Acta Polytechnica DOI:10.14311/AP.2020.60.0318 Acta Polytechnica 60(4):318–323, 2020 © Czech Technical University in Prague, 2020 available online at https://ojs.cvut.cz/ojs/index.php/ap CHAOTISED POLYMERIC HOLLOW FIBRE BUNDLE AS A CROSSFLOW HEAT EXCHANGER IN AIR-WATER APPLICATION Tereza Krouliková∗, Ilya Astrouski, Miroslav Raudenský Brno University of Technology, Faculty of Mechanical Engineering, Heat Transfer and Fluid Flow Laboratory, Technická 2896/2, 616 69 Brno, Czech Republic ∗ corresponding author: Tereza.Kroulikova@vut.cz Abstract. Fifteen years ago, polymeric hollow fibre heat exchangers were presented for the first time. Nowadays there are not only the shell-and-tube types as there were at the beginning. In this paper, six chaotised polymeric hollow fibre bundles with a different number of fibres were studied. The bundles presented varied in their fibre diameter, number and shape. These bundles were fixed into the module in such a way that the middle part serves as a cross-flow heat exchanger in an air tunnel. They were tested for air-water application with three different airflow rates. The overall heat transfer coefficients were determined, and the inner and outer heat transfer coefficients were derived. The modules presented achieved a heat transfer rate of up to 1309 W. The overall heat transfer coefficient reached a maximum of 339 Wm−2 K−1. Keywords: Polymer hollow fibre, chaotic structure, cross-flow heat exchanger, gas-liquid application. 1. Introduction Polymeric hollow fibre heat exchangers (PHFHE) are an alternative to common metal heat exchangers in low temperature applications. Their advantages are low cost, low weight and corrosion resistance. Their heat transfer surface consists of hundreds or even thousands of fibres with a small outside diameter, commonly 0.4 – 1.6 mm. Polymeric material provides such advantages as light weight, easy machining and forming, flexibility and corrosion resistance. The dis- advantage of polymeric material is its low thermal conductivity (0.1–0.4 Wm−1 K−1), which is 100–300 times lower than metals. The high thermal resis- tance can be overcome by the hollow fibres having a small diameter and thin wall. According to [1], the wall thickness for PHFHE should be kept below 100 µm. PHFHE are mainly made of polypropylene (PP), but also other polymers are used, such as polyamide (PA), polyetheretherketone (PEEK), polyphthalamide (PPA) or asymmetric polyethersulfone (PES). PHFHE were firstly presented by Zarkadas in [2] fifteen years ago as an alternative to conventional shell- and-tube heat exchangers. In his study, the overall heat transfer coefficients were 647− 1314 Wm−2 K−1 for water–water system and 414− 642 Wm−2 K−1 for ethanol–water system. As PHFHE are corrosion resis- tant, they are mainly studied in desalination process [3, 4]. But they were also studied for a cooling system of a solar panel [5] or the HVAC (heating, ventilation, air-conditioning) applications [6]. The gas-liquid application was studied in [7], where two PHFHE prototypes were presented as a possi- ble alternative to the aluminium automotive radiator. Two modules with a rectangle cross-section measur- ing 250 × 250 mm were prepared and tested. The 50/50% water glycol solution was used as a coolant. Both devices achieved similar heat transfer rate, up to 10.4 kW and high values of the overall heat transfer coefficient, up to 335 Wm−2 K−1. That study also mentions the significant influence of the diameter of the polymeric hollow fibre on heat transfer coefficients on both inner and outer surfaces of the fibre. In order to provide direct contact between the fibres and the surrounding stream of fluid, it is necessary to separate fibres from each other to let the fluid flow between them. Unseparated fibres have an extensive mutual contact, therefore, they are blocking them- selves and only a few of them are active. A method of separation of bundles is chaotisation, which was presented in [8]. The fibres are deformed by stretching and then thermal fixation of the shape generated by stretching is applied. This provides direct contact between fibres and the surrounding fluid and enables better heat transfer. Chaotised bundles were studied as immersed heat exchangers in a hot water reservoir [9]. Three bundles with two different diameters were put into a hot water reservoir. The results obtained showed a high value of heat transfer rate, up to 22.76 kW. Their analysis also indicates a strong dependence of overall heat transfer coefficient on the quality of the fibre distribution. The polymer material is generally more resistant to fouling than metals [1]. The fouling of PHFHEs was studied in [10] and [11]. The particulate fouling in [10] decreased the overall heat transfer coefficient by 20 % and defouling by air-bubbling was found as efficient. The fouling in shower wastewater and laundry wastew- ater was studied in [11]. In the first experiment, no significant decrease of overall heat transfer coefficient was observed during the two-week experiment. In the latter case, the overall heat transfer coefficient dropped from 1747 Wm−2 K−1 to 963 Wm−2 K−1 318 https://doi.org/10.14311/AP.2020.60.0318 https://ojs.cvut.cz/ojs/index.php/ap vol. 60 no. 4/2020 Chaotised polymeric hollow fibre bundle. . . and the surface of the fibres was covered by biofilm and solid particle deposit. Both mentioned papers concern liquid applications, there are no fouling data specifically for PHFHEs for a gas application. The yield strength of polymers is much lower than the yield strength of metals [1]. This can limit the min- imal wall thickness without compromising the physical integrity of the heat exchanger structure. PHFHE are a promising alternative in applications, which require non-corrosive and clean materials or where weight plays a major role, for example, heat re- covery systems or HVAC systems. Chaotised PHFHE are elastic and flexible, which allows for the use of some design variants, which can improve the overall heat transfer and utilise some spaces, which are not suitable for non-flexible heat exchangers. In the presented study the chaotised polymeric hol- low fibre bundles were used as the cross-flow heat exchanger. As material, the polypropylene was cho- sen, since it is easy to make the chaotised bundles from it. Six modules were made, tested and compared in an air-water application. 2. Experiment In the laboratory, six modules (M-200 – M-1200) with a different number of fibres were prepared from chao- tised polypropylene hollow fibre bundles. The bundles were put inside a box and plastered in such a way that there is a passage with a square cross-sectional area (100× 100mm), see Fig. 2. The only active heat transfer surface is in the passage. Three modules were made of fibre with an outside diameter of 0.6 mm and the other three with a diameter of 0.8 mm. In both cases, the wall thickness is 10 % of the fibre outer di- ameter. Those are the most common fibre dimensions used in gas-liquid applications. All the modules differ in the number of fibres used. The average over-length was defined as a way of measuring the rate of the chaotisation. It is the ratio of an average length of a fibre in the active area to the width of the tunnel passage. Over-length influences the ratio between the heat transfer area and the volume of the bundle. The parameters of the tested bundles are given in Table 1. Those modules were placed in an air tunnel, see Fig. 3. The air circulation inside the tunnel was ensured by a small fan and water flowed inside the hollow fibre modules. The flow rate and temperature of the water were controlled and maintained as required by external equipment (test circuit of the calorimetric room). The actual water flow rate was measured by the magnetic-inductive flowmeter Krohne OPTI- FLUX 4300 (measuring tube DN10) and recorded by a data acquisition system. The flowmeter is calibrated and ensures precision of ± 0.2%. The water and air temperature were measured by Pt100 1/3 class A temperature sensors (± 0.1 K precision in 0–80 ◦C temperature range). The water temperature sensors were connected to a data acquisition system of the testing circuit and the air temperature sensors were T T T T P H F H E C a lo r im e t r ic c h a m b e r 3× 6× Fan Figure 1. Scheme of the test rig. connected to National Instrument NI-9217 modules. The air flow rate was measured by the portable hot wire anemometer Omega HHF-SD1 (calibrated with precision ± 5 %) placed in the tunnel 30 cm behind the tested bundle. Three temperature sensors Pt100 were placed in front of the module and six behind the module. The average temperatures values of Tc,in and Tc,out, air inlet and air outlet respectively, are calculated from the temperatures measured. Then, the air and water flows were monitored. All modules were tested three times with different air and water flow rates. In all runs, the water flow rate was adjusted in such a way that the inlet tem- perature was 75 ◦C and the outlet temperature was 65 ◦C, thus the difference in water temperature was 10 K. 3. Data reduction Firstly, the thermophysical properties of the water and the air were calculated based on the average of the inlet and outlet temperature of liquids. The properties of water were considered to be temperature dependent and were calculated using the following formulas. The viscosity: µ = exp ( −6.358− 2.88 · 10−2T + 1.31 · 10−4T 2 −2.58 · 10−7T 3) , (1) the specific heat capacity: cp = 1.3410−9T 6 − 4.9506 · 10−7T 5+ 7.09647 · 10−5T 4 − 0.004864569T 3 + 0.16759809T 2 − 2.81027645351T + 4201.37207, (2) the thermal conductivity: k = 5.76 · 10− 1 + 1.77 · 10−3T − 6.37 · 10−6T 2 (3) and the density: ρ = 1001−0.0672T−4.04·10−3T 2+4.94·10−6T 3, (4) where in all the above formulas, T is temperature in ◦C. Thermophysical properties of the air were interpolated from the table in the appendix of [12]. 319 T. Krouliková, I. Astrouski, M. Raudenský Acta Polytechnica Module No. of Outer diameter Inner diameter Total outer Average active fibres of fibre [mm] of fibre [mm] surface area [m2] overlength M-200 186 0.8 0.64 0.111 2.38 M-300 199 0.6 0.48 0.064 1.7 M-400 320 0.8 0.64 0.193 2.4 M-600 440 0.6 0.48 0.167 2.01 M-800 530 0.8 0.64 0.318 2.39 M-1200 1049 0.6 0.48 0.402 2.03 Table 1. Characteristics of tested modules. Figure 2. One of the tested modules. Figure 3. Experimental tunnel. 320 vol. 60 no. 4/2020 Chaotised polymeric hollow fibre bundle. . . The overall heat transfer coefficient of the cross-flow PHFHE without phase change can be obtained from: Q = ṁhcp,h (Th,in − Th,out) (5) = ṁccp,c (Tc,out − Tc,in) , (6) Ui = Q 4TlmAiF , (7) where Q is the heat transfer rate, ṁh, ṁc are mass rates and cp,h, cp,c are specific heat capacities of hot water and cold air respectively, Ui is the overall heat transfer area based on the inner area, Ai is the inner area and 4Tlm is log mean difference temperature calculated from: 4Tlm = (Th,in − Th,out) ln ( Th,in − Tc,out Th,out − Tc,in ) (8) and F is the correction factor, which is calculated by: F = NTUcf NTU , (9) where NTUcf is the so called number of transfer units of counterflow heat exchanger and NTU is the num- ber of transfer units of the tested heat exchanger. The actual correction factor and overall heat transfer coefficient must be calculated simultaneously based on the experimental data, therefore, the NTU was computed by a formula for a single pass cross-flow heat exchanger with one fluid mixed and one unmixed [12, 13]. To calculate the tube-side and air-side heat trans- fer coefficients, hi and ho, the approach proposed by Hickman was used [2]. The formulas for Nusselt num- ber under the boundary condition of third kind NuT 3, wall Nusselt number Nuw and the reciprocal of the lumped resistance term 1/Uw, the wall heat transfer coefficient, are given by: NuT 3 = 48 11 + Nuw 1 + 59 220Nuw , (10) Nuw = UwDi ki (11) 1 Uw = Di Doho + Di 2kw , (12) where Do and Di are the outer and inner diameter of the hollow fibre, ki is the thermal conductivity of water, kw is the thermal conductivity of the polypropy- lene wall, 0.18 Wm−1 K−1. The overall Nusselt num- ber can be calculated using 1 Nuov = ki UiDi = 1 NuT 3 + 1 Nuw , (13) 1 Ui = 1 hi + 1 Uw (14) where the tube-side heat transfer coefficient is com- puted from the inside Nusselt number: hi = NuT 3ki di . (15) After some manipulation, a quadratic equation can be obtained( 1− 59 220Nuov ) Nu2 w+ ( 48 11 − 2Nuov ) Nuw − 48 11Nuov = 0 (16) and its solutions are a positive and a negative root. The positive one is considered to be the wall Nusselt number. This approach allows us to determine both heat transfer coefficients based on the experimentally obtained overall heat transfer coefficient. 4. Results and discussion All experiments were done with the tube-side Reynolds number in the range of 38 − 667, therefore, all ex- periments were done with the tube-side flow in the laminar regime. The air speed was relatively low, max. 4.5 m s−1. Tab. 2. contains a range of results of all 18 experiments, such as the Reynolds number Retube, the air speed vair, the heat transfer rate Q, the over- all heat transfer coefficient based on the outer fibre surface Uo, and air-side and tube-side heat transfer coefficients,ho and hi respectively. The Fig. 4 shows the heat transfer rate (the average of the hot and cold side) dependent on tube flow and air speed. The heat transfer rate was up to 1309 W for the module M-400, but also three others were over 1200 W. It can be seen that the heat transfer rate on the tube side depends on the tube flow since the difference between the inlet and outlet temperature was set to be the same for all runs. The heat transfer rate on the air side is obviously affected by the air speed and the surface area, i.e., the number of fibres and their diameters. The M-1200 module’s heat transfer surface is larger than M-800 module’s one, but its performance is the same as the M-800. The heat transfer area of the module M-1200 is too large and inefficient. Also, it is possible, that there are too many fibres in the volume that they block each other even though they are separated by chaotisation. The overall heat transfer coefficient was up to 339 Wm−2 K−1. But not only the maximum value is high, see Fig. 5. For example, [14] denotes 25− 55 Wm−2 K−1 as the approximate values of the overall heat transfer coefficient for a finned-tube heat exchanger. It is true that those values incorporate the fouling factor, which is not available for the PHFHE since there has been little study of this issue. The air-side heat transfer coefficient behaves simi- larly to the overall heat transfer coefficient, see Fig. 5. It is dependent on the air speed and the outer di- ameter. The modules with Do = 0.8 (M-200, M-400 321 T. Krouliková, I. Astrouski, M. Raudenský Acta Polytechnica Module vair [m s−1]