for example, are adsorbed while large molecules pass through the wheel. .... 20
m2 of vacuum tube collectors and two electrical heaters (18 kW and 24 kW) were
used ... drop of 150 Pa is installed at the outlet of the process air channel to drive
..... The process shown in the Mollier diagram (Figure 14) further shows that the.
Published in Applied Thermal Engineering (2012), doi:10.1016/ j.applthermaleng.2012.03.005
Experimental investigations on desiccant wheels Ursula Eicker 1 , Uwe Schürger 1 , Max Köhler1 , Tianshu Ge2 , Yanjun Dai 2 , Hui Li 2 , Ruzhu Wang 2 1 2
Research Centre zafh.net, University of Applied Sciences, Stuttgart D-70714, Germany
Institute of Refrigeration and Cryogenics, Shanghai Jiao Tong University, Shanghai, 200240, PR China
ABSTRACT: Experimental investigations on several co mmercially available and newly fabricated rotors are conducted in two different laboratories to evaluate performance t rends . Experimental uncertainties are analysed and the parameters determin ing the rotor performance are investigated. It is found that the optimal rotation speed is lower for lithiu m chlo ride or co mpound rotors than for silica gel rotors. Higher regeneration air temperatures lead to higher dehumidification potentials at almost equal dehumidification efficiencies, but with increasing regeneration specific heat input and enthalpy changes of the process air. The influence of the regeneration air hu mid ity was also notable and low relat ive humidit ies increase the dehumidification potential. Finally, the measurements show that rising water content in the amb ient air causes the dehumid ification capacity to rise, wh ile the dehumidification efficiency is not much affected and both specific regeneration heat input and latent heat change of the process air decrease. For desiccant cooling applications in humid climates this is a positive trend.
Keywords: Desiccative cooling, desiccant wheels, adsorption, dehumidification
1
Introduction to desiccant wheel characterisation Desiccative and Evaporative Cooling (DEC) systems are based on the principle of ad iabatic
evaporative cooling. The extent to wh ich the supply air can be cooled through humid ification is usually limited by the maximally allo wed water content of the supply air. Especially in hot and humid climates, the ambient air contains so much water that very high dehumidification rates are required. For a continuous dehumidificat ion of the process air the water adsorbed on the desiccant material has to be removed, which is done by allowing hot air to flow throug h the desiccant material (regeneration). A range of materials are used for today’s desiccant rotor constructions. The synthetically-produced silica gel is a fine pored solid silicic acid which consists of 99 % silicon dio xide. It can adsorb up to 40 % of its dry weight in water when in equilibriu m with air at saturation. Silica gel can wit hstand
temperatures up to 400 °C and is a solid, insoluble desiccant. No special precautions are required when it is exposed to air at 100 % relative hu mid ity. It is also possible to wash a wheel in water if dust or other particulate block the air passageways. Silica gel does not undergo any chemical or physical change during the adsorption process. It is inert, non-toxic, stable and resistant to most chemicals [1], [2] and [3]. Lithiu m ch loride can attract and hold over ten times its weight in water and is one of the most hygroscopic compounds that exist. Its ability to attract and hold water is due to the absorption of water through a chemical react ion. As lithiu m chloride is water soluble, precautions are required to protect the wheel fro m high relative hu midity. Lithiu m chloride prevents the growth of bacteria on the wheel surface. It can also significantly reduce the number of organis ms which may be carried in the air st ream. Test results show there is typically a 25 % to 50 % reduction in the bacteria content of the air as it passes through the wheel. Lithiu m chloride is unaffected by most air stream pollutants, and resistant to many contaminants such as petroleum vapour, solvents, etc. A mo lecular sieve or synthetic zeolite is a crystalline material of alu min iu m silicate which is capable of separating molecules of different sizes by sorption. Therefore, small mo lecules, such as water molecules for example, are adsorbed while large molecu les pass through the wheel. Molecular sieve materials are suitable for special applications that call for the dehumid ification of air to a very lo w level o f hu mid ity and extremely lo w dew points of about -40 °C to -60 °C. Fo r the same reason, the molecular sieve has a better sorption capacity at higher temperatures than other sorbents . Co mposite desiccant materials have been used to increase the dehumidification capacity [4]. Co mbin ing different host mat rices such as mesoporous or microporo us silica gels, alu mina, porous carbons or poly mers with inorganic s alts such as CaCl2, LiBr, MgCl2 or LiCl change the sorbent properties in a wide range. A compound silicagel–halo id desiccant consisting of a host matrix with open pores (silicagel) and a hygroscopic substance (lithiu m ch loride) was shown to have about 20 – 40 % higher dehumidification rates compared to silica gel [5]. Also composite materials with other inorganic salt solutions such as calciu m chloride were also shown to strongly increase t he sorption rate and diffusion constant [6]. The characteristic properties of the desiccant wheels are described mainly by the following figures of merit:
dehumidification capacity (Δx) in [g/kg] as a type of “performance” figure
Regeneration Specific Heat Input (RSHI) in [kJ/g] as an “energy efficiency” figure
dehumidification efficiency (η dehum ) as a type of "quality" figure
enthalpy change of process air (Δh) in [kJ/kg] as a type of "thermal quality" figure
The dehumid ification capacity (Δx) in [g/kg] is defined as the amount of moisture removed fro m the process airflow.
The RSHI is the thermal power supplied to the device for regeneration
Qreg in relation to the
dehumidification capacity flu x, which is the product of process mass flow
m process and
dehumidification capacity:
RSHI
The thermal power
Qreg x m process
kJ g
(1)
Qreg supplied for regeneration is obtained from the regeneration mass flow
mreg and the temperature difference between the regeneration air T reg and room exhaust air Troom . If ambient air is used for regeneration, the ambient air temperature replaces the room temperature.
Qreg c p ,air mreg Treg Troom
(2)
Alternatively the regeneration effectiveness can be used, which is given by the latent load, i.e the heat of evaporation of water (h latent) mu ltip lied with the mo isture removed by the wheel Δx, divided by the regeneration heat required per unit mass flow
reg
x hlatent Qreg / mreg
(3)
The dehumidificat ion efficiency η dehum is defined as the ratio of the reached dehumid ification capacity Δx to the theoretical possible dehumidification Δxmax.
dehum
x xmax
(4)
The calculation o f the dehumidification efficiency was performed using a simp lified model, where the maximu m dehumid ification capacity is obtained fro m the sorption isotherm of the adsorption material. The model is based on the assumption that the equilibriu m water charge is solely a function of the relative humidity and independent of temperature (X = f (φ)) and that the process is isenthalpic. With these simplificat ions it follows that the maximu m possible dehumid ification capacity is reached when the process airflo w with the relative hu midity φ0 and the absolute humid ity x0 reaches the relative hu midity of the regeneration airflow φreg , which g ives the minimu m possible charge. Fo r a given process air enthalpy the absolute humidity can then be calculated.
An advantage of this calculat ion model is that the functional form of the sorption isotherm must not be known because the "end" point of the drying process is determined only by the enthalpy of the process air and the relative humidity of the regeneration air. When comparing different desiccant wheels, the dehumidificat ion efficiency alone is not always sufficient to rate the quality of the wheel. Thus, for examp le, if the dehumidification efficiencies of t wo wheels is the same but one shows a higher temperature rise in the process air (e.g. caused by heat transfer fro m the regeneration air, a h igher binding enthalpy during adsorption or an exothermal chemical reaction during absorption) the dehumidificat ion efficiency is also identical. However, the wheel with the lower temperature rise is better for desiccant cooling applications. Therefore, in addit ion to the dehumidification efficiency, the enthalpy change (Δh) of the process air during dehumidification must also be considered in the quality rating. The ratio of enthalpy change to the process inlet air enthalpy has been used to define an adiabatic desiccant wheel effectiveness, which is 100 % for a completely adiabatic operation [7].
1
h hinlet
(5)
There are few co mparative experimental investigations of sorption rotors available. Often performance is only given at design rotational speed and wheel performance at different boundary conditio ns is derived fro m fundamental heat and mass transfer equations and then fitted with emp irical equations [8]. A small desiccant rotor with 32 cm d iameter was investigated by [7], wh ich had very low dehumidification efficiencies of 24 % maximu m, decreasing with increasing inlet air temperature. A 70 cm diameter silica gel wheel was recently tested by [9]. The dehumid ification efficiencies increased from 20 % at 45 °C regeneration temperature to 40 % at 70 °C regeneration temperature at amb ient air conditions of 32.7 °C and 13 g kg -1 humidity. At the same t ime the adiabatic efficiency slightly decreased with rising regeneration temperature similar to a stronger decrease of regeneration efficiency. Also small rotors with four part itions for double passes of air have been analysed. A 25 cm d iameter silicagel rotor tested by [10] gave dehumidification rates of 2 to 6 g kg -1 for regeneration temperatures between 35 and 65 °C. The optimu m rotation speed for this rotor was very low at less than five rotations per hour. In this work, several co mmercially available rotors have been extensively tested and compared to fabricated rotors. Experimental uncertainties were analysed and the main parameters determin ing the rotor performance were investigated.
2
Experimental set-up The two test plants at the University of Applied Sciences in Stuttgart and Shanghai Jiao Tong
University (SJTU) were designed to measure characteristic figures of merit of different desiccant wheels under changing operating conditions, so that comparative advantages and disadvantages of different wheels could be determined. In the Stuttgart test plant, all operation parameters such as temperature, humidity, flow rates and the rotation velocity of the wheel could be set to different values as required . The limits of the air flo ws were 0 to 2500 m3 h -1 for both the process and regeneration air flows. The limits for possible temperature and relative humidity values for the ambient air were 10 °C, 50 % to 40 °C, 70 % and for regeneration air 20 °C, 50 % up to 100 °C, 3 %. The temperature, relative hu midity and flow rates were measured near the sorption wheel on both the inlet and outlet of the process and regeneration air flows. The temperature and humid ity measurements were taken with Micatrone sensors (type: Micaflex MFHTT), wh ich use resistance thermo meters for temperature measurements (Pt1000) at an accuracy of ±0.5 °C and capacitive humidity sensors with ±2 % error in relative hu mid ity. The air flows were measured using flow meter Micatrone sensors (type: Micaflex M F-FD), wh ich determine the flow velocity v or volu me flow by measuring the pressure difference (v ~ p ). The measuring range of the sensors was fro m 0 Pa to 200 Pa, which corresponds to a volume flow up to about 3000 m3 h -1 or a flo w velocity of up to about 12 m s -1 at the test plant. The accuracy of the sensors is given as ±3 % of the measured pressure difference. 20 m2 of vacuum tube collectors and two electrical heaters (18 kW and 24 kW) were used for heating the regeneration air and a 16 kW electrical heater was used for heating the process air. The air was drawn using a 1.5 kW electric fan on both the process and regeneration sides. The relative humid ity in both air flows was set using steam humid ifiers. Long straight tubes were installed in as many p laces as possible at the test plant in order to get measuring points with a uniform airflo w. Essentially the goal was to design a test plant that meets the requirements of the test facilities described in the “Desiccant Dehu mid ification Wheel Test Guide” [11]. In o rder to fully meet these requirements, much more capital expenditure and space would have been necessary than was at disposal. The design of the test plant does not meet all the requirements specified by the Wheel Test Guide, but is much more co mp liant with the typical built-in situation of desiccant wheels in complete DEC-systems. Thus the outcome of the experiments does not show the maximu m possible performance of the wheel, but most likely provides better insight into the performance that can be expected of DEC-systems.
The SJTU test plant consists of two air ducts, one for process air and another for regeneration air. At the process air side, an air pre-conditioner equip ment with an electrical heater of 8 kW as well as an electrical hu mid ifier of 7.5 kW are installed to condition the process air to the required state. Afterwards the process air passes through the desiccant wheel and is dried. An axial-flow fan with a maximu m air flu x of 1000 m3 h -1 at a pressure drop of 150 Pa is installed at the outlet of the process air channel to drive the air. Additionally a wind damper is installed at the outlet to adjust the flow rate between 600 and 1000 m3 h -1 . For the regeneration air an electrical heater of 7.8 kW is utilized in the in let part to heat up the regeneration air to a set regeneration temperature (60 to 120 °C). Likewise an air damper and an axialflow fan (400 m3 h -1 , 150 Pa) are installed at the regeneration air outlet. The setup of the test plants is shown in Figure 1.
2.1
Investigated Wheels
At the University of Applied Sciences in Stuttgart three different desiccant wheels were available for comprehensive experiments. For this study primarily a desiccant wheel fro m th e Engelhard Hexcore LP company (Type: DES-H-Hexcore-DC15/N), and a wheel fro m the Klingenburg company (Type: SECO 1000) were tested because the third wheel fro m the Munters company had already been tested as part of a study by [12]. Additionally two desiccant wheels fabricated in Shanghai Jiao Tong Un iversity coated with silica gel and co mposite silica gel & lithiu m chloride were tested under different operation conditions (see Table 1 and Table 2).
2.2
Influence of measurement accuracy
For the calcu lation of the dehumid ification capacity, the absolute humidity of the process air in front of and behind the desiccant wheel has to be determined. As the sensors only measure relative hu mid ity and temperature, the absolute humidity has to be calculated using these two values. The maximu m error that can be caused by the inaccuracies of the meas urement devices can be obtained with an error propagation calculation shown in equation 6, where the derivatives of the absolute humidity are calculated.
x
dx( , psat ) dx( , psat ) psat d dpsat
(6)
This error propagation calculation shows that the deviation of the absolute humidity rises with rising temperatures and that the inaccuracies even can be more than six times higher at 70 °C than at 32 °C. Apart fro m the temperature and humidity measurements, the volume flow measurements can also cause errors resulting fro m sensor inaccuracy. With the above-mentioned flo w meter sensors, the volume or mass flow can be indirectly determined by the measurement of the pressure difference. The accuracy of
the sensors is given at ±3 % of the pressure difference. Due to the fact that the emp irical function used to calculate the volume flo w is proportional to the root of the pressure difference ( V ~
p ), the
accuracy of flow measurements is about ±1.7 %. The maximu m error o f the RSHI caused by the inaccuracy of the flow meter sensors is obtained again with an error propagation calculat ion. The assumption that only the inaccuracies of the flow measurements cause the error of the RSHI (which is obviously incorrect but nevertheless shows the influence of the volume flow measurements), leads to a proportional relationship between the RSHI and the ratio of the mass flows (see equations 7 and 8):
RSHI
Qreg x m process
c p ,air Treg Troom
RSHI
x
m reg m process
mreg m process
(7)
(8)
Thus a maximu m error limit of 3.4 % is obtained for the RSHI as a result of the volu me flo w measurements. Co mpared with the erro rs that occur with the humidity measurements this error can most likely be ignored. In summary the measuring accuracy of the humidity sensors has the main influence on the possible error in determin ing the figures of merit, especially at high temperatures. Fortunately, in our case, the usual temperatures of the process air flow lie within a very accurate measuring range of the sensors. The biggest problem with measurement accuracy (the determination of the relat ive humidity of the regeneration air) could be avoided by measuring the humidity of the “cold” regeneration air and using it for the absolute humidity calculation.
3
Results
3.1
Inertia of the test plant
The first item that was examined was the time required to regain steady state conditions after a change had been made to the operation conditions in order to evaluate the inertia of the test plant. In Figure 2 the time dependence of the measured values is shown for a change in regeneration air temperature. The determined time period needed to regain steady state conditions in this examp le was about 15 minutes. Thus the measurements were started at the earliest 15 to 20 minutes after a change had been made in the operation conditions and only if the observed monitored values clearly showed steady state conditions. Normally, the measurements were started 30 to 45 minutes after the change was made in the operating conditions.
3.2
Influence of rotation velocity
Co mpared to rotating heat exchangers, desiccant wheels operate at much lower rotation velocities. Rotating heat exchangers normally operate at 600 rotations per hour (RPH). Desiccant wheels in dehumidification operation run at between 15 and 100 RPH. Of course, at higher rotation velocities the desiccant wheels can also be us ed for heat recovery operation, but due to the lower specific heat capacity of the matrix material they are not as efficient as metallic heat exchangers. The comparison between a silica gel wheel fro m Engelhard HexCo re LP and a lithiu m ch loride wheel fro m Klingenburg was performed in Stuttgart (St) once under almost identical conditions and once under design conditions with the recommended operation parameters as specified by the manufacturers. The first comparison was carried out under the following conditions for both wheels: 32 °C / 40 % for the ambient air, 60 °C / 3.3 % [K-St 1] and 5.4 % [H-St4] for the regeneration air, a p rocess air volu me flow of 2000 m3 h -1 and a volume flow ratio between the process air and the regeneration air of 75%. These conditions are the design conditions recommended by the manufacturer Klingenburg for the lithiu m chloride wheel. For the HexCore wheel, a h igher regeneration air temperature of 75°C is recommended as design condition [H-St5]. For almost identical process conditions [K-St1] and [H-St4] the influence of the rotation velocity is different for each wheel. Fo r the Hexco re wheel the dehumid ification capacity increases as rotation velocities rise up to 85 to 100 RPH. The Klingenburg wheel however shows a decrease in t he dehumidification capacity as rotation velocities rise above about 25 RPH. Th is different behaviour is most likely caused by the different sorption mechanis ms of the used sorption materials. For the Hexcore wheel, the sorption material used is silica gel and the sorption mechanism is the adsorption process, for the Klingenburg wheel, the sorption material is lithiu m chlo ride and the sorption process is an absorption process. The results lead one to assume that the absorption process is slower than the ads orption process.
This can also be the reason that for the Klingenburg wheel at higher rotation velocities sometimes the dehumidification process becomes a humid ification process (negative dehumidificat ion capacity). In addition, the kind of matrix material, which is also different fo r the two wheels studied, can certainly influence the sorption dynamics. Figure 3 shows the comparison of the dehumid ification capacities. Whereas the Klingenburg wheel reaches its maximu m dehumidification capacity between 20 and 30 RPH, the Hexcore wheel does the same at about 100 RPH. The absolute maximu m values however do not deviate much for similar boundary conditions. For a higher regeneration temperature of 75°C, the dehumidificat ion capacity of the Hexcore wheel increases by about 1 g/kg [H-St5]. Considering that the water content of the regeneration air was slightly lower with the Klingenburg measurements than it was with the Hexcore measurements, the difference under absolute ly identical conditions would probably be even smaller. Also the maximu m values of the dehumid ification efficiency were reached at different rotation velocities (see Figure 4). The maximas of the Hexcore wheel are h igher at nearly 60% than the Klingenburg wheel with less than 50% efficiency. With increasing regeneration temperature the dehumidification efficiency of the Hexcore wheel slightly increases (from 0.54 to 0.58). Due to the almost equal dehu mid ification capacit ies and the almost equal regeneration air conditions in the measurement series [K-St1] and [H-St4], the minimu m RSHI values (Figure 5) must also be equal to one another, naturally at different rotation velocities . For the higher regeneration air temperature [H-St5], the RSHI of the Hexcore wheel is also higher, i.e. worse. The enthalpy change curves (Figure 6) show again the expected behaviour, which is a rising enthalpy change with rising rotation velocit ies. Considering only the grad ients of the curves, the enthalpy rise of the Hexco re wheel is slightly lower than that of the Klingenburg wheel. When examining the rotation velocity with the best dehumidificat ion capacity h owever, the enthalpy rise is more advantageous (smaller) for the Klingenburg wheel than it is for the Hexcore wheel. The kink on the curve [H-St5] is caused by an inadequacy in the volu me flo w control. The curves for this higher regeneration temperature show once more a slower rise of the Hexcore curve as the rotation velocity increases.
Experiments at Shanghai Jiao Tong University showed that compound silicagel desiccant wheels with the addit ion of lithiu m chloride improve the dehumidificat ion capacity b y about 3 g/kg for similar boundary conditions while doubling the dehumidificat ion efficiency and reducing the RSHI and enthalpy change [5]. For both silica gel and compound desiccant wheels with two thicknesses of 16.6 and 100 mm, the dehumidificat ion capacity increases with increasing rotation speed (see Figure 7). These experiments were conducted under the following conditions: for the thin wheels with the thickness of 16.6mm: 30 °C / 60 % r.H. for the ambient air, 80 °C / 7 % r.H. for the regeneration air, for the wheels with a thickness of 100mm: 28 °C / 40 % for the amb ient air, 80 °C / 3.5 % for the
regeneration air, a process air volu me flow of 36 0 m3 h -1 and a volume flow ratio between the regeneration and process air of 33%. Experimental results of dehumid ification efficiency η are presented in Figure 8. The results show that the trends of the dehumidification efficiency η under all the conditions are similar to the results of the dehumidification capacity, i.e. the moisture removal. Both dehumidification capacity and efficiency improve with higher wheel thickness. The RSHI change of the wheels fabricated at SJTU are shown in Figure 9. The RSHI decreases with increasing rotation speed for all DWs . For thicker wheels the enthalpy change is higher due to the longer path of process air. In terms of the best rotation velocity, the results shown above indicate that there are different values for different materials. For the Hexcore silicagel wheel, the best performance was obtained at about 85 to 100 RPH for all experiments, whereas fo r the Lithiu m ch loride Klingenburg wheel, the best performance was 20 to 30 RPH. For the desiccant wheels fabricated by the SJTU, the trends are similar. While the lithiu m chloride co mpound wheels have a low optimal rotation speed at about 20 to 24 RPH, the silica gel wheel dehumidification capacities still increase with rotation velocity in the range measured. This difference is caused by the different sorption and matrix materials with h igher dynamics for the adsorption processes. It is notable, that the optimu m rotation velocity almost does not vary with changes of the other operation parameters such as the volu me flo w ratio or the regeneration / amb ient air conditions. Thus the best rotation velocity of the wheels can regarded as independent of the other parameters. The rotation velocity could be used for the control of a DEC-system, e.g. if the maximu m dehumidification capacity is not required. However, in most cases it would be much more energy efficient to change other operation parameters such as the volume flow or the regeneration air temperature.
3.3
Influence of ambient air water content
A major factor which determines the dehumid ification process is the ambient air absolute humidity. Figure 11 shows the influence of the water content of the ambient air on the dehumid ification capacity as a function of rotation velocity fo r the silica gel wheel tested in Stuttgart (Figure 11 (a)) and the thin compound desiccant wheel tested in Shanghai (Figure 11 (b )). The regeneration temperatures were comparable (in Stuttgart 75°C for [H-St3], [H-St5] and [H-St 8] and in Shanghai 80°C for [Sh1], [Sh 2] and [Sh3]). The process air absolute water content varied from 9.4g/kg up to 22.8 g/kg. When the water content of the ambient air doubles, the dehumidification capacity also nearly doubles for the silica gel wheel and more than doubles for the compound wheel. The dehumidificat ion efficiency for the silica gel wheel is not much affected by the absolute humid ity, but increases for the co mpound wheel (see Figure 12 (a) and (b)). The regeneration specific heat input RSHI (see Figure 13 (a) and (b)) drops in both measurement series with rising water content of the ambient air. A lthough the boundary
conditions are not absolutely identical in the two laboratories, the trends in both sets of measurements are very comparable, which increases the confidence of measurement accuracy.
3.4
Influence of volume flow ratios
For the Hexcore silica gel rotor the dehumidification capacity increases from 3.8 g kg -1 to 5.1 g kg -1 when the volume flow rat io of regeneration to process air is increased fro m 0.5 [H-St6] to 0.75 [H-St5] (at 85 RPH and a fixed regeneration temperature of 75 °C). At the same time the dehumidificat ion efficiency increases fro m 43 to 58 %. Despite the greater energy demand of higher regeneration airflo w, the RSHI increases only slightly fro m 5.9 to 6.3 kJ g -1 . Due to the heat inhibition effect which heats up the desiccant wheel and decreases the dehumid ification capacity, a further increase of volu me flow ratio to 1.0 leads to higher outlet temperatures [H-St1]. Consequently the RSHI is substantially higher at 9.7 kJ g -1 . The process shown in the Mollier d iagram (Figure 14) further shows that the dehumidification process has almost the s ame enthalpy gradient up to a volume flow ratio of 0.75: the enthalpy change is 2.7 kJ kg -1 at a volume flow ratio of 0.5 and 3.2 kJ kg -1 at 0.75), wh ile the dehumidification capacity increases. With a volu me flo w ratio over 0.75, the enthalpy then increases marked ly (up to 12.3 kJ kg -1 at a volume flow ratio of 1.0). Th is increase of enthalpy of the supply air is not advantageous for the cooling function of DEC-systems, especially if there is no further improvement in dehumidification capacity. Thus the best volume flow rat io for the Hexco re wheel is 0.75. With less demand on the dehumidification capacity, the regeneration air volu me flow can be reduced even further, which would save energy twofold: first, less fan power is needed and secondly because the energy required to heat the regeneration air is diminished. The studies performed on the Klingenburg wheel (at 24 RPH optimu m rotation velocity) show similar results. The dehumid ification capacity increases fro m 1.2 g kg -1 to 2.7 g kg -1 (at 60 °C regeneration temperature) when the volume flo w ratio rises fro m 0.50 [K-St3] to 0.75 [K-St 9] and the dehumidification efficiency increases even more than for the Hexco re wheel (fro m 22 to 50 %). Due to the clearly h igher dehumidificat ion capacity, the RSHI even decreases as a result of a rise in the volu me flow ratio (fro m 12 kJ g -1 to 8 kJ g -1 ). A fu rther increase of the volume flo w ratio to 1.0 [K-St 2] does not result in a further improvement of the dehumidification capacity, but the RSHI is higher again at 10 kJ g -1 . For the Klingenburg wheel, the influence of the volume flow ratio on the change in enthalpy is low with values between 5 and 6 kJ kg -1 . Th is can be exp lained by the low rotation velocity (no significant heat transfer because of a very low heat capacity flow by the rotating matrix) and by the low regeneration air temperatures (60 °C). Here the best volu me flo w ratio is also 0.75 with the same possibility of energy saving with less demand on the dehumidification capacity by reducing the volu me flow ratio.
The experiments with the small silicagel and compound rotors at Shanghai Jiao Tong University were all carried out with a low volu me flo w rat io of 0.33. Only for lo w absolute process air hu mid ities there was an enthalpy increase during dehumid ification, at higher p rocess air hu mid ities the enthalpy change was very small (see Figure 15).
3.5
Influence of regeneration temperature
For the co mmercial silica gel Hexco re wheel, the measurements show that with regeneration air temperatures rising fro m 60 to 90 °C, the dehumidification capacity clearly increases fro m 4 to 5.7 g kg -1 while at the same t ime the dehumidification efficiency stays almost constant between 54 and 58 %. As expected, the RSHI also increases slightly from 5.6 to 7.9 kJ g -1 as the regeneration air temperature rises . The absolute rise in dehu mid ification capacity of 1.1 g kg -1 is higher for the increase fro m 60 °C to 75 °C than from 75 °C to 90 °C (0.6 g kg -1 ). Figure 16 shows the dehumid ification process at different regeneration air temperatures in a Mollier diagram. The increase of the regeneration air temperature fro m 60 °C to 75 °C causes a higher dehumidification without significantly increasing the enthalpy (between 1.4 and 2.7 kJ kg -1 ). An increase fro m 75 °C to 90 °C however, does not have much influence on the dehumid ification capacity but it increases the enthalpy of the supply air by 10.9 kJ kg -1 . This rise of the enthalpy is not desirable for cooling operation. Thus the best regeneration air temperature is about 75 °C. If the dehumidification process has the main priority, it is possible to increase the dehumidificat ion capacity by increasing the regeneration air temperature, but then a higher enthalpy of the supply air must be accepted. In order to reach very lo w supply air dew point temperatures, the regeneration air temperature can be increased up to 120 °C (manufacturer data). The outcomes for the Lithiu m chloride Klingenburg wheel show, that the dehumid ification capacity rises fro m 2 g kg -1 to 2.7 g kg -1 when the regeneration air temperature is increased fro m 45 °C to 60 °C (at 24 RPH), but that further increasing this temperature fro m 60 °C to 70 °C does not increase the dehumidification capacity. However, it must be considered, that the experiments were done with a constant relat ive hu mid ity of 10 % of the regeneration air, and as a result the water content was higher at a regeneration air temperature of 70 °C (19.3 g kg -1 ) than it was at 60 °C (13 g kg -1 ). A range of further measurements showed that at constant regeneration temperature of 60 °C, the dehumid ification capacity increased from 2.3 g kg -1 to 4.1 g kg -1 when the relative humidity of the regeneration air decreased from 20 % (24 g kg -1 ) to 3 % (4.3 g kg -1 ). With constant water contents in the regeneration air, a h igher dehumidification capacity and also higher dehumid ification efficiency could be expected, as was observed by the experiments with the Hexcore wheel. The increase in the regeneration air temperature fro m 45 °C to 60 °C causes a higher dehumidificat ion capacity without significantly increasing the enthalpy (from 4.1 to 4.9 kJ kg -1 ). A further rise fro m 60 °C to 70 °C does not lead to an imp rovement in dehu mid ification but also does not increase significantly the
enthalpy of the process air (5.9 kJ kg -1 ) as was observed with the Hexcore wheel. The best regeneration air temperature fo r the Klingenburg wheel seems to be 60 °C. However, with a constant absolute water content in the regeneration air (and not a constant relative hu mid ity) a further improvement of the dehumidification capacity could be expected up to 70 °C. Higher temperatures may not be used as this could damage the wheel. Therefore, the best regeneration temperature will lie between 60 °C and 70 °C for the Klingenburg lithium chloride wheel. The rotors tested at Shanghai Jiao Tong University were always regenerated at 80°C. Despite the high temperature levels the enthalpy increase of the process air was always below 6 kJ per kg and the RSHI below 5 kJ/g at optimum rotation speed.
Conclusions The evaluation of a large series of measurements in two different laboratories in Stuttgart and Shanghai leads to new detailed knowledge of the influence of different operating parameters on the p erfo rmance of sorption wheels. The main results can be summarized as follows:
The best rotation velocity was determined to be wheel specific. A co mmercial silica gel rotor (Hexcore ) reaches its best performance at 85 to 100 RPH, a co mmercial lithiu m ch loride rotor (Klingenburg) at lower rotation velocit ies of 20 to 30 RPH. Similar trends were observed with silica gel and co mpound rotors fabricated at Shanghai Jiao Tong University. Th is difference in behaviour can be mainly exp lained with the faster dynamics of a sorption process by adsorption compared to absorption.
The addition of lithiu m chloride to silica gel rotors significantly imp roves the dehumid ification capacity by about 3 g/kg and increases the dehumidification efficiency to nearly 90%.
Both for co mmercial and fabricated desiccant wheels, the higher the water content of the ambient air the higher the dehumidification capacity. Both the regeneration specific heat input and the enthalpy increase of the process air reduce with high process air water content, which is favourable for a desiccant cooling process.
The evaluation of the measurements with different volume flo w rat ios shows that a ratio of 75% between the regeneration air and the process air leads to the best performance for both commercial wheels. Higher flow ratios lead to increased enthalpy changes without improving the dehumidification capacity.
A higher regeneration air temperature leads principally to higher dehumidificat ion potentials at almost equal dehumid ification efficiencies. However, the energy efficiency (RSHI) slightly dimin ishes when the regeneration temperature rises and the enthalpy change of the process air increases.
The influence of the regeneration air humid ity is notable. The reduction of the relative humid ity within one series of experiments from 20 % to 6 % caused a rise of almost 100 % in the dehumidification capacity. Due to this, the regeneration air should be also as dry as possible in order to achieve good performance.
References
[1]
Gutermuth, W.: Untersuchung der gekoppelten Wärme- und Stoffübertragung in Sorptionsregeneratoren (Investigation of coupled heat and mass transfer in sorption regenerators) Dissertation, Technische Universität Darmstadt, 1980.
[2]
Kast, W.: Adsorption aus der Gasphase (Adsorption from gas phase). Weinheim: VCH Verlagsgesellschaft mbH, 1988.
[3]
Munters Cargocaire Engineering Corp.: Dehumidification for all requirements. (Internet publication: www.muntersamerica.co m/dh/htm/desiccant.htm). Munters Cargocaire Engineering Corp. USA, 2001.
[4]
Majumdar, P.: Heat and mass transfer in composite desiccant pore structures for dehumidification. Solar Energy 62(1), 1–10, 1998.
[5]
Jia, C.X.; Dai, Y.J.; Wu, J.Y.; Wang, R.Z.: Use of compound desiccant to develop high performance desiccant cooling system. International Journal of Refrigeration, 30, 345 – 353, 2007.
[6]
Aristov, Yu.I.; Glazneva, I.S.; Frenib, A.; Restuccia, G.: Kinetics of water sorption on SWS-1L (calcium chloride confined to mesoporous silica gel): Influence of grain size and temperature. Chemical Engineering Science 61, 1453 – 1458, 2006.
[7]
Mandegari, M. A.; Pahlavanzadeh, H.: Introduction of a new definition for effectiveness of desiccant wheels. Energy, 34, 797–803, 2009.
[8]
Jeong, J.-W.; Mumma, S.A.: Practical thermal performance correlations for molecular sieve and silica gel loaded enthalpy wheels. Applied Thermal Engineering, 25, 719 – 740, 2005.
[9]
Angrisani, G. et al.: Desiccant wheel regenerated by thermal energy from a microcogenerator: Experimental assessment of the performances. Applied Energy 88, 1354 – 1365, 2011.
[10]
Jeong, J.; Yamaguchi, S.; Saito, K.; Kawai, S.: Performance analysis of four-partition desiccant wheel and hybrid dehumidification airconditioning system. International Journal of refrigeration, 33, 496 – 509, 2010.
[11]
Slayzak, S. J.; Ryan, J. P.: Desiccant Dehumidification Wheel Test Guide. National Renewable Energy Laboratory (NREL), prepared under Task No. BET1.3001, December 2000.
[12]
Hoefker, G.: Desiccant cooling with solar energy (PhD thesis). Leicester: Institute of Energy and Sustainable Development, De Montfort University , 2011.
List of tables
Table 1: Desiccant wheels tested at the University of Applied Sciences, Stuttgart Table 2: Summary of test conditions
Table 1: Desiccant wheels tested at the University of Applied Sciences Stuttgart (St) and Shanghai Jiao Tong University Shanghai (Sh)
Manufacturer
Engelhard
Klingenburg
HexCore,LP
SECO 1000
Silica gel fabricated
Silica gel composite fabricated
DES-HHexcore-DC15/N H-St
K-St
S-Sh
SC-Sh
Cellulose
Ceramic based
Ceramic based
lithium chloride
Silica gel
Silica gel & lithium chloride
honeycomb
sinusoidal structure
honeycomb
honeycomb
Wheel depth
140 mm
250 mm
16.6mm/100mm
16.6mm/100mm
Wheel diameter
870 mm
895 mm
260 mm
260 mm
8-24 r h -1
8-24 r h -1
2.0-2.5 m/s
2.0-2.5 m/s
60 °C to 120 °C
60 °C to 90 °C
Abbreviation
Wheel specifications Matrix material
Hexcore NomexTM Hexcore ETS TM
Desiccant type
(titanium silicate)
Matrix structure
Revolutions hour face velocity Reg. temp.
per
15-100 r h -1 (winter) 2 to 2.6 m/s 55 °C to 85 °C (max. 120 °C)
20 r h -1 (summer) / up to 600 r h -1 (winter) 1.5 to 2 m/s 40 °C up to 70 °C
Table 2: Summary of test conditions
Number
Process air flow
Volume flow ratio
r.H (%) 40
Regeneration air flow Treg r.H (°C) (%) 75 3
H-St1
Tproc (°C) 32
H-St2
28
40
60
4
75
H-St3
36
60
75
4
75
H-St4
32
40
60
5
75
H-St5
32
40
75
3
75
H-St6
32
40
75
3
50
H-St7
36
50
60
7
75
H-St8
28
40
75
2.5
75
H-St9
32
40
90
1.5
75
K-St1
32
40
60
3
75
K-St2
32
40
60
11
100
K-St3
32
40
60
11
50
K-St4
28
40
60
10.5
75
K-St5
36
50
60
10
75
K-St6
32
40
70
10
75
K-St7
32
40
50
10.5
75
K-St8
32
40
45
12
75
K-St9
32
40
60
10.5
75
Sh1
30
40
80
7
33
Sh2
30
60
80
7
33
Sh3
30
80
80
7
33
Sh4
28
40
80
3.5
33
Vreg / Vproc (%) 100
List of Figures
Figure 1: Setup of test plant at the University of Applied Sciences, Stuttgart Figure 2: Time before steady state condition after a change of regeneration air temperature Figure 3: Co mparison of the dehumidification capacity between Hexcore [H-St4 and H-St 5] and Klingenburg [K-St1] wheel Figure 4: Co mparison of the dehumidificat ion efficiency between Hexcore [H-St4 and H-St5] and Klingenburg [K-St11] wheel Figure 5: Comparison of the RSHI between Hexcore [H-St4 and H-St5] and Klingenburg [K-St1] wheel Figure 6: Comparison of enthalpy change between Hexcore [H4] and Klingenburg [K1] wheel Figure 7: Comparison of the dehumidification capacity between silica gel and compound wheel. The test conditions for the 16.6 mm thin rotors were 30°C / 60% r.H. for the ambient air and 80°C / 7% for the regeneration air (Sh2) and for the thicker100 mm rotors 28 °C / 40 % (ambient) and 80 °C / 3.5 % (regeneration) (Sh4). Figure 8: Comparison of the dehumidification efficiency between silica gel and compound wheel Figure 9: Comparison of the RSHI between silica gel and compound wheel Figure 10: Comparison of the enthalpy change between silica gel and compound wheel Figure 11 (a) and (b): Dehu mid ification capacity as a function of ambient air conditions (a: Hexcore -DC 15/N, b: Compound wheel 16.6 mm) Figure 12 (a) and (b) Dehu midification efficiency for different amb ient air absolute humidity (a: HexcoreDC 15/N, b: Compound wheel 16.6 mm) Figure 13 (a) and (b): RSHI for different ambient air absolute humidity (a: Hexcore -DC 15/ N, b: Compound wheel 16.6 mm) Figure 14: Dehu midification of process air dependent on volume flow ratio in a Mollier d iagram for a constant rotation velocity of 85 RPH and 75 °C regeneration temperature for the Hexcore wheel Figure 15: Dehumidification of process air dependent on ambient air conditions in a Mollier diagram at 24 RPH and 80 °C regeneration temperature for the Compound wheel 16.6mm Figure 16: Dehu mid ification of process air dependent on regeneration air temperature in a M ollier diagram at 85 RPH for the Hexcore wheel
Figure1: Setup of test plant at the University of Applied Sciences, Stuttgart (St). The set-up at Shanghai Jiao Tong University (Sh) has no humidifier in the exhaust air stream and uses valves at the sup ply air inlet and waste air outlet to control the volume flow.
Figure 2: Time before steady state condition after a change of regeneration air temperature
Figure 3: Comparison of the dehumidification capacity between Hexcore [H-St4 and H-St5] and Klingenburg [K-St1] wheel
Figure 4: Comparison of the dehumidification efficiency between Hexcore [H-St4 and H-St5] and Klingenburg [K-St1] wheel
Figure 5: Comparison of the RSHI between Hexcore [H-St4 and H-St5] and Klingenburg [K-St1] wheel
Figure 6: Comparison of enthalpy change between Hexcore [H-St4 and H-St5] and Klingenburg [K-St1] wheel
Figure 7: Comparison of the dehumidification capacity between silica gel and compound wheels. The test conditions for the 16.6 mm thin rotors were 30°C / 60% r.H. for the ambient air and 80°C / 7% for the regeneration air (Sh2) and for the thicker100 mm rotors 28 °C / 40 % (ambient) and 80 °C / 3.5 % (regeneration) (Sh4).
Figure 8: Comparison of the dehumidification efficiency between silica gel and compound wheel for Sh2 conditions for thin rotors and Sh4 for thick rotors.
Figure 9: Comparison of the RSHI between silica gel and compound wheel for Sh2 conditions for thin rotors and Sh4 for thick rotors.
Figure 10: Comparison of the enthalpy change between silica gel and compound wheel for Sh2 conditions for thin rotors and Sh4 for thick rotors.
Figure 11 (a): Dehumidification capacity as a function of ambient air conditions for the Hexcore-DC 15/N wheel
Figure 11 (b): Dehumidification capacity as a function of ambient air conditions for the 16.6 mm compound wheel
Figure 12 (a): Dehumidification efficiency for different ambient air absolute humidity for the HexcoreDC 15/N wheel
Figure 12 (b): Dehumidification efficiency for different ambient air absolute humidity for the 16.6 mm compound wheel)
Figure 13 (a): RSHI for different ambient air absolute humidity for the Hexcore-DC 15/N wheel
Figure 13 (b): RSHI for different ambient air absolute humidity for the 16.6 mm compound wheel
10%
0.030 0.025 5%
0.020 0.015 0.010
1.0 [H1] 2%
0.5 [H6]
0.005
0.75 [H5] 0
10
20
Water content [g kg-1]
20 %
40 %
80%
60%
100%
0.035
1%
30 40 50 60 o Temperature [ C]
70
0.000 90
80
Figure 14: Dehumidification of process air as a a function of volume flow ratio in a Mollier diagram for a constant rotation velocity of 85 RPH and 75 °C regeneration temperature for the Hexcore wheel tested in Stuttgart.
10%
0.030
5%
0.025
21.8g/kg [Sh3]
0.020 0.015 0.010
16.2g/kg [Sh2]
2%
Water content [g kg-1]
20%
40%
60%
80%
100%
0.035
0.005
10.7g/kg[Sh1] 1%
0
10
20
30 40 50 60 o Temperature [ C]
70
80
0.000 90
Figure 15: Dehumidification of process air dependent on ambient air conditions in a Mollier diagram at 24 RPH and 80 °C regeneration temperature for the Compound wheel with 16.6mm thickness tested in Shanghai.
10%
0.030
5%
0.025 0.020 0.015 0.010
o
90 C [H9]
o
2%
60 C [H4]
0.005
o
75 C [H5] 0
10
20
30 40 50 60 o Temperature [ C]
Water content [g kg-1]
20%
40%
80%
60%
100%
0.035
1%
70
80
0.000 90
Figure 16: Dehumidification of process air dependent on regeneration air temperature in a Mollier diagram at 85 RPH for the Hexcore wheel tested in Stuttgart.
