1、OR-05-1 0-3 Thermal and Hydraulic Characteristics of Brazed Plate Heat Exchangers-Part II: Current Research on Evaporators at KTH Joachim Claesson ABSTRACT This paper summarizes recent research on plate heat exchangers used as evaporators in domestic heat pumps and refrigeration systems, carried out
2、 at the Royal Institute of TechnoloD, Sweden. The investigations have been focusing on issues relevant to the performance and to parameters influ- encing the performance for the application in mind. Thus, the adiabatic pressure drop in a plate heat exchanger has been investigated and a new correlati
3、on is suggested, based on the classical approach by Lockhart-Martinelli and the correlation by Chisholm. The boiling section heat transfer coescient was experimentally determined using TLC measurements. The resulting heat transfer coeficient was then plotted against mass flux and heat Jux in order t
4、o investigate the relative importance of the twoparameters. As it was found that the heat flux correlated the data much better than the mass flux, the applicability of using the LMTD assuming a heat flux- governed flow boiling correlation was investigated next. In addition, since the heat transfer c
5、oeficient seemed dependent on heatflux (wall superheat,) the impact of different brine mass flow rates and brine temperature projiles was investigated. Furthel; the possible improvement of running the evaporator in Co-current conjiguration was investigated. The influence of geometry (chevron angle)
6、and of using different inlet refriger- antflow distributor devices on the performance of a plate heat exchanger was also investigated. INTRODUCTION Brazed plate heat exchangers (CBEs) have several features making them suitable as heat exchangers in small domestic heat pumps. The external size is sma
7、ll compared to the heat transfer capacity, Le., the CBE is rather compact. The small size means that the entire heat pump may be fitted in an enclosure similar in size to a domestic refrigerator and is thus easy to install in the house of the customer. The compactness also results in small internal
8、volumes. This is desirable in view of the environmental effects in case of leakage. Even thought the external size is rather small, the available heat transfer area is comparably large, making the CBE rather effective as a heat exchanger, i.e., small tempera- ture differences may be obtainable. Fina
9、lly, the production of a CBE is highly rationalized, keeping the price tag on the heat exchanger very competitive to other heat exchanger geometries for this kind of application. In recent years CBEs have become the preferred choice of heat exchanger type in domestic ground-source heat pump applicat
10、ions, as evaporator, condenser, and subcooler, for the Swedish heat pump manufacturers. Domestic heat pump systems are exclusively electrically driven, and with increas- ing electricity prices, there is a constant demand for increased energy efficiency. The energy efficiency of a heat pump may be im
11、proved in several ways, e.g., by increased compressor efficiency and by decreased temperature differences in the condenser and evaporator. This paper presents a summary of recent research at the Royal Institute of Technology, Sweden, on plate heat exchang- ers operating as evaporators in ground-sour
12、ce heat pump applications. The purpose and aim of the research was to increase the understanding of the parameters affecting heat transfer and efficiency of the PHE evaporator and identifj methods of decreasing the temperature difference between the secondary refrigerant and the refrigerant. Joachim
13、 Claesson is pursuing his PhD degree in the Department of Energy Technology at the Royal Institute of Technology, Stockholm, Sweden. a34 02005 ASHRAE. CALCULATION PROCEDURE Most of the experimental data presented in the present article concerns compact brazed plate heat exchangers oper- ating as eva
14、porators under conditions similar to those in a heat pump system. As such, the evaporator was run with the refrig- erant flowing vertically upward and the secondary refrigerant in countercurrent. (In one set oftests the secondary refrigerant was run Co-current). For most cases, the refrigerant was s
15、uper- heated at the evaporator exit. The total area averaged heat transfer coefficients reported thus is an average over the boil- ing section and the superheating section. The data reduction procedure is presented below. For geometrical and dimension- less parameters, the reader is referred to Part
16、 I of this paper (Claesson 2005). In order to obtain the heat transfer coefficient on the refrigerant side from the overall heat transfer coefficient, the film heat transfer coefficient on the brine side is required. The measured mass flow rate of the brine fluid (secondary refiigerant) was used to
17、calculate the brine-side Reynolds number as where 2 Ij2b G, = - W.b.np The Nusselt number on the brine side was calculated accord- ing to Bogaert and Blcs (1995), In the present article, the last term (the viscosity ratio) has been neglected. The reason is the negligible impact of the term in heat p
18、ump applications, as the temperature differences between wall and bulk are small. All constants in Equation 3 were supplied to the author by the heat exchanger manufac- turer and are proprietary information. However, using the correlation by Martin (1996) would not significantly alter the findings i
19、n the present paper, as shown in Part I. From the Nusselt number, the film heat transfer coefficient on the brine side may be calculated as (4) In order to obtain the refrigerant-side film heat transfer Coefficient, the overall heat transfer coefficient is required. In the present analysis, the appr
20、oach by Dutto et al. (1 99 1) has been used. They defined the overall heat transfer coefficient as the area averaged value of the boiling and superheated sections; thus, UCBE.A = Uevap. Aevup + Usup. sup . For each term of Equation 5, the following apply: Hence, Equation 5 transforms into Qtot Qevap
21、 I Qsue QCBE Qevap sup - or where we have assumed a constant specific heat capacity on the brine side throughout the entire heat exchanger. The value tb” is the brine temperature at the location where the refriger- ant becomes saturated vapor, calculated from an energy balance on the superheated sec
22、tion of the evaporator. SevUp and SsUp, are the logarithmic mean temperature differences in each section of the evaporator, not accounting for pressure drop of the refrigerant. Using Equation 8 as the appropriate tempera- ture difference in the heat exchanger, the overall heat transfer coefficient m
23、ay be calculated as Qtot Atot. QCBE UCBE = (9) Now, using this area averaged overall heat transfer coef- ficient, a corresponding heat transfer coefficient on the refrig- erant side may be calculated as, since the heat transfer areas on both fluid sides are equal, The above analysis also holds for t
24、he case where the refrigerant leaves the evaporator at a saturated state. Equation 8 then reduces to the logarithmic mean temperature difference of the boiling section, which in that case is the entire heat exchanger. The physical properties of the refrigerant were evaluated using REFPROP 6.01 from
25、NIST (1998), and the physical properties of the brine fluid were evaluated using Melinder (1997). Five different test facilities have been used in this work. Two of them (denoted “Test rigg A” and “Test rigg B”), from which Figure 8 to Figure 12 originate, are quite similar, using the same kind of e
26、quipment for measuring mass flow rates, temperatures, and pressures. Therefore, the uncertainty of measurement in these two test facilities may be considered to be similar. The energy balance (refrigerant side to brine side) ASHRAE Transactions: Symposia 835 1 O0 $1 10 * 1 0.01 o. 1 1 X Figure 1 Two
27、-phase multiplier, estimated from adiabatic tests, vs. Lockhart-Martinelli parameter. in these measurements was within *2.5%. The test facility used in producing Figure 3 and Figure 4 had a flow meter on the secondary refrigerant that was less accurate, and the corre- sponding energy balance was wit
28、hin +5%. The test facility used for the data in Figure 1 was quite different compared to the others, using a pump to circulate the refrigerant instead of a traditional compressor system. However, the equipment used to conduct measurement is of the same type as for “Test rigg A” and “Test rigg B”. Fi
29、nally, the experimental results displayed in Figure 6 and Figure 7 were based on tests conducted at SWEP International AB. In addition, different areas as the base for the reported refrigerant-side heat transfer coefficient have been used. For instance, the measured boiling heat transfer area is use
30、d in Figure 3 and Figure 4, estimated (using single-phase film heat transfer coefficients on brine and water sides) boiling heat transfer area is used in Figure 7, and the total heat exchanger heat transfer area is used for the other figures reporting heat transfer coefficients. MIO-PHASE PRESSURE D
31、ROP, ADIABATIC CONDITIONS In order to determine the flow boiling heat transfer in plate heat exchangers, the accompanying pressure drop must also be known. In a separate set of tests, the adiabatic two- phase pressure drop in plate heat exchangers was investigated. The vapor quality, mass flux, and
32、pressure of the refrigerant were varied within a range of interest for heat pump and refiig- eration applications. Thus, the mass flux was vaned between 18 kg/s.mz and 48 kg/s.mz and the vapor quality ranged 250 200 150 c1 0 100 50 O O. v O. o E v O G, = 48 kg/m% o G, = 18 kg/m2.s R134a p = 3.6 - 6.
33、7 bar(a) x= 11% - 79% G, = 15 - 50 kg/m.s p = 470 mm W=113mm d.=4mm E o v 0.0 0.2 0.4 0.6 0.8 1.0 1.2 x -I Figure 2 Chisholm parameter C calculated using Equation 16 and experimental data. between 10% and 70%. Higher vapor qualities were not obtainable due to experimental limitations. R- 134a was us
34、ed as refrigerant. The plate heat exchanger used consisted of only one refrigerant channel, eliminating any mass flow maldistri- bution otherwise occurring in multichannel evaporators. However, at the inlet of the refngerant, the supplier ofthe plate heat exchanger installed a flow distributor devic
35、e. Thus, in order to obtain the channel frictional pressure drop, the pres- sure drop in the inlet flow distributor and the pressure drops in the inlet and outlet ports had to be estimated (see Claesson and Simanic 2003). Even though no heat was added to the refrigerant in the test section, one may
36、still expect a small acceleration pressure drop. Thus, these pressure drop were estimated as (11) 2 Vtp;outport Vtp,import APp, = 2 . outport Ainport where v tP = v,+x.(v,-v,) (12) is the specific volume of the two-phase mixture evaluated at the outlet and inlet ports. In addition, the gravitational
37、 contri- butions were calculated as Once the frictional pressure drop in the flow channel was obtained, the two-phase multiplier was calculated as 836 ASH RAE Transactions: Symposia 16000 14000 12000 Y E hl 10000 = 1 +c/x+ id, (1 6) where Chas different values depending on the flow regime of the two
38、 different phases. Chisholm and Laird (1958) suggested C = 2 1 for the turbulent-turbulent flow regime with air-water flow in horizontal circular tubes at atmospheric pres- sures. Chisholm (1967) included the values for the constant C for the three other possible combinations. These classical b2 Tot
39、al heat load of heat exchanger 8 O O ieynolds number is based on he entire flow as liquid. O 1 O0 200 300 400 500 Rerefr,io -I values do not fit the experimental data obtained in this inves- tigation. The values, not previously published, of the Chish- Olm parameter were fitted from the experimental
40、 data as (17) 27.7 0.8 C(X,Rel,) = - 67.6 + 0.1 . Re, + - - - A- 2 where Re, is the Reynolds number assuming the entire mass flow as liquid. This correlation resembles the correlation used for calculating the two-phase multiplier. The shape of the calculated values of C using Equation 16 and experim
41、ental data was similar to the two-phase multiplier when plotted against the Lockhart-Martinelli parameter (see Figure 2). The tested range for Equation 17 was 0.05 to 1 .O0 for the Lockhart- Martinelli parameter and 250 to 875 for the refrigerant Reynolds number, based on the entire flow being liqui
42、d. The obtained values ofthe Chisholm parameter are signif- icantly higher than reported for smooth tubes according to Chisholm (1967) and also higher than reported on similar geometries (see Part I ofthis paper). The best obtainable single value for the Chisholm parameter, using Equation 16, is C =
43、 154, as indicated in Figure 1. This high value is comparable to values reported by Sterner and Sundn (1997) in which ammo- nia was evaporated in a plate heat exchanger. ASHRAE Transactions: Symposia a37 DETERMINATION OF BOILING HEAT TRANSFER AREA One difficulty investigating the heat transfer mecha
44、nisms in plate heat exchangers is the problem of measuring the local heat transfer coefficient, which requires measurements of the wall temperature. A representative surface temperature is difficult to measure, first because the surface is inaccessible inside the narrow channels and second because t
45、he surface temperature (and the heat transfer coefficients) may vary locally depending on the location within the unit cell of the structure. If the local heat transfer coefficient is not readily obtainable, it is a simpler task to measure the average heat transfer coefficient over the area used for
46、 boiling. Since the refrigerant leaves the evaporator slightly superheated, the problem transforms into one of measuring the actual boiling heat transfer area. In a special test rig, the boiling heat transfer area was determined by means of thermochromic liquid crystals (TLC). The outer plate (with
47、zero heat flux) was painted with TLCs, which change color with temperature, and the color play was recorded. Then the number of pixels in the picture with a color corresponding to the saturation temperature was counted and compared to the total number of pixels (approximately 68,000) of the plate he
48、at exchanger. Two pictures of the TLC colorplay are shown in Figure 3, and the clear, distinct transi- tion between a low wall temperature, indicative of the boiling section, and a higher wall temperature, indicative of the super- heated section, may be noted. The major uncertainty with this method
49、is not in the deter- mination of the boiling and superheating areas, but in the assumption that all refrigerant is evaporated at the point where the temperatures change, i.e., in the amount ofheat transferred in the two regions. In addition, the apparent good wo-dimen- sional refrigerant flow distribution may also be noted in Figure 3. In the diagram of Figure 2, each point is determined from approximately 1 O0 pictures of the colorplay. As may be noted in Figure 3, a slight scatter of the refrigerant, all-liquid Reynolds number occurs for the “same” total heat load.
