Accepted Manuscript Title: Experimental Investigation of Two-Phase Flow Distribution in Plate-Fin Heat Exchangers Authors: Zhe Zhang, Sunil Mehendale, JinJin Tian, YanZhong Li PII: DOI: Reference: S0263-8762(17)30087-4 http://dx.doi.org/doi:10.1016/j.cherd.2017.02.003 CHERD 2564 To appear in: Received date: Revised date: Accepted date: 31-8-2016 3-1-2017 1-2-2017 Please cite this article as: Zhang, Zhe, Mehendale, Sunil, Tian, JinJin, Li, YanZhong, Experimental Investigation of Two-Phase Flow Distribution in Plate-Fin Heat Exchangers.Chemical Engineering Research and Design http://dx.doi.org/10.1016/j.cherd.2017.02.003 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Experimental Investigation of Two-Phase Flow Distribution in Plate-Fin Heat Exchangers Zhe Zhanga,, Sunil Mehendaleb, JinJin Tiana and YanZhong Lic a Tianjin Key Laboratory of Refrigeration Technology, Tianjin University of Commerce, Tianjin, China, 300134 b c School of Technology, Michigan Technological University, Houghton, USA, 49931 School of Energy and Power Engineering, Xi’an Jiaotong University, Xi’an, China, 710049 Corresponding author. Email: zhangzhe@tjcu.edu.cn, Tel: +8622 26686251 Graphical abstract Figure 14 Effect of non-uniform dryness flow distribution on heat exchanger effectiveness 0.74 ReG=1500, x=29.2%,ΔT=20℃:Th,in=42℃,Tc,in=22℃ 0.72 τ 0.70 0.68 0.66 0.64 0.35 0.40 Sx 1 0.45 0.50 Highlights Two-phase flow maldistribution is more pronounced than that of single-phase flow. Increasing Reynolds number causes more severe two-phase flow maldistribution. Non-uniformity of the liquid-phase flow is greater than that of the gas-phase flow. The new distributor design led to improved thermal performance. Abstract Flow maldistribution causes declining plate-fin heat exchanger thermal-hydraulic performance. A first-of-its kind experimental facility and the related data acquisition system were constructed for studying liquid-gas flow distribution in a plate-fin heat exchanger. The gas Reynolds numbers ranged from 1880 to about 2600 and the inlet dryness (i.e., quality) from 12% to 41%. Two-phase flow maldistribution among the heat exchanger passages was more widespread compared to that of single-phase flow. More specifically, the liquid-phase distribution was more uneven compared to the gas-phase distribution. The inlet flow rate and dryness were identified as the chief factors affecting the distribution of phases in the heat exchanger. For a given inlet dryness, the two-phase flow distribution became increasingly non-uniform with the inlet gas flow rate, consistent with the behavior observed for single-phase flow. Additionally, the non-uniformity in the gas flow distribution decreased and that in the liquid flow non-uniformity increased with increasing inlet dryness fraction. A novel distributor design, with a complementary fluid cavity was also built and tested. Experimental results show that improving the distributor design is very effective in improving the two-phase flow distribution in plate-fin heat exchangers. Based on heat transfer studies conducted at a single Reynolds number of about 1500 and a dryness of 29.2%, the heat exchanger effectiveness was also correlated as a function of the dryness distribution non-uniformity parameter Sx. The effectiveness was found to reduce as the flow distribution became more uneven, highlighting the importance of accounting for and controlling the flow 2 maldistribution through proper distributor design. Key words: plate-fin heat exchanger, two-phase flow, flow distribution, distributor, effectiveness. Nomenclature A area of the channel, m2 d hydraulic diameter of fin passage, m J measured value of flow rate or dryness N channel number, dimensionless P pressure drop, Pa Q flow rate, m3 h-1 Re average Reynolds numbers in heat exchanger ( Re Vave d ), dimensionless S degree of non-uniformity, dimensionless T temperature, ℃ ΔT temperature difference between the hot fluid and the cold fluid, ℃ u' velocity deviation in lateral direction, dimensionless u" velocity deviation in gross flow direction, dimensionless V flow velocity, m s-1 vave average flow velocity for all channels, m s-1 3 v' average flow velocities in gross flow direction, m s-1 v" average flow velocities in lateral direction, m s-1 x dryness, dimensionless Greek symbols distributor angle Δ error, dimensionless degree of two-phase flow non-uniformity, dimensionless kinematic viscosity, m2s-1 τ effectiveness of heat exchanger, dimensionless Subscripts ave average value c cold G gas h hot i serial number of flow channels in inlet J variable representing flow rate or dryness L liquid 4 out outlet Q flow rate x dryness 1. Introduction Plate-fin heat exchangers are a type of high performance compact heat exchanger. They enable significant amounts of heat transfer to take place between two fluid streams at a very small driving temperature difference between the two fluids. In other words, high heat exchanger effectiveness is possible with such heat exchangers. In the endeavor to attain the best possible heat exchanger performance, close attention needs to be paid to the flow maldistribution of the two fluids within the header and the heat exchanger passages. If the maldistribution issue is not intelligently addressed, it can lead to significant loss in heat transfer effectiveness for single phase heat exchangers. In the thermodynamic design and analysis of plate-fin heat exchangers, it is generally assumed that both fluids are uniformly distributed among all the parallel passages throughout the exchanger core. In practice, however, flow maldistribution of one or both fluids very likely exists. Possible causes of this flow maldistribution include a general lack of design data related to non-uniform flow distribution and lack of attention to the design of the distributor or header and the flow passages (Kays and London, 1984). Maldistribution of flow not only results in a penalty in thermal-hydraulic performance for single-phase heat exchangers, as described above. It also causes a significant deterioration of the heat transfer for heat 5 exchangers in which one or both fluids undergo a phase change (Shah and London, 1980). In particular, if the two phases of either fluid were distributed non-uniformly through the heat exchanger core, the following deviations from the ideal case, (i.e., perfectly uniform distribution of the phases), would occur: a. The vapor quality or dryness distribution would be non-uniform. b. The mass flow or mass flux distribution would be non-uniform, and c. Condensation and/or boiling heat transfer would potentially take place under different vapor-liquid equilibrium conditions. Hence, the pressure drop and heat transfer would be very different from that under “ideal” operation. Thus, maldistribution of the two phases in heat exchangers is one of the principal and least understood causes behind their poor thermal performance. Because the two-phase flow maldistribution has such a far-reaching influence on the heat exchanger’s thermal-hydraulic performance, it is essential to understand the factors governing gas-liquid phase distribution in the same. A study of the relevant published literature reveals that most authors have studied only the single-phase flow maldistribution in plate-fin heat exchangers, either experimentally (Jiao and Baek, 2005; Zhang et al., 2004; Wen et al., 2007) or numerically (Zhang and Li, 2003; Wen and Li, 2004; Zhang et al., 2002). Mueller and Chiou (1988) summarized various types of flow maldistribution in heat exchangers and discussed the reasons behind such maldistribution. Ranganayakulu and Seetharamu (1999) carried out an analysis of the effects of inlet fluid flow non-uniformity on the thermal performance and pressure drop of crossflow 6 plate-fin heat exchangers by using a finite element method. Shen and Bell (1987) studied the problem of the effect of two-phase flow maldistribution on the performance of a feed-effluent heat exchanger. Wu and Chen (1995) researched the two-phase flow distribution and performance of plate-fin heat exchangers with different inlet constructions. Kitto and Robertson (1989) gave the definition of two-phase flow maldistribution and pointed out that it would result in diminished thermal performance. Experimental results for the effects of flow maldistribution on the thermal performance of heat exchangers were presented by Lalot et al. (1999). Bai and Newell (2000) investigated two-phase flow field characteristics in flat plate heat exchangers and published some photographs of the flow distribution in a Chevron-style test section. Wang et al. (2010) investigated the distribution of two-phase flow in a plate-fin exchanger and concluded that very severe maldistribution of two-phase flow occurred due to improper inlet configuration. Muller-Menzel and Hecht (1995) observed various flow patterns in a plate-fin heat exchanger using a flow visualization test rig. They found that at high gas mass fluxes, the liquid and gas phases flowed uniformly upwards, i.e., the flow distribution improved. In a related study prior to the present one, Zhang et al. (2015a) found that for single-phase flow, the non-uniformity of temperature distribution was more pronounced than that of the flow, and led to varying degrees of loss in the heat exchanger effectiveness. They also demonstrated that a new header configuration could effectively enhance the performance of plate-fin heat exchangers by controlling the non-uniform distribution of flow. Al-Rawashdeh et al. (2012a) and Al-Rawashdeh et al. (2012b) proposed a design methodology and criteria for the design of a flow distributor to achieve even distribution of gas and liquid flows among parallel microchannels. The design methodology 7 determines the required hydraulic resistance in the barrier channels and their dimensions for the barrier-based distributor. The design criteria can also be applied to larger numbers of parallel microchannels. Al-Rawashdeh et al. (2012c) have successfully designed the barrier-based micro/milli reactor with modular flow distributor according to the methodology purposed by Al-Rawashdehet al. (2012a). With this design, a uniform flow distribution was achieved at varied conditions and even at larger fluid viscosity. Lee and Lee (2004) examined the distribution of two-phase annular flow at header-channel junctions of a compact plate heat exchanger. They focused on the effect of the intrusion depth of the channels to improve the liquid phase distribution. Yuan et al. (2013) proposed a new flow inlet structure, in which the gas and liquid phases enter the plate fin heat exchanger separately, and the distributor they designed can mix the two phases and distribute the mixture uniformly in the subsequent parallel channels. Saad et al. (2011); Saad et al. (2012); and Saad et al. (2014) utilized a gas-liquid distributor to improve the phase distribution performance of air-water flows. Their results showed that computational fluid dynamics effectively predicts the non-uniformity of the gas and liquid distribution across the experimental heat exchanger. Ha et al. (2006) carried out numerical studies for air-water two-phase flow distributions in multiple channels between two headers. It was found that with an increase in the liquid flow rate, the flow distribution became increasingly non-uniform. As one can see, there is scarcity of literature about the uneven distribution of two-phase flow in plate-fin heat exchangers. Moreover, the few previous works available mostly deal with the numerical prediction of the thermal performance of such heat exchangers for 8 non-uniform inlet temperatures. As described above, it is very important to develop a good understanding of this subject, since the thermal performance of this type of heat exchanger is very sensitive to flow maldistribution, particularly for the case of a liquid-gas (or vapor) mixture entering the heat exchanger. The present study, which is a further development of the single-phase results of Zhang et al. (2015b), describes the results of an experimental method to explore the influence of different inlet constructions on the distribution of two-phase flow in a plate-fin heat exchanger over a range of inlet dryness (or quality) and flow rates. Additionally, to achieve improved thermal performance, a novel distributor design with a complementary fluid cavity has been built and tested. 2. Experiment 2.1. Test bench and experimental procedure The test bench (Fig. 1), details of which can be found in Zhang et al. (2015a), includes an air circuit, a water circuit and a data acquisition system. The air circuit consists of an air compressor, a chiller, an electric heater, a test section (the plate-fin heat exchanger), a gas-liquid separator, and a passage-switching device. The water circuit includes a water tank, a water pump, filters, a stabilization tank and a gas-liquid mixer (Fig. 2). Additionally, the test bench also has a data acquisition system, gas and liquid turbine flow meters, thermocouples, and gage and differential pressure transducers. Experiments to measure the two-phase (water and air) flow distribution within the heat exchanger and its thermal-hydraulic performance have been conducted on the test bench described above. 9 To measure the distribution of the two phases in the heat exchanger, the flow rate of water pumped to the test section is measured by a turbine flowmeter. The water is mixed with air in the mixing section and the mixture then flows to the heat exchanger. The mixture at the heat exchanger outlet then flows first into the passage-switching device and then to the air-water separator. The flow rate of the separated air is measured by a second turbine flowmeter. The water separated from the mixture then flows down into collection vessels and its flow rate is measured by weighing the water collected in a known time interval. To measure the thermal performance of the heat exchanger, the air pumped by the air compressor is split into two branches. The portion which is heated by the electric heater enters the test section and the unheated stream goes into the air-water mixer. The hot air then exchanges heat with the cold air-water mixture in a counter flow configuration. A conventional distributor is used to distribute the hot fluid among the heat exchanger passages, whereas various distributor configurations are used for channeling the cold air-water mixture. 2.2. Instruments and their accuracy Details of the instrumentation and their accuracy have been described in detail in Zhang et al. (2015a) and Zhang et al. (2015b). Here, only a brief summary of the same has been provided. Turbine flow meters are used to measure the air and water flow rates in the gross (or header) passage and the various channel passages. The maximum relative error in the measured frequency is 1.3%, 0.7%, and 0.4% for the header and channel gas flow meters, and the gross liquid flow meter respectively. The digital multi-meter has an uncertainty in measured frequency of less than 0.1%. 10 The air and water temperatures in the header and individual channels were measured by copper-constantan thermocouples. The uncertainty in temperature measurement has been shown in Zhang et al. (2015a) and Zhang et al. (2015b) to be no greater than 0.24°C. 2.3. Plate-fin heat exchanger The aluminum plate-fin heat exchanger (see Fig. 3b) used in this study is similar to that used in a 2000Nm3·h-1air separation plant. The surface of the heat exchanger is wrapped with insulation to minimize heat losses. Per the schematic (see Fig. 3a), heated air enters the heat exchanger through the top header and leaves through the middle header. The low temperature air-water mixture enters the heat exchanger through the bottom header. The heat exchanger has dimensions of 200 mm × 250 mm × 178 mm. The header passage diameter and length are 40 mm and 250 mm, respectively. Details of the heat exchanger appear in Fig. 3c.The heat exchanger has 6.5 2.0 0.3 mm (length × width × thickness) plain fins (see Fig. 3d). As depicted in Fig. 4, the heat exchanger outlet has been divided into 30 equal channels (each 41.7 × 40 mm2) to measure the distribution of temperature, flow rate, and pressure. Each of the 30 channels includes several microscale passages, and the flow distribution within any of the channels is assumed to be uniform. 2.4. Improved distributor design A new distributor has been designed and tested in this work. The new design incorporates a complementary or buffer fluid cavity for improving flow distribution in the heat exchanger. Fig. 5 shows the schematic of the internal configuration of this new distributor. The geometrical variable h/H of the distributor, as shown in Fig. 5 took values from 0 to 0.3 in the 11 experimental investigation. The distributor angle was fixed at 45, the orifice diameter on the perforated distributor was 2.5 mm, and the orifice opening ratio (i.e., the orifice open area as a fraction of the total plate area) was 10%. 2.5. Distribution of Phases To evaluate the distribution of the flow, dryness, or temperature for the two-phase mixture in the heat exchanger, the statistical parameter (SJ) is employed, which is defined as follows: 1 2 N 2 1 1 SJ J J J Q, x, T i ave J ave N 1 i 1 (1) Additionally, the dryness fraction non-uniformity (λx) is defined as: x S x x (2) Two-phase dryness (x) is defined as the ratio of the mass flow rate of the gas to that of the two-phase mixture. The dimensionless standard deviation (SJ) of the measured data represents the non-uniformity in the flow, dryness, or temperature if J = Q, x, or T, respectively. If J = Q, for example, Ji and Jave represent the flow rate in any channel i and the average flow rate, respectively. The degree of flow non-uniformity (SQ), and non-uniformity in dryness (Sx) can be fruitfully employed to assess the extent of flow or dryness maldistribution. Using these parameters, the different distributor design options can be evaluated under various operating conditions. A small value of SJ means that the fluid is distributed relatively evenly, a large value of SJ implies highly non-uniform flow/dryness/temperature distribution and an ideal value of SJ = 0 represents a perfectly uniform flow distribution. The dryness non-uniformity (λx) is a measure of how uniformly the 12 liquid and gas phases are mixed. The smaller the value of λx, the better the two phases are mixed. 3. Results and discussion Fig. 6 shows the influence of the inlet dryness on the distribution of two-phase flow in the plate-fin heat exchanger for the conventional distributor configuration without the buffer cavity (h/H=0). It can be seen from Fig. 6 that in the experimental range, severe non-uniformities exist in the two-phase flow distribution. The complex interaction between the two phases makes it much more difficult to comprehend the flow distribution dynamics of two-phase flow compared to that of single phase flow. From the experimental results plotted in Fig. 6, the distribution of air and water flow in the lateral and the gross flow directions displays multiple peaks. In Fig. 6, it is seen that the flow distribution patterns for the gas and liquid show similar characteristics in that both phases undergo flow maldistribution. The interactive forces between the air and water exacerbate the maldistribution of the two phases compared to the situation of either air or water flowing by itself. It is apparent from Fig. 6 that at x=11.7%, the air is more non-uniformly distributed than at x=40.5%; Fig. 6 also shows that the uneven distribution of water at x=40.5% is more pronounced than at x=11.7%,The reason for this trend is that at greater inlet dryness, the liquid flow rate decreases, which makes it more difficult to distribute the liquid phase more uniformly in the heat exchanger. The data listed in Table 1 reveals that the air flow distribution becomes more uniform while the water flow distribution becomes increasingly non-uniform as the inlet dryness fraction increases. Fig. 6 also shows 13 that, in general, water flow maldistribution is more pronounced than that of air and that the two-phase flow distribution in the lateral direction is more strongly asymmetric compared to the gross flow direction. The inlet dryness has a significant influence on the flow distribution of the two phases. However, it should be noted that the inlet dryness does not comprehend the actual gas and liquid flow rate, since it only indicates the ratio of the air mass flow rate to the two-phase mass flow rate. Therefore, it is very necessary to investigate the effect of dryness as well as the two flow rates on the distribution patterns in the heat exchanger. Fig. 7 shows the patterns of air and water distribution at the heat exchanger outlet for various air Reynolds numbers. It is evident that the gas Reynolds numbers strongly affects the air-water distribution in the heat exchanger. For the four gas Reynolds numbers tested, the degree of non-uniformity in the dryness distribution (λx) is 1.997 (ReG=1880), 2.271 (ReG=2124), 2.524 (ReG=2358) and 2.777 (ReG=2593) respectively, as shown in Table 2. It is therefore clear that for fixed inlet dryness, the air Reynolds number profoundly influences the air-water distribution and that the maldistribution is exacerbated with increasing air Reynolds number. This observation is consistent with the conclusion obtained from single-phase flow measurements (Zhang et al., 2015b). Thus, increasing the air flow rate (or Reynolds number) aggravates the maldistribution of the two-phase flow, and the departure from perfectly homogeneously distributed flow becomes more pronounced. The above conclusions have been quantified in Table 2, which shows that in the experimental range considered, great non-uniformities in the air-water distribution exist. The two-phase flow maldistribution is much more pronounced than single-phase flow 14 maldistribution, as reported in (Zhang et al., 2015b). As listed in Table 2, the non-uniformity in dryness (Sx) is 0.583 (ReG=1880), 0.663 (ReG=2124), 0.737 (ReG=2358) and 0.811 (ReG=2593), respectively. Thus, for a fixed inlet dryness, it is seen that the two-phase flow distribution becomes increasingly more asymmetric as the inlet gas Reynolds number increases. Fig. 8 shows the effect of inlet dryness on the heat exchanger pressure drop for two different air Reynolds numbers. The inlet dryness can only reach a minimum value of about 11.7% because of limitations on the sensitivity of the gas-turbine flow meters and differential pressure transducers. It is evident from Fig. 8 that the pressure drop curve clearly follows an upward trend with increasing inlet dryness and increases more rapidly after x =23.5%. The reason for this behavior is that as the inlet dryness increases, the mass and volume of gas increase rapidly, causing the liquid flow distribution to be progressively more skewed (as also discussed above). Thus, the two-phase pressure drop across the heat exchanger is more adversely affected at higher dryness or quality, which is obviously not an issue for the flow of a single phase. This observation demands that careful attention be given to improving the two-phase flow distribution while designing two-phase flow heat exchangers. Again, from Fig. 8, it is seen that increasing Reynolds number results in larger pressure drop. Based on the experimental data for the various gas Reynolds number depicted in Fig. 8, the pressure drop ( P ) has been correlated as a function of the inlet dryness (x) and gas Reynolds number (ReG) in Eq. (3). P 2.53319 108 x 0.5664 ReG 3.1171 (3) 15 All the pressure drop data are predicted within ±25% by Eq. (3) with a mean absolute deviation of 12%. This fact indicates that the pressure drop can be adequately described by a single scaling relationship in terms of the gas Reynolds number and inlet dryness. Again, as pointed out above, this equation should be applied with caution outside the experimental range of gas Reynolds numbers and inlet dryness. To clearly and systematically study the distribution of the two phases in the plate-fin heat exchanger, the two-phase flow is divided into lateral and gross components, according to Eqs. (4) and (5): v‘i (Q5i 4 Q5i 3 Q5i 2 Q5i 1 Q5i ) / 5 A (i=1-6) v“i (Qi Qi 5 Qi 10 Qi 15 Qi 20 Qi 25 ) / 6 A (i=1-5) (4) (5) In the above equations, vi ' and vi " are the average lateral and gross direction flow velocities, respectively, while Qi and A represent the flow rate and the cross-sectional area of channel i, respectively. The flow velocity deviations u i ' and u i " which represent the lateral and gross direction non-uniformities in the heat exchanger, are defined as: ui ' (vi ' vave ) / vave , ui " (vi " vave ) / vave (6) where vave denotes the average flow velocity over all the channels. As is clear from the above description, the non-uniform distribution of two-phase flow in plate-fin heat exchangers is more severe and thus, more problematic, than single-phase flow distribution. So a new distributor design, with a buffer (or complementary) fluid cavity was 16 developed and tested in this research. Figs. 9-12 illustrate the lateral and gross direction air and water flow distributions for various values of inlet dryness. Figs. 9-12 clearly depict that the two-phase flow distribution manifests as different patterns at different inlet dryness. It should be noticed that although two-phase flow is an inherently oscillatory and unstable process, the data reported here are based on time-averaged quantities. A study of Figs. 9-12 clearly reveals how the dryness, flow direction, and the cavity parameter h/H influence the gas and liquid phase distribution among the heat exchanger channels: a. At ReG = 2593 and x = 17.9% (see Fig. 9), ui´ for the air ranges from about -0.17 to 0.15, while ui´´ varies from about -0.075 to 0.07. Thus, the gas phase is more unevenly distributed among the channels in the lateral direction than in the gross flow direction. h/H is also seen to affect the gas flow distribution, with h/H = 0.2 yielding the most uniform phase distribution. h/H = 0 is seen to result in the worst gas phase distribution. b. At ReG = 2593 and x = 17.9% (see Fig. 10), ui´ for the water ranges from about -0.6 to 1.4, while ui´´ varies from about -0.3 to 0.32. Thus, the liquid phase is more unevenly distributed compared to the gas phase. Moreover, like the gas phase, the liquid phase is more unevenly distributed among the channels in the lateral direction than in the gross flow direction. h/H is also seen to affect the liquid flow distribution, with h/H = 0.2 yielding the most uniform phase distribution, and h/H = 0 is seen to result in the worst liquid phase distribution. c. At the same ReG (= 2593), as the dryness is increased to 29.2% (see Fig. 11), ui´ for the air ranges from about -0.1 to 0.11, while ui´´ varies from about -0.11 to 0.09. In this case, 17 the uneven distribution of the gas phase is slightly greater in the lateral flow direction than in the gross flow direction. As before, the h/H = 0.2 distributor yields the most uniform gas phase distribution, and h/H = 0 is seen to result in the worst gas phase distribution. d. At the same ReG (= 2593) and x = 29.2% (see Fig. 12), ui´ for the water ranges from about -0.7 to 1.5, while ui´´ varies from about -0.6 to 0.4. Thus, once again, the liquid phase is more unevenly distributed compared to the gas phase at this increased dryness. Again, like the gas phase, the liquid phase is more unevenly distributed among the channels in the lateral direction than in the gross flow direction. h/H is also seen to affect the liquid flow distribution, with h/H = 0.2 yielding the most uniform phase distribution, and h/H = 0 is seen to result in the worst liquid phase distribution. Based on the above observations, the following general observations can be made: On the whole, the liquid phase is much more unevenly distributed among the heat exchanger channels compared to the gas phase. Again, in general, the effect of the distributor cavity parameter h/H on the gross flow direction flow distribution is significantly more pronounced than on that in the lateral direction. The flow non-uniformity is the maximum at h/H=0 (conventional distributor) because, in this case, there is no buffer or mixing space at all in the distributor. For h/H = 0.3, the mixing space has sub-optimal effect on the flow distribution, because the fluid has more than the necessary space to undergo the right amount of mixing in the distributor, before it enters the heat exchanger. However, even with this non-optimal design, the flow distribution for h/H = 0.3 is better than for h/H=0 in most of the cases considered here. Flow distribution measurements show that the flow is distributed most uniformly at h/H=0.2.This observation is also consistent with the conclusion obtained from 18 single-phase flow measurements. From Table 1 and Table 2 also, it can be readily concluded that the most uniform flow distribution was obtained when the improved distributor configuration parameter h/H=0.2 was employed. From Figs. 9 through 12, it is obvious that the configuration parameter h/H for the distributor is a key design factor which profoundly influences the two-phase flow distribution in the plate-fin heat exchanger. This is due to the fact that a complementary (or buffer) fluid mixing cavity, which mixes the two phases to different extents before the fluid enters the heat exchanger, is provided as part of the distributor design by altering the value of h/H. By appropriately pre-mixing and channelizing the fluid, this cavity thus allows the gas-phase and liquid-phase to be redistributed in the distributor due to the different pressure drops between the fluid passages. The fluid in the low flow rate and/or high pressure drop passages can thereby be compensated via the orifice on the perforated distributor. The flow distribution performance of the plate-fin heat exchanger can thereby be improved, more so in the gross flow direction. Table 1 and Table 2 show the two-phase flow non-uniformity in the heat exchanger for various distributor designs investigated at different x and ReG. It was found that the two-phase flow distribution varied with distributor design. In Table 1, λx changed from 1.175 to 12.684 (h/H=0), 0.941 to 11.803 (h/H=0.1), 0.802 to 11.376 (h/H=0.2) and 1.044 to 11.973 (h/H=0.3). In Table 2, S x changed from 0.583 to 0.811 (h/H=0), 0.526 to 0.735 (h/H=0.1), 0.483 to 0.673 (h/H=0.2) and 0.555 to 0.773 (h/H=0.3). The data in Table 2 also demonstrate that S Q ,G and S Q , L show a similar trend with increasing ReG. However, S Q ,G and S Q , L show the opposite trend with increasing x in Table 1. From Table 1 and Table 2, one can conclude that among 19 the distributor designs explored in the present study, the distributor with h/H=0.2 provides the most uniform two-phase flow distribution. Fig. 13 shows the temperature of the hot outlet fluid for the distributor configurations investigated here (h/H = 0, 0.1, 0.2, 0.3). It is very clear that distributor design profoundly affects the heat transfer. The hot fluid exit temperature is the lowest for distributor configuration h/H=0.2 and it is the highest for distributor h/H=0, i.e., the baseline distributor without any pre-mixing zone. This is because the distribution of thermal energy follows the pattern of the flow distribution. More uniform flow and thermal energy distribution results in optimal utilization of the heat transfer surface area, thus improving the thermal effectiveness. Fig. 13 also reveals the fact that the distributor configuration h/H = 0.2 is superior to the design with no pre-mixing buffer zone, h/H = 0. Thus, the novel distributor design leads to uniform flow and thermal energy distribution, which in turn results in better performing plate-fin heat exchangers. On the basis of the experimental results at about 1500 gas Reynolds number and a dryness of 29.2%, the effectiveness of the heat exchanger was also calculated for different dryness non-uniformity parameters. The heat exchanger effectiveness can then be plotted as a function of this parameter. Fig. 14 illustrates the correlation between τ and two-phase flow non-uniformity (Sx). The curve in Fig. 14 shows a downward trend with increasing two-phase flow non-uniformity (Sx), which implies that the loss in effectiveness τ is more probable in situations with higher flow non-uniformity. The more uniformly the fluid is distributed, the more uniformly the energy of the two phases is distributed. Thus, the effective area of the heat exchanger is optimally used and heat transfer occurs more efficiently. It is therefore 20 evident that the two-phase flow non-uniformity (Sx) is the single most important factor affecting the heat exchanger effectiveness. Hence, care should be taken to account for flow non-uniformity and to minimize the same in the design and operation of heat exchangers. In order to show the effect of the dryness distribution non-uniformity (Sx) on the heat exchanger effectiveness, a correlation (Eq. 7) between Sx and τ was established by calculating a least-squares minimized curve-fit to the experimental data in Fig. 14. The deviation of the data from the prediction of the correlation is less than ±3.2%: τ 0.272 2.737Sx 4.131Sx 2 (7) All of the above observations and data serve to highlight the fact that proper attention needs to be devoted to the distributor design to distribute the two-phase flow as uniformly as possible. Such a minimization of the flow and thermal energy maldistribution is essential to achieving optimal heat exchanger performance. 4. Conclusions Air-water flow maldistribution characteristics were studied for gas Reynolds numbers ranging from 1880 to about 2600 and inlet dryness (or quality) of 12% to 41%. Maldistribution of two-phase flow in plate-fin heat exchangers is found to be much more severe than the uneven distribution of single-phase flow. The gas (air) Reynolds number as well as dryness are the key factors affecting the distribution patterns of the two phases. For the experiments in this study, the non-uniformity of gas as well as liquid flow distribution are very large. Increasing inlet dryness causes the maldistribution in the gas flow to decrease and that in the liquid flow to increase. The two-phase flow maldistribution is exacerbated with 21 increasing gas Reynolds number, which leads to higher pressure drop in the heat exchanger. The liquid-phase distribution pattern exhibits greater non-uniformity than that of the gas-phase. Additionally, the non-uniform distribution of flow in the lateral direction is more severe than that in the gross direction. A novel distributor with a complementary or buffer cavity to provide pre-mixing and flow channeling was designed, built, and tested for mitigating the two-phase flow maldistribution in the plate-fin heat exchanger. It was found that the configuration parameter h/H was the main variable influencing the flow / phase distribution, particularly in the gross flow direction. The two-phase flow distribution in heat exchangers was improved by designing the pre-mixing cavity with h/H equal to about 0.2. This finding is consistent with experimental single-phase flow distribution results reported in an earlier work by the present authors. Heat exchanger effectiveness was also correlated as a function of the dryness non-uniformity Sx at a single gas Reynolds number of about 1500 and a dryness of 29.2%. The effectiveness is seen to drop as the non-uniformity of the dryness distribution increases, highlighting the necessity of accounting for and keeping under control the flow maldistribution through proper distributor design. Acknowledgements This work was supported by the Key project of Tianjin Natural Science Foundation (No. 15JCZDJC34200) and National Natural Science Fund of China(No. 11572223). 22 References Al-Rawashdeh, M., Nijhuis, T., Rebrov, E., Hessel, V., Schouten, J., 2012a. Design methodology for barrier-based two phase flow distributor. AIChE J. 58, 3482-3493. Al-Rawashdeh, M., Fluitsma, L., Nijhuis, T., Rebrov, E., Hessel, V., Schouten, J., 2012b. Design criteria for a barrier-based gas–liquid flow distributor for parallel microchannels. Chem. Eng. J. 181, 549-556. Al-Rawashdeh, M., Yu, F., Nijhuis, T.A., Rebrov, E.V., Hessel, V., Schouten, J.C., 2012c. Numbered-up gas-liquid micro/milli channels reactor with modular flow distributor. Chem. Eng. J. 207, 645-655. Bai, X., Newell, T.A., 2000. An investigation of two-phase flow characteristics in chevron-style flat plate heat exchangers. Proc. Eighth Int. Refrig. Conf., Purdue, West Lafayette. Ha, M.Y., Kim, C.H., Jung, Y.W., Heo, S.G., 2006. Two-phase flow analysis in multichannel. J. Mech. Sci. Technol. 20, 840-848. Jiao, A.J., Baek, S.W., 2005. Effects of distributor configuration on flow maldistribution in plate-fin heat exchangers. Heat Trans. Eng. 26, 19-25. Kays, W.M., London, A.L., 1984.Compact heat exchangers, 1st ed. McGraw-Hill. Kitto, J.B., Robertson, J.M., 1989. Effects of maldistribution of flow on heat transfer equipment performance. Heat Transf. Eng. 10, 18-25. Lalot, S., Florent, P., Lang, S.K., Berglles, A.E., 1999. Flow maldistribution in heat 23 exchangers. Appl. Therm. Eng. 19, 847-863. Lee, J.K., Lee, S.Y., 2004. Distribution of two-phase annular flow at header-channel junctions, Exp. Therm. Fluid Sci. 28, 217-222. Mueller, A.C., Chiou, J.P., 1988. Review of various types of flow maldistribution in heat exchangers. Heat Transf. Eng. 9, 36-50. Muller-Menzel, T., Hecht, T., 1995. Plate-fin heat exchanger performance reduction in special two-phase flow conditions. Cryogenics. 35 (5), 297-301. Ranganayakulu, C.H., Seetharamu, K.N., 1999. The combined effects of wall longitudinal heat conduction, inlet fluid flow nonuniformity and temperature nonuniformity in compact tube-fin heat exchangers: a finite element method. Int. J. Heat Mass Transf. 42, 263-273. Saad, S.B., Clement, P., Gentric, C., Fourmigue, J., Leclerc, J., 2011. Experimental distribution of phases and pressure drop in a two-phase offset strip fin type compact heat exchanger, Int. J. Multiphase Flow. 37 (6), 576-584. Saad, S.B., Clement, P., Fourmigue, J., Gentric, C., Leclerc, J., 2012. Single phase pressure drop and two-phase distribution in an offset strip fin compact heat exchanger, Appl. Therm. Eng. 49, 99-105. Saad, S.B., Gentric, C., Fourmigue, J., Clement, P., Leclerc, J., 2014. CFD and experimental investigation of the gas-liquid flow in the distributor of a compact heat exchanger, Chem. Eng. Res. Des. 92, 2361-2370. 24 Shah, R.K., London, A,L., 1980. Effects of nonuniform passage on compact heat exchanger performance. J. Eng. Power.102, 653-659. Shen, Z., Bell, K.J., 1987.Analysis of feed-effluent heat exchanger with maldistributed two-phase inlet flow and vaporizing/condensing streams. Proc. 1987 ASME/JSME Therm. Eng. Joint Conf., Purdue, West Lafayette. Wang, S.M., Li, Y.Z., Wen, J., Ma,Y.S., 2010. Experimental investigation of header configuration on two-phase flow distribution in plate-fin heat exchanger. Int. Commun. Heat Mass Transf. 37, 116-120. Wen, J., Li, Y.Z., 2004. Study of flow distribution and its improvement on the header of plate-fin heat exchanger. Cryogenics. 44, 823-831. Wen, J., Li, Y.Z., Wang, S.M.,Zhou, A.M.,2007. Experimental investigation of header configuration improvement in plate-fin heat exchanger. App. Therm. Eng. 27, 1761-1770. Wu, J.H., Chen, C.Q., 1995. Analysis of two-phase flow distribution in plate-fin heat exchanger and experimental research. J. Xi’an Jiaotong University. 29, 117-126. Yuan, P., Jiang, G.B., He, Y.L., Yi, X.L., Tao, W.Q., 2013. Experimental study on the performance of a novel structure for two-phase flow distribution in parallel vertical channels, Int. J. Multiphase Flow. 53, 65-74. Zhang, Z., Li, Y.Z., Tian, J.J., 2002. Numerical simulation on distributor’s construction of plate-fin heat exchanger. J. Chem. Ind. Eng. 53, 1311-1314. 25 Zhang, Z., Li, Y.Z., 2003. CFD simulation on inlet configuration of plate-fin heat exchangers. Cryogenics. 43, 673-678. Zhang, Z., Li, Y,Z,, Xu, Q., 2004. Experimental research on effects of distributor configuration on flow distribution in plate-fin heat exchangers. Heat Transf.-Asian Res. 33, 402-410. Zhang, Z., Mehendale, S., Tian, J.J., Li, Y.Z., 2015a. Fluid flow distribution and heat transfer in plate-fin heat exchangers. Heat Transf. Eng., 36, 806-819. Zhang, Z., Mehendale, S., Tian, J.J., Li, Y.Z., 2015b. Experimental investigation of distributor configuration on flow maldistribution in plate-fin heat exchangers. App. Therm. Eng. 85, 111-123. 26 List of Figure Captions Fig. 1-Schematic flow diagram of experimental system. Fig. 2-Schematic drawing of gas-liquid mixer. Fig. 3-(a) Schematic drawing of plate–fin heat exchanger, (b) Photograph of heat exchanger, (c) Schematic drawing of test section, (d) Geometry of the fin. Fig. 4-Channel distribution and direction. Fig. 5-Schematic drawing of improved distributor configuration. Fig. 6-Gas-phase flow and liquid-phase flow distribution for different values of the inlet dryness (ReG=2593). Fig. 7-Gas-phase flow and liquid-phase flow distribution for different values of ReG (x =29.2%). Fig. 8-Relationships between pressure drop and inlet dryness for different ReG. Fig. 9-Gas-phase flow distribution in lateral direction and in gross flow direction (x =17.9%, ReG=2593). Fig. 10-Liquid-phase flow distribution in lateral direction and in gross flow direction (x =17.9%, ReG=2593). Fig. 11-Gas-phase flow distribution in lateral direction and in gross flow direction (x =29.2%, ReG=2593). 27 Fig. 12-Liquid-phase flow distribution in lateral direction and in gross flow direction (x =29.2%, ReG=2593). Fig. 13-Outlet hot fluid temperature under different distributors. Fig. 14 - Effect of flow distribution on heat exchanger effectiveness. List of Table Captions Table 1-Two-phase flow non-uniformity in heat exchanger for different inlet x (ReG=2593). Table 2-Two-phase flow non-uniformity in heat exchanger for different inlet ReG (x=29.2%). 28 T T T Gas-liquid segregator Pp Switching Stabilization tank Pp T device Pp Liquid-turbine flow meter Electric heater T Liquid-turbine P flow meter Test Filter Heat exchanger section P Water tank Gas-liquid mixer Pp Pp Air compressor Chiller T Gas-turbine T T Pp T Gas-turbine flow meter flow meter Fig. 1- Schematic flow diagram of experimental system. 29 Gas-turbine T flow meters liquid gas-liquid mixture gas Fig. 2-Schematic drawing of gas-liquid mixer. 30 Outlet of cooling fluid Inlet of heating fluid Inlet of cooling fluid Outlet of heating fluid unit:mm (b) (a) separating plate unit:mm unit:mm fin gross passage header header A distributor guide vane (c) (d) Fig. 3- (a) Schematic drawing of plate–fin heat exchanger, (b) Photograph of heat exchanger, (c) Schematic drawing of test section, (d) Geometry of the fin. 31 Lateral direction 5 4 3 2 1 10 9 8 7 6 15 0 20 5 25 0 30 5 14 13 12 11 19 4 24 9 29 4 18 3 23 8 28 17 2 22 7 27 2 16 21 6 26 1 0 Gross 9 flow 8 direction 7 6 Fig. 4-Channel distribution and direction. 32 H fluid complementary cavity h Fig. 5-Schematic drawing of improved distributor configuration. 33 0.08 3.5 0.07 3.0 0.06 Vi (m/s) 0.09 4.0 2.5 2.0 0.05 0.04 1.5 0.03 1.0 0.02 on re di w lo w Later sf al dir ectio os s ction cti o re 0.00 flo al dir e n Gr Gr Later di 0.0 n 0.01 cti 0.5 os Vi (m/s) 4.5 (1) gas-phase (2) liquid-phase (a) x=11.7% 4.5 0.06 4.0 0.05 3.5 0.04 Vi (m/s) Vi (m/s) 3.0 2.5 2.0 1.5 0.03 0.02 1.0 0.01 w lo al dir (1) gas-phase sf ectio n Gr os w Later sf lo ction di 0.00 os l dire Gr Later a di 0.0 re re c cti tio on n 0.5 (2) liquid-phase (b) x=17.9% 4.0 0.035 3.5 0.030 0.025 Vi (m/s) 2.5 2.0 1.5 1.0 0.020 0.015 0.010 re on re di w sf al dir os n lo Later lo al dir ectio ectio Gr n sf di 0.000 w Later 0.005 os 0.0 cti cti on 0.5 Gr Vi (m/s) 3.0 (1) gas-phase (2) liquid-phase (c) x=23.5% 34 4.0 0.020 3.5 3.0 0.015 Vi (m/s) Vi (m/s) 2.5 2.0 1.5 1.0 0.010 0.005 di re Later al dir ss ss ro ectio n G ro ction flo w w 0.000 flo al dir e G Later di 0.0 ct re io ct n io n 0.5 (1) gas-phase (2) liquid-phase (d) x=29.2% 4.0 0.030 3.5 0.025 0.020 2.5 Vi (m/s) Vi (m/s) 3.0 2.0 1.5 0.015 0.010 1.0 (1) gas-phase sf Gr os ectio n Gr os n lo al dir sf ectio w di Later w al dir lo Later di 0.000 re 0.0 re cti cti on on 0.005 0.5 (2) liquid-phase (e) x=34.2% 3.5 0.035 3.0 0.030 2.5 0.025 0.020 0.015 1.0 0.010 0.5 0.005 cti on 1.5 w ction flo l dire Gr n Gr ectio os s di Later a flo w al dir os s Later di re 0.000 re 0.0 on 2.0 cti Vi (m/s) 0.040 Vi (m/s) 4.0 (1) gas-phase (2) liquid-phase (f) x=40.5% Fig. 6-Gas-phase flow and liquid-phase flow distribution for different values of the inlet dryness (ReG=2593). 35 2.5 0.025 2.0 0.020 Vi (m/s) 0.030 1.5 1.0 0.015 0.010 0.5 io n di w sf lo Later (1) gas-phase (2) liquid-phase ro ectio n ss al dir Gr n os al dir ectio flo di 0.000 w Later re re 0.0 ct cti o n 0.005 G Vi (m/s) 3.0 (a) ReG=1880 0.030 2.5 0.025 2.0 0.020 Vi (m/s) Vi (m/s) 3.0 1.5 0.015 0.010 0.5 0.005 ct di w lo Later flo ss ction G ro sf al dir e Gr ectio n os al dir re di re 0.000 w 0.0 Later io cti o n n 1.0 (1) gas-phase (2) liquid-phase (b) ReG=2124 3.2 0.030 2.8 0.025 0.020 Vi (m/s) Vi (m/s) 2.4 2.0 1.6 0.015 1.2 0.010 0.8 0.005 di w w Later ss ro G ectio n flo al dir Gr n re 0.000 lo ectio sf al dir os Later di 0.0 ct re io cti n on 0.4 (1) gas-phase (2) liquid-phase (c) ReG=2358 Fig. 7-Gas-phase flow and liquid-phase flow distribution for different values of ReG (x=29.2%). 36 Fig. 8-Relationships between pressure drop and inlet dryness for different ReG. 37 Fig. 9-Gas-phase flow distribution in lateral direction and in gross flow direction (x=17.9%, ReG=2593). 38 Fig. 10-Liquid-phase flow distribution in lateral direction and in gross flow direction (x =17.9%, ReG=2593). 39 Fig. 11-Gas-phase flow distribution in lateral direction and in gross flow direction (x=29.2%, ReG=2593). 40 Fig. 12-Liquid-phase flow distribution in lateral direction and in gross flow direction (x =29.2%, ReG=2593). 41 Fig. 13-Outlet hot fluid temperature under different distributors. 42 Fig. 14-Effect of flow distribution on heat exchanger effectiveness. 43 Table 1-Two-phase flow non-uniformity in heat exchanger for different inlet x (ReG=2593). S Q,L S Q ,G x λx Sx h/H 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3 11.70% 0.412 0.362 0.315 0.383 0.857 0.829 0.812 0.832 1.484 1.381 1.331 1.401 12.684 11.803 11.376 11.973 17.90% 0.343 0.315 0.276 0.322 0.871 0.841 0.823 0.852 1.330 1.294 1.217 1.320 7.430 7.229 6.799 7.374 23.50% 0.281 0.233 0.203 0.267 0.923 0.896 0.858 0.911 1.184 1.112 1.025 1.151 5.038 4.732 4.362 4.898 29.20% 0.278 0.217 0.181 0.242 0.928 0.901 0.899 0.916 0.811 0.735 0.673 0.773 2.777 2.517 2.305 2.647 34.20% 0.260 0.191 0.157 0.228 0.931 0.917 0.906 0.923 0.508 0.418 0.378 0.457 1.485 1.222 1.105 1.336 40.50% 0.224 0.165 0.109 0.186 0.951 0.925 0.912 0.932 0.476 0.381 0.325 0.423 1.175 0.941 0.802 1.044 44 Table 2-Two-phase flow non-uniformity in heat exchanger for different inlet ReG (x=29.2%). ReG S Q ,G λx Sx S Q,L h/H 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3 1880 0.194 0.151 0.130 0.173 0.668 0.649 0.647 0.660 0.583 0.526 0.483 0.555 1.997 1.801 1.654 1.901 2124 0.221 0.172 0.147 0.196 0.759 0.737 0.736 0.750 0.663 0.597 0.548 0.630 2.271 2.045 1.877 2.158 2358 0.246 0.191 0.164 0.218 0.844 0.819 0.817 0.833 0.737 0.664 0.609 0.700 2.524 2.274 2.086 2.397 2593 0.278 0.217 0.181 0.242 0.928 0.901 0.899 0.916 0.811 0.735 0.673 0.773 2.777 2.517 2.305 2.647 45