Measurement and Visualization of R410A Distribution in the Vertical Header of the Microchannel Heat Exchanger

Transcription

1 Purdue Universit Purdue e-pubs International Refrigeration and Air Conditioning Conference School of Mechanical Engineering 2012 Measurement and Visualization of R410A Distribution in the Vertical Header of the Microchannel Heat Exchanger Yang Zou Predrag S. Hrnjak llo this and additional orks at: Zou, Yang and Hrnjak, Predrag S., "Measurement and Visualization of R410A Distribution in the Vertical Header of the Microchannel Heat Exchanger" (2012). International Refrigeration and Air Conditioning Conference. Paper This document has been made available through Purdue e-pubs, a service of the Purdue Universit Libraries. Please contact epubs@purdue.edu for additional information. Complete proceedings ma be acquired in print and on CD-RM directl from the Ra W. Herrick Laboratories at Herrick/Events/orderlit.html

2 2465, Page 1 Yang ZU, Pega HRNJAK* Measurement and Visualization of R410A Distribution in the Vertical Header of the Microchannel Heat Exchanger Universit of Illinois at Urbana-Champaign, Department of Mechanical Science and Engineering, Urbana, IL, USA Contact Information ( angzou1@illinois.edu, pega@illinois.edu) * Corresponding Author ABSTRACT This paper presents the refrigerant adiabatic upard flo in the vertical header of microchannel heat exchanger and its effect on distribution. R410A is circulated into the header through the microchannel tubes (5 or 10 tubes) in the bottom pass and exits through tubes (5 or 10 tubes) in the top pass representing flo in the heat pump mode of reversible sstems. Three circular headers ere explored, each ith the microchannel tubes inserted to half depth. The qualit as tpicall varied from 0.2 to 0.8. Mass flo rate as from 1.5 to 4.5 kg/h per microchannel. The best distribution is found at high flo rate and lo qualit. Distribution is improved b doubling the number of microchannel tubes although elongation of the header has negative effect. Visualization reveals the effects of flo patterns in terms of homogeneit and liquid momentum. Refrigerant in the churn flo has better distribution than in the separated flo since the to-phase mixture is more homogeneous. The distribution is better at high mass flux in the header because the higher momentum liquid can be supplied to the top exit tubes. 1. INTRDUCTIN Microchannel heat exchangers have come to the frontier of stationar air conditioning and refrigeration application after its success in mobile air conditioning sstems, for their advantages in compactness and higher air-side heat transfer coefficient. In the reversible sstems, microchannel heat exchanger functions as both evaporator and condenser. The outdoor heat exchanger, tpicall a multipass microchannel heat exchanger, has vertical header to reduce cost and for easier bending hen U shape condenser is needed. The outdoor heat exchanger functions as condenser in the cooling mode and the overall flo is donard. In the heating mode, it functions as evaporator and refrigerant flos in the reversed direction. Then, refrigerant maldistribution due to to-phase flo in the vertical header deteriorates the heat exchanger s performance, and consequentl reduces sstem efficienc. Good design that provides reasonable distribution in heat pump mode is still a challenge. Several researchers have reported the distribution problems, mostl related to horizontal header ith vertical tubes. Hrnjak (2004) and Webb and Chung (2005) have revieed the distribution problems. Combined ith others results, it can be conceived that maldistribution is a ver complex problem that is affected b numerous parameters: header geometr and orientation, fluid properties, iet mass flux and qualit, etc. In their studies of horizontal header, Fei and Hrnjak (2004), Vist and Pettersen (2004), Boers et al. (2006) and Hang et al. (2007) shoed that the flo pattern in the iet tube affected the flo regimes in the header. Such flo regimes had a strong influence on liquid distribution in the to-phase flo. In the studies of vertical header, Watanabe et al. (1995) investigated the distribution of R11 in a vertical header ith five round branch tubes. Along the flo direction in the header, the flo pattern transited from annular to slug or froth flo. Song et al. (2002) manifested maldistribution of C2 in the outdoor microchannel coil in heating mode through frosting patterns. The related distribution in the header ith the balance beteen inertial, gravitational and shear forces. Some ork of Cho and Cho (2004) as devoted to distribution of R22 in the vertical header ith microchannel tubes. It as found that most liquid ere in the bottom

3 2465, Page 2 because of gravit regardless of the iet tpes. The visualization of air and ater upard flo in the vertical header from Lee (2009) shoed that three regions ere formed. The flo patterns in these three regions ere different. The had great impact on the distribution. Bun and Kim (2011) tested and visualized the distribution of R410A in both the iet and second pass headers at one qualit condition. r the iet header, a pool as formed at the bottom. r the second pass header, a liquid film is formed in the header due to to-phase jet iet. It is noteorth that the most of the tests ere done in lo qualities, i.e. xin < 0.5, because the iet header as usuall studied. Hoever, tpical microchannel outdoor heat exchanger has at least to passes. To-phase refrigerant is redistributed in the intermediate header in hich case distribution phenomenon is more complex but important. Thus, in this paper, refrigerant distribution in an intermediate header for a multipass heat exchanger (ith capacit of kw) as studied. The refrigerant to the header as provided b multi-parallel microchannel tubes from the iet header. 2. EXPERIMENTAL METHD The test loop as constructed to model R410A distribution in the microchannel heat exchanger, as shon Figure 1. The liquid refrigerant as pumped into the iet header hile the iet mass flo rate min as measured b mass flo meter (Micromotion D40, ±0.75%). The pump speed as controlled b a bpass valve and VFD. The temperature Tsub (immersed thermocouple, ±0.5 oc) and pressure Psub (Sensotec TJE, ±1% FS) at the iet to the first header ere measured to determine the subcooling. The subcooled liquid as assumed to distribute eve into microchannel tubes in the bottom pass, here the refrigerant as heated to the desired qualit hile the heaters are insulated. Set of six heaters per microchannel tube ere providing uniforml of heat suppl. The poer needed to determine iet qualit as measured b att transducer (hio Semitronics GW5-024CX5, ±0.2% FS). After the refrigerant entered into the loer part of the header, it turned 90o, floed upard and reached the upper part of the header. Due to maldistribution, different liquid amount exited through the microchannel tubes in the top pass. In these tubes, the refrigerant as heated again to provide equal superheat at the exit. Each tube as heated b six heaters and insulated. Each set of six heaters as individuall controlled to provide adequate poer to generate equal superheat, and as also measured b att transducer (hio Semitronics GW5-024CX5, ±0.2% FS) separatel to determine iet qualit into each exit tube. The temperature Tsup,i (immersed thermocouple, ±0.5 oc) in each tube and pressure Psup (Sensotec TJE, ±1% FS) ere measured to determine the superheating. The mass flo rate of single phase superheated vapor as measured individuall as the total mass flo rate for each microchannel tube b mass flo meter (Micromotion D06, ±0.15%). The vapor as then brought to the condenser. With the help of the receiver and subcooler, the subcooled liquid as provided to the pump. Figure 1: Sstem schematics

4 2465, Page 3 Enthalp at the superheated point isup,i as determined b Tsup,i and Psup. The outlet enthalp from the header or iet to each exit tube iout,i as then calculated as Q& out,i m& out,i (1) iout,i = isup,i The pressure in the header Pave is estimated as the average of subcooled and superheating pressures. The outlet qualit from the header xout,i can be found from iout,i and Pave. Then, the liquid mass flo rate in each tube can be obtained as m& l,out,i = m& out,i (1 xout,i ) (2) A high speed camera, Phantom v4.2, is used for visualizing the flo patterns in the transparent header. The exposure time of the camera as set from 80 μsec to 100 μsec. The framing rate as at 2200 frames per second. The resolution as set as 512x512 or 256x512 pixels. Three circular headers ere examined, each ith the microchannels inserted to half depth: 1) the aluminum header ith five iet and five exit microchannel tubes as used in industr (5+5 aluminum header), 2) the transparent replica of the first (5+5 transparent header), and 3) transparent header ith ten iet and ten exit microchannel tubes (10+10 transparent header) to stud the effect of increasing microchannel number and header length. The transparent ones ere made of PVC tube. The gap beteen microchannel tube and PVC tube as sealed b special epox (JBWELD). It is the black material on the right part of the header in the images. In addition, an epox block as made to increase the pressure tolerance hile still ensure the transparenc outside the PVC tube. The inner diameter of the aluminum header is mm, and that of the transparent header is mm. The tube pitch is 13 mm. The 5+5 header is 170 mm long, and the header is 300mm long. Each microchannel tube has 17 rectangular channels. The hdraulic diameter of the microchannel is calculated based on the manufacturer data to be 0.5 mm. To examine the case of ten iet and ten exit tubes using the same header diameter (different circuiting), modification of the facilit as made ith intention to maintain the same number of mass flo meters (five) and the same heaters and control strateg. Decision is made to combine to neighboring exit tubes in one ith a Y connection, assuming that ill affect resolution of the distribution parameters o, not the essence of the results. The refrigerant from the neighboring to branches as heated together to be superheated. Therefore, o five liquid flo rate results ere obtained, hereas actuall each one represented the liquid amount in the neighboring to branches. As shon in Figure 1, in order to facilitate the connection of the header, o five iet microchannel tubes ere connected and alloed refrigerant to pass through implicitl assuming insignificant effect of iet difference. The other five iet microchannel tubes ere blocked and just served as obstruction in the header. The average saturation temperature as about 5 oc. The iet qualit as from 0.2 to 0.8. The results of xin = 0 and 1 ere taken to provide to asmptotes, and those of xin = 0.95 ere to offer more tpical situation at high qualit. The mass flo rate as from 1.5 to 4.5 kg/h for each microchannel. Thus, the mass flo rate in the middle of the header as from 2.14 to 6.25 g/s for 5+5 header and 4.28 to g/s for header, hile the mass flux in the header from to kg/m2-s for 5+5 headers and to kg/m2-s for header. Tests at 7.27 g/s (38.82 kg/m2-s) for 5+5 header ere taken in attempt to reach uniform distribution. Tests at 2.22 g/s (11.86 kg/m2-s) for header ere taken to compare the results ith those of 5+5 header. Based on Boers et al. (2006), the mass flux defined b the smallest cross sectional area in the header is more representative hen the protrusion is present, so the maximum mass flux in the header, Gin, as shon in Figure 1, ill be used in the folloing analsis. And it as from to kg/m2-s for 5+5 headers and from to kg/m2-s for header. To metrics are used to evaluate the distribution. The first one is coefficient of variation σ. It is the dimensioess standard deviation, hich is defined as σ= 1 ml 1 n (m& l,out,i ml ) 2 n 1 (3)

5 2465, Page 4 LFi = m& l,out,i (4) n Uniform distribution is described b σ = 0. The orst distribution, for this case (five results), is σ = 2. The advantage of this metrics is that one number characterizes goodness of distribution. Hoever, in order to illustrate the distribution profile, it is necessar to appl the liquid fraction, m& l,out,i i r the case of five results, uniform distribution is described b LFi = 0.2 in ever tube. The uncertaint of measuring σ is ithin 4.5%. The absolute uncertaint of liquid fraction is from to RESULTS AND DISCUSSIN 3.1 Liquid fraction and coefficient of variation Figure 2 to Figure 4 sho the distribution profiles and the coefficient of variation of 5+5 aluminum (Figure 2) and transparent (Figure 3) headers and transparent header (Figure 4). The darkness of the bar color represents different branches (tubes), the pale being loest exit branch and the dark being highest exit branch. The best distribution indicated b σ is at xin = 0.2 (the loest) and the highest iet mass flux among the test conditions. The results of 5+5 aluminum and transparent headers are ver similar (See Figure 2 and 3). The small discrepanc ma be due to different internal areas. In Figure 2 (a) and 3, at fixed qualit, the liquid fraction of the top branch is higher as mass flo rate increases and at xin = 0.6, 0.8 and 0.95, the liquid fractions of the bottom branches reduce at the same time, hereas at xin=0.2 and 0.4, the liquid fractions of the bottom branches do not change much. At fixed mass flo rate, the liquid fraction of the highest branch reduces as the qualit increases. Meanhile, for the other branches, as qualit increases, 1) at min = 2.14 and 3.17 g/s, the bottom branches have higher and higher liquid fraction; 2) at min = 4.19 g/s, more liquid leaves at the middle tubes; 3) at min = 5.22, 6.25 and 7.27 g/s, the liquid fraction of branch 4 becomes highest hile the others are less and less. Based on the values of σ in Figure 2 (b), it is seen that: 1) at fixed mass flo rate, the distribution usuall deteriorates as qualit increases; 2) at lo qualities, the distribution improves ith increasing mass flux; 3) at high qualities, the distribution is better at some intermediate mass flo rate. It gets orse as min either increases or decreases. But compared to the lo qualit cases, the distribution is alas poor. What matters is just hether header at the higher or loer part has liquid to be entrained to branches (tubes). a) Liquid fraction b) Coefficient of variation Figure 2: Distribution results of 5+5 aluminum header

6 2465, Page 5 Figure 3: Distribution results of 5+5 transparent header Figure 4: Distribution results of 5+5 transparent header In Figure 4 for header, the trend of the results is ver similar as that of 5+5 header cases: for the same mass flo rate, hen iet qualit increases, the distribution is orse; and for the same iet qualit, hen mass flo rate increases, the distribution is improved. The mass flo rate in each iet microchannel tube is consequence of design mostl affected b the air side heat transfer calculation. When the tube number in each pass is doubled, so is the mass flo rate in the middle of the header. This ould help the top tubes to receive more liquid and hence improve the distribution. Hoever, the header is longer than that for 5+5 configuration. It consequentl requires higher inertia forces to suppl the liquid to the top branch (tube). All of these factors affect the distribution profiles. Hoever, experiment shos that distribution is usuall better for the header at the same mass flo rate in one iet microchannel. Comparing the distribution profiles beteen 5+5 header (Figure 3) and header (Figure 4) at the same mass flo rate in the header, the look ver similar, so the mass flux in the header ma be an important parameter in determining the distribution.

7 2465, Page 6 The coefficient of variation of all these distribution results can be related to liquid mass flux Gl = Gin(1-xin) in the middle of the header, as shon in Figure 5. r D=15 mm, 5 or 10 iet and 5 or 10 outlet microchannel tubes, R410A, an empirical correlation can be developed as: (5) σ = Gl Gl Gl Gl This preliminar correlation can help identif possible distribution issues if σ is higher than 0.2. Figure 5: Empirical correlation of σ as a function of liquid mass flux 3.2 Visualization To flo regimes are identified from the visualization: churn and separated flo. The flo patterns are also listed in Figure 3 and 4, here C denotes churn flo hile S denotes separated flo. Churn flo tpicall appears at lo qualities, hereas separated flo usuall occurs at high qualities. Increasing qualit at the same mass flo rate ill change churn to separated flo because of higher velocities and less liquid. Such transition is also found sometimes hen increasing the mass flo rate at the same qualit, e.g. xin = 0.4 for 5+5 transparent header. In churn flo, shon in Figure 6 (a), most of the header is liquid refrigerant ith bubbles, but at the top it is almost vapor o. Because the horizontal velocit is small, hen to-phase refrigerant enters into the header, bubbles turn and flo verticall immediatel due to buoanc effects. Liquid falls, forming a liquid pool or interacting ith upard floing vapor. Bubbles stir the liquid and form a local recirculation, although the mean velocit of liquid is upard. It is hard to distinguish the vapor and liquid interface; the are mixed almost homogenousl. a) Churn flo b) Separated flo Figure 6: Flo patterns in the header

8 2465, Page 7 Hoever, Figure 6 (b) shos that in separated flo, the void fraction in the header is ver high, and thus most the volume of the header is taken b vapor phase, but liquid is present in the form of film along the inner all. Although the velocit from the iet microchannel tube is high, it is still eas for vapor to turn and flo upard. Hoever, the liquid jet at the exit from the microchannel tube flos horizontall, hits the all and forms a liquid film. High speed vapor and the liquid film flo upard along the header. In the top exiting region, vapor phase ith lighter densit is much easier to turn 90o and branch out, but liquid ith larger densit and higher momentum tends to run through the header and bpassed the first fe microchannel tubes. The liquid film is separated from the all at certain height. This location h, as shon in Figure 6 (b), is defined as the liquid separation height. Above the separation height, the flo pattern is locall like churn flo. Some liquid flos horizontall and leaves through the outlet microchannel tubes. ther liquid falls don through the gap beteen tube and round header, so that creates recirculation in the header. Finall, due to the lo momentum, liquid cannot reach the top and the tubes (branches) get ver little if an liquid. The highest liquid level H, as shon in Figure 6 (b), is defined as liquid reach. Above the liquid reach, there is almost vapor o. The liquid reach is also observed from churn flo in Figure 6 (a). From Figure 6, it can be seen that the liquid distribution is better in churn flo. It is because the opportunit of liquid suppl to reach each branch is increased hen liquid occupies most of the header. Hoever, in separated flo, in o a small part of the header, liquid is easil available in front of the entrance to the tube. Therefore, generalization of these to flo patterns can be used to evaluate the homogeneit of liquid and vapor phases in the header and mabe further the distribution. The generalization is accomplished b considering the to-phase flo in the header as the deviation of to-phase flo in vertical pipe. It is because: 1) flo is developing; 2) mass flux is changing along the header; 3) tubes protruding in the header obstruct the flo. Thus, the superficial momentum fluxes of liquid and vapor phases ρl jl 2 = (6) ρl [Gin xin ]2 (7) ρv ρ v jv 2 = [Gin (1 xin )]2 as used b Heitt and Roberts (1969) to generalize to-phase flo in vertical pipe are applied in attempt to generalize current results. The flo regime map is shon in Figure 8 here the transition line is also identified. Figure 7: Flo regime map in the vertical header

9 2465, Page 8 Hoever, evaluation of the homogeneit of liquid and vapor phases in the header is not sufficient to evaluate distribution. Liquid momentum is another important factor. ne example is seen that at xin = 0.2 from Figure 4, the top tube has no liquid at all hen iet mass flo rate is lo hich is because the liquid reach is belo the top tube, as shon in Figure 8 (a). The liquid reach is higher hen iet mass flo rate increases, thus the distribution is better. The relative liquid reach, H/L, as defined in Figure 6 (b), is related to the liquid mass flux Gl in Figure 9. r 5+5 header, it is necessar to keep Gl above 20 kg/m2-s, so that the liquid can reach the top tube. Such condition is Gl > 60 kg/m2-s for header. This is a necessar but sufficient condition for good distribution. a) Liquid reach increases as flo rate increases b) Liquid separation increases as qualit increases Figure 8: Variation of liquid reach and liquid separation as flo rate and/or qualit changes Figure 9: Generalization of relative liquid reach ith liquid mass flux

10 2465, Page 9 As shon in Figure 8 (b), the liquid reach also changes as qualit increases at the same mass flo rate. Besides, the liquid separation height in the separated flo pattern increases hen the qualit is higher. The relative liquid separation, h/l, as defined in Figure 6 (b), is related to gas mass flux Gv = Ginxin in Figure 10. When Gv is belo 20 kg/m2-s, the flo pattern is churn flo, and no microchannel tubes ill be bpassed. This is another necessar but sufficient condition for good distribution. In order to achieve good distribution, both conditions for liquid reach and liquid separation height should be satisfied. Figure 10: Generalization of relative liquid separation ith vapor mass flux 4. CNCLUSINS The conclusions of this stud are summarized belo: 1) The good distribution as found at high iet mass flo rate and lo qualit among the test conditions. The distribution is orse as the qualit increases at the same mass flo rate. At fixed lo qualities (e.g. xin 0.6), the distribution usuall improves as the mass flo rate is higher. At fixed high qualities (e.g. xin 0.8), there is usuall a better distribution case at some intermediate mass flo rate, but generall speaking the distribution is poor. 2) Due to increase in mass flux, greater number of the microchannel tubes improves the distribution although longer header requires higher momentum for liquid to reach the top. Comparing the results of header ith those of 5+5 header, mass flux in the header is an important parameter in determining the distribution. 3) A developed correlation for coefficient of variation (σ is a function of Gl) shos good agreement ith experiment. It also shos regions of good distribution. 4) Churn and separated flo in the header ere identified. The flo patterns depend upon the mass flux and qualit. Churn flo transited to separated flo as the iet qualit and/or mass flux increases. Distribution is related to the flo patterns in the header. Churn flo generates better distribution because vapor and liquid phases are more homogeneous. These to flo patterns can be generalized using superficial liquid and vapor momentum fluxes. 5) Based on the visualization, axial momentum is another important factor that affects the refrigerant distribution in the header. The highest liquid level, defined as liquid reach, is generalized ith liquid mass flux. In order for liquid to reach the top tube, Gl > 20 kg/m2-s for 5+5 header and Gl > 60 kg/m2-s for header. In the separated flo, liquid separation height is generalized ith vapor mass flux. To avoid liquid bpassing the first tube, Gv < 20 kg/m2-s.

11 2465, Page 10 NMENCLATURE Header diameter Mass flux Liquid reach Liquid separation height (m) (kg/m2-s) (m) (m) Enthalp Superficial velocit Length of the header exit part Liquid fraction Mass flo rate Number of the outlet tubes Poer of the heaters Pressure Temperature Qualit Densit Coefficient of variation (kj/kg) (m/s) (m) (-) (g/s) (-) (kw) (kpa) (K) (-) (kg/m3) (-) Subscripts ave Average pressure i Branch number in At the smallest area in the middle of the header l Liquid out ut of the header sup Superheated sub Subcooled v vapor i j L LF m n Q P T x ρ σ D G H h REFERENCES Boers, C.D., Hrnjak, P.S., Neell, T.A., To-phase refrigerant distribution in a micro-channel manifold. Proc. 11th Int. Refrigeration Air Conditioning Conf. at Purdue, R161. Bun, H.W., Kim, N.H., Refrigerant distribution in a parallel flo heat exchanger having vertical headers and heated horizontal tubes. Exp. Thermal Fluid Sci. 35, Cho, H., Cho, K., Mass flo rate distribution and phase separation of R-22 in multi-microchannel tubes under adiabatic condition. Microscale Thermophsical Eng. 8 (2), Fei, P., Hrnjak, P.S., Adiabatic Developing To-Phase Refrigerant Flo in Manifolds of Heat Exchangers. Technical Report TR-225, Air Conditioning and Refrigeration Center, Univ. Illinois at Urbana-Champaign. Heitt, G.F., Roberts, D.N., Studies of to-phase flo patterns b simultaneous X-ra and flash photograph. AERE-M 2159, HMS. Hrnjak, P., Flo distribution issues in parallel flo heat exchangers. ASHRAE Annu. Meet. AN Hang, Y., Jin, D.H., Radermacher, R., Refrigerant distribution in minichannel evaporator manifolds. HVAC&R Res. 13 (4), Lee, J.K., To-phase flo behavior inside a header connected to multiple parallel channels. Exp. Thermal Fluid Sci. 33, Song, S., Bullard, C.W., Hrnjak, P.S., Frost deposition and refrigerant distribution in microchannel heat exchangers. ASHRAE Trans. 108 (2), Vist, S., Pettersen, J., To-phase flo distribution in compact heat exchanger manifolds. Exp. Thermal Fluid Sci. 28, Watanabe, M., Katsuta, M., Nagata, K., General characteristics of to-phase flo distribution in a multipass tube. Heat Transfer-Japanese Res. 24, Webb, R.L., Chung, K., To-phase flo distribution to tubes of parallel flo air-cooled heat exchangers. Heat Transfer Eng. 26 (4), ACKNWLEDGEMENT This project as completed in It as conducted at Air Conditioning and Refrigeration Center at the Universit of Illinois at Urbana-Champaign, sponsored b Daikin Industries, Ltd., and also supported b Creative Thermal Solutions. The authors are grateful to Hunong Kim from Daikin Industries, Ltd. for his support, help and advices during execution of this project. Poered b TCPDF (.tcpdf.org)