Session Thermalhydraulics: Fluidized and Packed Beds 1 DIRECT CONTACT AIR-WATER HEAT TRANSFER IN A COLUMN WITH STRUCTURED PACKING Sofoklis Kypritzis and Anastasios J. Karabelas Department of Chemical Engineering and Chemical Process Engineering Research Institute, Aristotle University of Thessaloniki, Univ. Box 455, GR 54 6 Thessaloniki, Greece Tel: (+3) 31 9961, Fax: (+3) 31 9969, E-mail: karabaj@cperi.certh.gr ABSTRACT This paper deals with the process of direct contact heat transfer between hot (relatively dry) and sub-cooled water flowing counter-currently in a column filled with structured packing (Sulzer, Melapak 25.Y). The main objective is to collect reliable experimental data for a type of process equipment (structured packed column) inadequately studied so far. The range of flow parameters studied was as follows: Air flux between 2.4 and 6.2 tn/hm 2 and Reynolds number between 8 and 215; water flux between 7.2 and 17.2 tn/hm 2 and Reynolds number between 1.4 and 25. The pressure varied from atmospheric to 1 bar gauge. A fully instrumented pilot scale unit was employed in the tests with a stainless steel column 7cm long and 15cm i.d. A significant amount of new data has been collected by measuring the local temperature at various levels of the column. Sensible heat exchange between and water in the column takes place simultaneously with mass transfer. Interpretation of data shows that the process strongly depends on the flow rate, as expected. The influence of liquid rate is noticed only in the upper section of the column. There is evidence that a very significant amount of heat, exchanged between the two streams is due to latent heat (mass) transfer. The bottom part of the packed bed operates essentially as an cooling section promoting uration of the supplied hot. Under some conditions, both humidification and subsequent de-humidification may take place in the bottom section i.e. within a rather short packing segment. Above this section, the device seems to operate as an ordinary direct contact heater of the liquid phase. For the middle and upper sections, common correlations provide predictions in rough agreement with measured heat transfer coefficients. INTRODUCTION There are many process applications of /liquid direct contact equipment where the main objective is transfer of heat. The preference for this type of equipment is due to its rather simple design (leading to reduced capital expenses) and ease of operation since heat transfer surfaces (and their usual corrosion and fouling problems) are absent, thus significantly reducing operating and maintenance expenses. Considerable work has been carried out on packed bed type direct-contact condensers (e.g. F, [1]). However, well-established methods for predicting heat transfer rates in such devices are still unavailable. The situation is worse in the case of structured packed beds, which are very attractive for a variety of applications. Indeed, there is a lack of heat transfer data in the open literature, helpful in clarifying some key issues involved and in providing guidance for design calculations [2], [3]. A complication frequently arising in studies of directcontact heat transfer is due to the simultaneous transfer of mass which may be in the same direction as the main heat flow or may proceed opposite to it, depending on process conditions. Therefore, experimental determination of overall heat transfer coefficients over the entire packed column may mask such effects and may not help improve our understanding of this process. In one of the few available direct-contact heat transfer studies on structured packings, Spigel et al [4] obtained data with /water and an /oil system. Overall transfer units and overall heat transfer coefficients U were reported for packed columns.63 and.675 m high. Correlations were also proposed for predicting heat transfer rates for structured packings. For the case of sub-cooled water brought in contact with hot, they reported that U varied with the Reynolds number to a power.8. Huang and F [5] h obtained a considerable amount of data on the /water system in various types of packed columns but not in structured packings. Bontozoglou and Karabelas [6] carried out experiments with the same structured packing employed in this work utilizing steam, with a small percentage of CO 2, flowing upwards in a column where sub-cooled water was fed at the top. They reported that the condenion coefficient was enhanced at the lower levels of the column (close to the steam feeding point), which was attributed to the fact that during the early stages of /liquid contact, the (mainly urated steam) is poor in non-condensables. However, as the distance from the feeding point increases (and so does the /liquid contact time) more steam condenses on the /liquid interface and the mixture becomes richer in non-condensables which retard condenion. The scope of this work is to collect reliable experimental data in a pilot scale column with structured packing that would enrich the literature and would offer the opportunity to improve understanding of the direct-contact process along the column. The common /water system is employed which (in addition to its practical usefulness) allows comparison with a few relevant literature studies. 1
1, ExHFT-5, Thessaloniki, Greece EXPERIMENTAL EQUIPMENT AND PROCEDURE The experimental system is shown in Figure 1. The condenser is a 1,5 m long,,15 m I.D. column made of stainless steel 316. A hot well, equipped with a magnetic floater forms the bottom of the column. The liquid outlet valves are electrically actuated through the floater in order to prevent the passage of in the drainage pipe. The column is filled with Mellapak 25.Y structured packing marketed by Sulzer, which appears to be one of the packing types favored for direct contact applications. Characteristics of this packing are listed in Table 1. Tap water is demineralized and delivered to a spray manifold at the top of the column. The water flow rate is controlled by a PID controller and a flowmeter which electronically actuate a Badger control valve. Dry is provided by the Laboratory compressor facility. Heating takes place in a special vessel equipped with electrical resistances (3,5 KW). Pressure gauges and flowmeters are installed to monitor flow conditions. The heating vessel, piping and packed column are thermally insulated to avoid heat losses and achieve adiabatic conditions. A temperature controller is installed at the outlet of the heating vessel to maintain constant temperature (1 o C) Temperatures are monitored with K-type thermocouples calibrated to ±,2 o C. Thermocouples are installed in the well (TI1), at the cold water inlet (TI7) and at the outlet (TI8). Five thermocouples are embedded in the packing (identified by TI2, TI3, TI4, TI5, TI6, in Figure 1) located at a distance 8, 1, 16, 375, 585 mm (respectively) from the bottom of the packing. The thermocouple tips are positioned close to the centerline of the column, in contact with the packing to provide the temperature of the local liquid film. Temperatures are indicated on a central panel and are also recorded by a data acquisition system for later processing. The range of conditions tested is as follows: Air flux between 2.4 and 6.2 tn/hm 2 and Reynolds number (based on the effective velocity of ) between 8 and 215. Water demineralization FI PI Air heating device PI TIC FIC Tap water inlet TI6 TI5 TI4 TI3 TI2 TI7 Hot & dry inlet, Water flux between 7.2 and 17.2 tn/hm 2 and Reynolds number (based on superficial water velocity) between 1.4 and 25. Pressure from atmospheric up to 1 bar gauge. To facilitate data interpretation the packed bed is considered to be comprised of five sections henceforth to be referred to as top, upper, middle, lower and uration sections, as indicated in Figure 1. The locations of thermocouples TI3, TI4, TI5 and TI6 serve as the boundaries of these sections. Thermocouple TI2 provides an estimate of uration section temperature. RESULTS Water Temperature Profiles Typical water temperature profiles measured in these tests are plotted in Figures 2,3 and 4. These data show that water is heated (as expected) while flowing through the top, upper, middle and lower sections. A modest water heating is achieved in the three top sections, possibly due to sensible heat transfer together with some humidity (mass) transport from the urated warm to the falling water films, under conditions of forced convection. A rather sharp temperature increase takes place (up to a maximum) within the lower section. At first look, it is doubtful whether convective heat transfer is the main mechanism responsible for this significant heating. Of greater interest is the reduction of water temperature within the so-called uration section. This is apparently caused by the humidification of the incoming hot dry, effected through evaporation and leading to water cooling. This explanation is supported by the observation that thermocouples TI1 and TI2 systematically measure water temperatures lower than TI3, as shown in figures 2,3,4. In summary, it appears that hot Top section Upper section Middle section Lower section Saturation section Cold water inlet TI8 7mm PI TI1 Cold & humid outlet to Atmosphere. LC Pressurized dry at 8 bar abs Heated water outlet to drain FIG. 1: The experimental device 2
Session Thermalhydraulics: Fluidized and Packed Beds 1 humidification of the urated up-coming by the colder falling water films which (aided by sensible /water convective heat transfer) leads to increasing water temperature. Water temperature ( C) Water temperature ( C) Water temperature ( C) 7 6 5 4 3 35 3 25 15 P 2. bar, Re(g) 13 P 1.5 bar, Re(g) 134 P 1.25 bar, Re(g) 128 P 1. bar, Re(g) 138 1 at well 4 6 8 3 25 15 Distance from the bottom of the packing (mm) FIG. 2: Typical water temperature profiles at constant water flux (12.2 tn/hm 2 ) and Re ~constant P 1.5 bar, Re(g) 215 P 1.25 bar, Re(g) 2 P 1. bar, Re(g) 3 1 at well 4 6 8 Distance from the bottom of the packing (mm) FIG. 3: Typical water temperature profiles at constant water flux (12.2 tn/hm 2 ) and Re ~constant T() P 2. bar, Re(g) 16 T() P 2. bar, Re(g) 16 T(), P 1.25 bar, Re(g) 159 T(), P 1.25 bar, Re(g) 159 1 at well 4 6 8 Distance from the bottom of the packed bed (mm) FIG. 4: Typical water and temperature profiles at constant water flux (12.2 tn/hm 2 ) and Re ~constant Air Temperature Profiles To further interpret the data, it is essential to determine the prevailing temperature profiles in the column under various conditions. In Figure 4 calculated temperatures are included which were obtained, assuming adiabatic column operation and urated eous phase, via the energy balance for water: Q Lc (T out, T in, ) (1) and for humid : Q G[c (T in, T out, ) + (H in H out )λ ] (2) In equations [1] and [2] all physical quantities are known; L, T out,, T in,, G and T out, are measured and H out can be obtained via a psychrometric chart. Equation [2] may be rearranged to solve for T in, for the corresponding H in. This procedure is repeated sequentially for all sections (except the uration section), starting from the top section where boundary conditions (T out, and T in, ) are obtained from the corresponding thermocouples (TI8 and TI7). Heat Transfer Coefficients Figures 5 through 1 depict calculated overall heat transfer coefficients versus Reynolds number for upper, middle and bottom sections. Various water fluxes in a range 7.2 to 17.22 m 3 /m 2 h are employed, for pressure 2. and 1.5 bar absolute. The overall heat transfer coefficient U, is obtained via equation [3] since all relevant quantities are calculated from equations [1] and [2]. In equation [3] the logarithmic mean temperature difference is approximated by the arithmetic mean difference, without introducing a significant error: Q Q U [3] aza ( T ) lm aza( T ) m Neither enhancement of interface compared to geometric packing area, as suggested by Henriques de Brito et al [7], nor reduction due to liquid maldistribution effects is considered here in order to specify the effective area A for heat/mass transfer between humid and water. Thus, the total effective interface area in the computations was assumed to be constant and equal to the nominal area of the packing (indicated in Table 1). Figures 5 to 1 show that the heat transfer coefficient (for all three column sections considered) is strongly affected by Reynolds number, as expected. It is also observed that, for the lower and middle sections (Figures 6, 7, 9, 1), the heat transfer coefficient U is almost independent of the liquid flow rate. Indeed, a relatively small variation of U for various liquid rates (for a fixed Re G ) is not systematic and appears to be within the experimental error (± 15%). This behaviour of U may be attributed to a controlling resistance to heat transfer residing at the side of the interface. Furthermore, the insignificant effect of liquid rate in these sections suggests that the condition of the falling liquid films does not influence the transfer rates at the side. Upon inspection of Figures 7 and 1 (for the lower section), corresponding to absolute pressure 2 and 1.5 bar respectively, one can make the following observations: 3
1, ExHFT-5, Thessaloniki, Greece The magnitude of U is much greater than that for the other two sections. Figure 1 for P1.5 bar depicts (at high Re G ) a decreasing U with increasing Re G. The latter trend appears to be physically impossible and may result from violation of assumptions involved in determining the transfer coefficient U; i.e., that only sensible heat transfer and condenion from a urated stream take place in this section. In reality, however, it is possible that (at sufficiently high velocities) the packing section designated as uration section (between TI2 and TI3, Figure 1) may not be thick enough, and the residence time there may be too short, to attain uration. Thus, uration may be completed in the next section (designated as lower in Figure 1). Furthermore, it is not unlikely that in the same ( lower ) section water evaporation ( humidification) may take place first followed by condenion. Under these conditions, Equ (2) would underestimate Q for that section leading to reduced U values. Apparently, with increasing Re G, U would keep decreasing as shown in Figure 1. 45 4 35 3 25 15 1 5 12 1,6 Heat transfer coefficient (W/mxmK 1 8 6 4 12 1,6 theoretical prediction 4 6 8 1 14 16 18 FIG. 5: Heat transfer coefficient vs Re of the upper section of the column (P 2. bar abs) 5 15 25 FIG. 8: Heat transfer coefficient vs Reynolds of the upper section of the column (P 1.5 bar abs) 1 8 6 4 12 1,6 Heat transfer coefficient (W/mxm 14 1 8 6 4 12 1,6 prediction thoretical FIG. 6: Heat transfer coefficient vs Re of the middle section of the column (P 2. bar abs) Heat transfer coefficient (W/mxmK 5 45 4 35 3 25 15 5 5 15 12 1,6 5 15 FIG. 7: Heat transfer coefficient vs Re of the lower section of the column (P 2. bar abs) 5 15 25 FIG. 9: Heat transfer coefficient vs Re of the middle section of the column (P 1.5 bar abs) 45 4 35 3 25 15 5 12 1,6 5 15 25 FIG. 1: Heat transfer coefficient vs Re of the lower section of the column (P 1.5 bar abs) 4
Session Thermalhydraulics: Fluidized and Packed Beds 1 The results in Figure 7 (P2. bar absolute) show that (unlike those in Figure 1) U increases monotonically with Re G. In line with the above explanation, this may be attributed to the higher pressure and the relatively smaller quantity of vapour required for uration, which may then be completed within the uration section. Consequently, Equation (2) would be isfied in the next ( lower ) section with the expected U versus Re G variation. The high U values in the lower section (e.g. Figure 7) are difficult to explain. One may attribute this to superuration of entering that section, which would lead to an amount of latent heat released greater than that accounted for in the procedure used for determining U (equ 1 and 2). Figures 5 and 8 as well as Figures 11 and 12 show that whereas for the lower and middle sections U is practically independent of water flow rate, for the upper section U is roughly inversely proportional to Re L. This rather curious trend of transfer coefficient U is not observed for the first time in direct-contact experiments. Bontozoglou and Karabelas [6] studying steam condenion in the same experimental setup, reported similar trends. Karapantsios et al [8] obtained experimental data of direct-contact steam condenion on falling water films, inside a vertical tube, in the presence of large amounts of non condensable es; it was also observed that, by increasing the liquid flow rate, reduced values of integral heat transfer coefficient were obtained. They hypothesized that (with increasing liquid rate) the enhanced liquid ws trapped noncondensable es at the interface (or they caused a steam deplection of the boundary layer at the side) promoting a reduction of the mass transfer coefficient. It is uncertain whether the above arguments apply to the present system, and this matter requires more attention in future studies. In view of the above observations that the resistance to heat transfer resides at the side, an attempt is made to predict integral heat transfer coefficients U in a section by employing common correlations. Considering sensible heat transfer, and latent heat transported to liquid films through turbulent convective mass transfer, one obtains U as follows: Q Q sens azaut + Q ( T ) azaht ( T ) + azakλ ( H H ) H U Kλ lat, T H T, + h,, Following F & Bravo [9], Schpigel & Meier [1] one may employ the same type of correlation for estimating transfer coefficients K and h..8.33 Nu.34 Re Pr (5a).8.33 (5b) Sh.34 Re Sc Predictions based on this approach are plotted in Figure 8 and 9 for the upper and middle sections where (as already discussed) the assumptions involved in estimating U may be better isfied than in other sections. Predictions are in rough (order of magnitude) agreement. It is interesting that the exponent of the Re G dependence of U data is closer to 1. than to.8 employed in the above correlations. (4) Heat transfer coefficient (W/mxm K) upper section middle section lower section 1 5 1 15 25 3 Liquid Reynolds number FIG. 11. Heat transfer coefficient vs Re(liq) (P 2. bar abs & Re(g) 16) upper section middle section lower section 1 5 1 15 25 3 Reynolds number of liquid FIG. 12: Heat transfer coefficient vs Re(liq) (P 1.25 bar abs & Re(g) 147) CONCLUDING REMARKS The well-known /water system employed here is characterized by a relatively large enthalpy of vaporization.therefore, if there is a large difference of inlet temperatures of the two streams, brought into direct contact, latent heat effects dominate in certain sections of the column. Moreover, if the incoming is dry, both humidification and de-humidification may take take place within rather short sections of the packing that require careful attention for data interpretation. Under the conditions studied here, the liquid flow rate appears to h no effect on the performance of the structured packed bed as direct-contact condenser, except at the top of the column. In intermediate sections, where latent heat transfer is not excessive, the integral heat transfer coefficient displays a dependence on Re G to a power near unity. In these sections, common convective mass/heat transfer correlations lead to acceptable (order of magnitude) predictions. A rather curious apparent effect (also observed in previous studies) of decreasing U with increasing liquid rate, at the top section, requires additional work to be confirmed and clarified. 5
1, ExHFT-5, Thessaloniki, Greece SYMBOLS A: Cross-sectional area of the column [] m 2 c : Specific heat of water [] J/Kg o C c : Specific heat of [] J/Kg o C D: Diffusivity [] m 2 /s d h : Hydraulic diameter of the packing (4/α) [] m G: Air flow rate [] Kg/s H in : Kg of water/kg of dry at the inlet H out : Kg of water/kg of dry at the outlet H H T,, T,, : Air uration humidity at T : Air uration humidity at T h: Heat transfer coefficient [] W/m 2 K: Mass transfer coefficient [] mol/m 2 s K g : Mass transfer coefficient [] mol/m 2 sbar L: Water flow rate, [] Kg/s Q: Rate of heat transferred, [] J/s Q sens : Rate of sensible heat transferred [] J/s Q lat : Rate of latent heat transferred [] J/s G Re : : Re Re liq : Liquid Reynolds number : µ u Re liq a T out, : Outlet temperature of water [] o C T in, : Inlet Temperature of water [] o C T out, : Outlet temperature of [] o C T in, : Inlet Temperature of [] o C Tin, + Tout, T 2 Tin, + Tout, T 2 u liq,sup : Superficial liquid velocity [] m/s U: Heat transfer coefficient [] W/m 2 K z: Height of packed section [] m / A cos45 o liq,supρliq pack µ liq d h REFERENCES 1. F J.R., Direct Contact Gas-Liquid Heat Exchange for Energy Recovery, Trans. of ASME, Journal of Solar Energy Engineering, vol. 112, (199), pp. 216 222 2. F J.R., Designing Direct Contact Coolers / Condensers, Chem. Eng., Vol. 12, (1972), pp. 91, June 3. F J.R., Process Heat Transfer by Direct Fluid-Phase Contact, AIChE Symp. Ser. No. 118, vol. 68, (1971) 4. Spigel L., Bomio P., Hunkeler R., Direct heat and mass transfer in structured packings, Chem. Eng. & Proc., 35, (1996), pp. 479 485 5. Huang Chen-Chia & F J.R., Direct Contact Gas- Liquid Heat Transfer in Packed Column, Heat Transfer Engineering, vol., no 2, (1989), pp. 19 28 6. Bontozoglou V. & Karabelas A.J., Direct Contact Steam Condenion with Simultaneous Noncondensable Gas Absorption, AIChE J., vol. 41, No. 2 (1995), pp. 241 25 7. Henriques de Brito M., von Stockar, A., Menendez Bangerter, A., Bomio, P. and Laso, M., Effective Mass-Transfer Area in a Pilot Plant Column Equipped with Structured Packings with Ceramic Rings, Ind. Eng. Chem. Res., vol 33 (1994), pp. 647-656 8. Karapantsios T.D., Kostoglou M. & Karabelas A.J., Direct Contact Condenion of Dilute Steam/Air Mixtures on Wavy Falling Films, Chem. Eng. Comm., vols. 141 142, (1996), pp. 261 285 9. F J.R, & Bravo J.L., Distillation columns containing structured packing, Chem Eng. Progr., vol 86(1), (199), pp. 19-29 1. Schpigel L., Meier W., Performance characteristics of various types of Mellapak packings (productivity, pressure differential, and deficiency), Chemical and Petroleum Engineering, vol 3, No. 3-4, (1994), pp.118 125 Greek Letters α: Specific area of the packing (25) [] m -1 λ: Latent heat of water evaporation [] J/Kg µ liq/ : Liquid/Gas viscosity respectively [] Kg/ms ρ liq/ : Liquid/Gas density respectively [] Kg/m 3 Experimental Heat Transfer, Fluid Mechanics, and Thermodynamics 1, Proceedings p.p. 1695-17. G.P. Celata, P.Di Marco, A. Goulas and A. Mariani 1 Edizioni ETS, Pisa. All rights reserved 6