- Email: [email protected]
Two-phase refrigerant distribution in an intermediate header of a parallel flow minichannel heat exchanger
Accepted Manuscript Title: Two-phase refrigerant distribution in an intermediate header of a parallel flow minichannel heat exchanger Author: Ho-Won Byun, Nae-Hyun Kim PII: DOI: Reference:
S0140-7007(15)00194-2 http://dx.doi.org/doi:10.1016/j.ijrefrig.2015.06.023 JIJR 3084
To appear in:
International Journal of Refrigeration
Received date: Revised date: Accepted date:
8-10-2014 30-3-2015 30-6-2015
Please cite this article as: Ho-Won Byun, Nae-Hyun Kim, Two-phase refrigerant distribution in an intermediate header of a parallel flow minichannel heat exchanger, International Journal of Refrigeration (2015), http://dx.doi.org/doi:10.1016/j.ijrefrig.2015.06.023. 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.
Two-phase refrigerant distribution in an intermediate header of a parallel flow minichannel heat exchanger
Ho-Won Byun, Nae-Hyun Kim*
University of Incheon, School of Mechanical System Engineering, Incheon 406-772, Korea
*Corresponding author School of Mechanical System Engineering, University of Incheon 12-1, Songdo-Dong, Yeonsu-Gu, Incheon 406-772 Korea Tel) +82-32-835-8420 Fax) +82-32-835-8410 e-mail) [email protected] (N.-H. Kim) [email protected] (H.-W. Byun)
1
Page 1 of 45
Highlights Two-phase distribution in an intermediate header is investigated. Insertion device does not affect the flow distribution in the second pass. Correlations are developed for liquid or gas ratio taken off by downstream channel. Header pressure drops are obtained from the calculated and measured pressure drops.
ABSTRACT Refrigerant R-410A distribution in a two pass evaporator with intermediate header was investigated. The number of tubes was 10 for the first pass and 12 for the second pass. Tests were conducted for the mass flux from 73 to 143 kg m-2s-1, quality from 0.4 to 0.6. In the intermediate header, two-phase mixture out of the first pass is merged and then re-distributed to the second pass. More liquid is forced downstream as mass flux or quality increases yielding better flow distribution. Effect of insertion device in the inlet header was also investigated. Efforts were made to develop correlations to predict the liquid or gas distribution in a header with limited success. Header pressure drop data are also provided.
Key words : Flow distribution, Heat exchanger, Header, Intermediate
2
Page 2 of 45
Nomenclature cp
specific heat, J kg-1K-1
CR
ratio of inlet and outlet channel number
d
tube inner diameter, m
D
header diameter, m
f
friction factor
Fr
Froude number
G
mass flux, kg m-2s-1
GFR
gas flow ratio
h
enthalpy, J kg-1
L
length, m
LFR
liquid flow ratio
m
mass flow rate, kg s-1
N
number of channels
P
pressure, Pa
Pc
critical pressure, Pa
R
function of measured variable
Q
rate of heat supply, W
Re
Reynolds number
T
temperature, K or tube
v
specific volume, m3kg-1
w
uncertainty of parameter
We
Weber number
x
quality or measured variable
3
Page 3 of 45
Greek notations
void fraction
P
pressure drop, Pa
viscosity, kg m-1s-1
2
two-phase flow multiplier
density, kg m-3
Surface tension, N m-1 or standard deviation
Subscripts a
acceleration
avg
average
ch
channel
cont
contraction
deg
degradation
exp
expansion
f
friction
ft
flat tube
g
gas or gravitation
go
gas only
H
header or homogeneous
i
inlet or ith
ideal
ideal
in
inlet
int
intermediate
l
liquid 4
Page 4 of 45
lg
latent heat
lo
all liquid
meas
measured
minor minor o
outlet
out
outlet
p
preheater
r
refrigerant
ran
random
rt
round tube
sat
saturation
sys
systematic
T
tube
w
cooling water
5
Page 5 of 45
1. Introduction
Brazed aluminum heat exchangers, which consist of flat minichannel tubes on the refrigerant-side and louver fins on the air-side, have long been used as condensers of automotive air conditioners due to superior thermal performance as compared with conventional fin-and-tube heat exchangers. Brazed aluminum heat exchangers may be categorized as parallel flow heat exchangers because a number of tubes are grouped to one pass using a header, and flows are parallel one another. Typical hydraulic diameter of the flat tube is 1 ~ 2 mm. Recently, brazed aluminum heat exchangers are considered as evaporators of automotive or residential air conditioners. In this case, it is very important to distribute the two-phase refrigerant (especially the liquid) evenly into each tube. Otherwise, the thermal performance is significantly deteriorated. According to Kulkarni et al. (2004), the performance reduction by flow mal-distribution could be as large as 20%. In general, brazed aluminum evaporators consist of more than twenty tubes to match the cooling requirement of the refrigeration system. Fig. 1 shows an example of refrigerant-side circuiting (one, two and four pass). Also shown in the figure are estimated vapor qualities for given inlet and exit quality of 0.2 and 1.0. Improved flow distribution is expected with increased number of refrigerant passes. The pressure drop will, however, also be increased. The literature survey reveals that most of the studies on two-phase distribution in a header-branch configuration have been conducted on a single pass configuration [Fig. 1 (a)]. For a single pass configuration, refrigerant is divided into channels at the inlet header, and is combined at the outlet header. Webb and Chung (2004), Hrnjak (2004), Lee (2006), Ahmad et al. (2009) provided recent reviews on this subject. For multi-pass configuration, refrigerant is first divided into channels at the inlet header. Refrigerant out of the first pass is combined, and then divided into channels in the intermediate header of the second pass. The process continues to the last header. It is expected that flow distribution characteristics in the intermediate header will be significantly different from that in the inlet or outlet header. However, literature shows very limited investigations on refrigerant distribution in the intermediate header. The literature survey will first briefly
6
Page 6 of 45
review the two-phase distribution in a single pass configuration, and then available studies on multi-pass configuration will be discussed. Watanabe et al. (1995) conducted a flow distribution study for a round header (20 mm I.D.) - four round tube (6 mm I.D.) upward flow configuration using R-11. Mass flux (based on the header cross sectional area) was varied from 40 to 120 kg m-2s-1, and inlet quality was varied up to 0.4. The flow at the inlet was stratified, and was supplied parallel to the header. The flow distribution was highly dependent on mass flux and quality. Vist and Pettersen (2004) investigated a round header (8mm and 16 mm I.D.) - ten round tube (4 mm I.D.) configuration using R-134a. Mass flux was varied from 12 to 21 kg m-2s-1, and quality was varied up to 0.5. The flow in the header inlet was mostly intermittent with some annular at high mass fluxes. For downward flow configuration, most of the liquid flowed through frontal part of the header. For upward configuration, on the contrary, most of the liquid flowed through the rear part of the header. Koyama et al. (2006) investigated the effect of tube protrusion depth for a horizontal round header (9 mm I.D.) - six vertical flat tube configuration using R-134a. The flow was supplied parallel to the header. Mass flux was fixed at 130 kg m-2s-1, and quality was varied up to 0.4. Tests were conducted for downward configuration, and the flow at header inlet was intermittent. Protrusion depth was systematically varied, and optimum configuration was found to be with front two tubes protruded to the center of the header and remaining four tubes flush-mounted. Better liquid distribution was obtained at a lower vapor quality. Bowers et al. (2006) investigated the effect of entrance length for a downward configuration using R-134a. Their test section composed of horizontal round header (20 mm I.D.) and fifteen vertical flat tubes. Mass flux was varied from 46 to 107 kg m-2s-1, and quality was varied up to 0.35. The apparatus was equipped an expansion valve, and expanded two-phase mixture was supplied to the test section through entrance tubes. For a short entrance length of 89 mm, liquid distribution was relatively uniform with minor influence of protrusion depth, mass flux or quality. For a long entrance length of 267 mm, however, better distribution was obtained as mass flux or protrusion depth increased. Hwang et al. (2007) investigated the effect of inlet configuration for a round header-thirty flat tube configuration using R-410A for upward flow. Two inlet configurations – parallel and normal – were tested, and
7
Page 7 of 45
better flow distribution was obtained for a normal inlet configuration. Kim et al. (2011) and Kim and Byun (2013) also investigated the effect of flow inlet direction for a round header (17 mm I.D.) - ten flat tube (Dh = 1.32 mm) configuration using R-134a for downward (2011) and upward (2013) flow. For downward flow (2011), normal or vertical inlet yielded approximately similar liquid distribution, although slightly better results were obtained for normal inlet at high mass fluxes or high qualities. For upward flow (2013) vertical inlet yielded the best flow distribution. Although some investigations were conducted on refrigerant distribution in a single pass configuration, very limited information is available for refrigerant distribution in a multi-pass configuration. Byun and Kim (2011) investigated the refrigerant distribution in a horizontal two pass evaporator having vertical intermediate header for upward flow of R-410A. Tubes were heated to yield a test section outlet superheat of 5oC with inlet quality of 0.3. Mass flux was varied from 50 kg m-2s-1 to 70 kg m-2s-1. Through flow visualization, they reported that two-phase jets, which exited the first pass, hit backside of the vertical intermediate header, and were deflected upward forming a thin liquid film. At the second pass, liquid was continuously sucked into the tubes, and got thinner as it traveled upstream. As mass flux increased, more liquid was supplied to upstream channels. Zou and Hrnjak (2013) also investigated the refrigerant distribution in a horizontal two pass evaporator having vertical intermediate header for upward flow of R-134a. Inlet quality at the first pass was varied from 0.2 to 0.8, and mass flux (based on branch tube) was varied from 93 kg m-2s-1 to 277 kg m-2s-1. Tubes were heated to yield a test section outlet superheat of 5oC. Two different flow pattern (churn and separated) was identified in the vertical header, where churn flow yielded better distribution due to more homogeneous nature of flow. Better flow distribution was obtained at high mass flux or low quality. The above literature survey reveals that virtually no systematic distribution study was conducted for a horizontal intermediate header. In this study, R-410A distribution in a two pass evaporator with upper horizontal intermediate header [and consequently inlet and outlet located at lower header as shown in Fig. 1(b)] was investigated. Tubes were heated to yield a test section outlet superheat of 5oC with inlet quality of 0.2. Then, the quality at the intermediate header is approximately 0.6. The quality of 0.4 at the intermediate header was
8
Page 8 of 45
also investigated to simulate the first intermediate header of four pass configuration [see Fig. 1(c)]. For each case, mass flux was varied from 73 kg m-2s-1 to 143 kg m-2s-1. The ratio of channel number (CR) was 10/12 (10 tubes for the first and 12 tubes for the second pass) with target evaporator capacity of 4.5 kW. The effect of insertion device located at lower inlet header on flow distribution was also investigated. In the test section, tubes were protruded to center of the header to simulate an actual heat exchanger. The flat tube had 1.32 mm hydraulic diameter and 12.24 mm2 flow cross-sectional area. Schematic drawing of the flat tube is shown in Fig. 2. The tube has eight ports of rectangular shape.
2. Experiments
The same apparatus of Kim et al. (2011) and Kim and Byun (2013) was used, and one may consult those papers for further details. A schematic drawing of the experimental apparatus is shown in Fig. 3. Detailed drawing and photo of the test section are shown in Fig. 4. The test section consists of the 17 mm I.D. round upper and lower headers, and 22 flat tubes inserted at 9.8 mm pitches simulating a one row heat exchanger. The vertical length of the flat tube is 90 cm. Refrigerant was supplied from the inlet header, passed through the intermediate header, and exited from outlet header. Refrigerant was supplied through copper tube having smaller inner diameter (7.92 mm) than that of the header. At outlet of the test section, copper tube with 11.1 mm I.D. was used. The tube diameters were chosen to simulate an actual parallel flow evaporator. The headers were made from transparent PVC rods for flow visualization. A 17 mm round hole was machined longitudinally in a square (60 mm x 55 mm) PVC rod with flat holes machined at the bottom of the rod for insertion of flat tubes. An aluminum plate, which had matching flat holes, was installed underneath the header. Flat tubes were protruded to center of the header, which were secured using O-rings between the header and the aluminum plate. Two transition blocks were installed at the central part of the test section, which connected the flat tubes and the 6.0 mm I.D. round tubes (see Fig. 5). The round tubes served as flow measurement lines. For an actual heat exchanger, fins are attached to the tube, providing heat from the process
9
Page 9 of 45
air. In this study, specially-made plate heaters having the same width and length with those of flat tubes were attached to every flat tubes as shown in Fig. 4(c). Thermal epoxy was applied between the tube and the heater, which were then tightly wrapped with nylon wire. The power to the heater was adjusted using a voltage transformer. To prevent heat loss to the environment, the test section was heavily insulated. When the supplied heat was compared with the acquired heat calculated from the refrigerant enthalpy change, the supplied heat was approximately 5% larger than the acquired heat. The heat loss occurred from the headers (not insulted for flow visualization) and many copper tube lines at the transition section. Thus, acquired heat was used for the calculation of local quality. For an actual heat exchanger, fins are attached to the tube, providing heat from the process air. During the tests, the inlet quality (xin) was fixed at 0.2. When the quality of the intermediate header (xint) was 0.6 simulating a two pass configuration [see Fig. 1(b)], exit superheat was maintained at 5oC. When xint was 0.4 simulating a four pass configuration [see Fig. 1(c)], exit quality was maintained at 0.6. The amount of heat to be supplied was determined considering the enthalpy difference and heat loss. Before the refrigerant flowed into the test section, quality of the refrigerant was controlled at the pre-heater (5 kW capacity) section by changing the power supply to the heater. The inlet quality is determined from the measured mass flow rate and supplied electric power as follows.
xi
1 hlg
Qp m r
c pr (T sat T p , i )
(1)
The refrigerant out of the test section was directed to a condenser, and the condensed refrigerant was pumped to the pre-heater. Mass flow rate was controlled by by-passing an appropriate amount of liquid from the magnetically coupled gear pump, and measured using a mass flow meter (Oval, CN101C-SS, 0 - 400 kg hr-1) which was installed before the pre-heater. The branch channel flow rate was measured for every other channel (channel 2, 4, 6, 8, 10 for first pass, and channel 12, 14, 16, 18, 20, 22 for second pass) by directing the refrigerant to the flow measurement section. Channel was numbered from inlet of the header. As shown in Fig.
10
Page 10 of 45
5, three manual valves – one at the main stream (valve 1), the other one at the bypass stream (valve 2) and the last one at the measurement line (valve 3) – were installed at every other channel. Normally, main stream valves (valve 1) were opened, and bypass stream valves (valve 2) were closed. To measure the flow rate at a certain channel, main stream valve was closed, and bypass valve was open. Pressure measurement valves (valve 3) were open all the time. During the measurement, differential pressure between inlet of the lower header and Tjunction of the transition section was maintained the same by controlling the valve 2 in the transition section. The differential pressure was measured using a high precision pressure transducer (Setra, Model 230, 0 - 34 kPa). Generally, two phase pressure drop is a function of mass flux and vapor quality. In the present study, total mass flux and vapor quality in addition to the channel pressure drop were maintained the same. It is thus anticipated that flow patterns would be the same before and during the measurement. During the test, pressure fluctuation occurred due to the nature of two phase flow. The fluctuation was within
10% of the average
value for most of the case. However, when the channel flow rate was small (for example, almost no liquid in the channel), the fluctuation increased up to
30% of the average value. To minimize the effect of pressure
fluctuation, data were taken more than 10 minutes after the system was stabilized. The liquid and vapor fraction (quality) of the flow was thermally determined using a double tube heat exchanger. The tube-side refrigerant was condensed by cooling water flowing at the annular side. The double tube heat exchanger was divided into five sub-sections to cover a wide range of thermal loads from different channels. Temperatures were measured at four locations of the double tube heat exchanger; refrigerant temperatures at inlet and outlet of the tube, cooling water temperatures at inlet and outlet of the annulus. Thermo-wells having five thermocouples (T-type) each were used to measure local temperatures. During the test, saturation temperature was maintained at 10oC. Two refrigerant pressures were also measured - one at the inlet of the heat exchanger, and the other at the outlet of the heat exchanger. The outlet refrigerant was maintained sub-cooled to insure sub-cooled flow into the flow meter. As shown in Eqs. (2) and (3), vapor quality of each channel at the measurement location ‘x’ is then determined from the heat balance between the
11
Page 11 of 45
refrigerant and cooling water. Channel inlet quality ‘xin’ is obtained by considering the heat added to the channel by plate heaters as shown in Eq. (4). Liquid or gas flow rate is determined from the channel inlet quality as shown in Eq. (5). .
x ( hr ,i hr ,l ) / hlg
(2)
h r , i h r , o m w c pw (T w , o T w , i ) / m r
(3)
x in x Q ch / hlg
(4)
m l (1 xin ) m r ,
m g xin m r
(5)
Here, Qch is the heat added to the channel from inlet to the measurement point. The refrigerant outlet enthalpy was determined from the measured temperature and pressure. The channel refrigerant flow rate was measured using a precision mass flow meter (Oval, CN003C-SS, 0 - 120 kg hr-1). The water flow rate was measured by weighing the drained water. During the whole series of tests, several runs were made to check the repeatability of the data. The data were repeatable within 10%. When the channel flow rates were added and compared with the supplied flow rate (for the channels where flow rates were not measured, the average values of the upstream and downstream channel flow rates were used), they agreed within 10%. In this study, flow distribution data were presented as liquid flow ratio (LFR) or gas flow ratio (GFR), which is defined as the ratio of liquid or gas flow rate in each channel to the average values as shown in Eq. (6).
LFR
ml
ml / N
,
GFR
mg
mg / N
(6)
For example, consider a inlet pass and assume that the total liquid and gas flow rate in the header is 1.0 kg s-1 and 0.1 kg s-1. Then, the average liquid and gas flow rate in each channel (for 5 measured channels in inlet pass)
12
Page 12 of 45
is 0.2 kg s-1 and 0.02 kg s-1. If the measured liquid and gas flow rate in the branch tube is 0.05 kg s-1 and 0.02 kg s-1, then the liquid flow ratio is 0.25 and the gas flow ratio is 1.0. Experimental uncertainties were analyzed following the method by Kline and McClintock (1953), and are listed in Table 1. According to Kline and McClintock (1953), when a parameter ‘R’ is a function of measured variables ( x1 , x 2 , , x n ) , the uncertainty on the parameter ‘wR’ is obtained from the following equation.
wR
R x1
w1
2
w2
R x2
2
R x n
wn
2
1/ 2
(7)
Here, ‘w1’ is the uncertainty on variable ‘x1’, ‘w2’ is the uncertainty on variable ‘x2’, etc. As an example, the systematic uncertainty on liquid flow ratio of channel 1 is obtained as follows. From Eqs. (2) to (7),
wLFR 1
L F R1 sys
2
1
ml / N
m l ,1 / N
m
l
/N
2
2 w m l ,1 LFR 1
wm
l
ml
ml ,2 / N
m
wm
2
/N
l
2
2
wh
r ,i
hr , i
c p (T w o T w i ) w m w
hr , i m r
2
ml,N / N m /N l
wm
2
l ,N
LFRN
1/ 2
(8)
1/ 2
wx mr x r
l ,2
L F R2
(9)
wx 1 wh x ( hr , i hr , l )
2
2
wm
m w c p wT w ,i
hr , i m r
2
(10)
r ,i
m w c p wT
w ,o
hr , i m r
2
m w c p (T w , o T w , i ) w m
hr , i m r 2
r
2
1/ 2
(11)
Introducing the measurement uncertainties listed in Table 1, the systematic uncertainty on the liquid flow ratio ranged 6.4% ~ 504.3%, and that on gas flow ratio from 8.6% ~ 544.0%, which increased as liquid or gas flow rate decreased. The total uncertainty may be obtained as follows, including the random uncertainty on liquid flow ratio.
w L F Ri
L F R i
tot
w L F Ri
L F R i
2
sys
w L F Ri
L F R i
( w LFR ) ran 2 i
13
2
(12) ran
(13)
Page 13 of 45
where
is standard deviation of liquid flow ratio. Including the random uncertainty, total uncertainty on the
liquid ratio ranged 11.9%~ 504.4% and that on the gas flow ratio raged 13.2~ 544.1%. Large uncertainty on liquid and gas ratio corresponded to channels with negligible liquid or gas flow. To ensure the correctness of the measurement, error bars are drown in the next figures.
3. Results and discussions
3.1 General flow pattern and effect of mass flux
Fig. 6(a) shows the liquid and gas flow ratio at inlets of each (first and second) pass for 10/12 configuration. The inlet quality of the first pass is 0.2 and that of the second pass is 0.4. For the first pass, Fig. 6(a) shows that, in general, liquid flow ratio increases as the channel number increases with less liquid at frontal channels. The effect of mass flux is not significant. The corresponding photos and sketches show that a twophase jet is supplied from the inlet tube, part of which hits the protruded tubes, and is deflected downward. Rest of the jet is delivered downstream bypassing protruded tubes. During the process, liquid film gets thicker along flow direction, reaching protruded tubes at a certain channel. From there on, the liquid level remains approximately the same. The actual flow, however, was quite unsteady. The liquid interface was wavy, and the liquid level changed intermittently. The still photo shown in the figure is a flow pattern just at one moment of the time. It was observed that liquid was intermittently sucked from the wavy interface. If the suction pressure at the inlet of the channel is large enough for the liquid suction from the interface, liquid is sucked into the channel. With the liquid in the channel, the suction pressure decreases. If the channel suction pressure is smaller than that needed for liquid suction, gas is sucked into the channel. With the gas in the channel, the suction pressure increases again sucking liquid into the channel. The process occurred repeatedly.
14
Page 14 of 45
Two-phase mixture out of the first pass is first merged and then redistributed to the second pass in the intermediate header. For the second pass, Fig. 6(a) shows that liquid distribution is strongly dependent on the mass flux. At a low mass flux of 73 kg m-2s-1, more liquid is supplied into frontal channels and less liquid is supplied into downstream channels. As mass flux increases, more liquid is supplied into downstream channels. Accompanying photos in Fig. 6(c) show that two-phase jets exits the first pass, hit top of the header, and are deflected downstream along topside of the header forming a thin liquid film with lots of mist drops in the core flow. The liquid film gets thicker as it travels downstream by added two-phase jets. At the inlet of the second pass, the liquid film (and mist drops as well) is sucked into the tubes and gets thinner as it travels downstream until all the liquid is sucked into the channels. Photos and sketches in Fig. 6(c) show that liquid pool with clockwise recirculating flow is formed between protruded tubes. Liquid is also sucked into tubes from these liquid pools. As mass flux increases, the strength of two-phase jet out of the first pass increases as evident from Fig. 5(c), providing stronger axial momentum, which delivers the liquid film further downstream.
Fig. 6
shows that gas distribution generally yields reversed trend to liquid distribution. It appears that combined effects of separated flow pattern and equal pressure drop constraint of each channel have yielded the reversed distribution pattern between liquid and gas. In Fig. 7, flow distribution data taken maintaining the inlet quality at the second pass (xint) to be 0.6 are shown. Similar trends to the previous xint = 0.4 case are shown. At the first pass, liquid flow ratio increases with the channel number with minimal effect of mass flux on liquid distribution. At the second pass, liquid is forced downstream as mass flux increases. One thing to be noted is that data were not obtainable at the highest mass flux (143 kg m-2s-1) due to limitation of experimental facility. Degradation of heat transfer capacity due to flow mal-distribution may be assessed by comparing the actual heat transfer rate (Q) and the ideal heat transfer rate (Qideal) (Byun and Kim, 2011). Heat transfer rate of each elementary channel can be calculated from the measured inlet, exit quality and channel mass flow rate using adequate air-side and tube-side correlation. Then, actual heat transfer rate (Q) is thus obtained by adding calculated channel heat transfer rates. Ideal heat transfer rate (Qideal) is obtained assuming uniform refrigerant
15
Page 15 of 45
distribution. Heat transfer degradation by flow mal-distribution may then be obtained from the following equation.
% deg
Q ideal Q Q ideal
(14)
Using the channel flow rate, inlet and outlet quality, heat transfer degradation was calculated. For the tubeside heat transfer coefficient, Shah (1976) correlation was used. Air-side thermal resistance [ 1 /( ho Ao ) ] was selected to yield 5oC superheat when uniform refrigerant distribution is assumed. Detailed description on this procedure is provided in Byun and Kim (2011). Results are shown in Table 2. Table 2 shows that heat transfer degradation decrease as mass flux increases. It has been shown that liquid is forced downstream as mass flux increases, and eventually improves flow distribution. The uniformity of a distribution may also be judged by a standard deviation. Standard deviations of liquid and gas flow ratio were calculated for the whole pass, and are listed in Table 3. The trends of standard deviation on liquid and gas flow ratio are approximately similar to the trend of heat transfer degradation shown in Table 2. The standard deviation generally decreases as mass flux increases.
3.2 Effect of vapor quality
Fig. 8 shows the effect of vapor quality on liquid and gas flow ratio in the intermediate header (xint). Data are shown for two different mass flux (G = 73 kg m-2s-1 and 108 kg m-2s-1). At the first pass, the effect of xint on liquid distribution is not significant, although slightly better distribution is obtained at high xint and high mass flux. At the second pass, the effect of xint is more evident. Liquid is forced downstream as xint increases, and as it happens, the strength of two-phase jet out of the first pass increases, providing stronger axial momentum, which delivers liquid further downstream. Fig. 8 shows that gas distribution generally yields reversed trend to liquid distribution. Heat transfer degradation was calculated as before, and the results are summarized in Table 2.
16
Page 16 of 45
Standard deviations of liquid and gas flow ratio were calculated for each pass, and are listed in Table 3. From Table 3, it can be concluded that standard deviation generally decreases as xint increases.
3.3 Effect of insertion device
Kim et al. (2013) have shown that refrigerant distribution in a parallel flow heat exchanger can be significantly improved by inserting a tube with small holes at the inlet of the header. The optimum hole geometry was dependent on the refrigerant flow direction. For the upward configuration, uniform hole configuration shown in Fig. 9 has been suggested. Thus, tests were conducted with the tube inserted at the inlet header. Representative results are shown in Fig. 10. These figures show that insertion device significantly improves the flow distribution at the first pass. Almost uniform distribution is obtained. At the second pass, however, the effect of insertion device on flow distribution is negligible. This suggests that, in the intermediate header, dividing characteristics of two-phase flow is not affected by the combining characteristics.
3.4 Flow distribution correlations
For two-phase flow distribution in a header/branch tube configuration, only limited models are available. Data and Majumdar (1983) extended the single-phase model of Bajura and Jones (1976) assuming a homogeneous two-phase flow. Ablanque et al. (2010) adopted T-junction models by Seeger et al. (1986) and Hwang et al. (1988) to predict the phase spilt at header/branch tube junction. Tuo et al. (2012) calculated the pressure drop at the header/branch junction using the experimentally determined pressure loss coefficient. All the above models require an iterative procedure, which satisfies equal pressure drop constraint along every header-channel-header pass, to obtain flow distribution in branch channels. In contrast, a simple correlating method was proposed by Watanabe et al. (1995) from their upward branching study using R-11.They noted that fraction of liquid taken off by downstream channel can be correlated by the header gas Reynolds number at
17
Page 17 of 45
immediate upstream. Zou and Hrnjak (2013) adopted the correlating method by Watanabe et al. (1995), and proposed a correlation for parallel branching of R-134a from a vertical header. Following the suggestion by Watanabe et al. (1995), correlations were developed to predict the fraction of liquid or gas taken off by downstream channel. The correlations were developed through a multiple linear regression analysis. Commercial statistical package (Datafit, 1996) was used to perform a multiple regression analysis, which provided the best fit of data. The accuracy of the correlation is evaluated using the “R-squared” value. Higher R2 value implies better correlation. Header gas Reynolds number (Reg) were chosen as an independent variable. Separate correlations were developed for the first and the second pass. One thing to note is that liquid and gas flow rate at odd numbered channels are obtained by interpolating those of the even numbered channels, because flow rates were measured only at even numbered channels. Listed below are the correlations.
first pass :
m Tl , i m H l ,i
second pass :
34.2 R e g
m Tl , i m H l ,i
0.63
4709 R e g
(liquid),
1.01
m T g ,i m H g ,i
(liquid) ,
6.9 R e g
m T g ,i m H g ,i
0.34
1479 R e g
(gas) ; (100 ≤ Reg ≤ 4ⅹ104)
0.92
(gas) ; (2ⅹ103 ≤ Reg ≤ 5ⅹ104)
(15)
(16)
The correlations are compared with the present data in Fig. 11. Fig. 11 along with Eqs. (15) - (16) shows that, for first pass, liquid or gas fraction taken off by downstream branch tube increases as gas Reynolds number decreases. In contrast, for the second pass, liquid or gas fraction increases as gas Reynolds number. Also listed in Fig. 11 are R2 values. Higher R2 value implies better correlation. In Fig. 12, the available upward flow data [vist and Pettersen (2004), Hwang et al. (2007), Watanabe et al. (1995)] are compared with the present correlation [Eq. (15)]. Large scatter of data is observed, although relatively good prediction is shown for Watanabe et al. (1995) data. The root mean square errors (RMSE) are 18
Page 18 of 45
listed in Table 4. The header/tube diameter ratio dimensional tube protrusion depth (D/d), header/tube crosssectional flow area ratio (AH/AT) and non-dimensional tube protrusion depth (h/D) of the present geometry (D/d=12.9, AH/AT=1.9 and h/D=0.5) are significantly different from those of Watanabe et al. (1995), Vist and Pettersen’s (2004) and Hwang et al. (2007). Both Watanabe et al.’s (1995) and Vist and Pettersen’s (2004) header had round branch tubes (compared to flat branch tubes of the present study), which yielded smaller header/branch tube diameter ratio [D/d=3.3 for Watanabe et al. (1995) and 2.4 for Vist and Pettersen (2004)]. Hwang et al.’s (2007) header had thirty flat tubes (compared to ten flat tubes of the present study), which resulted in smaller header/branch tube flow area ratio (AH/AT=0.33). In addition, tested mass flux ranges are significantly different from the present one. The above discussion suggests that the present correlation has limited applicability, which is understandable from the fact that only one variable (gas Reynolds number) was used for development of the correlation. More data are necessary, which account for the effect of geometric variables (non-dimensional tube protrusion depth, header/branch tube diameter ratio, header/branch tube flow area ratio, header shape, etc.) as well as operating variables (mass flux, quality, temperature, etc.). Refined correlation should consider both geometric and operating variables.
3.5 Header pressure drop
During flow distribution tests, pressure drops were also measured from inlet of the header to the T junctions of the transition section (as illustrated in Fig. 5 and 13). Measurements were made both at first and second pass. For the first pass, the measured pressure drop consists of expansion loss at header inlet, pressure drop in the inlet header, pressure drop across the flat tubes, expansion loss in the transition section from flat tube to round tube, pressure drop across round tube to the T junction as illustrated in Fig. 13. The expansion loss and round tube pressure drops are expected to be small compared to other pressure drops, and addition of them is termed as a ‘minor pressure drop’.
19
Page 19 of 45
PH , in Pm ea s ,1 st P ft ,1 st _ 1 Pm in o r ,1 st
(17)
Pm inor ,1 st Pexp , inlet Pexp ,1 st Prt ,1 st _ 1
(18)
Header pressure drop at the first pass may be determined from Eq. (17) if other pressure drops are obtained from appropriate models or correlations. Similarly, for the second pass, the measured pressure drop consists of expansion loss at header inlet, pressure drop in the inlet header, pressure drop across flat tubes, expansion loss in the transition section from flat tube to round tube, pressure drop across round tube, contraction loss in the transition section from flat tube to round tube, pressure drop across flat tubes, pressure drop in the intermediate header, pressure drop across the flat tube at second pass, expansion loss in the transition section from flat tube to round tube and pressure drop across round tube to the T junction as illustrated in Fig. 13. Pressure drop in the intermediate header is then obtained from the following equation. Average pressure drop in the first pass to be used with Eq. (19) is obtained from the averaged value of all the channels of inlet pass as shown in Eq. (20).
PH ,
Pm
i n t
P
e a s, 2n d
P
1 s t p a, s s . a v g
, 2
fPt
n d
, 2m i n o r
n d
(19)
N 1 st
P1 st pass , avg .
(P
m eas ,1 st , i
P ft ,1 st _ 2 , i Pcont ,1 st , i Prt ,1 st _ 2 , i ) / N 1 st
(20)
i 1
Pm in o r , 2 n d Pexp , 2 n d Prt , 2 n d
(21)
In Eq. (20), ‘N1st’ is a number of tubes where measurement was made at the first pass, which is 5. The pressure drop across the flat tube is composed of frictional pressure drop, accelerational pressure drop and gravitational pressure drop. For the round tube, accelerational pressure drop is not included because it is not heated.
P ft P ft , f P ft , a P ft , g
Prt Prt , f Prt , g
20
(22) (23)
Page 20 of 45
For the frictional pressure drop across the flat tube, Zhang and Webb (2001) correlation was used because of the geometric similarity between the present flat tube and Zhang and Webb’s samples. For the frictional pressure drop across the round tube, Friedel (1979) correlation was used. Acceleration pressure drop and gravitational pressure drop across the flat or round tube, expansion and contraction loss were calculated following Collier and Thome (1994). For the void fraction, Zivi’s (1964) equation was used. The pressure drops and corresponding correlations are summarized in Table 5. The header pressure drop of each channel was obtained from Eq. (17) and (19) using measured channel mass flux and quality, and typical results are shown in Fig. 14. Fig. 14 shows the pressure drops of the first and second pass at G = 108 kg m-2s-1 and xint = 0.4. Fig. 14 (a) shows that inlet header pressure drop decreases as channel number increases. Fig. 14 (b) shows that pressure drops in intermediate header are relatively uniform irrespective of channel number. Such trend was also observed all mass fluxes and qualities. At the present time, the reason of the pressure drop trend is not clear. Complex flow pattern in the headers may be responsible. The measured pressure drop is denoted as lines with symbol ‘x’. One thing to be noted from Fig. 14 is that, between channels, measured pressure drops are not significantly different. Measured and calculated inlet header pressure drops at other mass fluxes and qualities are summarized in Table 6. Measured and calculated intermediate header pressure drops are summarized in Table 7. For the first pass (Table 6), the measured pressure drop in each channel increases as mass flux increases. Same trend is shown for the calculated inlet header pressure drop. As for the effect of quality, no general trend is observed both on the measured pressure drop and the calculated inlet header pressure drop. On the contrary, for the second pass (Table 7), the measured pressure drops as well as the intermediate header pressure drop increases as mass flux or quality increases. Table 6 shows that calculated intermediate pressure drops of channel 8 and 10 at G=73 kg m-2s-1, xint=0.4 are negative. This means that pressure is continuously recovered in the header, which is realized by branching of the flow in previous channels.
21
Page 21 of 45
4. Conclusions Refrigerant R-410A two-phase flow distributions were experimentally studied for a two pass evaporator with intermediate header simulating a brazed aluminum heat exchanger. Tests were conducted flow for the mass flux from 73 to 143 kg m-2s-1 and quality (xint) from 0.4 to 0.6. Listed below are major conclusions.
(1) In the intermediate header, two-phase mixture from the first pass is merged and then re-distributed to the second pass. More liquid is forced downstream as header mass flux or quality increases yielding better flow distribution (or smaller heat transfer degradation). (2) With the insertion device installed in the inlet header, flow distribution at the first pass is significantly improved. However, at the second pass, the effect of insertion device is negligible. (3) Correlations are developed to predict the fraction of liquid or gas taken off by downstream or upstream channel as a function of header gas Reynolds number. (4) Header pressure drops are obtained by subtracting flat tube pressure drop and other minor pressure drops from measured pressure drops. For the inlet header, the header pressure drop increases as mass flux increases. For the intermediate header, the pressure drop increases as mass flux or quality increases.
Acknowledgements This work was supported by Incheon National University Research Grant in 2014.
22
Page 22 of 45
References Ablanque, N., Oliet, C., Rigola, J., Perez-Segarra, C.D., Oliva, A., 2010. Two-phase flow distribution in multiple parallel tubes, Int. J. Thermal Sci., 49, 909-921. Ahmad, M., Berthoud, G., Mercier, P., 2009. General characteristics of two-phase flow distribution in a compact heat exchanger, Int. J. Heat Mass Transfer, 52, 442-450. Bowers, C.S., Hrnjak, P.S., Newell, T.A., 2006. Two-phase refrigerant distribution in a micro-channel manifold, Proceedings of the 11th Int. Refrigeration and Air Conditioning Conference at Purdue, R161. Byun, H.W., Kim, N.H., 2011. Refrigerant distribution in a parallel flow heat exchanger having vertical headers and heated horizontal tubes, Exp. Thermal Fluid Sci., 35, 920-932. Collier, J.G., Thome, J.R., 1994. Convective Boiling and Condensation, 3rd ed., Oxford University Press. Data, A.B., Majumdar, A.K., 1983. A calculation procedure for two phase flow distribution in manifolds with and without heat transfer, Int. J. Heat Mass Transfer, 26 (9), 1321-1328. Datafit, 1996. Oakdale Engineering, http://www.oakdaleengr.com/, Version 8.2. Friedel, L., 1979. Improved pressure drop correlations for horizontal and vertical two-phase pipe flow, 3R Int. 18, 485-492. Hrnjak, P., 2004. Flow distribution issues in parallel flow heat exchangers, ASHRAE Annual Meeting, AN-041-2. Hwang, S.T., Soliman, H.M., Lahey, R.T., 1988. Phase separation in dividing two-phase flow, Int. J. Multiphase Flow, 14 (4) 439-458. Hwang, Y., Jin, D.H., Radermacher, R., 2007. Refrigerant distribution in minichannel evaporator manifolds, HVAC&R Research, 13 (4) 543-555. Kim, N.H., Byun, H.W., 2013. Effect of inlet configuration on upward branching of two-phase refrigerant in parallel flow heat exchanger, Int. J. Ref., 36, 1062-1077. Kim, N.H., Kim, D.Y., Byun, H.W., 2011. Effect of inlet configuration on the refrigerant distribution in a parallel flow minichannel heat exchanger, Int. J. Ref., 34, 1209-1221.
23
Page 23 of 45
Kim, N.H., Lee, E.J., Byun, H.W., 2013. Improvement of two-phase refrigerant distribution in a parallel flow minichannel heat exchanger using insertion devices, Applied Thermal Engineering, 59, 116-130. Kline, S.J., McClintock, F.A., 1953. The description of uncertainties in single sample experiments, Mechanical Engineering, 75, 3-9. Koyama, S., Wijayanta, A.T., Kuwahara, K., Ikuda, S., 2006. Developing two-phase flow distribution in horizontal headers with downward micro-channel branches, Proceedings of the 11th Int. Refrigeration and Air Conditioning Conference at Purdue, R142. Kulkarni, T., Bullard, C.W., Cho, K., 2004. Header design tradeoffs in microchannel evaporators, Applied Thermal Engineering, 24, 759-776. Lee, S.Y., 2006. Flow distribution behaviour in condensers and evaporators, Proceedings of the 13th International Heat Transfer Conference, KN-08, Sydney, Australia. Seeger, W., Reimann, J., Muller, U., 1986. Two-phase flow in a T-junction with a horizontal inlet, part 1: phase saparation, Int. J. Multiphase Flow, 12 (4) 575-585. Shah, M.M., 1976. A new correlation for heat transfer during boiling flow through pipes, ASHRAE Trans., 82 (2), 66-86. Tuo, H., Bielskus, A., Hrnjak, P., 2012. An experimentally validated modeling of refrigerant distribution in a parallel microchannel evaporator, ASHRAE Trans., 118 (1) CH-12-C-48. Vist, S., Pettersen, J., 2004. Two-phase flow distribution in compact heat exchanger manifolds, Exp. Thermal Fluid Sci., 28, 209-215. Watanabe, M., Katsuda, M., Nagata, K., 1995. Two-phase flow distribution in multi-pass tube modeling serpentine type evaporator, ASME/JSME Thermal Engineering Conf., 2, 35-42. Webb, R.L., Chung, K., 2004. Two-phase flow distribution in tubes of parallel flow heat exchangers, Heat Transfer Engineering, 26, 3-18. Zhang, M., Webb, R.L., 2001. Correlation of two-phase friction for refrigerants in small-diameter tubes, Exp. Thermal Fluid Sci., 25, 131-139.
24
Page 24 of 45
Zivi, S.M., 1964. Estimation of steady-state steam void fraction by means of the principle of minimum entropy production, J. Heat Transfer 86, 247-251. Zou, Y., Hrnjak, P., 2013. Experiment and visualization on R-134a upward flow in the vertical header of microchannel heat exchanger and its effect on distribution, Int. J. Heat Mass Transfer, 62, 124-134.
25
Page 25 of 45
Figure captions
Figure 1 Refrigerant-side circuiting (one, two and four pass) investigated in this study Figure 2 Schematic drawing of the flat tube used in this study (unit: mm) Figure 3 Schematic drawing of the experimental apparatus Figure 4 Schematic drawing of the test section showing and photos and sketches of the test section Figure 5 Schematic illustrating of flow measurement principle Figure 6 Flow distribution data and photos and sketches showing the effect of mass flux on flow distribution (xint=0.4) Figure 7 Flow distribution data showing the effect of mass flux on flow distribution (xint=0.6) Figure 8 Flow distribution data showing the effect of inlet vapor quality at second pass (xint) Figure 9 Insertion device used by Kim et al.(2013) (unit: mm) Figure 10 Effect of insertion device on flow distribution Figure 11 Correlations for liquid and gas fraction taken off by downstream branch tube using gas Reynolds number of upstream header Figure 12 Available upward flow distribution data compared with present correlation [Eq. (13)] Figure 13 Schematic drawing illustrating pressure losses from the header to the flow measurement location Figure 14 Typical pressure loss distributions along each channel for the first and second pass (G=108 kg m-2s-1, xint=0.4)
26
Page 26 of 45
Table 1 Experimental uncertainties Measurement
Uncertainty
heater power
±0.2%
mass flow rate
±0.1% of full scale
differential pressure
±0.3% of full scale
absolute pressure
±0.2% of full scale
temperature
±0.5K
liquid flow ratio
± 504.4%
gas flow ratio
± 544.1%
27
Page 27 of 45
Table 2 Heat transfer degradation by flow mal-distribution G (kg m-2s-1) 73
108 143
xint
%deg
0.4
10.5
0.6
31.0
0.4
6.1
0.6
6.1
0.4
1.8
28
Page 28 of 45
Table 3 Standard deviations of liquid or gas flow ratio G (kg m-2s-1) 73
108 143
xint
liquid
gas
0.4
0.57
0.79
0.6
0.51
0.44
0.4
0.49
0.58
0.6
0.33
0.32
0.4
0.45
0.39
29
Page 29 of 45
Table 4 RMSE of the available upward flow data predicted by the present correlation
Investigators
liquid
gas
Vist and Pettersen (2004)
4.28
4.39
Hwang et al. (2007)
2.95
4.47
Watanabe et al. (1995)
0.30
0.49
30
Page 30 of 45
Table 5 Pressure drops and corresponding correlations Pressure
Correlations
drop
Reference
p ft, f p l lo 2
p l = 2 f l ( L /d ) l G
ΔPft,f
R e lo
GD
2
0 .0 7 9
, fl
l
R e lo
0 .2 5
lo (1 x ) 2 .8 7 x ( 2
2
(2300 ≤Relo) , f l
P
2
)
1
1 .6 8 x
0 .8
Zhang and
16 R e lo
(1 x )
0 .2 5
(Relo<2300) P
(
Pc
)
Webb (2001)
1 .6 4
Pc
p rt, f p l lo 2
p l = 2 f l ( L /d ) l G
R e lo
GD
2
Fr
R e lo
Fr
0 .0 4 5
We
2
2
0.78
G d
, We
2
C F 1 (1 x ) x (
l
2
(1 x )
x
g
0.24
(
f go
)(
(Relo<2300)
l g
)
0.91
(
g l
)
0.19
(1
ΔPrt,g
(1 (
1 x x
)(
g l
(1 ) l
)
2/3
Friedel (1979)
)
g l
)
0.7
2
P ft , g or Prt , g g g (1 ) l L
g
1
)
ΔPft,g
2
f lo
(1 x )
2
, H
H
p ft, a G [
ΔPcont
R e lo
x 1 x l g
ΔPft,a
ΔPexp
16
0 .0 3 5
2
gd H
CF2 x
(2300 ≤Relo) , f l
0 .2 5
3 .2 4 C F 2
lo = C F 1 +
G
0 .0 7 9
, fl
l
ΔPrt,f
2
Collier and Thome (1994)
]
Collier and Thome (1994)
1
Zivi (1964)
(1 x ) 2 g x 2 2 Pexp G in (1 ) l (1 ) l
Pcont
2 g G l 1 1 1 1 2 1 2 C c l 2 out
31
f
x
Collier and Thome (1994)
Page 31 of 45
Table 6 Measured and calculated inlet header pressure drop in each channel for the first pass G (kg m-2s-1) 73
108 143
ΔP(kPa)
xint 2
4
6
8
10
0.4
3.31(1.48)*
3.52(0.55)
4.27(0.43)
4.14(-0.53)
4.55(-0.42)
0.6
3.17(1.81)
3.17(0.63)
3.79(0.86)
4.14(1.13)
4.21(1.18)
0.4
4.34(3.42)
4.41(1.41)
4.62(0.96)
5.45(0.96)
5.31(0.44)
0.6
4.62(2.60)
4.89(1.92)
4.55(1.46)
4.96(1.73)
5.03(1.64)
0.4
6.69(3.58)
6.48(3.30)
6.27(2.53)
6.41(2.27)
6.89(1.38)
*Inlet header pressure drop in parenthesis
32
Page 32 of 45
Table 7 Measured and calculated intermediate header pressure drop in each channel for the second pass G (kg m-2s-1)
73
108 143
ΔP(kPa) xint 12
14
16
18
20
22
0.4
9.65(4.60)*
9.65(4.83)
9.65(4.32)
9.58(2.84)
9.86(3.45)
10.00(3.84)
0.6
10.13(3.55)
10.27(3.31)
10.20(3.28)
9.93(3.62)
10.20(4.47)
10.41(4.84)
0.4
14.13(5.44)
14.13(6.04)
14.13(6.04)
13.99(6.22)
14.48(5.72)
14.27(6.56)
0.6
16.75(9.13)
16.61(8.97)
16.48(7.60)
16.13(7.66)
17.17(8.28)
16.82(8.08)
0.4
22.82(11.16)
23.03(10.74)
22.20(10.59)
21.85(10.75)
23.16(11.82)
23.30(11.34)
*intermediate header pressure drop in parenthesis
33
Page 33 of 45
34
Page 34 of 45
35
Page 35 of 45
36
Page 36 of 45
37
Page 37 of 45
38
Page 38 of 45
39
Page 39 of 45
40
Page 40 of 45
41
Page 41 of 45
42
Page 42 of 45
43
Page 43 of 45
44
Page 44 of 45
45
Page 45 of 45
S0140-7007(15)00194-2 http://dx.doi.org/doi:10.1016/j.ijrefrig.2015.06.023 JIJR 3084
To appear in:
International Journal of Refrigeration
Received date: Revised date: Accepted date:
8-10-2014 30-3-2015 30-6-2015
Please cite this article as: Ho-Won Byun, Nae-Hyun Kim, Two-phase refrigerant distribution in an intermediate header of a parallel flow minichannel heat exchanger, International Journal of Refrigeration (2015), http://dx.doi.org/doi:10.1016/j.ijrefrig.2015.06.023. 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.
Two-phase refrigerant distribution in an intermediate header of a parallel flow minichannel heat exchanger
Ho-Won Byun, Nae-Hyun Kim*
University of Incheon, School of Mechanical System Engineering, Incheon 406-772, Korea
*Corresponding author School of Mechanical System Engineering, University of Incheon 12-1, Songdo-Dong, Yeonsu-Gu, Incheon 406-772 Korea Tel) +82-32-835-8420 Fax) +82-32-835-8410 e-mail) [email protected] (N.-H. Kim) [email protected] (H.-W. Byun)
1
Page 1 of 45
Highlights Two-phase distribution in an intermediate header is investigated. Insertion device does not affect the flow distribution in the second pass. Correlations are developed for liquid or gas ratio taken off by downstream channel. Header pressure drops are obtained from the calculated and measured pressure drops.
ABSTRACT Refrigerant R-410A distribution in a two pass evaporator with intermediate header was investigated. The number of tubes was 10 for the first pass and 12 for the second pass. Tests were conducted for the mass flux from 73 to 143 kg m-2s-1, quality from 0.4 to 0.6. In the intermediate header, two-phase mixture out of the first pass is merged and then re-distributed to the second pass. More liquid is forced downstream as mass flux or quality increases yielding better flow distribution. Effect of insertion device in the inlet header was also investigated. Efforts were made to develop correlations to predict the liquid or gas distribution in a header with limited success. Header pressure drop data are also provided.
Key words : Flow distribution, Heat exchanger, Header, Intermediate
2
Page 2 of 45
Nomenclature cp
specific heat, J kg-1K-1
CR
ratio of inlet and outlet channel number
d
tube inner diameter, m
D
header diameter, m
f
friction factor
Fr
Froude number
G
mass flux, kg m-2s-1
GFR
gas flow ratio
h
enthalpy, J kg-1
L
length, m
LFR
liquid flow ratio
m
mass flow rate, kg s-1
N
number of channels
P
pressure, Pa
Pc
critical pressure, Pa
R
function of measured variable
Q
rate of heat supply, W
Re
Reynolds number
T
temperature, K or tube
v
specific volume, m3kg-1
w
uncertainty of parameter
We
Weber number
x
quality or measured variable
3
Page 3 of 45
Greek notations
void fraction
P
pressure drop, Pa
viscosity, kg m-1s-1
2
two-phase flow multiplier
density, kg m-3
Surface tension, N m-1 or standard deviation
Subscripts a
acceleration
avg
average
ch
channel
cont
contraction
deg
degradation
exp
expansion
f
friction
ft
flat tube
g
gas or gravitation
go
gas only
H
header or homogeneous
i
inlet or ith
ideal
ideal
in
inlet
int
intermediate
l
liquid 4
Page 4 of 45
lg
latent heat
lo
all liquid
meas
measured
minor minor o
outlet
out
outlet
p
preheater
r
refrigerant
ran
random
rt
round tube
sat
saturation
sys
systematic
T
tube
w
cooling water
5
Page 5 of 45
1. Introduction
Brazed aluminum heat exchangers, which consist of flat minichannel tubes on the refrigerant-side and louver fins on the air-side, have long been used as condensers of automotive air conditioners due to superior thermal performance as compared with conventional fin-and-tube heat exchangers. Brazed aluminum heat exchangers may be categorized as parallel flow heat exchangers because a number of tubes are grouped to one pass using a header, and flows are parallel one another. Typical hydraulic diameter of the flat tube is 1 ~ 2 mm. Recently, brazed aluminum heat exchangers are considered as evaporators of automotive or residential air conditioners. In this case, it is very important to distribute the two-phase refrigerant (especially the liquid) evenly into each tube. Otherwise, the thermal performance is significantly deteriorated. According to Kulkarni et al. (2004), the performance reduction by flow mal-distribution could be as large as 20%. In general, brazed aluminum evaporators consist of more than twenty tubes to match the cooling requirement of the refrigeration system. Fig. 1 shows an example of refrigerant-side circuiting (one, two and four pass). Also shown in the figure are estimated vapor qualities for given inlet and exit quality of 0.2 and 1.0. Improved flow distribution is expected with increased number of refrigerant passes. The pressure drop will, however, also be increased. The literature survey reveals that most of the studies on two-phase distribution in a header-branch configuration have been conducted on a single pass configuration [Fig. 1 (a)]. For a single pass configuration, refrigerant is divided into channels at the inlet header, and is combined at the outlet header. Webb and Chung (2004), Hrnjak (2004), Lee (2006), Ahmad et al. (2009) provided recent reviews on this subject. For multi-pass configuration, refrigerant is first divided into channels at the inlet header. Refrigerant out of the first pass is combined, and then divided into channels in the intermediate header of the second pass. The process continues to the last header. It is expected that flow distribution characteristics in the intermediate header will be significantly different from that in the inlet or outlet header. However, literature shows very limited investigations on refrigerant distribution in the intermediate header. The literature survey will first briefly
6
Page 6 of 45
review the two-phase distribution in a single pass configuration, and then available studies on multi-pass configuration will be discussed. Watanabe et al. (1995) conducted a flow distribution study for a round header (20 mm I.D.) - four round tube (6 mm I.D.) upward flow configuration using R-11. Mass flux (based on the header cross sectional area) was varied from 40 to 120 kg m-2s-1, and inlet quality was varied up to 0.4. The flow at the inlet was stratified, and was supplied parallel to the header. The flow distribution was highly dependent on mass flux and quality. Vist and Pettersen (2004) investigated a round header (8mm and 16 mm I.D.) - ten round tube (4 mm I.D.) configuration using R-134a. Mass flux was varied from 12 to 21 kg m-2s-1, and quality was varied up to 0.5. The flow in the header inlet was mostly intermittent with some annular at high mass fluxes. For downward flow configuration, most of the liquid flowed through frontal part of the header. For upward configuration, on the contrary, most of the liquid flowed through the rear part of the header. Koyama et al. (2006) investigated the effect of tube protrusion depth for a horizontal round header (9 mm I.D.) - six vertical flat tube configuration using R-134a. The flow was supplied parallel to the header. Mass flux was fixed at 130 kg m-2s-1, and quality was varied up to 0.4. Tests were conducted for downward configuration, and the flow at header inlet was intermittent. Protrusion depth was systematically varied, and optimum configuration was found to be with front two tubes protruded to the center of the header and remaining four tubes flush-mounted. Better liquid distribution was obtained at a lower vapor quality. Bowers et al. (2006) investigated the effect of entrance length for a downward configuration using R-134a. Their test section composed of horizontal round header (20 mm I.D.) and fifteen vertical flat tubes. Mass flux was varied from 46 to 107 kg m-2s-1, and quality was varied up to 0.35. The apparatus was equipped an expansion valve, and expanded two-phase mixture was supplied to the test section through entrance tubes. For a short entrance length of 89 mm, liquid distribution was relatively uniform with minor influence of protrusion depth, mass flux or quality. For a long entrance length of 267 mm, however, better distribution was obtained as mass flux or protrusion depth increased. Hwang et al. (2007) investigated the effect of inlet configuration for a round header-thirty flat tube configuration using R-410A for upward flow. Two inlet configurations – parallel and normal – were tested, and
7
Page 7 of 45
better flow distribution was obtained for a normal inlet configuration. Kim et al. (2011) and Kim and Byun (2013) also investigated the effect of flow inlet direction for a round header (17 mm I.D.) - ten flat tube (Dh = 1.32 mm) configuration using R-134a for downward (2011) and upward (2013) flow. For downward flow (2011), normal or vertical inlet yielded approximately similar liquid distribution, although slightly better results were obtained for normal inlet at high mass fluxes or high qualities. For upward flow (2013) vertical inlet yielded the best flow distribution. Although some investigations were conducted on refrigerant distribution in a single pass configuration, very limited information is available for refrigerant distribution in a multi-pass configuration. Byun and Kim (2011) investigated the refrigerant distribution in a horizontal two pass evaporator having vertical intermediate header for upward flow of R-410A. Tubes were heated to yield a test section outlet superheat of 5oC with inlet quality of 0.3. Mass flux was varied from 50 kg m-2s-1 to 70 kg m-2s-1. Through flow visualization, they reported that two-phase jets, which exited the first pass, hit backside of the vertical intermediate header, and were deflected upward forming a thin liquid film. At the second pass, liquid was continuously sucked into the tubes, and got thinner as it traveled upstream. As mass flux increased, more liquid was supplied to upstream channels. Zou and Hrnjak (2013) also investigated the refrigerant distribution in a horizontal two pass evaporator having vertical intermediate header for upward flow of R-134a. Inlet quality at the first pass was varied from 0.2 to 0.8, and mass flux (based on branch tube) was varied from 93 kg m-2s-1 to 277 kg m-2s-1. Tubes were heated to yield a test section outlet superheat of 5oC. Two different flow pattern (churn and separated) was identified in the vertical header, where churn flow yielded better distribution due to more homogeneous nature of flow. Better flow distribution was obtained at high mass flux or low quality. The above literature survey reveals that virtually no systematic distribution study was conducted for a horizontal intermediate header. In this study, R-410A distribution in a two pass evaporator with upper horizontal intermediate header [and consequently inlet and outlet located at lower header as shown in Fig. 1(b)] was investigated. Tubes were heated to yield a test section outlet superheat of 5oC with inlet quality of 0.2. Then, the quality at the intermediate header is approximately 0.6. The quality of 0.4 at the intermediate header was
8
Page 8 of 45
also investigated to simulate the first intermediate header of four pass configuration [see Fig. 1(c)]. For each case, mass flux was varied from 73 kg m-2s-1 to 143 kg m-2s-1. The ratio of channel number (CR) was 10/12 (10 tubes for the first and 12 tubes for the second pass) with target evaporator capacity of 4.5 kW. The effect of insertion device located at lower inlet header on flow distribution was also investigated. In the test section, tubes were protruded to center of the header to simulate an actual heat exchanger. The flat tube had 1.32 mm hydraulic diameter and 12.24 mm2 flow cross-sectional area. Schematic drawing of the flat tube is shown in Fig. 2. The tube has eight ports of rectangular shape.
2. Experiments
The same apparatus of Kim et al. (2011) and Kim and Byun (2013) was used, and one may consult those papers for further details. A schematic drawing of the experimental apparatus is shown in Fig. 3. Detailed drawing and photo of the test section are shown in Fig. 4. The test section consists of the 17 mm I.D. round upper and lower headers, and 22 flat tubes inserted at 9.8 mm pitches simulating a one row heat exchanger. The vertical length of the flat tube is 90 cm. Refrigerant was supplied from the inlet header, passed through the intermediate header, and exited from outlet header. Refrigerant was supplied through copper tube having smaller inner diameter (7.92 mm) than that of the header. At outlet of the test section, copper tube with 11.1 mm I.D. was used. The tube diameters were chosen to simulate an actual parallel flow evaporator. The headers were made from transparent PVC rods for flow visualization. A 17 mm round hole was machined longitudinally in a square (60 mm x 55 mm) PVC rod with flat holes machined at the bottom of the rod for insertion of flat tubes. An aluminum plate, which had matching flat holes, was installed underneath the header. Flat tubes were protruded to center of the header, which were secured using O-rings between the header and the aluminum plate. Two transition blocks were installed at the central part of the test section, which connected the flat tubes and the 6.0 mm I.D. round tubes (see Fig. 5). The round tubes served as flow measurement lines. For an actual heat exchanger, fins are attached to the tube, providing heat from the process
9
Page 9 of 45
air. In this study, specially-made plate heaters having the same width and length with those of flat tubes were attached to every flat tubes as shown in Fig. 4(c). Thermal epoxy was applied between the tube and the heater, which were then tightly wrapped with nylon wire. The power to the heater was adjusted using a voltage transformer. To prevent heat loss to the environment, the test section was heavily insulated. When the supplied heat was compared with the acquired heat calculated from the refrigerant enthalpy change, the supplied heat was approximately 5% larger than the acquired heat. The heat loss occurred from the headers (not insulted for flow visualization) and many copper tube lines at the transition section. Thus, acquired heat was used for the calculation of local quality. For an actual heat exchanger, fins are attached to the tube, providing heat from the process air. During the tests, the inlet quality (xin) was fixed at 0.2. When the quality of the intermediate header (xint) was 0.6 simulating a two pass configuration [see Fig. 1(b)], exit superheat was maintained at 5oC. When xint was 0.4 simulating a four pass configuration [see Fig. 1(c)], exit quality was maintained at 0.6. The amount of heat to be supplied was determined considering the enthalpy difference and heat loss. Before the refrigerant flowed into the test section, quality of the refrigerant was controlled at the pre-heater (5 kW capacity) section by changing the power supply to the heater. The inlet quality is determined from the measured mass flow rate and supplied electric power as follows.
xi
1 hlg
Qp m r
c pr (T sat T p , i )
(1)
The refrigerant out of the test section was directed to a condenser, and the condensed refrigerant was pumped to the pre-heater. Mass flow rate was controlled by by-passing an appropriate amount of liquid from the magnetically coupled gear pump, and measured using a mass flow meter (Oval, CN101C-SS, 0 - 400 kg hr-1) which was installed before the pre-heater. The branch channel flow rate was measured for every other channel (channel 2, 4, 6, 8, 10 for first pass, and channel 12, 14, 16, 18, 20, 22 for second pass) by directing the refrigerant to the flow measurement section. Channel was numbered from inlet of the header. As shown in Fig.
10
Page 10 of 45
5, three manual valves – one at the main stream (valve 1), the other one at the bypass stream (valve 2) and the last one at the measurement line (valve 3) – were installed at every other channel. Normally, main stream valves (valve 1) were opened, and bypass stream valves (valve 2) were closed. To measure the flow rate at a certain channel, main stream valve was closed, and bypass valve was open. Pressure measurement valves (valve 3) were open all the time. During the measurement, differential pressure between inlet of the lower header and Tjunction of the transition section was maintained the same by controlling the valve 2 in the transition section. The differential pressure was measured using a high precision pressure transducer (Setra, Model 230, 0 - 34 kPa). Generally, two phase pressure drop is a function of mass flux and vapor quality. In the present study, total mass flux and vapor quality in addition to the channel pressure drop were maintained the same. It is thus anticipated that flow patterns would be the same before and during the measurement. During the test, pressure fluctuation occurred due to the nature of two phase flow. The fluctuation was within
10% of the average
value for most of the case. However, when the channel flow rate was small (for example, almost no liquid in the channel), the fluctuation increased up to
30% of the average value. To minimize the effect of pressure
fluctuation, data were taken more than 10 minutes after the system was stabilized. The liquid and vapor fraction (quality) of the flow was thermally determined using a double tube heat exchanger. The tube-side refrigerant was condensed by cooling water flowing at the annular side. The double tube heat exchanger was divided into five sub-sections to cover a wide range of thermal loads from different channels. Temperatures were measured at four locations of the double tube heat exchanger; refrigerant temperatures at inlet and outlet of the tube, cooling water temperatures at inlet and outlet of the annulus. Thermo-wells having five thermocouples (T-type) each were used to measure local temperatures. During the test, saturation temperature was maintained at 10oC. Two refrigerant pressures were also measured - one at the inlet of the heat exchanger, and the other at the outlet of the heat exchanger. The outlet refrigerant was maintained sub-cooled to insure sub-cooled flow into the flow meter. As shown in Eqs. (2) and (3), vapor quality of each channel at the measurement location ‘x’ is then determined from the heat balance between the
11
Page 11 of 45
refrigerant and cooling water. Channel inlet quality ‘xin’ is obtained by considering the heat added to the channel by plate heaters as shown in Eq. (4). Liquid or gas flow rate is determined from the channel inlet quality as shown in Eq. (5). .
x ( hr ,i hr ,l ) / hlg
(2)
h r , i h r , o m w c pw (T w , o T w , i ) / m r
(3)
x in x Q ch / hlg
(4)
m l (1 xin ) m r ,
m g xin m r
(5)
Here, Qch is the heat added to the channel from inlet to the measurement point. The refrigerant outlet enthalpy was determined from the measured temperature and pressure. The channel refrigerant flow rate was measured using a precision mass flow meter (Oval, CN003C-SS, 0 - 120 kg hr-1). The water flow rate was measured by weighing the drained water. During the whole series of tests, several runs were made to check the repeatability of the data. The data were repeatable within 10%. When the channel flow rates were added and compared with the supplied flow rate (for the channels where flow rates were not measured, the average values of the upstream and downstream channel flow rates were used), they agreed within 10%. In this study, flow distribution data were presented as liquid flow ratio (LFR) or gas flow ratio (GFR), which is defined as the ratio of liquid or gas flow rate in each channel to the average values as shown in Eq. (6).
LFR
ml
ml / N
,
GFR
mg
mg / N
(6)
For example, consider a inlet pass and assume that the total liquid and gas flow rate in the header is 1.0 kg s-1 and 0.1 kg s-1. Then, the average liquid and gas flow rate in each channel (for 5 measured channels in inlet pass)
12
Page 12 of 45
is 0.2 kg s-1 and 0.02 kg s-1. If the measured liquid and gas flow rate in the branch tube is 0.05 kg s-1 and 0.02 kg s-1, then the liquid flow ratio is 0.25 and the gas flow ratio is 1.0. Experimental uncertainties were analyzed following the method by Kline and McClintock (1953), and are listed in Table 1. According to Kline and McClintock (1953), when a parameter ‘R’ is a function of measured variables ( x1 , x 2 , , x n ) , the uncertainty on the parameter ‘wR’ is obtained from the following equation.
wR
R x1
w1
2
w2
R x2
2
R x n
wn
2
1/ 2
(7)
Here, ‘w1’ is the uncertainty on variable ‘x1’, ‘w2’ is the uncertainty on variable ‘x2’, etc. As an example, the systematic uncertainty on liquid flow ratio of channel 1 is obtained as follows. From Eqs. (2) to (7),
wLFR 1
L F R1 sys
2
1
ml / N
m l ,1 / N
m
l
/N
2
2 w m l ,1 LFR 1
wm
l
ml
ml ,2 / N
m
wm
2
/N
l
2
2
wh
r ,i
hr , i
c p (T w o T w i ) w m w
hr , i m r
2
ml,N / N m /N l
wm
2
l ,N
LFRN
1/ 2
(8)
1/ 2
wx mr x r
l ,2
L F R2
(9)
wx 1 wh x ( hr , i hr , l )
2
2
wm
m w c p wT w ,i
hr , i m r
2
(10)
r ,i
m w c p wT
w ,o
hr , i m r
2
m w c p (T w , o T w , i ) w m
hr , i m r 2
r
2
1/ 2
(11)
Introducing the measurement uncertainties listed in Table 1, the systematic uncertainty on the liquid flow ratio ranged 6.4% ~ 504.3%, and that on gas flow ratio from 8.6% ~ 544.0%, which increased as liquid or gas flow rate decreased. The total uncertainty may be obtained as follows, including the random uncertainty on liquid flow ratio.
w L F Ri
L F R i
tot
w L F Ri
L F R i
2
sys
w L F Ri
L F R i
( w LFR ) ran 2 i
13
2
(12) ran
(13)
Page 13 of 45
where
is standard deviation of liquid flow ratio. Including the random uncertainty, total uncertainty on the
liquid ratio ranged 11.9%~ 504.4% and that on the gas flow ratio raged 13.2~ 544.1%. Large uncertainty on liquid and gas ratio corresponded to channels with negligible liquid or gas flow. To ensure the correctness of the measurement, error bars are drown in the next figures.
3. Results and discussions
3.1 General flow pattern and effect of mass flux
Fig. 6(a) shows the liquid and gas flow ratio at inlets of each (first and second) pass for 10/12 configuration. The inlet quality of the first pass is 0.2 and that of the second pass is 0.4. For the first pass, Fig. 6(a) shows that, in general, liquid flow ratio increases as the channel number increases with less liquid at frontal channels. The effect of mass flux is not significant. The corresponding photos and sketches show that a twophase jet is supplied from the inlet tube, part of which hits the protruded tubes, and is deflected downward. Rest of the jet is delivered downstream bypassing protruded tubes. During the process, liquid film gets thicker along flow direction, reaching protruded tubes at a certain channel. From there on, the liquid level remains approximately the same. The actual flow, however, was quite unsteady. The liquid interface was wavy, and the liquid level changed intermittently. The still photo shown in the figure is a flow pattern just at one moment of the time. It was observed that liquid was intermittently sucked from the wavy interface. If the suction pressure at the inlet of the channel is large enough for the liquid suction from the interface, liquid is sucked into the channel. With the liquid in the channel, the suction pressure decreases. If the channel suction pressure is smaller than that needed for liquid suction, gas is sucked into the channel. With the gas in the channel, the suction pressure increases again sucking liquid into the channel. The process occurred repeatedly.
14
Page 14 of 45
Two-phase mixture out of the first pass is first merged and then redistributed to the second pass in the intermediate header. For the second pass, Fig. 6(a) shows that liquid distribution is strongly dependent on the mass flux. At a low mass flux of 73 kg m-2s-1, more liquid is supplied into frontal channels and less liquid is supplied into downstream channels. As mass flux increases, more liquid is supplied into downstream channels. Accompanying photos in Fig. 6(c) show that two-phase jets exits the first pass, hit top of the header, and are deflected downstream along topside of the header forming a thin liquid film with lots of mist drops in the core flow. The liquid film gets thicker as it travels downstream by added two-phase jets. At the inlet of the second pass, the liquid film (and mist drops as well) is sucked into the tubes and gets thinner as it travels downstream until all the liquid is sucked into the channels. Photos and sketches in Fig. 6(c) show that liquid pool with clockwise recirculating flow is formed between protruded tubes. Liquid is also sucked into tubes from these liquid pools. As mass flux increases, the strength of two-phase jet out of the first pass increases as evident from Fig. 5(c), providing stronger axial momentum, which delivers the liquid film further downstream.
Fig. 6
shows that gas distribution generally yields reversed trend to liquid distribution. It appears that combined effects of separated flow pattern and equal pressure drop constraint of each channel have yielded the reversed distribution pattern between liquid and gas. In Fig. 7, flow distribution data taken maintaining the inlet quality at the second pass (xint) to be 0.6 are shown. Similar trends to the previous xint = 0.4 case are shown. At the first pass, liquid flow ratio increases with the channel number with minimal effect of mass flux on liquid distribution. At the second pass, liquid is forced downstream as mass flux increases. One thing to be noted is that data were not obtainable at the highest mass flux (143 kg m-2s-1) due to limitation of experimental facility. Degradation of heat transfer capacity due to flow mal-distribution may be assessed by comparing the actual heat transfer rate (Q) and the ideal heat transfer rate (Qideal) (Byun and Kim, 2011). Heat transfer rate of each elementary channel can be calculated from the measured inlet, exit quality and channel mass flow rate using adequate air-side and tube-side correlation. Then, actual heat transfer rate (Q) is thus obtained by adding calculated channel heat transfer rates. Ideal heat transfer rate (Qideal) is obtained assuming uniform refrigerant
15
Page 15 of 45
distribution. Heat transfer degradation by flow mal-distribution may then be obtained from the following equation.
% deg
Q ideal Q Q ideal
(14)
Using the channel flow rate, inlet and outlet quality, heat transfer degradation was calculated. For the tubeside heat transfer coefficient, Shah (1976) correlation was used. Air-side thermal resistance [ 1 /( ho Ao ) ] was selected to yield 5oC superheat when uniform refrigerant distribution is assumed. Detailed description on this procedure is provided in Byun and Kim (2011). Results are shown in Table 2. Table 2 shows that heat transfer degradation decrease as mass flux increases. It has been shown that liquid is forced downstream as mass flux increases, and eventually improves flow distribution. The uniformity of a distribution may also be judged by a standard deviation. Standard deviations of liquid and gas flow ratio were calculated for the whole pass, and are listed in Table 3. The trends of standard deviation on liquid and gas flow ratio are approximately similar to the trend of heat transfer degradation shown in Table 2. The standard deviation generally decreases as mass flux increases.
3.2 Effect of vapor quality
Fig. 8 shows the effect of vapor quality on liquid and gas flow ratio in the intermediate header (xint). Data are shown for two different mass flux (G = 73 kg m-2s-1 and 108 kg m-2s-1). At the first pass, the effect of xint on liquid distribution is not significant, although slightly better distribution is obtained at high xint and high mass flux. At the second pass, the effect of xint is more evident. Liquid is forced downstream as xint increases, and as it happens, the strength of two-phase jet out of the first pass increases, providing stronger axial momentum, which delivers liquid further downstream. Fig. 8 shows that gas distribution generally yields reversed trend to liquid distribution. Heat transfer degradation was calculated as before, and the results are summarized in Table 2.
16
Page 16 of 45
Standard deviations of liquid and gas flow ratio were calculated for each pass, and are listed in Table 3. From Table 3, it can be concluded that standard deviation generally decreases as xint increases.
3.3 Effect of insertion device
Kim et al. (2013) have shown that refrigerant distribution in a parallel flow heat exchanger can be significantly improved by inserting a tube with small holes at the inlet of the header. The optimum hole geometry was dependent on the refrigerant flow direction. For the upward configuration, uniform hole configuration shown in Fig. 9 has been suggested. Thus, tests were conducted with the tube inserted at the inlet header. Representative results are shown in Fig. 10. These figures show that insertion device significantly improves the flow distribution at the first pass. Almost uniform distribution is obtained. At the second pass, however, the effect of insertion device on flow distribution is negligible. This suggests that, in the intermediate header, dividing characteristics of two-phase flow is not affected by the combining characteristics.
3.4 Flow distribution correlations
For two-phase flow distribution in a header/branch tube configuration, only limited models are available. Data and Majumdar (1983) extended the single-phase model of Bajura and Jones (1976) assuming a homogeneous two-phase flow. Ablanque et al. (2010) adopted T-junction models by Seeger et al. (1986) and Hwang et al. (1988) to predict the phase spilt at header/branch tube junction. Tuo et al. (2012) calculated the pressure drop at the header/branch junction using the experimentally determined pressure loss coefficient. All the above models require an iterative procedure, which satisfies equal pressure drop constraint along every header-channel-header pass, to obtain flow distribution in branch channels. In contrast, a simple correlating method was proposed by Watanabe et al. (1995) from their upward branching study using R-11.They noted that fraction of liquid taken off by downstream channel can be correlated by the header gas Reynolds number at
17
Page 17 of 45
immediate upstream. Zou and Hrnjak (2013) adopted the correlating method by Watanabe et al. (1995), and proposed a correlation for parallel branching of R-134a from a vertical header. Following the suggestion by Watanabe et al. (1995), correlations were developed to predict the fraction of liquid or gas taken off by downstream channel. The correlations were developed through a multiple linear regression analysis. Commercial statistical package (Datafit, 1996) was used to perform a multiple regression analysis, which provided the best fit of data. The accuracy of the correlation is evaluated using the “R-squared” value. Higher R2 value implies better correlation. Header gas Reynolds number (Reg) were chosen as an independent variable. Separate correlations were developed for the first and the second pass. One thing to note is that liquid and gas flow rate at odd numbered channels are obtained by interpolating those of the even numbered channels, because flow rates were measured only at even numbered channels. Listed below are the correlations.
first pass :
m Tl , i m H l ,i
second pass :
34.2 R e g
m Tl , i m H l ,i
0.63
4709 R e g
(liquid),
1.01
m T g ,i m H g ,i
(liquid) ,
6.9 R e g
m T g ,i m H g ,i
0.34
1479 R e g
(gas) ; (100 ≤ Reg ≤ 4ⅹ104)
0.92
(gas) ; (2ⅹ103 ≤ Reg ≤ 5ⅹ104)
(15)
(16)
The correlations are compared with the present data in Fig. 11. Fig. 11 along with Eqs. (15) - (16) shows that, for first pass, liquid or gas fraction taken off by downstream branch tube increases as gas Reynolds number decreases. In contrast, for the second pass, liquid or gas fraction increases as gas Reynolds number. Also listed in Fig. 11 are R2 values. Higher R2 value implies better correlation. In Fig. 12, the available upward flow data [vist and Pettersen (2004), Hwang et al. (2007), Watanabe et al. (1995)] are compared with the present correlation [Eq. (15)]. Large scatter of data is observed, although relatively good prediction is shown for Watanabe et al. (1995) data. The root mean square errors (RMSE) are 18
Page 18 of 45
listed in Table 4. The header/tube diameter ratio dimensional tube protrusion depth (D/d), header/tube crosssectional flow area ratio (AH/AT) and non-dimensional tube protrusion depth (h/D) of the present geometry (D/d=12.9, AH/AT=1.9 and h/D=0.5) are significantly different from those of Watanabe et al. (1995), Vist and Pettersen’s (2004) and Hwang et al. (2007). Both Watanabe et al.’s (1995) and Vist and Pettersen’s (2004) header had round branch tubes (compared to flat branch tubes of the present study), which yielded smaller header/branch tube diameter ratio [D/d=3.3 for Watanabe et al. (1995) and 2.4 for Vist and Pettersen (2004)]. Hwang et al.’s (2007) header had thirty flat tubes (compared to ten flat tubes of the present study), which resulted in smaller header/branch tube flow area ratio (AH/AT=0.33). In addition, tested mass flux ranges are significantly different from the present one. The above discussion suggests that the present correlation has limited applicability, which is understandable from the fact that only one variable (gas Reynolds number) was used for development of the correlation. More data are necessary, which account for the effect of geometric variables (non-dimensional tube protrusion depth, header/branch tube diameter ratio, header/branch tube flow area ratio, header shape, etc.) as well as operating variables (mass flux, quality, temperature, etc.). Refined correlation should consider both geometric and operating variables.
3.5 Header pressure drop
During flow distribution tests, pressure drops were also measured from inlet of the header to the T junctions of the transition section (as illustrated in Fig. 5 and 13). Measurements were made both at first and second pass. For the first pass, the measured pressure drop consists of expansion loss at header inlet, pressure drop in the inlet header, pressure drop across the flat tubes, expansion loss in the transition section from flat tube to round tube, pressure drop across round tube to the T junction as illustrated in Fig. 13. The expansion loss and round tube pressure drops are expected to be small compared to other pressure drops, and addition of them is termed as a ‘minor pressure drop’.
19
Page 19 of 45
PH , in Pm ea s ,1 st P ft ,1 st _ 1 Pm in o r ,1 st
(17)
Pm inor ,1 st Pexp , inlet Pexp ,1 st Prt ,1 st _ 1
(18)
Header pressure drop at the first pass may be determined from Eq. (17) if other pressure drops are obtained from appropriate models or correlations. Similarly, for the second pass, the measured pressure drop consists of expansion loss at header inlet, pressure drop in the inlet header, pressure drop across flat tubes, expansion loss in the transition section from flat tube to round tube, pressure drop across round tube, contraction loss in the transition section from flat tube to round tube, pressure drop across flat tubes, pressure drop in the intermediate header, pressure drop across the flat tube at second pass, expansion loss in the transition section from flat tube to round tube and pressure drop across round tube to the T junction as illustrated in Fig. 13. Pressure drop in the intermediate header is then obtained from the following equation. Average pressure drop in the first pass to be used with Eq. (19) is obtained from the averaged value of all the channels of inlet pass as shown in Eq. (20).
PH ,
Pm
i n t
P
e a s, 2n d
P
1 s t p a, s s . a v g
, 2
fPt
n d
, 2m i n o r
n d
(19)
N 1 st
P1 st pass , avg .
(P
m eas ,1 st , i
P ft ,1 st _ 2 , i Pcont ,1 st , i Prt ,1 st _ 2 , i ) / N 1 st
(20)
i 1
Pm in o r , 2 n d Pexp , 2 n d Prt , 2 n d
(21)
In Eq. (20), ‘N1st’ is a number of tubes where measurement was made at the first pass, which is 5. The pressure drop across the flat tube is composed of frictional pressure drop, accelerational pressure drop and gravitational pressure drop. For the round tube, accelerational pressure drop is not included because it is not heated.
P ft P ft , f P ft , a P ft , g
Prt Prt , f Prt , g
20
(22) (23)
Page 20 of 45
For the frictional pressure drop across the flat tube, Zhang and Webb (2001) correlation was used because of the geometric similarity between the present flat tube and Zhang and Webb’s samples. For the frictional pressure drop across the round tube, Friedel (1979) correlation was used. Acceleration pressure drop and gravitational pressure drop across the flat or round tube, expansion and contraction loss were calculated following Collier and Thome (1994). For the void fraction, Zivi’s (1964) equation was used. The pressure drops and corresponding correlations are summarized in Table 5. The header pressure drop of each channel was obtained from Eq. (17) and (19) using measured channel mass flux and quality, and typical results are shown in Fig. 14. Fig. 14 shows the pressure drops of the first and second pass at G = 108 kg m-2s-1 and xint = 0.4. Fig. 14 (a) shows that inlet header pressure drop decreases as channel number increases. Fig. 14 (b) shows that pressure drops in intermediate header are relatively uniform irrespective of channel number. Such trend was also observed all mass fluxes and qualities. At the present time, the reason of the pressure drop trend is not clear. Complex flow pattern in the headers may be responsible. The measured pressure drop is denoted as lines with symbol ‘x’. One thing to be noted from Fig. 14 is that, between channels, measured pressure drops are not significantly different. Measured and calculated inlet header pressure drops at other mass fluxes and qualities are summarized in Table 6. Measured and calculated intermediate header pressure drops are summarized in Table 7. For the first pass (Table 6), the measured pressure drop in each channel increases as mass flux increases. Same trend is shown for the calculated inlet header pressure drop. As for the effect of quality, no general trend is observed both on the measured pressure drop and the calculated inlet header pressure drop. On the contrary, for the second pass (Table 7), the measured pressure drops as well as the intermediate header pressure drop increases as mass flux or quality increases. Table 6 shows that calculated intermediate pressure drops of channel 8 and 10 at G=73 kg m-2s-1, xint=0.4 are negative. This means that pressure is continuously recovered in the header, which is realized by branching of the flow in previous channels.
21
Page 21 of 45
4. Conclusions Refrigerant R-410A two-phase flow distributions were experimentally studied for a two pass evaporator with intermediate header simulating a brazed aluminum heat exchanger. Tests were conducted flow for the mass flux from 73 to 143 kg m-2s-1 and quality (xint) from 0.4 to 0.6. Listed below are major conclusions.
(1) In the intermediate header, two-phase mixture from the first pass is merged and then re-distributed to the second pass. More liquid is forced downstream as header mass flux or quality increases yielding better flow distribution (or smaller heat transfer degradation). (2) With the insertion device installed in the inlet header, flow distribution at the first pass is significantly improved. However, at the second pass, the effect of insertion device is negligible. (3) Correlations are developed to predict the fraction of liquid or gas taken off by downstream or upstream channel as a function of header gas Reynolds number. (4) Header pressure drops are obtained by subtracting flat tube pressure drop and other minor pressure drops from measured pressure drops. For the inlet header, the header pressure drop increases as mass flux increases. For the intermediate header, the pressure drop increases as mass flux or quality increases.
Acknowledgements This work was supported by Incheon National University Research Grant in 2014.
22
Page 22 of 45
References Ablanque, N., Oliet, C., Rigola, J., Perez-Segarra, C.D., Oliva, A., 2010. Two-phase flow distribution in multiple parallel tubes, Int. J. Thermal Sci., 49, 909-921. Ahmad, M., Berthoud, G., Mercier, P., 2009. General characteristics of two-phase flow distribution in a compact heat exchanger, Int. J. Heat Mass Transfer, 52, 442-450. Bowers, C.S., Hrnjak, P.S., Newell, T.A., 2006. Two-phase refrigerant distribution in a micro-channel manifold, Proceedings of the 11th Int. Refrigeration and Air Conditioning Conference at Purdue, R161. Byun, H.W., Kim, N.H., 2011. Refrigerant distribution in a parallel flow heat exchanger having vertical headers and heated horizontal tubes, Exp. Thermal Fluid Sci., 35, 920-932. Collier, J.G., Thome, J.R., 1994. Convective Boiling and Condensation, 3rd ed., Oxford University Press. Data, A.B., Majumdar, A.K., 1983. A calculation procedure for two phase flow distribution in manifolds with and without heat transfer, Int. J. Heat Mass Transfer, 26 (9), 1321-1328. Datafit, 1996. Oakdale Engineering, http://www.oakdaleengr.com/, Version 8.2. Friedel, L., 1979. Improved pressure drop correlations for horizontal and vertical two-phase pipe flow, 3R Int. 18, 485-492. Hrnjak, P., 2004. Flow distribution issues in parallel flow heat exchangers, ASHRAE Annual Meeting, AN-041-2. Hwang, S.T., Soliman, H.M., Lahey, R.T., 1988. Phase separation in dividing two-phase flow, Int. J. Multiphase Flow, 14 (4) 439-458. Hwang, Y., Jin, D.H., Radermacher, R., 2007. Refrigerant distribution in minichannel evaporator manifolds, HVAC&R Research, 13 (4) 543-555. Kim, N.H., Byun, H.W., 2013. Effect of inlet configuration on upward branching of two-phase refrigerant in parallel flow heat exchanger, Int. J. Ref., 36, 1062-1077. Kim, N.H., Kim, D.Y., Byun, H.W., 2011. Effect of inlet configuration on the refrigerant distribution in a parallel flow minichannel heat exchanger, Int. J. Ref., 34, 1209-1221.
23
Page 23 of 45
Kim, N.H., Lee, E.J., Byun, H.W., 2013. Improvement of two-phase refrigerant distribution in a parallel flow minichannel heat exchanger using insertion devices, Applied Thermal Engineering, 59, 116-130. Kline, S.J., McClintock, F.A., 1953. The description of uncertainties in single sample experiments, Mechanical Engineering, 75, 3-9. Koyama, S., Wijayanta, A.T., Kuwahara, K., Ikuda, S., 2006. Developing two-phase flow distribution in horizontal headers with downward micro-channel branches, Proceedings of the 11th Int. Refrigeration and Air Conditioning Conference at Purdue, R142. Kulkarni, T., Bullard, C.W., Cho, K., 2004. Header design tradeoffs in microchannel evaporators, Applied Thermal Engineering, 24, 759-776. Lee, S.Y., 2006. Flow distribution behaviour in condensers and evaporators, Proceedings of the 13th International Heat Transfer Conference, KN-08, Sydney, Australia. Seeger, W., Reimann, J., Muller, U., 1986. Two-phase flow in a T-junction with a horizontal inlet, part 1: phase saparation, Int. J. Multiphase Flow, 12 (4) 575-585. Shah, M.M., 1976. A new correlation for heat transfer during boiling flow through pipes, ASHRAE Trans., 82 (2), 66-86. Tuo, H., Bielskus, A., Hrnjak, P., 2012. An experimentally validated modeling of refrigerant distribution in a parallel microchannel evaporator, ASHRAE Trans., 118 (1) CH-12-C-48. Vist, S., Pettersen, J., 2004. Two-phase flow distribution in compact heat exchanger manifolds, Exp. Thermal Fluid Sci., 28, 209-215. Watanabe, M., Katsuda, M., Nagata, K., 1995. Two-phase flow distribution in multi-pass tube modeling serpentine type evaporator, ASME/JSME Thermal Engineering Conf., 2, 35-42. Webb, R.L., Chung, K., 2004. Two-phase flow distribution in tubes of parallel flow heat exchangers, Heat Transfer Engineering, 26, 3-18. Zhang, M., Webb, R.L., 2001. Correlation of two-phase friction for refrigerants in small-diameter tubes, Exp. Thermal Fluid Sci., 25, 131-139.
24
Page 24 of 45
Zivi, S.M., 1964. Estimation of steady-state steam void fraction by means of the principle of minimum entropy production, J. Heat Transfer 86, 247-251. Zou, Y., Hrnjak, P., 2013. Experiment and visualization on R-134a upward flow in the vertical header of microchannel heat exchanger and its effect on distribution, Int. J. Heat Mass Transfer, 62, 124-134.
25
Page 25 of 45
Figure captions
Figure 1 Refrigerant-side circuiting (one, two and four pass) investigated in this study Figure 2 Schematic drawing of the flat tube used in this study (unit: mm) Figure 3 Schematic drawing of the experimental apparatus Figure 4 Schematic drawing of the test section showing and photos and sketches of the test section Figure 5 Schematic illustrating of flow measurement principle Figure 6 Flow distribution data and photos and sketches showing the effect of mass flux on flow distribution (xint=0.4) Figure 7 Flow distribution data showing the effect of mass flux on flow distribution (xint=0.6) Figure 8 Flow distribution data showing the effect of inlet vapor quality at second pass (xint) Figure 9 Insertion device used by Kim et al.(2013) (unit: mm) Figure 10 Effect of insertion device on flow distribution Figure 11 Correlations for liquid and gas fraction taken off by downstream branch tube using gas Reynolds number of upstream header Figure 12 Available upward flow distribution data compared with present correlation [Eq. (13)] Figure 13 Schematic drawing illustrating pressure losses from the header to the flow measurement location Figure 14 Typical pressure loss distributions along each channel for the first and second pass (G=108 kg m-2s-1, xint=0.4)
26
Page 26 of 45
Table 1 Experimental uncertainties Measurement
Uncertainty
heater power
±0.2%
mass flow rate
±0.1% of full scale
differential pressure
±0.3% of full scale
absolute pressure
±0.2% of full scale
temperature
±0.5K
liquid flow ratio
± 504.4%
gas flow ratio
± 544.1%
27
Page 27 of 45
Table 2 Heat transfer degradation by flow mal-distribution G (kg m-2s-1) 73
108 143
xint
%deg
0.4
10.5
0.6
31.0
0.4
6.1
0.6
6.1
0.4
1.8
28
Page 28 of 45
Table 3 Standard deviations of liquid or gas flow ratio G (kg m-2s-1) 73
108 143
xint
liquid
gas
0.4
0.57
0.79
0.6
0.51
0.44
0.4
0.49
0.58
0.6
0.33
0.32
0.4
0.45
0.39
29
Page 29 of 45
Table 4 RMSE of the available upward flow data predicted by the present correlation
Investigators
liquid
gas
Vist and Pettersen (2004)
4.28
4.39
Hwang et al. (2007)
2.95
4.47
Watanabe et al. (1995)
0.30
0.49
30
Page 30 of 45
Table 5 Pressure drops and corresponding correlations Pressure
Correlations
drop
Reference
p ft, f p l lo 2
p l = 2 f l ( L /d ) l G
ΔPft,f
R e lo
GD
2
0 .0 7 9
, fl
l
R e lo
0 .2 5
lo (1 x ) 2 .8 7 x ( 2
2
(2300 ≤Relo) , f l
P
2
)
1
1 .6 8 x
0 .8
Zhang and
16 R e lo
(1 x )
0 .2 5
(Relo<2300) P
(
Pc
)
Webb (2001)
1 .6 4
Pc
p rt, f p l lo 2
p l = 2 f l ( L /d ) l G
R e lo
GD
2
Fr
R e lo
Fr
0 .0 4 5
We
2
2
0.78
G d
, We
2
C F 1 (1 x ) x (
l
2
(1 x )
x
g
0.24
(
f go
)(
(Relo<2300)
l g
)
0.91
(
g l
)
0.19
(1
ΔPrt,g
(1 (
1 x x
)(
g l
(1 ) l
)
2/3
Friedel (1979)
)
g l
)
0.7
2
P ft , g or Prt , g g g (1 ) l L
g
1
)
ΔPft,g
2
f lo
(1 x )
2
, H
H
p ft, a G [
ΔPcont
R e lo
x 1 x l g
ΔPft,a
ΔPexp
16
0 .0 3 5
2
gd H
CF2 x
(2300 ≤Relo) , f l
0 .2 5
3 .2 4 C F 2
lo = C F 1 +
G
0 .0 7 9
, fl
l
ΔPrt,f
2
Collier and Thome (1994)
]
Collier and Thome (1994)
1
Zivi (1964)
(1 x ) 2 g x 2 2 Pexp G in (1 ) l (1 ) l
Pcont
2 g G l 1 1 1 1 2 1 2 C c l 2 out
31
f
x
Collier and Thome (1994)
Page 31 of 45
Table 6 Measured and calculated inlet header pressure drop in each channel for the first pass G (kg m-2s-1) 73
108 143
ΔP(kPa)
xint 2
4
6
8
10
0.4
3.31(1.48)*
3.52(0.55)
4.27(0.43)
4.14(-0.53)
4.55(-0.42)
0.6
3.17(1.81)
3.17(0.63)
3.79(0.86)
4.14(1.13)
4.21(1.18)
0.4
4.34(3.42)
4.41(1.41)
4.62(0.96)
5.45(0.96)
5.31(0.44)
0.6
4.62(2.60)
4.89(1.92)
4.55(1.46)
4.96(1.73)
5.03(1.64)
0.4
6.69(3.58)
6.48(3.30)
6.27(2.53)
6.41(2.27)
6.89(1.38)
*Inlet header pressure drop in parenthesis
32
Page 32 of 45
Table 7 Measured and calculated intermediate header pressure drop in each channel for the second pass G (kg m-2s-1)
73
108 143
ΔP(kPa) xint 12
14
16
18
20
22
0.4
9.65(4.60)*
9.65(4.83)
9.65(4.32)
9.58(2.84)
9.86(3.45)
10.00(3.84)
0.6
10.13(3.55)
10.27(3.31)
10.20(3.28)
9.93(3.62)
10.20(4.47)
10.41(4.84)
0.4
14.13(5.44)
14.13(6.04)
14.13(6.04)
13.99(6.22)
14.48(5.72)
14.27(6.56)
0.6
16.75(9.13)
16.61(8.97)
16.48(7.60)
16.13(7.66)
17.17(8.28)
16.82(8.08)
0.4
22.82(11.16)
23.03(10.74)
22.20(10.59)
21.85(10.75)
23.16(11.82)
23.30(11.34)
*intermediate header pressure drop in parenthesis
33
Page 33 of 45
34
Page 34 of 45
35
Page 35 of 45
36
Page 36 of 45
37
Page 37 of 45
38
Page 38 of 45
39
Page 39 of 45
40
Page 40 of 45
41
Page 41 of 45
42
Page 42 of 45
43
Page 43 of 45
44
Page 44 of 45
45
Page 45 of 45