higher electronic power consumption of about 150 W. Though, the liquid metal driving ejector has a drawback that its driving fluid, GaInSn alloy, will be easily oxidized when it contacts with oxygen. This problem will not be a concern if it is used for nonoxidative medium pumping. And if generated, alkaline solution can be applied to efficiently remove the oxides. 4. Discussions The ejector apparatus is an assembly of several variable crosssection parts. The flow patterns of a typical cross-section flow is shown in Fig. 9. The flow regime of the cross-section flow is governed by the equations of continuity, momentum and energy. The change in flow area of the flow channel will result in changes of flow parameters such as flow velocity and static pressure which indicate flow work converting between kinetic energy and potential energy. The process will also be accompanied with flow work losses which mainly manifest as friction loss and local dissipation. The friction loss is a result of fluid viscosity and wetting surface
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
would lead to estimations with large derivation. Consequently, conclusions can be made that under normal operations, the performance of liquid ejector can be well predicted by laws of fluid mechanics while under extreme conditions (i.e., high temperature, high vacuum), thermodynamic principle may become dominant. The theory that the ultimate vacuum pressure of the ejector is subject to the saturated pressure of the driving fluid with respect to its working temperature is consistent with the experimental result that NaI solution driving case obtains a lower no-entrainment vacuum pressure than that of water driving case. It has also been demonstrated that owing to the inherent nature of ultra-low vapor pressure, liquid metal provides an exception for those who want to break the thermodynamic restriction. A simple mathematical model can be developed to describe the influence of pressure gradient and volume flow rate of the driving fluid on the ejector’s entrainment behavior. Since the driving fluid and the entrained fluid are immiscible, the driving fluid and the entrained gas should both satisfy the law of continuity. Let’s recall Fig. 3(b) and consider the flow conditions at the mixing section and the diffuser exit. The continuity equation gives:
Fig. 9. Flow patterns and loss mechanisms of cross-section flow.
roughness while the local dissipation is the energy loss consumed by processes such as vortex evolution. Bernoulli’s equation can be applied to describe the flow patterns at cross-sections of the ejector:
1 1 p1 þ qu21 þ qgz1 ¼ p2 þ qu22 þ qgz2 þ DW 1—2 2 2
25
ð1Þ
where the subscripts 1 and 2 are locations indicated in Fig. 9; p is the static pressure of the driving fluid; q is the density; u is the velocity; g is the gravitational acceleration, z is the height, DW 1—2 ¼ f ðq; l; u; D; L; dÞ is the total flow work loss as a function of q, u, and the driving fluid dynamic viscosity l, the flow channel diameter D, the flow distance length L, as well as the surface roughness d. Complexities will arise when two fluids interact with each other. The flow patterns inside the ejector arising from the interaction of the driving fluid and the entrained fluid exhibits highly unstable and complex two-phase flow characteristics, though, qualitative analysis of the density and viscosity influence on ejector mass transfer process can be made according to Eq. (1). Density q determines the specific kinetic energy of the carrier fluid. For a horizontally arranged ejector as in the current case, it implies that fluid with higher density requires smaller change in velocity to produce the same amount of static pressure drop. When flow velocity decreases, the required pumping speed is also reduced accordingly, which is favorable for low power consumption. This explains why the liquid metal driving case has the smallest power consumption. Ejector system frequently suffers from low efficiency, components wear and noise emitting under high velocity operation. Therefore, reducing operating velocity by using high density driving fluid could hopefully solve these problems while still meet the performance requirements. The influence of driving fluid viscosity on ejector system usually manifests as friction loss and wear mechanism inside the system. So driving fluid should also have an acceptable viscosity to minimize flow work loss and component wear. The static pressure drop at the outlet of the ejector nozzle produces the vacuum pumping effect. The vacuum pressure is mainly controlled by the operating parameters based on specific structural design. But for conditions that the vacuum pressure reaches the saturated pressure level, the influence of vapor evaporation arises. This is because when the pressure is comparable to the vapor pressure of the driving fluid, the vaporization of the driving fluid itself would be significant and thus prevent higher vacuum ejecting. Under such circumstances, simply applying Bernoulli’s equation
_ d ðlÞ and m _ m ðgÞ ¼ m _ d ðgÞ _ m ðlÞ ¼ m m
ð2Þ
_ _ where mðlÞ and mðgÞ denote the mass flow rate of driving fluid and entrained gas, respectively. The subscripts m and d represent the flow positions of the mixing section entrance and the diffuser exit, respectively. As for the volume flow rate, since the driving fluid is incompressible, its volume flow rate will remain as a constant. Though no phase transition will take place for nitrogen, the volume flow rate of the entrained gas will be influenced by its local pressure and temperature. So we have:
v_ m ðlÞ ¼ v_ d ðlÞ and v_ m ðgÞ ¼ f ðP; TÞv_ d ðgÞ
ð3Þ
where v_ ðlÞ and v_ ðgÞ denote the volume flow rate of driving fluid and entrained gas, respectively. By introducing the gas state equation _ where R is the gas constant, T is the temperaPv_ ðgÞ ¼ mðgÞRT=M, ture, and M is the molar mass of the gas, we get:
v_ m ðgÞ ¼
PD v_ d ðgÞ ¼ nc v_ d ðgÞ PV
ð4Þ
where nc ¼ PD =PV is the compression ratio of the ejector. Since _ _ n ðlÞ and the temperature of the fluid is maintained as a mðgÞ m constant in the present case, the temperature change of the entrained gas is neglected in Eq. (4). For a qualitative comparison, the ideal gas model is used, and note that for more accurate calculation, there are real gas models available. According to Eq. (4), high compression ratio will lead to a high volume entrainment at the suction chamber. However, the volume flow rate of the entrained gas will be constrained by the structural parameters of the flow channel. Further assumptions can be made that upon mixing, the velocity change of the driving fluid is small while the entrained gas can still obtain the same velocity as the driving fluid based on the law of momentum conservation. Thus we have:
v_ m ðgÞ /
D2m D2n D2n
v_ m ðlÞ
ð5Þ
where Dm and Dn are the diameters of the mixing section and nozzle exit, respectively. Since the standard state volume flow rate of the nitrogen measured by G2 is close to the working condition at the diffuser exit, the volume entrainment ratio of the ejector kappav has the form ofv_ d ðgÞ ¼ jv v_ d ðlÞ, so the combination of Eqs. (4) and (5) gives:
jv /
D2m D2n nc D2n
ð6Þ
26
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
Eqs. (5) and (6) show the relationship between the volume entrainment and established structural parameters as well as measureable working condition parameters. Eq. (5) indicates that the gas entrainment is proportional to the volume flow rate of the driving fluid. From a microscopic perspective, the mass transfer process is subject to the collisions between the two fluid particles especially in high vacuum conditions. So driving fluid with larger contacting surface, i.e., larger volume flow rate, will increase the colliding opportunities which result in larger entrainment. Eq. (6) implies that the gas entrainment is proportional to the inverse of the compression ratio which means high compression ratio may cause a decline to the gas entrainment. The decline in volume flow rate of liquid metal driving case comparing with the other two cases is contributed to both of the above-mentioned reasons. The entrainment ratio jv is a parameter that is influenced by structural design, working conditions as well as the nature of the driving fluid. In general, it reflects the combined effect of its vacuum capacity (vacuum pressure) and transportation capacity (volume flow rate) of the ejector system. For high vacuum pumping cases (the liquid metal driving case), the latter is the entrainment-determine factor while for low vacuum pumping cases (NaI solution and water driving cases), the former becomes dominant. As has been shown, NaI solution driving cases have a better vacuum capacity than water driving cases, so it is reasonable for the NaI solution driving cases to produce higher entrainment ratio. 5. Conclusions A liquid–gas ejector vacuum system is designed and a series of experiments are conducted to investigate the influence of fluid properties on its mass transfer performance. Liquid metal, NaI solution and water are prepared with their properties characterized for experimental investigation. The system manages to work under different conditions and the results show that a driving fluid with high density, low viscosity and low vapor pressure is favorable for high vacuum ejecting. The entrainment of the ejector is contributed to both the compression ratio and volume flow rate of the driving fluid. The room-temperature liquid metal actuated ejector has demonstrated its potential for oil-free vacuum pumping as well as high temperature operation. But when entrainment is the major concern, low density driving fluids such as aqueous solution or water are recommended. This study has disclosed a method to improve ejector performance from the perspective of driving fluid properties and its applications are predictable especially when structural modifications are restrained. Acknowledgment This study is partially supported by the National Natural Science Foundation of China (Grant No. 51176198).
References [1] H. Bannwarth, Liquid Ring Vacuum Pumps, Compressors and Systems, Conventional and Hermetic Design Translated by Christine Ahner, WILEYVCH Verlag GmbH & Co. KGaA, Weinheim, 2005. [2] Kanjanapon Chunnanond, Satha Aphornratana, Ejectors: applications in refrigeration technology, Renewable Sustainable Energy Rev. 8 (2004) 129– 155. [3] Stefan Elbel, Predrag Hrnjak, Ejector Refrigeration: An Overview of Historical and Present Developments with an Emphasis on Air-Conditioning Applications, in: Int. Refrig. Air Conditioning Conf. Purdue, 2008, pp. 14–17. Paper 884. [4] S.B. Riffat, L. Jiang, G. Gan, Recent development in ejector technology – a review, Int. J. Ambient Energy 26 (1) (2005) 13–26, http://dx.doi.org/10.1080/ 01430750.2005.9674967. [5] B. Gourich, N. El Azher, C. Vial, M. Belhaj Soulami, M. Ziyad, A. Zoulalian, Influence of operating conditions and design parameters on hydrodynamics and mass transfer in an emulsion loop–venturi reactor, Chem. Eng. Process. 46 (2007) 139–149. [6] S. Balamurugan, Mayank D. Lad, Vilas G. Gaikar, Ashwin W. Patwardhan, Hydrodynamics and mass transfer characteristics of gas–liquid ejectors, Chem. Eng. J. 131 (2007) 83–103. [7] P.H.M.R. Cramers, A.A.C.M. Beenackers, Influence of the ejector configuration, scale and the gas density on the mass transfer characteristics of gas–liquid ejectors, Chem. Eng. J. 82 (2001) 131–141. [8] G.M. de Billerbeck, J.S. Condoret, C. Fonade, Study of mass transfer in a novel gas–liquid contactor: the aero-ejector, Chem. Eng. J. 72 (1999) 185–193. [9] Yoon Kee Kim, Dong Yeop Lee, Hyun Dong Kim, Joo Ha Ahn, Kyung Chun Kim, An experimental and numerical study on hydrodynamic characteristics of horizontal annular type water–air ejector, J. Mech. Sci. Technol. 26 (9) (2012) 2773–2781, http://dx.doi.org/10.1007/s12206-012-0724-9. [10] Hyo Min Jeong, Han Shik Chung, Kang Youl Bae, Se Hyun Kim, You Sik Shin, Water cooling characteristics in an enclosed vacuum tank by water driven ejector, J. Mech. Sci. Technol. 19 (1) (2005) 164–172. [11] M.D. Butterworth, T.J. Sheer, High-pressure water as the driving fluid in an ejector refrigeration system, Appl. Therm. Eng. 27 (2007) 2145–2152. [12] A. Elgozali, V. Linek, M. Fialov, O. Wein, J. Zahradn, Influence of viscosity and surface tension on performance of gas–liquid contactors with ejector type gas distributor, Chem. Eng. Sci. 57 (2002) 2987–2994. [13] B.D. Power, N.T.M. Dennis, An unusual mercury ejector pump, in: E. Thomas (Ed.), Fundamental Problems in Vacuum Techniques Ultra-High Vacuum: Proceedings of the First International Congress on Vacuum Techniques, Namur, Belgium, June 10–13 1958, Pergamon Press, 1960, pp. 197–201. [14] Werner H. Haak, Frederick Vrátny´, Annular Jet Pump, Rev. Sci. Instrum. 30 (1959) 296, http://dx.doi.org/10.1063/1.1716609. [15] N.B. Morley, J. Burris, L.C. Cadwallader, M.D. Nornberg, GaInSn usage in the research laboratory, Rev. Sci. Instrum. 79 (2008) 056107, http://dx.doi.org/ 10.1063/1.2930813. [16] Gökhan Perçin, Laurent Levin, Butrus T. Khuri-Yakub, Piezoelectrically actuated droplet ejector, Rev. Sci. Instrum. 68 (1997) 4561, http://dx.doi.org/ 10.1063/1.1148431. [17] A. Miner, U. Ghoshal, Cooling of high-power-density microdevices using liquid metal coolants, Appl. Phys. Lett. 85 (2004) 506, http://dx.doi.org/10.1063/ 1.1772862. [18] I. Silverman, A.L. Yarin, S.N. Reznik, A. Arenshtam, D. Kijet, A. Nagler, High heat-flux accelerator targets: cooling with liquid metal jet impingement, Int. J. Heat Mass Transfer 49 (2006) 2782–2792. [19] Peiper Li, Jing Liu, Harvesting low grade heat to generate electricity with thermosyphon effect of room temperature liquid metal, Appl. Phys. Lett. 99 (2011) 094106, http://dx.doi.org/10.1063/1.3635393. [20] Jianbo Tang, Yuan Zhou, Jing Liu, Junjie Wang, Wenxiu Zhu, Liquid metal actuated ejector vacuum system, Appl. Phys. Lett. 106 (2015) 031901, http:// dx.doi.org/10.1063/1.4906098. [21] GB/T 10247-2008, Methods of Viscosity Measurement, 2008. [22] NIST Standard Reference Database 23, Version 8.0, The U.S. Secretary of Commerce, USA, 2007.
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Influence of driving fluid properties on the performance of liquid-driving ejector Jianbo Tang a,b, Zhiyi Zhang c, Luyao Li a,b, Junjie Wang a, Jing Liu a, Yuan Zhou a,⇑ a
Key Laboratory of Cryogenics, Technical Institute of Physics and Chemistry, Chinese Academy of Sciences, Beijing 100190, China University of Chinese Academy of Sciences, Beijing 100049, China c Tianjin Key Laboratory of Refrigeration Technology, Tianjin University of Commerce, Tianjin 300134, China b
a r t i c l e
i n f o
Article history: Received 14 September 2015 Received in revised form 7 April 2016 Accepted 10 April 2016 Available online 25 May 2016 Keywords: Ejector Liquid metal Vacuum Bernoulli’s principle
a b s t r a c t This study focuses on the influence of driving fluid properties on the mass transfer process of ejector apparatus. A liquid-actuated ejector system is firstly designed and three experimental cases working with different kinds of driving fluids, i.e., gallium-based liquid metal, aqueous NaI solution and water, are successively investigated. The results show that the ejector system working with liquid metal has high-vacuum ejecting capacity and good temperature stability. With a system power consumption of about 50 W, the liquid metal driving ejector obtains a no-entrainment vacuum pressure of 33 Pa which is three orders of magnitudes lower than that of the water driving case. When the gas-suction pressure exceeds 40 kPa, NaI solution driving case reveals the highest entrainment. The reasons for the differences in the ejector performance are later discussed. Based on the experimental results and theoretical analyses, the method of using high density, low viscosity and low vapor pressure working fluid to improve ejector performance is proposed. Other than conventional methods which focus on optimizing the geometrical design and operating parameters of ejectors, the method discussed here puts forward a new perspective to improve ejector performance. Moreover, the ejector system working with liquid metal demonstrates its advantages such as compact system design, oil-free vacuum pumping as well as high temperature operation. Ó 2016 Published by Elsevier Ltd.
1. Introduction An ejector is a simple-structure device that provides multiple capacities such as vacuum-pumping, compressing, transporting, mixing. It has been widely used in the fields of refrigeration, food industry, as well as chemical, biochemical and medical processes ever since its invention [1–3]. The utilization history of ejector witnesses its continuous evolutions and enormous efforts have been made to improve the performance of different types of ejectors [4]. The performance of the ejector is sensitive to its geometrical design, operating parameters and driving fluid properties. The first two factors are extensively investigated while few studies have been focused on the driving fluid properties [5–8]. As for liquiddriving ejectors, water, as an abundant and easy-handling candidate, is used in most cases [9–11]. Among the few examples that can be found in the open literature, Elgozali et al. [12] studied the effect of viscosity and surface tension on performance of liquid ejector by adding polymeric thickener, sucrose and alcoholic foam ⇑ Corresponding author. E-mail address: [email protected] (Y. Zhou). http://dx.doi.org/10.1016/j.ijheatmasstransfer.2016.04.028 0017-9310/Ó 2016 Published by Elsevier Ltd.
breaker to water. Power et al. [13] and Haak et al. [14] reported experimental studies on high-density mercury vapor jet pumps in 1958 and 1959, respectively. Studies on ejector refrigeration cycles and ejector-absorption cycles encounter a variety of driving fluids such as vaporized halocarbon compound (CFC, HCFCs and HFCs), water, ammonia–water solution, and LiBr–water solution. However, thermodynamic properties (enthalpy, entropy, specific heat, etc.) of the driving fluids which have significant influence on the refrigeration performance are of general interest while others are usually neglected [2,4]. Transport properties (density, viscosity and vapor pressure, etc.) have strong effects on the momentum transfer and mixing process of the ejector and should be considered as prior factors when designing an ejector system or choosing operating parameters. Besides, by simply replacing conventional driving fluids with candidates which offer more desirable transport properties may significantly improve the ejecting performance. Theoretically, driving fluids with high density, low viscosity and low vapor pressure would be ideal candidates. Liquid metal and inorganic salt solution represent two common types of high density, low vapor pressure fluids. Novel functional material GaInSn
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
21
(eutectic gallium–indium–tin) is a nontoxic room-temperature liquid metal which offers many advantages such as ultra-high density and ultra-low vapor pressure while still exhibits good fluidity [15]. Recent years have witnessed its emergence in many unique applications [16–19]. Adding inorganic salt to water is an easy way to create a solution with higher density and lower vapor pressure. In this study, three kinds of driving fluids, namely liquid metal, aqueous solution and water were firstly prepared with their transport properties measured for comparison purpose. The details related to GaInSn handling and oxidation prevention have been described in our previous work [20]. Excellent solubility, good temperature stability and material compatibility are the main reasons for us to choose NaI aqueous solution among all the other inorganic salt solutions that have been tested. The performance of the ejector system working with different driving fluids is later compared in order to study the influence of driving fluid properties on the system performance.
2. Experimental setups and procedures The properties of the driving fluids vary with temperature change. So after the driving fluids are prepared, separate experiments are carried out to measure their saturated pressure and viscosity under different temperatures. The density of each fluid may also change from temperature to temperature. Since temperatureinduced density change is relatively small compared to the density difference between different kinds of fluids, this kind of density change is not considered in this study. As shown in Fig. 1(a), for saturated pressure measurement, a partially filled stainless steel sample fluid container is firstly vacuumed by an auxiliary vacuum pump. The check valve is turned off thereafter so that the container becomes airtight. Then the container is put into a water bath for temperature control. Capillary viscometer is used for viscosity measurement. The setups are illustrated in Fig. 1(b) and the method can be found in Ref. [21]. The system arrangements for ejector performance study are shown in Fig. 2. The driving fluid is firstly pumped to a high pumping pressure before it enters the ejector where it draws in nitrogen from the reservoir. Then the two flows interact with each other inside the ejector where they eventually become a liquid–gas mixture. The mixture is later separated due to gravitational difference in the separator. Nitrogen is expelled to the atmosphere while the driving fluid flows back to the pump to complete the ejector loop. Since the driving fluid properties are temperature dependent,
Fig. 2. Schematic drawing of the test rig (not to scale).
a water coil is installed inside the separator for temperature control. The rotation speed x and electric power input EP of the pump can be controlled with an adjustable controller. The absolute pumping pressure PP of the driving fluid at the ejector entrance is measured by pressure sensor P1 (Model number: PX429S15150A5V, Omega Engineering Inc. Accuracy: ±0.08%). The absolute vacuum pressure inside the reservoir PV and the absolute discharge pressure at the separator PD are measured by pressure senor P2 (Model number: PX429-005A5V, Omega Engineering Inc. Accuracy: ±0.08% and P3 (Model number: PX429-100A5V, Omega Engineering Inc. Accuracy: ±0.08%), respectively. The mass flow rate of the driving fluid G1 is measured by Flowmeter 1 (Model number: DMF-1-3-A, Beijing Sincerity Automatic Equipment Co., Ltd. Accuracy: ±0.2%) and the volume flow rate (standard state) of nitrogen entrainment G2 is measured by Flowmeter 2 (Model number: D0719BM, Beijing Sevenstar Electronics Co., Ltd. Accuracy: ±1%). The temperature of the driving fluid is measured by a sheathed thermocouple (Accuracy: ±0.5 °C) mounted inside the separator. The vacuum pumping capacity, nitrogen entrainment as well as the pump power consumption are the main parameters that are
Fig. 1. Schematic drawing of the experimental setups for (a) saturated pressure measurement; (b) kinematic viscosity measurement. 1-Water bath, 2-sample fluid container, 3-thermal couple, 4-pressure sensor, 5-check valve, 6-to vacuum pump, 7-thermometer, 8-capillary viscometer.
22
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
used to evaluate the system performance. The no-entrainment vacuum pressure is considered as the ultimate vacuum pumping capacity of the ejector under each working condition. To measure the no-entrainment vacuum pressure, the reservoir is precharged by the nitrogen source with check valve V1C open and V2C closed. Check valve V2C is turned off later to cut off nitrogen supply. When the ejector loop gets started, V1C is turned on to expel the nitrogen from the reservoir. After a certain period of time, the reservoir will reach a steady no-entrainment vacuum pressure. The entrainment of the system is measured by another procedure. In this case, V2C remains open throughout the experiment. The pressure of the nitrogen supply at the inlet of Flowmeter 2 is maintained as a constant value by metering valve V2M. Another metering valve V1M is installed between the reservoir and Flowmeter 2 to control the vacuum pressure inside the reservoir. The entrainment at an arbitrary vacuum pressure PV is taken to be equal to the nitrogen flow rate G2 measured by Flowmeter 2. Based on the system arrangement, system parts including the ejector, separator, reservoir as well as their connections are designed using SolidWorks software. Fig. 3(a) shows the assembled
Computer Aided Design (CAD) system prototype. Transparent PMMA (polymethyl methacrylate) is used to construct the ejector parts, separator and the reservoir for flow visualization. The capacity of the reservoir and separator is 800 mL and 900 mL, respectively. The connecting tubes, water coil, wetting part of the valves and sensors are all made of stainless steel since GaInSn alloy is incompatible with materials like aluminum and copper. The ejector is a crucial part of the system and its structure and parameter details are shown in Fig. 3(b). Given the liquid-driving nature of the ejector, a converging-type ejector nozzle is chosen. Also considering the existence of the highest system velocity at the ejector nozzle, an arc-transitional channel is fabricated to reduce flow loss. The nozzle is also designed to ensure that the tangent line at the nozzle exit is horizontal. At the mixing section, a constant area mixing model is used. The mixing section is 6 mm in diameter with respect to the nozzle throat diameter of 2 mm. The mechanical pump used in this study is a volumetric rotary vane pump. Its structure and working mechanism are demonstrated in Fig. 3(c). The wetting parts of the pump are either made of stainless steel or graphite, both materials have good compatibility with all the three kinds of driving fluids.
Fig. 3. (a) CAD system model: 1-pump; 2-controller; 3-ejector; 4-separator; 5-reservoir; 6-nitrogen source; 7-water coil; 8,9-flowmeters; 10,11-check valves, 12,13-metering valves, 14,15,16-pressure sensor; 17-thermometer; (b) structure and parameter details (in millimeter) of the ejector; (c) structure of the pump head.
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
23
Table 1 Comparison on driving fluids properties [15]. Property 3
Density (kg/m ) Melting point (°C) Boiling point (°C)
Ga67In20.5Sn12.5
NaI (aq)
Water
6360 10.5 >1300
1530 <0 >100
1000 0 100
3. Experimental results The comparisons of density and phase-change temperature of the three driving fluids can be found in Table 1 and properties such as saturated pressure and viscosities are presented in Fig. 4. The dash lines represent data of water adapted from the NIST database [22]. The measured data points of water demonstrate good fitting relation with the reference data which indicate the reliability of the measurement. The saturated pressure of both water and NaI solution shows an increase as temperature rises. The difference is that the pressure value as well as the pressure increment of NaI solution is lower than that of water. Since the vapor pressure of the liquid metal is below 106 Pa even though it is over 600 °C [15], it’s safe to neglect its effect in our experiment. Different from the saturated pressure evolution, the viscosities of the fluids show inverse trends as temperature rises. The measured kinematic viscosity v of NaI solution is slightly smaller than that of water. And the liquid metal shows a much smaller kinematic viscosity and notably, it is also found to be more stable when temperature changes. When comparing the dynamic viscosity l which takes fluid density into consideration, the liquid metal shows a much higher value. As can be expected, blending solute in water increases its density and dynamic viscosity, while it reduces the vapor pressure of the solvent. Fig. 5 gives a comparison of the no-entrainment vacuum pressure inside the reservoir under different working conditions. The temperature of the driving fluid for each case is maintained at a constant value of 15 °C. The ejector system working with water reaches a vacuum level of 10 kPa with a power consumption of about 120 W. When we use NaI solution to replace water, lower no-entrainment vacuum pressure is obtained and the vacuum level drops to 1 kPa level while the power consumption rises to 170 W. The liquid metal driving case shows a significant improvement. The vacuum level achieved under liquid metal driving condition is much higher than the other two cases, and better, it consumes the lowest amount of electric power. A vacuum pressure of 33 Pa, which is the lowest vacuum pressure that has ever been reported in the open literature achieved by a single-stage ejector, is detected with about 50 W power input. The no-entrainment vacuum pressure for each case under constant power input and different temperature is also tested and the result is presented in Fig. 6. For water and NaI solution driving cases, when temperature increases form 15 °C to 45 °C, the no-entrainment vacuum pressure rises gradually. The water driving case experiences a 7.3 kPa vacuum pressure increment while for the NaI driving case, a smaller value of 5.8 kPa is detected. The temperature-induced no-entrainment vacuum pressure rise and the pressure–temperature saturation relationship are closely matched when we compare Figs. 6 with 4. So we conclude that the temperature-induced vapor pressure rise of the driving fluid is the main reason for the no-entrainment vacuum pressure rise as the temperature increases. This also explains why liquid metal driving case shows such a small vacuum fluctuation. The ultra-low vapor pressure nature of the liquid metal makes the no-entrainment vacuum pressure not sensitive to temperature variation. The fluctuation detected in the experiment is contributed to other reasons such as flow instability and measurement error. Conclusion can also be made that mercury, a well-known
Fig. 4. Comparison of saturated pressure and viscosities of the different driving fluid under different temperature.
Fig. 5. No-entrainment vacuum pressure of different driving fluids as a function of pump power input. Icons: experimental data points; dash lines: fitting curves.
liquid metal which has higher density (13,600 kg/m3) than GaInSn alloy, may be less competitive in high vacuum ejecting because it has a much higher vapor pressure than GaInSn alloy.
24
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
Fig. 6. The influence of temperature on the non-entrainment vacuum pressure of the ejector.
The nitrogen entrainment of different fluids driving conditions are measured to further study the pumping characteristics of the ejector system. Comparisons are made under the same electric power input of about 60 W. The main working parameters of the ejector system are listed in Table 2. It shows that driving fluid with higher density requires higher pumping pressure and usually has a higher mass flow rate, while driving fluid with lower density obtains a higher volume flow rate and a higher nozzle velocity. Due to its high density, the typical nozzle throat velocity of the liquid metal driving case drops to an incredibly low value of 6.5 m/s. When it comes to the entrainment ratio which also represents an important performance criterion of the ejector, the volume entrainment ratio, rather than mass entrainment ratio, is generally used for liquid–gas ejectors given the great density difference between the driving fluid and the entrained fluid. As shown in Fig. 7, both the nitrogen entrainment and the volume entrainment ratio show a similar trend. In the vacuum pressure range above 40 kPa, the entrainment of NaI solution and water driving cases surpass the liquid metal driving case. But when the vacuum pressure of the reservoir drops below 40 kPa, the entrainment of NaI and water driving cases decrease rapidly to zero. On the contrary, liquid metal driving case shows its advantage in this low vacuum pressure range. The reasons for these different behaviors will be explained in Section 4. A more detailed plot of vacuum pumping performance of the liquid metal driving case against operating time can be found in Fig. 8. As has been shown, by adjusting the electric power input of the pump, different vacuum levels can be obtained. And the ejector system works well as a rough vacuum system with tens of watts power consumption. One major advantage of the liquidring vacuum system is that no lubrication or sealing oil is required so it is desirable for clean vacuum pumping. A commercially available mechanical vacuum pump is also used to compare with the current ejector system. The mechanical vacuum pump can produce
Table 2 Comparison on working parameters under 60 W electric power input. Working parameter
Ga67In20.5Sn12.5
NaI (aq)
Water
Mass flow rate (kg/min) Volume flow rate (L/min) Pumping pressure (Pa) Nozzle throat velocity (m/s)
7.83 1.23 4.87 105 6.5
6.08 3.97 2.59 105 21.1
4.97 4.97 2.52 105 26.4
Fig. 7. Comparison of nitrogen entrainment under the electric power input of about 60 W and driving fluid temperature of 20 °C. Icons: experimental data points of nitrogen entrainment (hollow) and volume entrainment ratio (filled); dash lines: fitting curves.
Fig. 8. Vacuum pressure development of the reservoir as a function of time under different electric power input. Icons: experimental data points; dash lines: fitting curves.
a comparable vacuum pressure of about 50 Pa only with a much higher electronic power consumption of about 150 W. Though, the liquid metal driving ejector has a drawback that its driving fluid, GaInSn alloy, will be easily oxidized when it contacts with oxygen. This problem will not be a concern if it is used for nonoxidative medium pumping. And if generated, alkaline solution can be applied to efficiently remove the oxides. 4. Discussions The ejector apparatus is an assembly of several variable crosssection parts. The flow patterns of a typical cross-section flow is shown in Fig. 9. The flow regime of the cross-section flow is governed by the equations of continuity, momentum and energy. The change in flow area of the flow channel will result in changes of flow parameters such as flow velocity and static pressure which indicate flow work converting between kinetic energy and potential energy. The process will also be accompanied with flow work losses which mainly manifest as friction loss and local dissipation. The friction loss is a result of fluid viscosity and wetting surface
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
would lead to estimations with large derivation. Consequently, conclusions can be made that under normal operations, the performance of liquid ejector can be well predicted by laws of fluid mechanics while under extreme conditions (i.e., high temperature, high vacuum), thermodynamic principle may become dominant. The theory that the ultimate vacuum pressure of the ejector is subject to the saturated pressure of the driving fluid with respect to its working temperature is consistent with the experimental result that NaI solution driving case obtains a lower no-entrainment vacuum pressure than that of water driving case. It has also been demonstrated that owing to the inherent nature of ultra-low vapor pressure, liquid metal provides an exception for those who want to break the thermodynamic restriction. A simple mathematical model can be developed to describe the influence of pressure gradient and volume flow rate of the driving fluid on the ejector’s entrainment behavior. Since the driving fluid and the entrained fluid are immiscible, the driving fluid and the entrained gas should both satisfy the law of continuity. Let’s recall Fig. 3(b) and consider the flow conditions at the mixing section and the diffuser exit. The continuity equation gives:
Fig. 9. Flow patterns and loss mechanisms of cross-section flow.
roughness while the local dissipation is the energy loss consumed by processes such as vortex evolution. Bernoulli’s equation can be applied to describe the flow patterns at cross-sections of the ejector:
1 1 p1 þ qu21 þ qgz1 ¼ p2 þ qu22 þ qgz2 þ DW 1—2 2 2
25
ð1Þ
where the subscripts 1 and 2 are locations indicated in Fig. 9; p is the static pressure of the driving fluid; q is the density; u is the velocity; g is the gravitational acceleration, z is the height, DW 1—2 ¼ f ðq; l; u; D; L; dÞ is the total flow work loss as a function of q, u, and the driving fluid dynamic viscosity l, the flow channel diameter D, the flow distance length L, as well as the surface roughness d. Complexities will arise when two fluids interact with each other. The flow patterns inside the ejector arising from the interaction of the driving fluid and the entrained fluid exhibits highly unstable and complex two-phase flow characteristics, though, qualitative analysis of the density and viscosity influence on ejector mass transfer process can be made according to Eq. (1). Density q determines the specific kinetic energy of the carrier fluid. For a horizontally arranged ejector as in the current case, it implies that fluid with higher density requires smaller change in velocity to produce the same amount of static pressure drop. When flow velocity decreases, the required pumping speed is also reduced accordingly, which is favorable for low power consumption. This explains why the liquid metal driving case has the smallest power consumption. Ejector system frequently suffers from low efficiency, components wear and noise emitting under high velocity operation. Therefore, reducing operating velocity by using high density driving fluid could hopefully solve these problems while still meet the performance requirements. The influence of driving fluid viscosity on ejector system usually manifests as friction loss and wear mechanism inside the system. So driving fluid should also have an acceptable viscosity to minimize flow work loss and component wear. The static pressure drop at the outlet of the ejector nozzle produces the vacuum pumping effect. The vacuum pressure is mainly controlled by the operating parameters based on specific structural design. But for conditions that the vacuum pressure reaches the saturated pressure level, the influence of vapor evaporation arises. This is because when the pressure is comparable to the vapor pressure of the driving fluid, the vaporization of the driving fluid itself would be significant and thus prevent higher vacuum ejecting. Under such circumstances, simply applying Bernoulli’s equation
_ d ðlÞ and m _ m ðgÞ ¼ m _ d ðgÞ _ m ðlÞ ¼ m m
ð2Þ
_ _ where mðlÞ and mðgÞ denote the mass flow rate of driving fluid and entrained gas, respectively. The subscripts m and d represent the flow positions of the mixing section entrance and the diffuser exit, respectively. As for the volume flow rate, since the driving fluid is incompressible, its volume flow rate will remain as a constant. Though no phase transition will take place for nitrogen, the volume flow rate of the entrained gas will be influenced by its local pressure and temperature. So we have:
v_ m ðlÞ ¼ v_ d ðlÞ and v_ m ðgÞ ¼ f ðP; TÞv_ d ðgÞ
ð3Þ
where v_ ðlÞ and v_ ðgÞ denote the volume flow rate of driving fluid and entrained gas, respectively. By introducing the gas state equation _ where R is the gas constant, T is the temperaPv_ ðgÞ ¼ mðgÞRT=M, ture, and M is the molar mass of the gas, we get:
v_ m ðgÞ ¼
PD v_ d ðgÞ ¼ nc v_ d ðgÞ PV
ð4Þ
where nc ¼ PD =PV is the compression ratio of the ejector. Since _ _ n ðlÞ and the temperature of the fluid is maintained as a mðgÞ m constant in the present case, the temperature change of the entrained gas is neglected in Eq. (4). For a qualitative comparison, the ideal gas model is used, and note that for more accurate calculation, there are real gas models available. According to Eq. (4), high compression ratio will lead to a high volume entrainment at the suction chamber. However, the volume flow rate of the entrained gas will be constrained by the structural parameters of the flow channel. Further assumptions can be made that upon mixing, the velocity change of the driving fluid is small while the entrained gas can still obtain the same velocity as the driving fluid based on the law of momentum conservation. Thus we have:
v_ m ðgÞ /
D2m D2n D2n
v_ m ðlÞ
ð5Þ
where Dm and Dn are the diameters of the mixing section and nozzle exit, respectively. Since the standard state volume flow rate of the nitrogen measured by G2 is close to the working condition at the diffuser exit, the volume entrainment ratio of the ejector kappav has the form ofv_ d ðgÞ ¼ jv v_ d ðlÞ, so the combination of Eqs. (4) and (5) gives:
jv /
D2m D2n nc D2n
ð6Þ
26
J. Tang et al. / International Journal of Heat and Mass Transfer 101 (2016) 20–26
Eqs. (5) and (6) show the relationship between the volume entrainment and established structural parameters as well as measureable working condition parameters. Eq. (5) indicates that the gas entrainment is proportional to the volume flow rate of the driving fluid. From a microscopic perspective, the mass transfer process is subject to the collisions between the two fluid particles especially in high vacuum conditions. So driving fluid with larger contacting surface, i.e., larger volume flow rate, will increase the colliding opportunities which result in larger entrainment. Eq. (6) implies that the gas entrainment is proportional to the inverse of the compression ratio which means high compression ratio may cause a decline to the gas entrainment. The decline in volume flow rate of liquid metal driving case comparing with the other two cases is contributed to both of the above-mentioned reasons. The entrainment ratio jv is a parameter that is influenced by structural design, working conditions as well as the nature of the driving fluid. In general, it reflects the combined effect of its vacuum capacity (vacuum pressure) and transportation capacity (volume flow rate) of the ejector system. For high vacuum pumping cases (the liquid metal driving case), the latter is the entrainment-determine factor while for low vacuum pumping cases (NaI solution and water driving cases), the former becomes dominant. As has been shown, NaI solution driving cases have a better vacuum capacity than water driving cases, so it is reasonable for the NaI solution driving cases to produce higher entrainment ratio. 5. Conclusions A liquid–gas ejector vacuum system is designed and a series of experiments are conducted to investigate the influence of fluid properties on its mass transfer performance. Liquid metal, NaI solution and water are prepared with their properties characterized for experimental investigation. The system manages to work under different conditions and the results show that a driving fluid with high density, low viscosity and low vapor pressure is favorable for high vacuum ejecting. The entrainment of the ejector is contributed to both the compression ratio and volume flow rate of the driving fluid. The room-temperature liquid metal actuated ejector has demonstrated its potential for oil-free vacuum pumping as well as high temperature operation. But when entrainment is the major concern, low density driving fluids such as aqueous solution or water are recommended. This study has disclosed a method to improve ejector performance from the perspective of driving fluid properties and its applications are predictable especially when structural modifications are restrained. Acknowledgment This study is partially supported by the National Natural Science Foundation of China (Grant No. 51176198).
References [1] H. Bannwarth, Liquid Ring Vacuum Pumps, Compressors and Systems, Conventional and Hermetic Design Translated by Christine Ahner, WILEYVCH Verlag GmbH & Co. KGaA, Weinheim, 2005. [2] Kanjanapon Chunnanond, Satha Aphornratana, Ejectors: applications in refrigeration technology, Renewable Sustainable Energy Rev. 8 (2004) 129– 155. [3] Stefan Elbel, Predrag Hrnjak, Ejector Refrigeration: An Overview of Historical and Present Developments with an Emphasis on Air-Conditioning Applications, in: Int. Refrig. Air Conditioning Conf. Purdue, 2008, pp. 14–17. Paper 884. [4] S.B. Riffat, L. Jiang, G. Gan, Recent development in ejector technology – a review, Int. J. Ambient Energy 26 (1) (2005) 13–26, http://dx.doi.org/10.1080/ 01430750.2005.9674967. [5] B. Gourich, N. El Azher, C. Vial, M. Belhaj Soulami, M. Ziyad, A. Zoulalian, Influence of operating conditions and design parameters on hydrodynamics and mass transfer in an emulsion loop–venturi reactor, Chem. Eng. Process. 46 (2007) 139–149. [6] S. Balamurugan, Mayank D. Lad, Vilas G. Gaikar, Ashwin W. Patwardhan, Hydrodynamics and mass transfer characteristics of gas–liquid ejectors, Chem. Eng. J. 131 (2007) 83–103. [7] P.H.M.R. Cramers, A.A.C.M. Beenackers, Influence of the ejector configuration, scale and the gas density on the mass transfer characteristics of gas–liquid ejectors, Chem. Eng. J. 82 (2001) 131–141. [8] G.M. de Billerbeck, J.S. Condoret, C. Fonade, Study of mass transfer in a novel gas–liquid contactor: the aero-ejector, Chem. Eng. J. 72 (1999) 185–193. [9] Yoon Kee Kim, Dong Yeop Lee, Hyun Dong Kim, Joo Ha Ahn, Kyung Chun Kim, An experimental and numerical study on hydrodynamic characteristics of horizontal annular type water–air ejector, J. Mech. Sci. Technol. 26 (9) (2012) 2773–2781, http://dx.doi.org/10.1007/s12206-012-0724-9. [10] Hyo Min Jeong, Han Shik Chung, Kang Youl Bae, Se Hyun Kim, You Sik Shin, Water cooling characteristics in an enclosed vacuum tank by water driven ejector, J. Mech. Sci. Technol. 19 (1) (2005) 164–172. [11] M.D. Butterworth, T.J. Sheer, High-pressure water as the driving fluid in an ejector refrigeration system, Appl. Therm. Eng. 27 (2007) 2145–2152. [12] A. Elgozali, V. Linek, M. Fialov, O. Wein, J. Zahradn, Influence of viscosity and surface tension on performance of gas–liquid contactors with ejector type gas distributor, Chem. Eng. Sci. 57 (2002) 2987–2994. [13] B.D. Power, N.T.M. Dennis, An unusual mercury ejector pump, in: E. Thomas (Ed.), Fundamental Problems in Vacuum Techniques Ultra-High Vacuum: Proceedings of the First International Congress on Vacuum Techniques, Namur, Belgium, June 10–13 1958, Pergamon Press, 1960, pp. 197–201. [14] Werner H. Haak, Frederick Vrátny´, Annular Jet Pump, Rev. Sci. Instrum. 30 (1959) 296, http://dx.doi.org/10.1063/1.1716609. [15] N.B. Morley, J. Burris, L.C. Cadwallader, M.D. Nornberg, GaInSn usage in the research laboratory, Rev. Sci. Instrum. 79 (2008) 056107, http://dx.doi.org/ 10.1063/1.2930813. [16] Gökhan Perçin, Laurent Levin, Butrus T. Khuri-Yakub, Piezoelectrically actuated droplet ejector, Rev. Sci. Instrum. 68 (1997) 4561, http://dx.doi.org/ 10.1063/1.1148431. [17] A. Miner, U. Ghoshal, Cooling of high-power-density microdevices using liquid metal coolants, Appl. Phys. Lett. 85 (2004) 506, http://dx.doi.org/10.1063/ 1.1772862. [18] I. Silverman, A.L. Yarin, S.N. Reznik, A. Arenshtam, D. Kijet, A. Nagler, High heat-flux accelerator targets: cooling with liquid metal jet impingement, Int. J. Heat Mass Transfer 49 (2006) 2782–2792. [19] Peiper Li, Jing Liu, Harvesting low grade heat to generate electricity with thermosyphon effect of room temperature liquid metal, Appl. Phys. Lett. 99 (2011) 094106, http://dx.doi.org/10.1063/1.3635393. [20] Jianbo Tang, Yuan Zhou, Jing Liu, Junjie Wang, Wenxiu Zhu, Liquid metal actuated ejector vacuum system, Appl. Phys. Lett. 106 (2015) 031901, http:// dx.doi.org/10.1063/1.4906098. [21] GB/T 10247-2008, Methods of Viscosity Measurement, 2008. [22] NIST Standard Reference Database 23, Version 8.0, The U.S. Secretary of Commerce, USA, 2007.