On the theoretical and experimental energy efficiency analyses of a vacuum-based dehumidification membrane

Journal of Membrane Science 539 (2017) 76–87

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

On the theoretical and experimental energy efficiency analyses of a vacuumbased dehumidification membrane T.D. Bui, Y. Wong, M.R. Islam, K.J. Chua

MARK

⁎

Department of Mechanical Engineering, National University of Singapore, 9 Engineering Drive 1, Singapore 117576, Singapore

A R T I C L E I N F O

A B S T R A C T

Keywords: Vacuum membrane dehumidification Thermodynamic analysis Membrane dehumidification prototype Dehumidification energy efficiency Dehumidification performance

Vacuum-based membrane dehumidification (VMD) has attracted much research interest due to its potential in increasing energy efficiency for air conditioning systems. However, there is a lack of reports on practical VMD systems and their performance efficiency. In this work, VMD was systematically studied via both theoretical and experimental analyses. Thermodynamics analysis was carried out to evaluate the energy efficiency of a VMD with varying feed air temperature, humidity, and membrane selectivity. It is shown that coefficient of performance (COP) of 2–3 is achievable in a VMD system under the least efficient operation of isentropic compression. A compact VMD prototype with an effective membrane area of 2.45 m2 was developed based on a thin film composite membrane. Apparent membrane permeance and selectivity as high as 11,900 GPU and 1780, were respectively attained. The thermodynamic analysis is validated with experimental results. Energy efficiency of the membrane's dehumidification process increases with higher temperature and humidity. Vacuum pump's efficiency markedly affects the overall VMD system. A survey of commercial vacuum pumps was subsequently conducted and the practical limit of the current vacuum pump technology was observed. A COP value close to the thermodynamic limit is obtainable with a few selected vacuum pumps which possess pumping speeds that are higher than 2000 m3/h.

1. Introduction Heating, Ventilation and Air Conditioning (HVAC) has been widely employed to control both supply air temperature and humidity in order to provide indoor thermal comfort as well as sustain a hospitable environment for storing goods and equipment in residential, industrial and commercial buildings. In tropical countries, the total energy consumed by the HVAC system is mainly due to the direct cooling process carried out by vapor compression chillers. Outdoor air is passed over the cooling coils of the Air Handling Unit (AHU) to remove both sensible and latent heats. The cooled and dehumidified air is then reheated or mixed with the return air to raise its temperature to the human thermal comfort level. Because of the air's high humidity, chillers need to work to ensure heat exchangers operate below the air's dew point temperature in order to condense the moisture. Excessive energy consumption during these deep cooling steps makes the present coupled cooling and dehumidification process an inefficient one [1]. It is apparent that decoupling the latent duty from the chiller's duty would significantly improve chiller efficiency where chillers are applied only for the sole purpose of sensible cooling which accounts for only 10–20% of the total cooling load. Thus far, attempts have been

⁎

Corresponding author. E-mail address: [email protected] (K.J. Chua).

http://dx.doi.org/10.1016/j.memsci.2017.05.067 Received 1 March 2017; Received in revised form 16 May 2017; Accepted 25 May 2017 Available online 25 May 2017 0376-7388/ © 2017 Elsevier B.V. All rights reserved.

conducted to handle the latent load using solids and liquid desiccant dehumidifiers [2–5]. The overall efficiencies of these processes are low because these systems require excessive energy to regenerate the desiccant at high temperature. Additionally, the air can be contaminated with undesired desiccating particles that are potentially entrained in the air stream. Recently, vacuum-based membrane dehumidification (VMD) has gained significant research attention [6–14]. In this process, the air passes over a membrane surface at normal pressure. A vacuum pressure is applied on the opposite side of the membrane to create a driving force for water to permeate through the membrane. Humid air is dehumidified without any temperature change. The dried air is then cooled down to the thermal comfort level with minimal energy consumption via a conventional vapor compression chiller. This isothermal dehumidification process is considered to be “green” since no heat source is needed for thermal regeneration, resulting in minimal environmental emission [8]. Over the last two decades, there have been many attempts to develop highly permselective membranes for air or gas dehumidification [9,10,14,15]. Many types of polymer [10–14,16–26], inorganic [8,10,27], liquid [6,28–30] and mixed matrix [31,32] membranes have

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

2. Thermodynamic analysis

been explored. Among them, dense polymeric membranes have received a significant level of attention due to their low cost, lightness, physical robustness and ease of fabrication and modification. Mass transport in dense polymer materials is based on the solution diffusion mechanism [9,10,12,16,33–35]. Water molecules selectively permeate through the membranes due to both its smaller kinetic diameter as well as greater condensability compared to other gases [13]. For low temperature working conditions in ventilation and air conditioning, stable performance of the polymeric membrane can be achieved [9,10,15]. Many of the existing works focus on making and evaluating water permeability and selectivity of new membranes [6–8,10–32,34,36]. Only a few of them evaluate the efficiency of VMD by means of theoretical analysis [7–9,11]. The energy efficiency of a dehumidification process is commonly characterized by its COP (dimensionless):

COP =

∆Hlt W

A cross-flow VMD and its pressure profile are shown in Fig. 1(a) and (b), respectively. Humid air is introduced to the feed side of the membrane at ambient pressure, pamb. A vacuum pressure, pvac, is applied on the permeate side of the membrane to create a driving force for water vapor to be selectively sieved out of the air stream, compressed, and discharged to the ambient by a vacuum pump. The water vapor pressure of the feed air stream ( pwf ) is gradually lowered as it passes over the membrane from its input value ( pwf / in ) to its output value ( pwf / out ). Lowering pvac leads to more water vapor being removed and hence a drier product air. Water vapor permeance and selectivity are two important mass transport properties of a membrane. For a membrane with water vapor and air permeances of Pw and Pa, respectively, the membrane's selectivity, S, is

(1)

Pw Pa

S=

where, ∆Hlt is latent heat removed and W is work input. It is worth noting that COP is the most suitable term to describe the energy efficiency of VMD because it can have a magnitude that is greater than 1 and is commonly used in the heating, ventilation and air conditioning (HVAC) community. It is apparent that without involving any phase change, VMD potentially has higher COP than condensing water vapor by conventional processes. Theoretical input work were estimated from the permeate flow using the isothermal compression work equation [7–9,11]. VMD's COP of 2–3.5 has been reported with different membranes and operation conditions [7–9,11]. This energy efficiency is significantly higher than that of dehumidification by desiccants, which have reported COPs of less than 1 [4,5,8,37,38]. Hitherto, there are several theoretical energy analysis based on the isothermal compression work equation [7–9,11,39,40], but there is a lack of reports on a practical working vacuum membrane dehumidification system comprising a high performance membrane module and an efficient vacuum pump to realize system performance that approaches the theoretical COP. Thus, the motivation arises for the need to conduct a systematic energy-efficiency study of VMD through thermodynamic and experimental analyses in order to determine the limits of the technology as far as the development of a practical system is concerned. In this work, a fundamental thermodynamic approach is carried out to study and analyse the energy consumption of VMD. Accordingly, the efficiency limit of VMD is determined. A membrane module is developed based on a highly water vapor permeable thin film composite membrane. The membrane module is connected to an appropriate vacuum pump to achieve water vapor dehumidification. The effect of concentration polarization on the performance of the system is evaluated. The experimental COP of the system is studied and practical limit of the dry vacuum pump technology is deduced.

(2)

Assuming that the contents of water and air are constant on both sides of the membrane along a small length increment of dx (Fig. 1(b)), the respective water vapor and air fluxes are [39]:

dfw = P w(pwf − pwp )dx

dfa =

Pa (paf

−

(3a)

pap )dx

(3b)

where pwf and pwp are water vapor partial pressures in feed and permeate streams, respectively, and paf and pap are air partial pressures in feed and permeate streams, respectively. Conducting a simple mass balance, the ratio of water vapor partial pressure to air partial pressure on the permeate side is equal to the ratio of water vapor flux to air flux shown as

pwp pap

=

p f − pwp dfw = S wf dfa pa − pap

(4)

From Eq. (4) and the boundary conditions for water vapor partial pressures at the entrance and exit, a 1D model was developed. Water vapor and air fluxes through the entire membrane are:

Fw =

∫ dfw dx = Pw ∫ (pwf

Fa =

∫ dfa dx = Pa ∫ (paf

− pwp ) dx

(5a)

− pap ) dx

(5b)

The dehumidification performance in terms of percentage of moisture removed can be computed as:

Dehumidification performance =

pwf / in − pwf / out pwf / in

⋅100% (6)

The effect of membrane selectivity (S) on dehumidification performance as a function of vacuum pressure is shown in Fig. 2(a). At lower

Fig. 1. Schematics of (a) cross-flow VMD; (b) the pressure profile along the membrane.

77

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

Fig. 2. (a) The effect of membrane selectivity (S) on dehumidification performance as a function of vacuum pressure, pvac, with the vertical dashed line indicating the input water vapor pressure; (b) COPisen as a function of dehumidification performance with different membrane selectivity. The inlet air's temperature and relative humidity are typical outdoor air condition in Singapore, at 31 °C and 60% (17 g/kg dry air), respectively.

isentropic conditions, far higher values than existing desiccant dehumidifiers. It is noteworthy that the theoretical limits depicted as the curves in these graphs are the maximal COPs that a VMD system is able to attain when the efficiency of the pump is assumed to be 100%. In practice, a pump's efficiency is always lower than 100% due to energy losses arising from friction or heat loss. Therefore, the practical COP is always less than the theoretical limit. Nonetheless, the COP limits shown in these graphs demonstrate the potential of a VMD system that is able to realize improved performance.

membrane selectivity, a larger fraction of air permeates through the membrane, reducing pwp . This results in a higher driving force for water vapor permeation. Therefore, a high dehumidification performance is obtained even at pvac > pwf / in and lower S results in higher dehumidification performance. Similarly, higher dehumidification performance is also achieved with the use of a sweep gas in permeate flow [7,11]. The work required in the separation process is the electricity to compress the water vapor from pvac to pamb. Theoretically, the pump works most efficiently in an isothermal process, which requires a maximal cooling; and least efficient in an isentropic process, which involves no cooling. Practically, the pump operates in a polytropic process that involves an intermediate degree of cooling due to heat loss to ambient air. In order to understand the lower COP limit of VMD, the isentropic operation (PV k w = constant and PV ka = constant, with kw and ka denoting the specific heat ratio, Cp/Cv, of water vapor and air, respectively) is considered. With 100% pumping efficiency, the maximal obtainable VMD's COP in isentropic compression can be determined by Eq. (7):

3. Fabrication of membrane module 3.1. Membrane fabrication and characterization To experimentally evaluate the performance of VMD, a lab-scale prototype was fabricated and tested. In one of our previous works, we demonstrated that adding hygroscopic substances such as triethylene glycol (TEG) in polyvinyl alcohol (PVA) markedly improves water vapor permeability of the PVA membrane [16]. A composite membrane with the top layer made of a PVA/TEG blend is environmental friendly, highly durable and suitable for dehumidification applications. In this work, the membrane sample was prepared by coating a thin layer of PVA/TEG blend on a flat-sheet thin film composite polymeric substrate. The substrate is nanofiltration membrane (model NFW) obtained from Synder Filtration [41]. This substrate possesses an asymmetric pore structure that grants high water vapor permeation rate. Its polyester backing layer enables the membrane to be mechanically stable. The substrate thickness is 150 µm. Maximal operation pressure is upto 40 bar. The top layer of the substrate is made of fine sphere-shaped polyamide grains with diameter of 30–40 nm. An SEM image showing the morphology and porosity of the substrate surface is shown in Fig. 3(a). A substrate with an effective dimension of 20 cmL × 11 cmW was clammed in a membrane holder and then dipped in aqueous PVA/TEG solution with the weight concentrations of PVA and TEG of 1.5% and 15%, respectively, following our previous study which showed that the membrane with PVA:TEG = 1:10 possesses high water vapor permeance [16]. Through this approach, only one side of the membrane was coated. The coated membranes were dried in an oven at 70 °C for 30 min. After 5 times of coating and drying, a thin layer of PVA/TEG blend forms over the pores of the polyamide layer as shown in Fig. 3(b). The PVA/TEG top layer is around 3–5 µm thick (Fig. 3(c)). Water vapor and air permeations of the obtained membranes were evaluated employing the instrumental testing setup shown in Fig. 3(d). The membrane support frame was connected to a vacuum pump. Permeate pressure was set at ~ 0 mbar. A leak-proof test was done

J

Fw⋅45,000( mol )

COPisen = Fw kw RT kw −1

kw −1 ⎤ ⎡ ⎥ ⎢ ⎛ Pamb ⎞ kw −1 ⎟ ⎜ ⎥+ ⎢ ⎝ Pvac ⎠ ⎥⎦ ⎢⎣

Fa k a RT k a −1

k a −1 ⎤ ⎡ ⎥ ⎢ ⎛ Pamb ⎞ a −1 ⎟ ⎜ ⎥ ⎢ ⎝ Pvac ⎠ ⎥⎦ ⎢⎣

(7)

COPisen as the functions of dehumidification performance are plotted for different S in Fig. 2(b). Detailed derivation of the functions is shown in Appendix A. COPisen constitutes the lower thermodynamic COP limits of VMD systems. Practical polytropic operation always results in higher COP than this limit. COPisen for S = ∞ varies around 1–3 and decreases when a higher percentage of the moisture is being removed. This indicates that the drier the product air becomes, the less efficient the VMD is. This is because a lower vacuum pressure, which incurs more pumping energy, is required to drive the separation in order to achieve a drier product air. As shown in Eqs. (4)–(7), the dehumidification COP is independent of the membrane's water vapor permeance. Instead it depends strongly on the membrane selectivity. The effect of membrane selectivity (S) on maximal COP via isentropic compression as a function of dehumidification performance is shown in Fig. 2(b). The fact that a higher dehumidification performance is obtained with lower S or a permeate sweep gas [7,11], shown in Fig. 2(a), does not mean that a poorer selectivity implies a higher energy efficiency. It is because a greater amount of energy is required to pump the permeated air. Therefore, the dehumidification COP decreases with lower S. Additionally, COP peaks as a function of dehumidification performance; an observation consistent with reported results [11]. From Fig. 2(b) it is apparent that when the membrane selectivity exceeds 1000 (attainable with current polymer membranes [9,10]), COP that can be achieved is about 2–3 in 78

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

Fig. 3. SEM images of (a) surface of bare substrate; (b) surface of membrane with 5 PVA/TEG dips; (c) cross section of the membrane's top-layers; (d) gas/vapor permeation test rig; and (e) apparent water vapor and permeances of the membranes with different number of PVA/TEG dips.

Fig. 3(e). The substrate itself with polyamide top layer has a high water vapor permeance. However, because of its large pore size which is favorable for air permeation, it has low selectivity which is around 200. When number of coating layers is increased, the measured water vapor permeance is marginally lowered, while the air permeance decreased significantly. It means that the coating layer is thin enough not to affect much the membrane water vapor permeation but effective enough to cover the pores of the substrate to hinder the air permeation. After 5 dips, the measured membrane water vapor permeance and selectivity were stable at 8800 GPU and 2650 respectively. This observed selectivity is consistent with the data of common selectivity of polymer membranes [9,10] thereby ensuring a high energy efficiency for the developed prototype according to Fig. 4(b).

prior to the start of each experiment. During the leak-proof test, a dense polypropylene sheet (0.5 mm thick) was put in the position of membrane in the membrane holder, as a stopper. The vacuum pump was then turn on to get the minimum vacuum pressure of 0 mbar. After that, the valve at the vacuum pump inlet was closed and the vacuum pump was turned off. No change in vacuum pressure was observed within 1 h. This meant that the vacuum system was completely leak proof, and there was only one way for air and water vapor to enter the system, which is through the membrane. The membrane support frame was placed in a humidity and temperature chamber set at 90% and 30 °C, respectively. The permeate stream is highly rich in water vapor. Water fraction in permeate stream is 0.9–0.99 depending the membranes’ selectivity. When it comes out from the vacuum pump, the water vapor quickly condenses to liquid. The RH sensor are not suitable for measuring supersaturated air like this because condensed water can damage the sensor tip. Instead, a cold trap commonly used by other researchers [6,8,27,36], is employed in such a situation. Cold water of 0 °C absorbs the rejected heat and accelerates the condensation of water vapor. As the content of air in the permeate stream is small, the retention time of the permeate stream is long enough to ensure a complete water condensation. The water permeated through the membrane was weighed. The non-condensable air was measured by a measuring cylinder. Water vapor and air permeances of the membranes with different number of PVA/TEG coating layers are shown in

3.2. Membrane module development The obtained membrane was then up-scaled to develop a membrane module. Square substrates (Fig. 4(a)) with dimensions of 40 cm × 40 cm were glued onto both sides of a metal frame with 2.5 cm border width and 1 cm thick. The space within the frame forming the permeate channel was filled with highly porous spacer to support the membrane. There are 4 vacuum outlets at two opposite sides of the frame for connecting with a vacuum pump. A picture of the membrane frame is 79

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

Fig. 4. (a) NFW nanofiltration membrane obtained from Synder Filtration; (b) a coated membrane frame and (c) membrane module setup.

indoor condition with temperature of 22 ± 1 °C and relative humidity of 60 ± 8% and an outdoor condition with temperature of 31.5 ± 1 °C and relative humidity of 57 ± 8%. As the feed air passes through the membrane module, its humidity decreases. The water vapor removal rates in these experiments are shown in Fig. 5(a). Water vapor removal rates are higher in the outdoor experiments than in the indoor ones. However, dehumidification in terms of the percentage of water vapor removed is the same for both indoor and outdoor experiments, as shown in Fig. 5(b). The water vapor partial pressure of the feed streams in the module varies within the shaded area which is limited by the input and output water vapor partial pressures as shown in Fig. 5(c) and (d) for outdoor and indoor conditions, respectively. It is noteworthy that the vacuum pump runs at its maximum pumping speed in all the experiments. The pumping volume flowrates are almost the same in the two experiments. The permeate pressure (pvac) is proportional to total permeation rate of water and air through the membrane. Therefore, higher permeation rates lead to higher permeate pressures in the outdoor experiments. Lower permeate pressures do not necessarily produce higher driving forces for water permeation, because the water partial pressure in the feed stream is also lower in the indoor experiments. It has been reported in the literature that there exists concentration polarization practically in all membrane separation processes [17,36,42,43]. In this case of water vapor permeable membranes, the concentration polarization is the depletion of water vapor in the air boundary layer on the feed side and the accumulation of water vapor in the boundary layer on the permeate side of the membrane. This is attributed to the resistances of the boundary layers to the water vapor diffusion. The resistance (k) of air boundary layer at pressure p and temperature T can be determined from water vapor diffusivity (D) in the air and the thickness of the boundary layer (μ) [43]:

shown in Fig. 4(b). The membrane frames were dip-coated 5 times with PVA/TEG. 10 coated membrane frames, each with a total effective membrane area of 0.245 m2, were then stacked in an enclosure with gaps of 1 cm for air flow. The complete membrane module is shown in Fig. 4(c). A blower Vents TT 100 with variable speed connects to the membrane module in order to push the air through the membrane module. The energy consumption of the blower is 21–33 W, depending on the feed flowrate used. Humidity and temperature of the air stream before and after passing through the module were measured using Vaisala HMT338 humidity and temperature sensors with the accuracy of 2% and 0.2 °C, respectively. The flowrate of the air stream was measured using a Kanomax climomaster 6501 with an accuracy of 2%. An appropriate vacuum pump (Edwards XDS46i) was connected to the membrane module to enable the separation process to take place. The pump was selected based on the performance of the membrane module to ensure that the system operates at its maximum energy efficiency according to Fig. 4(b). Vacuum pressure of the permeate stream was monitored using a pressure gauge (Omega DPG1000B, accuracy error < ± 0.35%). The volume of the non-condensable air was measured by a measuring cylinder. The water removal was computed from the humidity drop in the feed stream. The vacuum pump power consumption was measured by using a wattage meter with an accuracy of ± 5 W. Experimental COP of the system was obtained using Eq. (1) based on the latent heat removal rate and the vacuum pump power consumption. The power consumption of the feed fan is not considered during the COP calculation, because of two key reasons, namely, (1) the membrane module is not supposed to be standalone system; it is plugged to the ducting module of a HVAC system that typically includes a blower; and (2) with our proposed design, the feed air can easily pass through the membrane module. The energy consumption for the fan is negligible - about 5% of the vacuum pump's power consumption.

k= 4. Result and discussion 4.1. Dehumidification performance of the membrane module

μRT = D

μRT ⎛ 101325(Pa) ⎞ ⎛ T ⎞1.81 2.1910−5 ⎜ ⎟ ⎜ 273.15(K ) ⎟ p ⎠ ⎠⎝ ⎝

(8)

As the resistance is proportional to pressure, it is negligible on the permeate side which is under vacuum pressure (pvac). Therefore, the concentration polarization in VMD is primarily due to the depletion of

The VMD prototype was tested under two specific conditions, an 80

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

Fig. 5. The changes in (a) the water and air collection rates and (b) dehumidification performances, at varying feed air velocity; and the change in water pressures in experiments carried out in (c) outdoor and (d) indoor conditions.

Fig. 5(a). From the air and water permeation rates, the water vapor pressure ( pwp ) at vacuum pump's inlet is obtained and displayed in Fig. 5(c) and (d) for the two experimental conditions. Concentration polarization also complicates the evaluation of membrane permeance [36,43]. This is because the actual effective water vapor partial pressure on the feed-side surface of the membrane is unknown. As a result, the driving force for water vapor permeation is unable to be exactly determined. Therefore, it is possible that one underestimates a membrane's apparent water vapor permeance from the known parameters such as water vapor pressures of input humid air, output dried air and permeate flow at vacuum pump's inlet [43]. Using the permeate water vapor pressure at vacuum pump's inlet for calculations will result in the largest driving force and the smallest membrane's water vapor permeance, compared to using water vapor pressure inside the permeate channel. Fig. 6 shows the apparent water permeances and selectivities estimated based on the assumption that the feed water vapor pressure is the average of the inlet and outlet water vapor pressures. The water vapor permeances and selectivities of the membranes for the outdoor experiments are observed to be higher than those of indoor ones. Due to a reduction in concentration polarization, both apparent water permeances and selectivities increase at higher feed air velocities. It is noteworthy that the water vapor permeances of 8800–10,500 GPU shown in Fig. 3(e) and 5000–11,000 GPU shown in Fig. 6(a) are apparent values and not actual values. Due to different experimental

water vapor on the feed boundary layer. The water vapor pressure at the membrane surface is at an equilibrium pressure, at which the water vapor flux through the boundary layer is equal to that through the membrane. When the membrane has higher water vapor permeance (less resistance to water vapor permeation), higher water vapor flux is obtained. This requires higher driving force for water vapor diffusion, and causes higher water vapor depletion. The concentration polarization is negligible when the membrane has high water vapor resistance. It becomes serious when the membrane's resistance has the same order of magnitude as the boundary layer's resistance. It significantly reduces the driving force for water vapor permeation resulting in a lower overall separation efficiency, higher capital and operation costs. Increasing the feed humid air velocity leads to a thinner boundary layer which consequently reduces the concentration polarization. As a result, water vapor removal rate increases with higher air speed as portrayed in Fig. 5(c). However, although more amount of water is being removed, the water removal fraction in the feed stream lowers as shown in the reduction of the membrane's dehumidification performance when subjected to a higher feed air flowrate in Fig. 5(b) [12]. While concentration polarization impacts on water vapor permeation, it has marginal influence on air permeation. It is because the driving force for air permeation is much larger than that for water vapor permeation. Therefore, the air permeation rate does not change with both feed air flowrate and feed air conditions, as shown in 81

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

Fig. 6. (a) Apparent water vapor permeance and (b) apparent selectivity of the membranes determined in outdoor and indoor conditions.

4.2. Energy efficiency analysis

conditions, the calculated apparent permeances are different in these experiments. Theoretically, the apparent values approach the actual values if the feed air velocity continues to increase [43]. It means that, under normal operation conditions, the membrane typically is operating below its potential.

The vacuum pump's energy consumption was monitored during the experiments. It is apparent from Fig. 7(a) that it increases linearly with higher pumping load comprising both gas and vapor pumping rates.

Fig. 7. (a) Vacuum pump power consumption versus total vapor and gas pumping flowrate; (b) experimental dehumidification COP at varying air flow rate; and (c) estimated pump efficiency and dehumidification COPisen for several commercial vacuum pumps.

82

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

pumping speeds normally have low efficiency, which result in low estimated dehumidification COP. Larger pumps with high pumping speed may have higher efficiency. As a result, the maximum estimated COPisen increases with higher pumping speed as seen from the dash line in Fig. 7(c). Therefore, the employment of VMD is more advantageous in a large scale system, which is comprised of a sizable pump with a high pumping speed and scaled-up membrane modules. A COP value that is close to the thermodynamic limit can be attained when the vacuum pump possesses a pumping speed that is higher than 2000 m3/h.

However, under a zero load condition where the inlet valve to the pump was closed, the pump's motor still consumed a large amount of energy. Power consumption of the pump spanned 488–604 W, close to the manufacturer's motor rating of 520 W. Based on the energy consumption of the vacuum pump and the latent heat removed, the experimental dehumidification COP values were determined employing Eq. (1) and shown in Fig. 7(b). Comparatively, the dehumidification COP is higher for experiments conducted under outdoor conditions. This observation is consistent with the results shown in Fig. 2(b) and (c) which indicated that dehumidifying hot and humid air is more efficient than dehumidifying cool and dry air. Higher dehumidification COPs are obtained with higher feed air flowrates. As dehumidification performance decreases with higher feed air flowrate shown in Fig. 5(d), there exists an intrinsic compromise between dehumidification energy efficiency and performance [11]. The dehumidification COP obtained from the experiments varied from 0.2 to 0.5. It is much smaller than the theoretical COP limit, which is 2–3 in isentropic condition as seen in Fig. 2(b). The theoretical COP limit is derived when the membrane has a selectivity of ~1000 and dehumidification performance in the range of 10–30%. The low obtained COP is attributed to the low efficiency of the present vacuum pump arising from its high zero-load energy consumption. In practice, the VMD's energy efficiency depends strongly on the vacuum pump efficiency. As far as the vacuum pump is concerned, its motor power rate (W) and pumping speed (m3/h) are two important parameters that determine its efficiency. A dry vacuum pump with a low power rate and high pumping speed at the vacuum pressure of 10–15 mbar is highly desirable for VMD application. In order to assess the performance of the current vacuum pump technology, a survey of several common commercial dry pumps with a wide range of pumping speed, from 3 to almost 4000 m3/h, has been carried out. The pump employed in this present work, with 520 W and 40 m3/h, is considered to be small in size. The pump models and their motor power rates and pumping speeds at 12.5 mbar are shown in Appendix B. The pump efficiency is defined as the ratio of its motor power rating to the theoretical work with the assumption of least-energy-efficient isentropic compression. From the pump's efficiency, the dehumidification COPisen was estimated. The estimated COPisen is proportional to the pump's efficiency, as shown in Fig. 7(c). The result shows that there is no standard efficiency for all the commercial vacuum pumps in the market. It varies widely from 4% to 97%. The efficiency of a pump depends on its brand, model, and particularly its pumping speed. Small pumps with low

5. Conclusion A systematic study has been conducted to determine the theoretical and experimental limits of VMD efficiency. Results have indicated that VMD is suitable for environments with high temperature and humidity. According to the theoretical model, it is possible to achieve VMDs' COP of 2–3 by adopting the assumption of the least-efficient isentropic compression. High performing PVA-TEG membranes coated on TFC substrates were fabricated and up-scaled to develop a compact VMD prototype unit. The experimentally obtained apparent water vapor permeance of higher than 11,900 GPU and selectivity of 1780 are among the highest achievable values for membranes thus far developed. It is also observed that higher air temperature and humidity are favorable to achieving higher dehumidification COP. However, a trade-off exists between dehumidification performance and energy efficiency with regulating feed air velocity. A review of existing vacuum pumps has indicated the existence of a practical efficiency limit of current vacuum pump technology. Specifically, a higher pump efficiency can be obtained with a higher capacity pump. A COP value close to the system's thermodynamic limit can be attained with certain vacuum pump types which possess pumping speeds that are higher than 2000 m3/h. The study on the lab-scale VMD prototype contributes great knowledge to the development and optimization of the technology. Acknowledgment The authors gratefully acknowledge the generous funding from the National Research Foundation (NRF) Singapore under the Energy Innovation Research Programme (EIRP) Funding Scheme (R-265-000543-279) managed on behalf by Building and Construction Authority (BCA).

Appendix A. Derivation of COPiso, S =∞ and COPisen, S =∞ plotted in Fig. 2(b) When S = ∞, there is no air flux through the membrane (Fa = 0) and the permeate water vapor pressure is equal to vacuum pressure ( pwp = pvac). Pressure profile along the membrane is shown in Fig. A1. The water vapor partial pressure of the feed air stream ( pwf ) is gradually lowered as it passes

Fig. A1. Pressure profile along the membrane when S = ∞.

83

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

over the membrane. The separation is most efficient if the feed water vapor partial pressure reach vacuum pressure at the exit ( pwf / out = pvac). This can be achieved by maintaining sufficient contact time between feed air and membrane by either increasing the membrane length or decreasing the feed flowrate. Lowering pvac leads to more water vapor being removed and hence a drier product air. Eq. (6) for the dehumidification performance now becomes:

Dehumidification performance = x =

⇨ Pvac

pwf / in − pvac pwf / in

⋅100% (A1)

x = pwf / in (1− ) 100

(A2)

Eq. (7) for maximal obtainable VMD's COP in isentropic compression now becomes:

COPisen, S =∞ =

⎛ J ⎞ 45,000 ⎜ ⎟ ⎝ mol ⎠ kw −1 ⎡ ⎤ ⎞ kw ⎢⎛ ⎥ ⎜ ⎟ k w RT ⎢ Pamb ⎥ −1 ⎜ ⎟ ⎥ k w−1 ⎢ ⎜ p f (1− x ) ⎟ ⎢ ⎝ w / in ⎥ 100 ⎠ ⎣ ⎦

(A3)

For typical outdoor condition (T = 31 °C and RH=60%), COPiso, S =∞ and COPisen, S =∞ are:

COPiso, S =∞ =

COPisen, S =∞ =

45,000 ⎛ ⎞ ⎜ ⎟ 4496 8.314⋅304.15⋅ln ⎜ x ⎟ ⎜ 2698(1− )⎟ ⎝ 100 ⎠

(A4)

45,000 1.32−1 ⎡ ⎤ ⎞ 1.32 ⎢⎛ ⎥ ⎟ 1.32⋅8.314⋅304.15 ⎢ ⎜ 101325 −1⎥ ⎜ ⎟ x ⎢ ⎥ 1.32 − 1 ⎜ 2698(1− )⎟ ⎢⎝ ⎥ 100 ⎠ ⎣ ⎦

(A5)

COPiso, S =∞ and COPisen, S =∞ as functions of dehumidification performance (x) are plotted in Fig. 2(b). For instance, if the dehumidification performance: x = 50%. Then

COPiso, S =∞ =

COPisen, S =∞ =

45,000 = 14.78 ⎛ ⎞ ⎜ ⎟ 4496 ⎟ 8.314⋅304.15⋅ln ⎜ ⎜ 2698(1− 50 ) ⎟ ⎝ 100 ⎠ 45,000 = 2.33 1.32−1 ⎡ ⎤ ⎞ 1.32 ⎢⎛ ⎥ ⎟ 1.32⋅8.314⋅304.15 ⎢ ⎜ 101325 ⎥ ⎜ ⎟ −1⎥ ⎢ 50 ⎟ 1.32 − 1 ⎜ ) ⎢ 2698(1− ⎥ 100 ⎠ ⎢⎣ ⎝ ⎥⎦

This means that, if the pump efficiency is 100% and the membrane selectivity is infinite (S = ∞), the VMD system will achieve a COP of at least 2.33, depending on whether the pump works in isentropic, polytropic or isothermal conditions. Appendix B. Several commercial vacuum pumps and their estimated COPs when they are used for VMD. The estimated COPs are plotted versus pumping speed in Fig. 7(c)

Brand

Models

Pumping speed (m3/h) *

Motor power rating (kW)

Efficiency (%)

Estimated COPisen

Edward

XDS35i XDS46i nXDS15i nXDS20i nXDS10i nXDS6i GXS160 GXS160/1750 GXS250 GXS250/2600 GXS450 GXS450/2600

35 39 15 22 11 6 140 500 240 550 440 1500

0.52 0.52 0.3 0.26 0.28 0.26 5 7.4 9 9.7 17.3 20

18.34 20.43 13.62 23.05 10.70 6.29 7.63 18.41 7.27 15.45 6.93 20.43

0.51 0.57 0.38 0.64 0.30 0.17 0.21 0.51 0.20 0.43 0.19 0.57

84

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

Agilent

Ebara

Busch

GXS450/4200 GXS750 GXS750/2600 GXS750/4200 iH80 iH160 iH600 iH1000 Drystar 80 Drystar 80 PFPE Drystar 80/EH500 Drystar 80/EH500 PFPE IDX1000 IDX1300 EPX180L EPX180LE EPX180N EPX180NE EPX500L EPX500LE EPX500N EPX500NE IDP-3 IDP-15 SH-110/SH-112 ESR20N ESR30 ESR80W ESR200W ESR300W EST25N EST100WN EST200WN EST300WN EST500WN EV-S20 EV-S50 EV-S100 EV-S200 COBRA NX 0450 A COBRA NX 0650 A COBRA NC 0100 B COBRA NC 0200 B COBRA NC 0300 B COBRA NC 0400 B COBRA NC 0630 C COBRA NC 1000 B COBRA NC 2500 B COBRA DS 0600 D COBRA DS 3010 D COBRA DS 8161 D COBRA DS 8162 D COBRA BA 0100 C COBRA BC 0100 F COBRA BC 0100 F COBRA BC 0101 F COBRA BC 0101 F COBRA BC 0600 F COBRA BC 0600 F COBRA BC 1000 F COBRA BC 2000 F COBRA BC 1000 F Panda WZ 2000 A Puma WP 0500 D2 Puma WP 0500 D4 Puma WP 0250 D4

1300 740 2100 2000 100 150 400 200 68 68 255 255 900 1290 60 60 60 60 60 60 60 60 3 13 6 78 108 240 600 1800 120 540 1200 1800 1800 60 198 540 1140 355 650 110 190 320 350 630 840 2000 595 2000 1000 3000 85 90 88 77 72 75 75 250 400 230 600 340 325 210

21.1 37 40 40 3.5 5 6.1 6.1 4 4 6.6 6.7 30 30 3 3 3 3 3 3 3 3 0.12 0.56 0.19 2.5 3.5 4.5 5 7.5 5.5 7.7 7.8 9.5 11.3 2.2 3.6 4.6 5.1 7.5 12.5 3.5 6 7.5 7.5 15 22 55 15 15 15 30 1.8 1.8 1.8 1.5 1.5 1.5 1.5 1.5 2.9 1.8 5.5 2.2 1.8 1.1 85

16.79 5.45 14.30 13.62 7.78 8.17 17.87 8.93 4.63 4.63 10.53 10.37 8.17 11.72 5.45 5.45 5.45 5.45 5.45 5.45 5.45 5.45 6.81 6.33 8.60 8.50 8.41 14.53 32.70 65.39 5.94 19.11 41.92 51.62 43.40 7.43 14.99 31.98 60.90 12.90 14.17 8.56 8.63 11.63 12.71 11.44 10.40 9.91 10.81 36.33 18.16 27.25 12.87 13.62 13.32 13.99 13.08 13.62 13.62 45.41 37.58 34.81 29.72 42.11 49.19 52.02

0.47 0.15 0.40 0.38 0.22 0.23 0.50 0.25 0.13 0.13 0.29 0.29 0.23 0.33 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.15 0.19 0.18 0.24 0.24 0.23 0.40 0.91 1.82 0.17 0.53 1.16 1.43 1.21 0.21 0.42 0.89 1.69 0.36 0.39 0.24 0.24 0.32 0.35 0.32 0.29 0.28 0.30 1.01 0.50 0.76 0.36 0.38 0.37 0.39 0.36 0.38 0.38 1.26 1.04 0.97 0.83 1.17 1.37 1.44

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

Puma WP 0700 D2 Puma WP 1000 D2 Puma WP 1000 D4 Puma WP 1250 D2 Puma WP 2000 D2 Puma WP 4500 B2 Puma WP A040 A Puma WP A055 A Puma WP A075 A Puma WP A080 A Puma WP A095 A Panda WV 0250 C Panda WV 0500 C Panda WV 1000 C Panda WV 1500 C Panda WV 2000 C Panda WV 4500 B Puma WY 0500 C Puma WY 0700 B Puma WY 1250 B Puma WY 2000 B Puma WY 4500 B Seco SD 1010 C Seco SD 1016 C Seco SD 1025 C Seco SD 1040 C COBRA NX 0450 A

110 110 490 125 1450 500 2800 850 850 850 850 220 350 400 1250 1550 3900 70 80 80 150 600 10 16 25 40 355

3.5 3.5 3 4.2 6 11 11 15 18.5 18.5 22 1.1 2.2 3.5 4.2 6 11 2.2 3 4 5.5 11 0.37 0.55 0.9 1.25 7.5

8.56 8.56 44.50 8.11 65.85 12.38 69.35 15.44 12.52 12.52 10.53 54.49 43.35 31.14 81.09 70.39 96.60 8.67 7.27 5.45 7.43 14.86 7.36 7.93 7.57 8.72 12.90

0.24 0.24 1.24 0.23 1.83 0.34 1.93 0.43 0.35 0.35 0.29 1.51 1.20 0.86 2.25 1.96 2.68 0.24 0.20 0.15 0.21 0.41 0.20 0.22 0.21 0.24 0.36

[19] S. Liu, F. Wang, T. Chen, Synthesis of poly(ether ether ketone)s with high content of sodium sulfonate groups as gas dehumidification membrane materials, Macromol. Rapid Commun. 22 (2001) 579–582. [20] H. Sijbesma, K. Nymeijer, R. van Marwijk, R. Heijboer, J. Potreck, M. Wessling, Flue gas dehydration using polymer membranes, J. Membr. Sci. 313 (2008) 263–276. [21] J.R. Du, L. Liu, A. Chakma, X. Feng, Using poly(N,N-dimethylaminoethyl methacrylate)/polyacrylonitrile composite membranes for gas dehydration and humidification, Chem. Eng. Sci. 65 (2010) 4372–4381. [22] W. Xie, G.M. Geise, B.D. Freeman, H.S. Lee, G. Byun, J.E. McGrath, Polyamide interfacial composite membranes prepared from m-phenylene diamine, trimesoyl chloride and a new disulfonated diamine, J. Membr. Sci. 403–404 (2012) 152–161. [23] J. Potreck, K. Nijmeijer, T. Kosinski, M. Wessling, Mixed water vapor/gas transport through the rubbery polymer PEBAX® 1074, J. Membr. Sci. 338 (2009) 11–16. [24] S.J. Metz, J. Potreck, M.H.V. Mulder, M. Wessling, Water vapor and gas transport through a poly(butylene terephthalate) poly(ethylene oxide) block copolymer, Desalination 148 (2002) 303–307. [25] Y. Li, H. Jia, F. Pan, Z. Jiang, Q. Cheng, Enhanced anti-swelling property and dehumidification performance by sodium alginate-poly(vinyl alcohol)/polysulfone composite hollow fiber membranes, J. Membr. Sci. 407–408 (2012) 211–220. [26] F. Pan, H. Jia, Z. Jiang, X. Zheng, Enhanced dehumidification performance of PVA membranes by tuning the water state through incorporating organophosphorus acid, J. Membr. Sci. 325 (2008) 727–734. [27] Y. Shin, W. Liu, B. Schwenzer, S. Manandhar, D. Chase-Woods, M.H. Engelhard, et al., Graphene oxide membranes with high permeability and selectivity for dehumidification of air, Carbon N. Y. 106 (2016) 164–170. [28] A. Ito, Dehumidification of air by a hygroscopic liquid membrane supported on surface of a hydrophobic microporous membrane, J. Membr. Sci. 175 (2000) 35–42. [29] F.F. Krull, C. Fritzmann, T. Melin, Liquid membranes for gas/vapor separations, J. Membr. Sci. 325 (2008) 509–519. [30] P. Scovazzo, Testing and evaluation of room temperature ionic liquid (RTIL) membranes for gas dehumidification, J. Membr. Sci. 355 (2010) 7–17. [31] Q. Cheng, F. Pan, B. Chen, Z. Jiang, Preparation and dehumidification performance of composite membrane with PVA/gelatin-silica hybrid skin layer, J. Membr. Sci. 363 (2010) 316–325. [32] F.H. Akhtar, M. Kumar, K.-V. Peinemann, Pebax®1657/graphene oxide composite membranes for improved water vapor separation, J. Membr. Sci. (2016). [33] J.G. Wijmans, R.W. Baker, The solution-diffusion model: a review, J. Membr. Sci. 107 (1995) 1–21. [34] S.J. Metz, N.F. a Van Der Vegt, M.H.V. Mulder, M. Wessling, Thermodynamics of water vapor sorption in poly(ethylene oxide) poly(butylene terephthalate) block copolymers, J. Phys. Chem. B 107 (2003) 13629–13635. [35] A.R.D. Verliefde, P. Van der Meeren, B. Van der Bruggen, E.M. V Hoek, V.V Tarabara, Solution-diffusion processes, in: Encycl. Membr. Sci. Technol., John Wiley & Sons, Inc., 2013. [36] S.J. Metz, W.J.C. Van De Ven, J. Potreck, M.H.V. Mulder, M. Wessling, Transport of water vapor and inert gas mixtures through highly selective and highly permeable polymer membranes, J. Membr. Sci. 251 (2005) 29–41.

References [1] K.J. Chua, S.K. Chou, W.M. Yang, J. Yan, Achieving better energy-efficient air conditioning – a review of technologies and strategies, Appl. Energy 104 (2013) 87–104. [2] M.M. Rafique, P. Gandhidasan, H.M.S. Bahaidarah, Liquid desiccant materials and dehumidifiers – a review, Renew. Sustain. Energy Rev. 56 (2016) 179–195. [3] K.S. Rambhad, P.V. Walke, D.J. Tidke, Solid desiccant dehumidification and regeneration methods—a review, Renew. Sustain. Energy Rev. 59 (2016) 73–83. [4] M. Sahlot, S.B. Riffat, Desiccant cooling systems: a review, Int. J. Low-Carbon Technol. (2016) (ctv032). [5] E.M. Mina, T.A. Newell, A.M. Jacobi, A generalized coefficient of performance for conditioning moist air, Int. J. Refrig. 28 (2005) 784–790. [6] J. Li, A. Ito, Dehumidification and humidification of air by surface-soaked liquid membrane module with triethylene glycol, J. Membr. Sci. 325 (2008) 1007–1012. [7] P. Scovazzo, A.J. Scovazzo, Isothermal dehumidification or gas drying using vacuum sweep dehumidification, Appl. Therm. Eng. 50 (2013) 225–233. [8] R. Xing, Y. Rao, W. TeGrotenhuis, N. Canfield, F. Zheng, D.W. Winiarski, et al., Advanced thin zeolite/metal flat sheet membrane for energy efficient air dehumidification and conditioning, Chem. Eng. Sci. 104 (2013) 596–609. [9] J. Woods, Membrane processes for heating, ventilation, and air conditioning, Renew. Sustain. Energy Rev. 33 (2014) 290–304. [10] B. Bolto, M. Hoang, Z. Xie, A review of water recovery by vapour permeation through membranes, Water Res. 46 (2012) 259–266. [11] T.D. Bui, F. Chen, A. Nida, K.J. Chua, K.C. Ng, Experimental and modeling analysis of membrane-based air dehumidification, Sep. Purif. Technol. 144 (2015) 114–122. [12] T.D. Bui, A. Nida, K.J. Chua, K.C. Ng, Water vapor permeation and dehumidification performance of poly(vinyl alcohol)/lithium chloride composite membranes, J. Membr. Sci. 498 (2016) 254–262. [13] B. Zhao, N. Peng, C. Liang, W.F. Yong, T.-S. Chung, Hollow fiber membrane dehumidification device for air conditioning system, Membranes (Basel) 5 (2015) 722–738. [14] B. Yang, W. Yuan, F. Gao, B. Guo, A review of membrane-based air dehumidification, Indoor Built Environ. 24 (2013) 11–26. [15] A. Wolińska-Grabczyk, A. Jankowski, Membranes for vapour permeation: preparation and characterization BT – pervaporation, vapour permeation and membrane distillation, in: Woodhead Publication Series in Energy, Woodhead Publishing, Oxford, 2015, pp. 145–175. [16] T.D. Bui, Y. Wong, K. Thu, S.J. Oh, M. Kum, Ja, K.C. Ng, et al., Effect of hygroscopic materials on water vapor permeation and dehumidification performance of polyvinyl alcohol membranes, J. Appl. Polym. Sci. 44765 (2017) 1–9. [17] H. Lin, S.M. Thompson, A. Serbanescu-Martin, J.G. Wijmans, K.D. Amo, K.A. Lokhandwala, et al., Dehydration of natural gas using membranes. Part I: composite membranes, J. Membr. Sci. 413–414 (2012) 70–81. [18] P. Scovazzo, A. Hoehn, P. Todd, Membrane porosity and hydrophilic membranebased dehumidification performance, J. Membr. Sci. 167 (2000) 217–225.

86

Journal of Membrane Science 539 (2017) 76–87

T.D. Bui et al.

membrane air drying and evaporative cooling, Trans. IChemE 78 (2000) 999–1009. [41] Synder Nanofiltration Microfiltration Ultrafiltration Membrane, Synder Filtr. Inc., 2017. synderfiltration.com (Accessed 2 February 2017). [42] G. He, Y. Mi, P.L. Yue, G. Chen, Theoretical study on concentration polarization in gas separation membrane processes, J. Membr. Sci. 153 (1999) 243–258. [43] T.D. Bui, Y. Wong, K.J. Chua, K.C. Ng, Evaluating water vapor permeance measurement techniques for highly permeable membranes, Comput. Mater. Contin. 47 (2015) 89–105.

[37] J. Taweekun, V. Akvanich, The experiment and simulation of solid desiccant dehumidification for air-conditioning system in a tropical humid climate, Engineering 5 (2013) 146–153. [38] T.S. Ge, Y.J. Dai, R.Z. Wang, Z.Z. Peng, Experimental comparison and analysis on silica gel and polymer coated fin-tube heat exchangers, Energy 35 (2010) 2893–2900. [39] C. Vallieres, E. Favre, Vacuum versus sweeping gas operation for binary mixtures separation by dense membrane processes, J. Membr. Sci. 244 (2004) 17–23. [40] H.T. El-Dessouky, H.M. Ettouney, W. Bouhamra, A novel air conditioning system-

87