Experimental investigation and numerical modelling

10 downloads 0 Views 4MB Size Report
Nov 28, 2017 - The test configurations include varied tilt angle, thermal effec- tiveness ... Air-to-air plate heat exchangers (PHEs) are used in a wide num-.
Applied Thermal Engineering 131 (2018) 89–101

Contents lists available at ScienceDirect

Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Experimental investigation and numerical modelling of a compact wet air-to-air plate heat exchanger S. Lowrey ⇑, Z. Sun Department of Physics, University of Otago, PO Box 56, Dunedin, New Zealand

h i g h l i g h t s  A compact air-to-air heat exchanger is investigated under two-phase conditions.  Thermal effectiveness is investigated for a range of by-pass configurations.  Moisture extraction rate optimisation is investigated for tilt and airflow rate.  A numerical model is validated for duct by-pass and aspect ratio configurations.

a r t i c l e

i n f o

Article history: Received 12 August 2017 Revised 25 October 2017 Accepted 24 November 2017 Available online 28 November 2017

a b s t r a c t Air-to-air plate heat exchangers are widely used for domestic and industrial scale HVAC applications. The wide use of plate heat exchangers makes their control and optimisation critical for improving overall system performance. The application of plate heat exchangers has recently been demonstrated in domestic scale dehumidifier and heat pump clothes dryer systems. In the case of the domestic dehumidifier, the plate heat exchanger, referred to as an evaporator economizer, reduces the sensible load of moist air, and raises the drying efficiency of the system. However, they have shown to contribute to the latent cooling process also. There is little information regarding control and optimisation of air-to-air plate heat exchangers operating under wet operating conditions typical of domestic scale dehumidification and heat pump clothes drying. In this work, a plate heat exchanger was designed and constructed to experimentally investigate several different conditions for optimizing the moisture extraction rate for application of a plate heat exchanger as an energy recovery device. The test configurations include varied tilt angle, thermal effectiveness control, varied duct aspect ratio and varied air volume flow rate and moist air conditions that are applicable to domestic scale dehumidification systems. The research findings show that a modest tilt angle of up to 20°, relative to a reference setting of the hot-side face area being parallel to the horizontal, the pressure drop can be reduced by 33% with a corresponding decrease in the moisture extraction rate of 8%. Furthermore, a study of different ducting configurations shows that deactivating cold-side ducts, while fixing the active hot-side, allows for a greater range of hot-side temperature difference control, while maintaining a relatively high MER compared with deactivating hot-side ducts while keeping the active cold-side ducts fixed. The experimental data obtained in this work has been used to extend the validity of a recently developed numerical model for a wet air-to-air plate heat exchanger expanding its working range of hot-side and cold-side moist air inlet conditions, airflow conditions and ducting configurations and duct aspect ratios. This model provides for a quantitative approach when engineering emerging advanced domestic scale dehumidification and heat pump clothes drying systems. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction

⇑ Corresponding author. E-mail address: [email protected] (S. Lowrey). https://doi.org/10.1016/j.applthermaleng.2017.11.127 1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.

Air-to-air plate heat exchangers (PHEs) are used in a wide number of HVAC applications at the domestic and industrial scale [1]. PHEs are used in condensing clothes dryers [2] and are being extended to other domestic systems such as the domestic dehumidifier and heat pump clothes dryer [3,4]. As of 2016, the

90

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

Nomenclature A A AFACE b b Cf Dh ev F Ff h hm I j Jq m _ m ^ n NCS NHS Nu P Pr qc Q_ Q_ ðmÞ Re

duct height [m] area [m2] face area of heat exchanger [m2] duct width [m] ADUCT/AFACE [–] friction factor [m2] hydraulic diameter [m] contraction loss factor [–] forces that the system exerts at the control volume surface [N] net force acting on control volume boundaries [N] specific enthalpy [J kg1] mass transfer coefficient [m s1] identity matrix [–] molecular mass flux [kg m2 s1] total heat flux vector [W m2] mass [kg] mass flow rate [kg s1] unit normal vector [–] number of cold-side ducts [–] number of hot-side ducts [–] nusselt number [–] pressure [Pa], Perimeter [m] Prandtl number [–] pure heat flow vector [W m2] heat transfer rate [W] enthalpy flux accompanying mass transfer [W] Reynolds number [–]

U.S. Department of Energy is looking to adopt greater energy conservation standards around domestic dehumidifier products [3], and have indicated that they expect pre-cooling air-to-air PHE to improve domestic dehumidification technology [3]. An energy efficient hybrid heat pump clothes dryer was recently demonstrated which incorporated an air-to-air PHE [4] showing that PHEs are extending into new domestic drying technology applications. Air-side energy recovery, in conjunction with the refrigerant evaporator, is a well established means of increasing dehumidification capacity [1] as it has the potential to reduce the sensible cooling load of the refrigerant evaporator, promoting greater latent cooling of the humid air at this component [5–7]. This can be carried out using heat-pipes [7,8] or air-to-air plate heat exchangers [9–12]. However, the application of air-side energy recovery in domestic scale equipment has only recently been tested experimentally for low household temperatures [10–11] where these systems are typically operated in New Zealand and Britain [13,14]. In this application, it has been shown that the ratio of the PHE condensation rate to the total system condensation rate can be as high as 19% in some ambient conditions [10]. To describe the performance of a wet economiser in this application, a numerical model was developed [15], however, the model was restricted to a single air volume flow rate and a single duct geometry. In this experimental study, the experimental data has been obtained and used to extend the validity of the numerical model developed in reference [15] for a range of airflow rates, ambient temperature and humidity conditions, duct configurations (that allow various degrees of air flow by-pass) and new plate heat exchanger duct aspect ratios. In addition, we have experimentally investigated means for improving condensate drainage and reducing air-side pressure drop in an air-to-air plate heat exchanger under two-phase flow conditions. There is little research on the effect of tilting an air-to-air plate heat exchanger [16] and this

S S t T Tdb Twb DT u; v

v

x; y; z ef

q

/

xk

surface extra stress tensor [Pa] time [s] temperature [°C, K] dry-bulb temperature [°C] wet-bulb temperature [°C] temperature difference [–] x-component, y-component of velocity velocity [m s1] spatial coordinates [m] enhancement factor [–] density [kg m3] relative humidity [–] mass fraction [kg-k/kg-mixture]

Subscripts CS cold-side db dry-bulb D/DUCT duct k, a, v, w species-k, dry-air, water-vapour, liquid-water HS hot-side P control volume central node s surface S interface W wall wb wet-bulb

investigation looks at a range of PHE tilt conditions for reducing pressure drop and for enhancing the vapour condensation. In addition, a wide range of duct blocking configurations have been studied to control the PHE’s thermal effectiveness and enhance the condensation rate. Finally, different duct aspect ratios were studied to determine how they perform under relatively high latent cooling conditions. We emphasise that the validated numerical model results show that this model can be applied for the design and investigation of an air-to-air PHE for a wide range of HVAC applications but with a specific focus on domestic scale applications such as dehumidification and heat pump clothes drying technology. The size of the compact plate heat exchanger investigated in this study, and the conditions of moist air investigated in this study, have been guided by our previous work into geared dehumidification technology. In particular the inlet conditions tested on the hot and cold sides of the plate heat exchanger used in this work are guided by the measured data presented in reference [11] in which a plate heat exchanger of a similar size was used as an evaporator economizer.

2. System and methods 2.1. Heat exchanger design and construction In this experimental study, an air-to-air PHE was designed and constructed to investigate the performance of domestic scale PHEs. In order to keep the new PHE’s capacity similar to the PHE investigated in reference [10], a similar number of HS and CS ducts were used and the heat exchanger internal heat transfer wall area was kept similar to the previous PHE’s wall area, which was 0.023 m2. However, to reduce the complexity of the numerical modelling of the new PHE, the turbulence inducing ribs in the

91

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

PHE in reference [10] were not constructed. In addition, the new PHE in this work has been designed such that the duct aspect ratios are different on the hot and cold sides of the heat exchanger, as is shown in Fig. 1. Fig. 1(a) depicts the overall plate heat exchanger and Fig. 1(b) and (c) show plan-view images of side-faces 1 and 2, respectively. Fig. 1(d) and (e) show schematics of the areas inside the red boxes where (d) corresponds to Fig. 1b and (e) corresponds to Fig. 1c. The plan-view images show the difference in the individual duct aspect ratios for side-faces 1 and 2 where the side-face 1 duct width is 0.9 mm and the side-face 2 duct width is 2.5 mm. The heat exchanger plates are aluminium, having a thickness of 0.1 mm and a thermal conductivity of 204 W/(m-K) at 20 °C [17]. Each plate was separated by aluminium spacers, which are shown schematically in Fig. 1d and e. Spacers used for the hot-side ducts had dimensions 250  10  0.9 mm3 and spacers used for the cold-side ducts had dimensions 120  10  2.5 mm3. Plastic spacers would have provided thermal insulation with air external to the PHE but were not used as thicknesses similar to what was used in this work were found to not be rigid enough and it was expected that constructing a heat exchanger using thin plastic spacers would have resulted in a wide variation in the duct spacing. The geometry of the PHE and the individual ducts are summarised in Table 1. The aspect ratio, defined as the ratio of ductheight to duct-width, of the HS and CS ducts were set such that the inlet air velocity would remain similar to the PHE presented in reference [15]. The total number of ducts used in this PHE was

Table 1 Parameters for the heat exchangers used in this work and in previous work [10]. Reference PHE (both the hot-side and cold-side)

Aluminium wall thickness [mm] PHE width [mm] Duct height (a) [mm] Duct width (b) [mm] Number of HS ducts [–] Number of CS ducts [–] Single duct aspect ratio (a/b) Heat transfer area [m2]

0.1 233 150 1.6 63 63 94 0.0225

Variable aspect ratio PHE Sideface 1

Sideface 2

0.1 243 230 0.9 63 63 256 0.023

0.1 243 100 2.5 63 63 40 0.023

126 (i.e. 63 on the hot-side and 63 on the cold-side) and the aspect ratio on one face was 256 and 40 on the other. We emphasise that either the small area heat exchanger face or large area heat exchanger face can act as the hot-side, and likewise for the coldside of the heat exchanger. 2.2. Heat exchanger test rig and instrumentation The PHE test rig is shown schematically in Fig. 2. The rig was designed to accommodate two plate heat exchanger configurations such that either the small or the large plate heat exchanger side-

Fig. 1. (a) photo of the varied aspect ratio plate heat exchanger. (b) photo of the plate heat exchanger showing side-face 1. (c) photo of the plate heat exchanger showing sideface 1. (d) schematic of red enclosed area in (b). (e) schematic of red enclosed area in (c). The plate heat exchanger dimensions for side-faces 1 and 2 are given in Table 1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

92

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

Fig. 2. Schematic of the heat exchanger rig.

faces (Fig. 1) could be tested with the hot-side moist air conditions. The rig ducting is stainless steel, with external 19 mm thick sheet insulation material (thermal conductivity approximately 0.035 W m1 K1). HVAC flexible ducting (item 8, Fig. 2) was used to duct air from a port at the wall dividing the chambers to the rig inlet. Air is drawn through both the hot-side and cold-side of the rig via centrifugal fans (item 1) each with a full-flow capacity of 50 L s1. The cold-side air-stream enters the HVAC ducting (item 8) from climate chamber 1 at inlet 2 and moves towards the coldside PHE inlet. The air then passes through the PHE where heat is transferred from the hot air-stream that enters at inlet 1. The heated cold-side air stream then exits the system via a centrifugal fan. The hot-side airstream condenses water vapour onto the PHE duct walls, which is received at a receptacle (item R1). The cooled hot-side air-stream then exits the rig via a centrifugal fan. The hot-side and cold-side pressure drops were recorded during each test using digital pressure transducers (shown with label DP Fig. 2). The green dashed line represents the pressure transducer ports. Air distribution and mixing devices were constructed using a stainless steel mesh grill (item 7, Fig. 2) and were placed downstream of both the HS and CS exits and before the corresponding thermocouple arrays. The test system was equipped with Sensiron digital pressure transducers, with a measurement uncertainty of ±1 Pa, to monitor the HS and CS pressure drop. Air temperatures were monitored using Type-T thermocouples housed in cylindrical stainless steel metal tubes. Type-T Temperature sensor arrays were set up at

the inlet and outlet of the cold-side of the PHE and at the outlet of the HS of the PHE to measure dry-bulb temperature (the thermocouple arrays are given the label A in Fig. 2). The thermocouples were calibrated against a platinum resistance thermometer with a measurement uncertainty of ±0.013 °C over a temperature range of 0–60 °C, conforming with ASHRAE [18]. The thermocouples and pressure transducers were connected to a DT800 dataTaker logger. The uncertainty in a thermocouple measurement is given by an error propagation associated with the calibration against a platinum resistance thermometer, which complies with the ASHRAE standards of temperature measurement [18]. The uncertainty in hot-side or cold-side temperature differences was taken as the addition of the absolute errors associated with the two temperatures.

2.3. Performance testing procedure The heat exchanger rig was operated in a controlled test chamber with the PHE hot-side moist air conditions controlled in climate chamber 1 and the PHE cold-side moist air conditions controlled in chamber 2 (as shown in Fig. 2). The dry- and wetbulb temperatures were continuously monitored in each chamber with ventilated psychrometers (item 2) (ASHRAE, 1986) and the relative humidity was calculated from these temperatures [19]. The speed of the centrifugal fans (item 1) is controlled using variacs (item 3).

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

The air volume flow rate for the HS and CS heat exchanger rig ducting was measured by timing the inflation of a 100 lm thick polythene bag, with a volume of 10.86 m3, using a digital stopwatch [20]. The reference used in this work is based on a commercial, domestic dehumidifier system airflow rate of 0.0385 ± 0.0002 m3/s. Both variacs were adjusted until the air volume flow rates on the HS and CS were balanced. The PHE’s moisture extraction rate was determined by weighing the mass of water in receptacle R1 (Fig. 1) and dividing the condensate mass (mw) by the test run time (Dt, in hours), i.e. MER = mw/ Dt. The condensate mass was measured using digital scales having a resolution of ±0.01 g and a capacity of 2000 g. MER measurements were taken three times for each condition with at least a one hour duration between each MER measurement. The MER measurement uncertainties reported here are the standard errors in the mean. 2.4. Heat exchanger tilting The heat exchanger rig was situated on an adjustable stand such that the entire ducting system could be rotated vertically, allowing for different tilt angles to be investigated. In this work, a tilt angle of zero is when the plate heat exchanger is oriented with either side 1 (or side 2) positioned at 90° to the horizontal. The rig could be tilted at angles between 30° and +20° with respect to its reference angle. Fig. 3 shows schematics of the PHE set at different tilt angles. In this work, the PHE reference position was taken to be with the hotside inlet face to be perpendicular to the incoming air-stream, with the air-stream moving in a downward vertical direction. A positive tilt angle is clockwise rotation relative to the reference position and a negative tilt angle is an anti-clockwise rotation relative to the reference position. The blue region is shown to indicate the expected resultant shape of the condensation field in the hot-side ducts.

93

2.5. Thermal effectiveness control The PHE was tested using variable ducting configurations to determine what thermal effectiveness control can be obtained for a system having both the HS and CS air volume flow rate fixed at 0.0385 m3/s. The thermal effectiveness was tested under four different configurations as shown schematically in Fig. 4(a) the reference configuration where both the HS and CS ducts are all active (N HS ¼ N CS ¼ 63); (b) the number of active HS ducts are varied (using settings of NHS = 50, 40, 30 and 20) while keeping the number of CS ducts fixed at 63; (c) the number of active CS ducts are varied (using settings of NCS = 50, 40, 30 and 20) while keeping the number of HS ducts fixed at 63; (d) the number of active HS and CS ducts kept equivalent and testing duct numbers of NHS = NCS = 50, 40, 30 and 20. The various ducting configurations were set-up by sealing off certain ducts using aluminium tape, and then covering the blocked regions with sheet insulation. It was envisaged that these duct configurations would allow for a range of HS and CS inlet and exit face temperature differences (i.e. modification to the heat exchanger thermal effectiveness) to be achieved. We note that when testing the various ducting configurations described here, the hot-side and/or cold-side inlet ducting was not modified which could lead to some airflow maldistribution. However, the experimental and numerical results, discussed below, indicate that these entrance maldistribution effects do not have a significant effect on the air-stream entering the open ducts or the air-stream inside the heat exchanger ducts. 2.6. Numerical modelling In this work a numerical model is used that simulates the performance of an air-to-air PHE where latent cooling occurs in the hot-side moist air-stream. The model solves gas-phase energy, mass and momentum balance equations and liquid-phase energy

Fig. 3. Schematics of the plate heat exchanger at its reference position (centre) along various tilt angle positions (far left schematic and far right schematic).

Fig. 4. Schematics of the duct blocking configurations.

94

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

and mass balance equations, allowing the HS and CS outlet moist air states to be determined. This model was previously validated for a single air volume flow rate (AVFR; 0.0385 m3/s) and for a single duct aspect ratio [15]. The data gathered in this work is used to extend the validity of the model for different duct aspect ratios, a range of AVFRs and duct blocking configurations. The model does not solve liquid phase momentum balance and therefore tilting is not investigated via numerical simulation. The model is two-dimensional (2D) and is applied to the HS and CS plate heat exchanger ducts. Fig. 5(a)–(d) shows the system boundary (red dashed lines) applied to both the HS and CS ducts where the mass flow and the heat and mass transfer streams are represented with arrows and the label S denotes each bounding surface (these control volumes are ultimately discretised into smaller control volumes). It can be seen from Fig. 5b that the effects of condensation from the moist air stream occur only in the HS ducts. The increase in the boundary layer thickness is ignored, i.e. condensation after the onset-point is treated as having a uniform liquid film thickness, and local phase equilibrium is assumed at the interface between the gas-phase and liquid-phase where condensation occurs. Control volume numbers of 8 (in x)  8 (in y) were used for all the numerical simulations in this study as this control volume number has been validated and shown to give sufficient numerical accuracy [15]. Convergence criteria for the plate heat exchanger model were set such that convergence would be met when the relative errors of the overall plate heat exchanger energy and mass balances were 8 (where a and b denote duct-height and duct-width, respectively), a Nusselt number of 7.54 is used [22] for the two duct aspect ratios in the PHE presented in this work. The friction factor C f ¼ 96:00=Re also applies in the case of a/b > 8 and is used here [22]. In the case of fully developed turbulent fluid flow through a duct, the Dittus-Boelter equation is employed to evaluate the heat transfer Nusselt number. This is expressed as [22]

Nu ¼ 0:023 Re4=5 Prn for Re P 10; 000

The numerical model has been improved by accounting for the effects of sudden contraction and expansion frictional losses at the inlets and exits and these effects have been integrated into the gasphase momentum balance model. The sudden contraction loss factor ‘ ev ’ is given by [24]

0:45ð1  bÞ ð5Þ

95

ð8Þ

where the superscript n is set to 0.4 in the case of a ducted fluid that is undergoing heating and 0.3 in the case that the fluid is being cooled [23]. The above expression applies to a circular tube but can be used in the case of noncircular tubes [22]. Therefore, the hydraulic radius (Dh = 4A/P) is used in place of the circular tube diameter, where A is the duct face area and P is the duct perimeter. The Chilton-Colburn analogy has been employed, in conjunction with the Nusselt number correlations above, to obtain the mass transfer coefficient in the absence of empirical expressions.

DP ¼

  1 2 qv ADUCT ev 2

ð11Þ

Gas-phase friction enhancement factors were included in the model’s momentum balance routine to scale the friction coefficient as using the above correlations were found to not precisely reproduce the experimental pressure drop data – note that the HS and CS pressure drop was not measured in reference [15]. Enhancement factors were applied to the hot-side duct model (efHS) and the cold-side duct model (efCS). These were treated as tuning factors to improve the agreement of the simulated model results with the experimental results. 3. Results 3.1. Airflow tests In this section, we present experimental and numerical results for the testing of the PHE at a number of moist air conditions and for a number of airflow conditions. The PHE was tested under moist air conditions that correspond to temperatures under which a PHE would operate in the application of an evaporator economizer for enhanced dehumidification. The HS inlet dry-bulb temperatures were 15 °C, 20 °C and 25 °C for corresponding CS inlet dry-bulb temperatures of 3 °C, 5.5 °C and 11 °C, respectively. The cold-side inlet dry-bulb temperatures used here are similar to the dry-bulb conditions of air after a refrigerant evaporator in a geared dehumidification system as shown in reference [11]. The HS inlet relative humidity (RH) was set to 70% and the CS relative humidity was set to 90% for all tests. The CS inlet RH was chosen as the PHE in the described dehumidifier application is located immediately after the refrigerant evaporator which means the temperature of air off the evaporator is close to saturation. At each combination of moist air conditions, the PHE was tested at air volume flow rates similar to those used in some domestic dehumidifier technology. The high airflow setting in domestic dehumidification can be 0.0385 m3/s [10,11], and therefore we have used AVFRs of 0.01925 m3/s (0.50  0.0385 m3/s), 0.02888 m3/s (0.75  0.0385 m3/s), 0.0385 m3/s and 0.04813 m3/s (1.25  0.0385 m3/s) – we will refer to these four AVFR conditions as V1, V2 V3 and V4, respectively. The HS and CS AVFRs were kept the same for each test within experimental uncertainty. Fig. 6 shows numerical condensation rate field results for the PHE-HS ducts (left images) and photographs of the PHE’s HS inlet (right images) during steady-state tests for the PHE oriented with its large face area exposed to the HS moist air conditions and its small face area exposed to the CS moist air conditions (see Fig. 1). The HS inlet condition for these test was 15 °C (a), 20 °C (b) and 25 °C (c), and the HS and CS AVFRs were both set to

96

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

Fig. 6. Left: numerical condensation rate fields for the hot-side PHE ducts where the magnitude of moisture extraction rate is given by the colour bar. Right: PHE hot-side inlet face photographs showing condensation formation on the left hand side of the inlet face. Images labelled A are for hot-side moist air inlet conditions of 15 °C, 70% RH; images labelled B are for hot-side moist air inlet conditions of 20 °C, 70% RH; image labelled C are for hot-side moist air inlet conditions of 25 °C, 70% RH. All numerical and experimental results are for an AVFR of 0.0385 m3/s. The red dashed line shows the extent of the condensation along the exterior wall of the cold side ducts. The coordinate system is to display the orientation of the heat exchanger in order to compare the simulated condensation fields on the left with the photographs on the right. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

0.0385 m3/s. In each of the numerical condensation rate field images, it appears that condensation forms immediately at the left hand side of the HS inlet image. The condensation rate is also largest for the 20 °C test. Each condensation rate distribution also shows that condensation will form along the entire hot-side exit face for each moist air condition. The photographs on the right hand side support the numerical results as condensation can be seen to form on the spacers that separate the CS ducts from the environment. Furthermore, the 20 °C test shows the largest extent of condensation along the CS spacers which stretch on average 26 mm along the exterior of the cold duct spacers. The numerical results for the 20 °C test show that condensation could occur approximately 36 mm along the external CS duct spacers at the HS face (each pixel has a width of 14 mm and height of 6 mm). At the 15 °C and 25 °C tests, on average the condensation rate does not extend as far as for the 20 °C test and this agrees with the trends given by the numerical results. We emphasise that we anticipated that using aluminium spacers would lead to some heat exchanger with the environment, but plastic spacers that had the thicknesses used in this work were not rigid enough and would have potentially led to a large degree of variation in duct width on both sides of the heat exchanger. The photographs of Fig. 6 show that there is possibly airflow maldistribution in both the HS and CS ducts given the variation of the condensation line along the HS inlet side-face (indicated by the red dashed lines where the condensation is on the left side of the red lines). We believe that this is in part due to slight variation in both HS and CS ducting width which may result in nonuniform distributions in the pressure field and velocity field in each duct. The condensation on the exterior of the CS duct spacers indicates that both sensible and latent heat transfer is occurring along the exterior of the metal spacers. The corresponding sensible

heat gain into the cold ducts was not accounted for in the numerical model. Fig. 7 shows numerical and experimental results for the PHE tests at the moist air conditions described at the start of this section. The left hand side graphs show the ducting moist air pressure drop, the middle graphs show the hot-side ducting temperature difference between the inlet and outlet of the heat exchanger (HSTD) and cold-side ducting temperature difference between the inlet and outlet of the heat exchanger (CSTD) and the RHS graphs show the moisture extraction rate in the HS ducts. The top, middle and lower row of graphs are for a HS inlet moist air temperature of 15 °C, 20 °C and 25 °C, respectively, where the RH is 70% in each case. The HS pressure drop (HSPD) is relatively large compared with the CS pressure drop (CSPD) and this was anticipated given that condensation in the HS ducts will further resist the airflow through the HS ducts. In each case, Fig. 7 shows that the hot-side temperature difference (between the heat exchanger inlet and outlet) is less than the cold-side temperature difference. We attribute this to both latent and sensible cooling in the hot-side ducts and only sensible heating in the CS ducts. In addition, and as mentioned earlier, due to condensation forming on the exterior of the cold-duct spacers at the HS inlet, this indicates that there is both latent and sensible cooling on the exterior of the walls that will contribute to sensible heat gain into the cold ducts, further contributing to the larger CS temperature difference compared with the hot-side temperature difference – we also note that there will be heat loss through the aluminium walls (that separate the aluminium plates) of the HS ducts so without simulating heat loss/gain, it is difficult to know how much of an effect it has on the PHE’s HS temperature difference and CS temperature difference.

97

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

Fig. 7. Left graphs: HS (black), CS (blue) and total (red) pressure drop versus AVFR. Middle graphs: HS (black) and CS (blue) temperature difference versus AVFR. Right graphs: MER in the HS ducts versus AVFR. Solid curves represent measured data and dashed curves represent simulated data. The HS and CS inlet moist air conditions are displayed on each graph. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

enhancement factor did not require adjustment for the low airflow rate tests but was increased with increasing airflow in order to match the simulated and measured data. As shown in Fig. 7, there is reasonable agreement between the simulated and measured trends for the 15 °C and 20 °C tests despite the small disparity in the absolute quantities. The 25 °C test shows a very good match between the measured and simulated MER results at the highest airflow test but the numerical results then over predict the MER at lower airflows. While the relative error is large (as much as 138% at the lowest airflow test), the absolute errors are relatively small, being at most 0.049 kg/hour for the lowest airflow test. Fig. 7 also shows there is good agreement between the measured and simulated hot-side temperature difference between

When running the numerical model with no HS or CS friction enhancement factors in place, in most cases, there was large disagreement between the measured and simulated pressure drop, temperature difference and condensation rate results. Table 2 shows the friction enhancement factors used for each numerical test run. The factor efHS was largest for the lowest airflow rate tests and was reduced as the airflow rate was raised where the enhancement factor was 2.6 for the V1 (0.50  0.0385 m3/s) tests and was between 1.10 and 1.20 for the V4 (1.25  0.0385 m3/s) tests. This suggests that the condensation contributes more to the increase in the HSPD from condensate bridging at low airflow, whereas at higher airflows, less condensate bridging forms (and the film thickness is likely to be reduced), meaning that the air may see a more uniform rectangular path with less obstructions. The CS

Table 2 Friction enhancement factors for various HS inlet dry-bulb settings and VFR settings. V1

V2

V3

V4

Tdb

efHS

efCS

efHS

efCS

efHS

efCS

efHS

efCS

15 20 25

2.60 2.60 2.74

1.00 1.00 1.00

1.90 1.70 1.91

1.35 1.30 1.42

1.40 1.35 1.45

1.25 1.25 1.65

1.16 1.25 1.17

1.90 1.90 2.05

98

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

the heat exchanger inlet and outlet for all of the moist air conditions and airflow conditions tested where no tuning was necessary for the heat transfer coefficients, corresponding to the heat transfer rate from a hot-duct to the adjacent cold-ducts, of the gas-phase energy balance expressions. The simulated cold-side temperature increase between the inlet and outlet is over-predicted for all tests carried out with the average over-predicted temperature increase being 5.04 °C, 6.79 °C and 6.68 °C for tests at 15 °C, 20 °C and 25 °C, respectively. The trend with airflow rate however is upheld by the simulated data with respect to the measured data for each test condition. We would expect that the cold-side temperature difference is larger than the hot-side temperature difference due to the combined latent and sensible cooling in the HS ducts. However, we also think that heat gain to the CS ducts may have inflated the measured cold-side temperature difference data. In addition, some degree of plate delamination on the PHE may contribute to HS and CS air mixing inside the PHE, which may also contribute to a higher measured CSTD. An enhancement factor was applied to the sensible heat transfer term for heat transfer from the metal wall to the cold duct air stream but this led to model instability and was not investigated further. For future work, the relative humidity at the exit of the CS ducts should be measured in order to further investigate the shortcomings of the model. Comparing the measured MER generated under the three sets of inlet conditions, with an identical air volume flow rate on both the hot and cold-sides of the PHE, the value of MER is the largest over the entire air volume flow rate test range for the hot-side inlet condition of 20 °C, 70% relative humidity. This is attributed to the difference in the hot-side inlet and cold-side inlet temperatures. The hot-side inlet temperatures of 15 °C, 20 °C and 25 °C correspond to cold-side inlet temperatures of 3 °C, 5.5 °C and 11 °C, respectively. Therefore, the temperature differences between the inlet hot-side and inlet cold-side are 12 °C, 14.5 °C and 14 °C, respectively. This indicates that the temperature difference between the hot-side and the cold-side corresponding to the hot-side inlet temperature condition of 20 °C is bigger than the other two cases and this is also the case when inspecting the individual control volume temperature differences for the hot-side and cold-side, thus more latent cooling would be generated under this condition. The PHE was also tested over the same set of airflow rates as above with its small area face exposed to the hot-side moist air conditions and the large area face exposed to the CS moist air conditions. The PHE was tested in this configuration for the HS inlet conditions of 20 °C and 25 °C (both with a relative humidity of 70%) and the MER gave a maximum of 0.26 kg/hour for the 20 °C tests and 0.21 kg/hour for the 25 °C tests for an airflow rate range of 0.50  0.0385–1.25  0.0385 m3/s. Therefore, this PHE configuration was not investigated further due to the low MER.

3.2. Tilting Here we present experimental results for the plate heat exchanger set up with a tilt relative to its reference configuration of the HS ducts orientated vertically and the CS ducts oriented horizontally. These tests were carried out on the PHE in the configuration with the large area face exposed to the HS air-stream and the small area face exposed to the CS air-stream. Fig. 8 shows the HS and CS pressure drop (left graph), the HS and CS temperature difference (between the inlet and outlet of the heat exchanger, middle graph) and the HS MER (right graph) all versus the tilting angle for the PHE tested with HS moist air conditions of 20 °C, 70% RH and CS moist air conditions of 3 °C, 90% RH. The airflow was kept fixed at 0.0385 m3/s for each test. The CS pressure drop (CSPD) shows weak sensitivity to duct tilting either side of the 0° tilt angle. This is expected given there is no condensate in the CS ducts to interfere with the air-stream. The HS pressure drop (HSPD) however shows an abrupt change either side of the 0° tilting angle (or reference position). At a tilt of -10°, the HSPD is reduced by 40 Pa (a 33% reduction) and an 8% reduction in the MER. Further negative tilting further reduces the HSPD, however the MER starts to drop abruptly once the tilt angle reaches 30°. We believe that the modest tilt angle of 10° aids coalescence of condensate at one end of the hot-side duct, allowing the coalesced condensate to be detach faster from the PHE, while not exacerbating condensate bridging and/or flooding in the bottom left quadrant of the HS ducts (relative to the normal operating position). We attribute the abrupt drop in MER at a tilt of 30° (which is reduced by 43% compared with a tilt angle of 0°) to the exacerbation of HS duct flooding at the lower end of the PHE hot-side ducts. A similar trend in the HSPD is seen when tilting the PHE at positive angles, however, the MER shows an abrupt drop at a tilt of 20° of 70%. We attribute the lower HSPD at 20° to the greater spreading of the condensate over more of the HS duct heat transfer surface. However, this will come with an increased thermal resistance due to the condensate film now spreading over previously dry HS duct wall, which will lower the amount of sensible and latent cooling. It’s possible that having the PHE at positive tilt angles, while reducing the HSPD with only a reduction in the MER (3% at 10°) could lead to condensate carryover given that a thin film will be spread across heat transfer surface area where the wall temperature is not at or below the dew-point temperature. These results suggest that a tilt between 20° and 0° is appropriate for reducing the HSPD and raising the MER for this particular PHE configuration operating under two-phase flow conditions. The CSTD shows little variation over the range of tilt angles. The HS temperature difference does seem to agree with the trend in the MER where the HS temperature difference decreases either side of the 0° operating condition.

Fig. 8. Left: HS (black), CS (blue) and total (red) PHE pressure drop versus tilt angle. Middle: HS (black) and CS (blue) temperature difference versus tilt angle. Right: HS MER versus tilt angle. The air volume flow rate was set to 0.0385 m3/s for each test. The HS moist air inlet conditions were 20 °C, 70% RH, and the CS moist inlet conditions were 5.5 °C, 90% RH. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

Our numerical model was not designed to account for duct tilting. For future work, development of a liquid-phase momentum balance is suggested in order to compute the condensate thickness over the HS duct walls and steady-state flow under the conditions tested here. 3.3. Ducting configurations Here we present experimental and numerical results for the PHE tested under a number of different duct configurations. For each test, an airflow of 0.0385 m3/s was used and the HS moist air conditions were kept at 20 °C, 70% RH and the CS moist air conditions were kept at 5.5 °C and 90% RH. Fig. 9 shows graphs for the HS and CS pressure drops (left graphs), the HS and CS temperature differences (middle graphs) and the HS MER (right graphs) versus the number of active ducts. The ducting configurations tested here include reducing the total number of active ducts keeping the active HS and CS ducts equivalent (lower row; configuration C), reducing the number of active HS ducts (NHS) with the active CS ducts fixed at 63 (top row; configuration A), and reducing the active CS ducts (NCS) keeping the active HS ducts fixed at 63 (middle row; configuration B). For configuration C, the HSPD increases by a smaller amount compared with the CSPD as the NHS is reduced and is eventually

99

the HSPD is equivalent to the CSPD when NHS = NCS = 20, which is attributed to the reduction in the MER i.e. less condensate will mean less HS duct airflow resistance. Both the HSTD and CSTD get smaller as the number of active ducts is reduced, which is expected given the AVFR remains fixed for each test. The experimental CSTD still remains high relative to the HSTD for each test condition and this is again attributed to the latent and sensible cooling in the HS ducts and also heat gain from the ambient to the CS ducts. The experimental MER for configuration C decreases steadily with decreasing number of active ducts as expected. Configuration A shows that the HSPD increase as the active HS ducts is reduced whereas the CSPD shows only a gradual decrease with NHS reduction, given NCS remains fixed at 63. The difference between the experimental HSTD and CSTD starts to decrease with the reduction of NHS as less CS ducts interact with warm air, leading to reduced latent cooling and more sensible heat transfer from the HS ducts to the CS ducts. The small amount of MER with NHS = 20 still shows a HSTD and CSTD difference of 2.5 °C, which further indicates that the CS duct heat gain contributes to the HSTD and CSTD difference, rather than just latent cooling. In configuration B, the HSPD reduces as NCS is reduced, and this attributed to the decreasing amount of latent cooling in the HS ducts which would lead to less condensate restricting airflow in the HS ducts. On the other hand, the CSPD increases with the

Fig. 9. Left graphs: HS (black), CS (blue) and total (red) PHE pressure drop versus duct configurations. Middle graphs: HS (black) and CS (blue) temperature difference versus duct configurations. Right graphs: HS MER versus duct configurations. Top row: CS ducts fixed at 63 with HS ducts varied. Middle row: HS ducts fixed at 63 with CS ducts varied. Bottom row: both HS and CS ducts varied. The air volume flow rate was set to 0.0385 m3/s for each test. The HS moist air inlet conditions were 20 °C, 70% RH, and the CS moist inlet conditions were 5.5 °C, 90% RH. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

100

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

Table 3 Friction enhancement factors for various HS inlet dry-bulb settings and duct configuration settings. NHS

NCS

efHS

efCS

NHS

NCS

efHS

efCS

NHS

NCS

efHS

efCS

50 40 30

50 40 30

1.11 1.00 0.95

1.50 1.65 1.72

50 40 30

63 63 63

1.11 0.96 0.91

1.40 1.15 1.15

63 63 63

50 40 30

1.35 1.20 1.10

1.80 1.75 1.75

reduced NCS, as expected given that the AVFR remains fixed for each test. The HSTD decreases with decreasing NCS as the amount of HS duct cooling is decreased. The CSTD on the other hand does not change much over the entire NCS test range again suggesting that CS duct heat gain is a considerable issue in our test PHE. An interesting outcome of configuration B is that the MER stays equivalent to the MER produced in configuration C for the given number of active HS ducts. Configuration B gives the advantage of retaining a higher MER compared with configuration A while giving the same total PHE pressure drop as that of configuration A, while also allowing the option of reducing the HSTD. Compared to the reference PHE configurations of 63 active hot-side and cold-side ducts, in configuration A, the MER drops by 87% when the active hot-side ducts is set to 20 with the active cold-side ducts fixed at 63, whereas for configuration B, the MER only drops by 62%. Furthermore, in configuration A, the hot-side temperature difference is reduced by 1.4 °C with the active hot-side ducts set at 20 and in configuration B, the hot-side temperature difference is reduced by 2.5 °C with the active cold-side ducts set at 20. This shows that configuration B would be most suitable in the application of airside energy recovery in a heat pump dehumidifier, as the saturated suction temperature can be kept high enough to avoid evaporator frosting when the system operates at relatively low ambient temperatures, while retaining a relatively high MER at the PHE. We attribute this higher MER with less active CS ducts to the higher air speeds through the CS ducts, while the HS ducts air speed remains fixed, which in turn will give a longer residence time of air in the HS ducts, allowing the condensation rate to be increased. The numerical model was used to produce simulated data for the duct configurations shown here and the dashed curves on Fig. 9 show the numerical results. Once again, the numerical model was tuned using the efHS and efCS enhancement factors and the tuned factors are given in Table 3 for the given test condition. Once good agreement was achieved between the simulated and measured pressure drop data with the tuned ef values, the simulated HSTD showed good agreement with the measured HSTD results for configuration A and C. For configuration B, the simulated data does not change with decreasing NCS. The simulated CSTD trend shows agreement with the measured CSTD trend for configurations A and C, where the simulated trend in configuration A shows the CSTD decreasing with about the same gradient as the experimental data. Similarly, with configuration C, the simulated CSTD shows a decreasing trend with decreasing NCS, but the decrease is only subtle. The simulated and measured MER show good agreement for configuration A. While configuration C does not show good absolute agreement between the measured and simulated data, the simulated trend is reasonably well matched with the measured data. For configuration B, while the absolute simulated and measured data are relatively similar, the simulated trend appears to slightly increase with decreasing NCS. It is uncertain why the simulated HSTD remains fixed with decreasing NCS, but the authors believe that a more accurate MER would be simulated if the correct trend if the HSTD could be simulated. Simulated data for duct settings of 20 for either the HS, CS or both the HS and CS, was not obtained. The model could not achieve stability in the gas-phase momentum balance and further work is required to achieve converged results for NHS and NCS values below 30.

4. Conclusions & future work In this paper we have presented experimental results for an airto-air PHE tested under a range of moist air states and airflow conditions typical of domestic dehumidifier operation. The PHE was also tested under a range of tilt angles relative to a control orientation and under a number of ducting configurations in order to investigate thermal effectiveness control. Furthermore, the PHE was designed with two inlet faces having differing duct aspect ratios. Experimental data was used to extend the utility of a previously developed numerical plate heat exchanger model to simulate a range of moist air states, duct aspect ratios, air volume flow rates and ducting configurations. This work has shown that with appropriate tuning of friction enhancement factors, our numerical PHE model can closely simulate the behaviour of a PHE with varied duct geometries on both the hot-side and cold-side of the heat exchanger, for a range of moist airflow conditions. Under all test conditions the numerical model over predicts the cold-side temperature difference. We attribute this to the numerical model not calculating the liquid film thickness as the liquid-phase model does not solve the liquidphase momentum balance equation. We suggest development of this model for future work as it may also reduce the need for gas-phase friction enhancement factors. The PHE was tested under a range of tilt angles (30 to 20°) to see if the air-side pressure drop could be reduced while maintaining a relatively high MER. Tilting the PHE by ±10° gave a significant reduction in the hot-side pressure drop while retaining a relatively high MER. Experimental results for this particular PHE geometry showed that a negative tilt angle between 20° and 0° gave the greatest reduction in the HSPD. A tilt of 10° gave an MER reduction of 8%, while the HSPD was reduced by 33%. Further tilting to 20° gave a relative percentage reduction in HSPD compared with 10° of 3% with an associated 6% reduction in the MER. Tilt angles less than 20° led to a further reduction in MER but with only a small decrease in the HSPD. Tilting in the opposite (positive) direction resulted in a HSPD reduction of 19% for an MER reduction of 3% at 10° but the MER drops off rapidly at 20° suggesting that the liquid film has spread over a greater area of the heat exchanger’s heat transfer surface. It is also possible that having the liquid film spread over heat transfer area where the wall temperature is greater than the dew-point temperature may lead to some condensate re-evaporating into the air-stream, however further experimental work would be needed to verify this. This suggests relatively small negative tilt angles (between 0 and 20°) are preferable for PHE operation under two phase flow conditions in order to obtain a significant HSPD reduction with only a minor reduction in MER. The PHE was tested for operation with various ducting configurations by blocking off ducts on either side of the heat exchanger but keeping the air volume flow rate fixed on either side of the heat exchanger. Good qualitative and reasonable quantitative agreement was demonstrated between the numerically simulated data and experimental data for the different ducting configurations, however, for configuration B – where the number of active coldside ducts was reduced while keeping the number of active hotside ducts fixed at 63 – the trends between the MER and hot- and cold-side temperature differences were not demonstrated.

S. Lowrey, Z. Sun / Applied Thermal Engineering 131 (2018) 89–101

The experimental results for configurations A and B showed that both methods of thermal effectiveness control can reduce the hot-side temperature difference, where the hot-side temperature difference is reduced with decreasing number of active ducts on one side of the heat exchanger (hot-side in the case of configuration A and cold-side in the case of configuration B), while fixing the number of active ducts on the adjacent side of the heat exchanger. Configuration A showed a hot-side difference reduction from 7.2 °C (all hot-side ducts active, 63 cold-side ducts active) to 5.8 °C (20 hot-side active, 63 cold-side ducts active) where the MER drops by 87%. Configuration B showed a hot-side difference reduction from 7.2 °C (63 hot-side ducts active, 63 cold-side ducts active) to 4.7 °C (63 hot-side active, 20 cold-side ducts active) where the MER drops by 62%. It’s important to note that the overall pressure drop (hot-side pressure drop plus cold-side pressure drop) in both configurations A and B is comparable. Therefore, these results show that controlling the thermal effectiveness by modulating the cold-side active ducts is favourable compared to modulating the hot-side ducts as configuration B gives a larger range of hot-side duct temperature difference control, while maintaining a higher MER. This is a very useful result for controlling thermal effectiveness of PHE’s used for energy recovery in dehumidification systems. Energy recovery in dehumidifiers that operate at relatively low temperatures using an air-side economizer (in the form of an air-to-air plate heat exchanger) has shown that the economizer can contribute to around 20% of the system’s total dehumidification rate and also, that evaporator frosting can be exacerbated when introducing an economizer to the system [10]. The results from configurations A and B show that configuration B is ideal for modulating thermal effectiveness in the application of mechanical dehumidification as the dehumidification rate can be kept high, while the hot-side temperature difference can be reduced by more than 1 °C compared with configuration A, allowing evaporator frosting to be avoided and the dehumidification rate to be kept relatively high down to lower ambient temperatures. The experimental results from this paper can be useful for optimisation and control of compact, condensing, air-to-air plate heat exchangers for air-side energy recovery in applications such as advanced domestic dehumidification, closed air cycle condensing clothes dryers and advanced heat pump clothes dryers. The numerical model presented in this work also provides for a quantitative approach when engineering emerging and advanced domestic scale dehumidification and heat pump clothes drying systems. Acknowledgements The authors would like to thank Peter Stroud and Rowan Davies for the construction of the equipment used in this work.

101

References [1] Handbook-Heating, ASHRAE, Ventilating, and Air-Conditioning Systems and Equipment (IP Edition), American Society of Heating, Refrigerating and AirConditioning Engineers, Inc, (2008) 25.9–25.10. [2] M. Cochran, J. Goodnight, B. Babin, S. Eckels, Condensing dryers with enhanced dehumidification using surface tension elements, Appl. Therm. Eng. 29 (2009) 723–731. [3] L. Mattison, D. Korn, Dehumidifiers: A Major Consumer of Residential Electricity, The Cadmus Group, Inc. August. ACEEE Summer Study on Energy Efficiency in Buildings, 2012. [4] W.E. TeGrotenhuis, A. Butterfield, D. Caldwell, A. Crook, A. Winkleman, Modeling and design of a high efficiency hybrid heat pump clothes dryer, Appl. Therm. Eng. 124 (2017) 170–177. [5] E. Doderer, C. Clower, Efficient mechanical dehumidification through sensible cooling recovery 81-WA/Sol-33 Am. Soc. Mech. Eng. (1981). [6] J. Dieckmann, K. McKenney, J. Brodrick, Energy-efficient dehumidification, ASHRAE J. (2009) 78–80. [7] P. Bannister, C.G. Carrington, Q. Liu, Influence of enhancing features on dehumidifier performance: laboratory measurements, Int. J. Energy Res. 19 (1995) 397–406. [8] Y.H. Yau, A.S. Tucker, The performance study of a wet six-row heat-pipe heat exchanger operating in tropical buildings, Int. J. Energy Res. 27 (2003) 187– 202. [9] MSP Technology, MSP Technology: The Next Generation of Green Technology. http://www.msptechnology.com/msp-hybrid-dehumidification-technology (accessed: 05.08.17) [10] S. Lowrey, G. Carrington, Z. Sun, M. Cunningham, Experimental Investigation of geared domestic refrigerative dehumidifier performance in New Zealand household climates, Int. J. Refrig 35 (2012) 750–756. [11] S. Lowrey, G. Carrington, Z. Sun, Adapting the geared domestic refrigerative dehumidifier for low-temperature operation, Int. J. Refrig 41 (2014) 137–146. [12] C.A.B. Pereira, R.H. Pereira, R.P. Marquas, J.A.R. Parise, J.R. Sodre, Experimental analysis of a heat pump assisted recuperative air dehumidifier, Therm. Eng. 5 (2004) 56–61. [13] G. Galbraith, C.H. Sanders, C. Allison, Portable dehumidifiers for the control of condensation in housing, Build. Serv. Eng. Res. Technol. 7 (1986) 1–10. [14] M.J. Cunningham, C.G. Carrington, Domestic Dehumidifiers in Cool Conditions I – Performance Factors, Proceedings of Indoor Air 2005: 10th International Conference on Indoor Air Quality and Climate, Beijing, PR China. 2005. [15] S. Lowrey, Z. Sun, A numerical model for a wet air-side economiser, Int. J. Refrig 60 (2015) 38–53. [16] M.A. Kedzierski, Effect of inclination on the performance of a compact brazed plate condenser and evaporator, Heat Transfer Eng. 18 (2007) 25–38. [17] The Engineering Tool Box. http://www.engineeringtoolbox.com/thermalconductivity-metals-d_858.html, 2017 (accessed: 07.08.17) [18] ASHRAE. Standard method for temperature measurement. Standard Method for Temperature Measurement. (1986) [19] A. Wexler, R.W. Hyland, R.B. Stewart, Thermodynamic properties of dry-air, moist-air and water and SI psychrometric charts, Reports from ASHRAE research projects (1984) 216-RP and 257-RP. ASHRAE J. [20] C.G. Carrington, A. Marcinowski, W.J. Sandle, A simple volumetric method for measuring airflow, J. Phys. E: Sci. Instrum. 15 (1982) 275–276. [21] S. Patankar, Numerical Heat Transfer and Fluid Flow, McGraw-Hill Book Company, New York, 1981. [22] Y.A. Cengal, A.J. Ghajar, Heat and mass transfer: fundamentals and applications –, 4th edition., McGraw-Hill Book Company, New York, 2011. [23] F. Incropera, D. Dewitt, Fundamentals of Heat Transfer, John Wiley & Sons Inc, New York, 1981. [24] R. Bird, W. Stewart, E. Lightfoot, Transport Phenomena, 2nd ed., Wiley, New York, 2007.