Abstract

Membrane distillation is a rate-governed non-isothermal membrane separation technique that utilizes trans-membrane temperature difference for evaporating water and thereby separating it from brackish feed for reproducing fresh water. A novel design of a cylindrical air gap membrane distillation module is presented. The module is fabricated in a way similar to a shell and tube heat exchanger. A PTFE hydrophobic membrane is used and is formed in a cylindrical shape. Design of experiments (DOE) is used to design the experiments statistically and to identify the significant operating parameters. Experiments were performed according to the Taguchi design approach using an L16 orthogonal array. Optimization of the whole process is performed by response surface methodology. It is shown that the feed temperature and feed flow rate have a positive effect, whereas the salinity has a negative impact on flux. The maximum value of flux achieved with this system is 3.6 kg/m2 hr. A high value of flux of 2.6 kg/m2 hr was achieved under optimum conditions at a temperature of 45 °C and a flow rate of 1.5 lpm with a salinity of 5 g/litre.

INTRODUCTION

The population explosion has resulted in a scarce supply of potable water and also contaminated the sources of freshwater. Most of the Earth's surface is covered with sea, and thus desalination is a possible solution for the world's potable water needs. Energy economy is another issue that needs to be taken into consideration when designing any utility, where minimal and efficient energy use is always the primary significance of any system. Membrane distillation (MD) combines both of these aspects. It can use waste heat for its operation and purifies the seawater with an excellent salt rejection rate, and therefore, it presents the viable choice for potable water solution that is also energy efficient (Alsalhy et al. 2018; Deshmukh et al. 2018; Swaminathan et al. 2018). The driving force in MD is a trans-membrane vapor pressure difference. A semi-permeable hydrophobic membrane is generally used that does not allow water or other solutes to pass but allows vapors to pass through the membrane. Here the solution to be treated is maintained in direct contact with the membrane on one side. The temperature of the solution is maintained slightly higher than the other side of the membrane. Due to this temperature difference, there exists a vapor pressure difference on both sides of the membrane. As the hydrophobic membrane does not allow water to pass through, liquid/vapor interfaces are formed at the membrane surface. Vapors pass through the membrane and encounter a condensation region on the other side, where they are condensed and form distillate. For different MD configurations, the condensation region is different. The basic MD configurations are classified based on their difference in condensation regions, as shown in Table 1 below.

Table 1

Basic membrane distillation configurations

Name of MD techniqueDirect contact membrane distillation (DCMD)Air gap membrane distillation (AGMD)Sweeping gas membrane distillation (SGMD)Vacuum membrane distillation (VMD)
MD configuration     
Condensation zone The cold permeate is in direct contact with the membrane. The vapors are quenched as soon as they encounter the permeate stream after the membrane (Grossi et al. 2020). An air gap is imposed between the membrane and the condenser surface. This air gap reduces heat transfer through the membrane and facilitates high trans-membrane temperature difference and thus high vapor pressure difference (Alkhudhiri & Hilal 2017). A gas stream is swept over the membrane that carries away the incoming vapors. This air stream is cooled outside in a separate condenser, forming permeate (Shahu & Thombre 2019). A vacuum is applied on the other side of the membrane with the help of a vacuum pump. The incoming vapors are transported and condensed in an external condenser (Shahu & Thombre 2019). 
Name of MD techniqueDirect contact membrane distillation (DCMD)Air gap membrane distillation (AGMD)Sweeping gas membrane distillation (SGMD)Vacuum membrane distillation (VMD)
MD configuration     
Condensation zone The cold permeate is in direct contact with the membrane. The vapors are quenched as soon as they encounter the permeate stream after the membrane (Grossi et al. 2020). An air gap is imposed between the membrane and the condenser surface. This air gap reduces heat transfer through the membrane and facilitates high trans-membrane temperature difference and thus high vapor pressure difference (Alkhudhiri & Hilal 2017). A gas stream is swept over the membrane that carries away the incoming vapors. This air stream is cooled outside in a separate condenser, forming permeate (Shahu & Thombre 2019). A vacuum is applied on the other side of the membrane with the help of a vacuum pump. The incoming vapors are transported and condensed in an external condenser (Shahu & Thombre 2019). 

It has always been of prime importance for every MD system to increase the flux as well as to have higher thermal efficiency. Only air gap membrane distillation (AGMD) allows latent heat recovery without any external heat exchanger and thus reduces its thermal energy consumption that results in higher thermal efficiency (Swaminathan et al. 2016; Lokare et al. 2018; Swaminathan et al. 2018; Gao et al. 2019). For improving the flux as well as the energy efficiency of MD systems, a number of studies have modified membrane properties (Woo et al. 2016; Attia et al. 2017; Feng et al. 2018; Goh et al. 2018), whereas others have changed the traditional design of MD systems (Khalifa & Alawad 2018; Lee et al. 2019). Numerous modifications are suggested in the design of condensers to improve performance. Francis et al. (2003) presented a new design of material gap MD system where they incorporated sand and DI water between the membrane and the condensation plate. They made the gap more conductive and thus they reported an increase of 820% in water flux under the same conditions compared with AGMD, water gap MD as well as when the sand filled in the air gap. Swaminathan et al. (2016) evolved a novel design of the permeate and conductive gap MD systems. They also mentioned that the improved air gap conductivity results in increased permeate production as well as higher gained output ratios (GOR). They reported that conductive gap MD is the most effective configuration in terms of energy efficiency followed by permeate gap MD and subsequently AGMD. In another study multiple cooling channels were provided in a hollow fiber AGMD system. Stainless steel was used for cooling channels instead of polymeric fibers, and greater flux and thermal efficiency were reported with values of 12.5 kg/m2 hr and 81.7% respectively (Aryapratama et al. 2016). Warsinger et al. (2015) changed the morphology of the condensation regime by using superhydrophobic, mildly hydrophobic, and untreated condenser surfaces. They also checked the effect of support mesh on the distillate production rate. They showed that superhydrophobic condenser surfaces and highly conductive support meshes provide significant flux improvement. Liu et al. (2016) developed an innovative double pipe AGMD system with PVDF hollow fiber membranes enveloped by copper tubes instead of polymeric hollow fibers for heat exchange. They introduced a new way to determine the system performance by combining the terms MD flux and gained output ratio and addressed it as equivalent membrane distillation flux (JAGMD). Maximum J was 11.4 kg/m2 hr and GOR was 6.6 while JAGMD was 29.6 kg/m2 hr (Liu et al. 2016). Using design of experiments (DOEs), He et al. (2014) developed a regression model by response surface methodology (RSM) to predict and optimize flux and GOR. The maximum value of J reached 5.07 l/m2 hr, and GOR reached 8.78. RSM and Taguchi methods were compared in a study to determine and optimize permeate flux and to check the influence of feed temperature, feed flow rate, and salt concentration. The thermal efficiency achieved was 96% while the GOR was 4.87 (Alsalhy et al. 2018). A detailed review of omniphobic membranes for treatment of industrial wastewater, saline solution as well as feed solutions containing surfactants was presented by Lu et al. (2019). They suggested that omniphobic membranes are the best candidates for MD systems as they show super-wetting resistance and also exhibit anti-fouling properties towards humid acid and surfactant stabilized oil/water emulsions (Lu et al. 2019). A comprehensive review on AGMD is presented by Shahu & Thombre (2019). They suggested that condenser design modification is a prospective field for improving the performance of AGMD systems. They also suggested that hybrid AGMD systems should be developed that present cost- and energy-efficient solutions for potable and treated water needs (Shahu & Thombre 2019). The present study aims at developing a new and economical arrangement for desalting sea water and producing fresh water with minimum cost and utility requirements. A cylindrical air gap membrane distillation (CAGMD) module is designed with a resemblance to a shell and tube heat exchanger design. The main aim is to analyze the performance of a new design in the form of a cylindrical module compared with the conventional design of AGMD systems. The system results in an economical MD module as the area of the condenser is less than that of the membrane in contrast to the conventional designs and previous results (Aaryapratama et al. 2016). The effect of the reduced condenser area is discussed and ways are suggested to produce the maximum possible flux under this low-cost condition. The study also suggests the conditions to be maintained to achieve the highest possible flux with minimum energy and operational cost with much lower temperature feed availability. DOEs is the statistical tool used for analyzing the effects of the main parameter on permeate flux and to determine the optimum condition for higher flux with minimum energy cost.

EXPERIMENTAL SETUP AND METHODOLOGY

Cylindrical AGMD module

The design of the cylindrical air gap membrane distillation module is cylindrical in shape and is similar to a kind of shell and tube heat exchanger. The schematic is shown in Figure 1. A hollow copper tube is used as a condenser surface. This tube is recessed on its outer surface. A hydrophobic polytetrafluoroethylene (PTFE) membrane has been used as a separating medium. The membrane is purchased from Hangzhou Co. Ltd, China, having a pore size of 0.2 μm, thickness of 160 μm and porosity 80%. The membrane is of flat sheet type and is wrapped over the cooling tube and formed into a cylindrical shape by pasting the ends together. A polypropylene net is used as the support for the membrane so that the membrane withstands the water flow pressure and remains in the proper cylindrical shape under all fluid flow conditions. An air gap of 3 mm is maintained throughout the length of the membrane and is fixed for all the experimental runs. The recess between membrane and copper tube creates the air gap. The copper tube with membrane is inserted inside the shell. The feed water flows through the annulus space between the outer shell and the membrane. Cold water flows through the copper tube. The membrane experiences the difference in vapor pressure on either side. To balance this difference, feed evaporates at the membrane surface and travels through it and on the other side encounters an air gap followed by the condenser surface where it is condensed. The permeate is collected from the bottom of the air gap. The detail of the cylindrical module as well as the mechanism of heat and mass transfer is shown in Figure 1. The figure shows the vertical sectional view of the CAGMD. The CAGMD is identical about its vertical axis and thus to explain the mechanism schematically only the left side of the symmetry is explained in the extra enlarged view, within the red circle, for better understanding, while the right side of the symmetry shows the actual view of the module design. The enlarged view shows the membrane, air gap and condensation film and associated heat and mass transfer processes. The advantage of the CAGMD module is that it presents a very compact system and design flexibility with a cylindrical shape over conventional AGMD modules.

Figure 1

Vertical sectional view of cylindrical air gap membrane distillation module.

Figure 1

Vertical sectional view of cylindrical air gap membrane distillation module.

CAGMD experimental setup

The experimental arrangement is shown in Figure 2. The primary objective is to make the setup cost-effective. A new method of operation is discussed in the present study, as the CAGMD module is placed vertically to gain the advantage of gravity-driven flow through the module and thus no feed water recirculation pump is used, which would otherwise impose higher cost on the system. The feed is heated using a temperature- controlled thermostatic electrical geyser. Feed temperature (T1) is varied from 40 to 80°C. Feed water flows by gravity through the shell of the cylindrical module and is collected in a collection tank at the exit. Cold water is supplied with the help of a centrifugal pump, and after circulation, it is sent to a cold water collection tank. RTD thermocouples are used for temperature measurement, and a rotameter and manual flow measurement is used for confirming flow rates. The whole setup is insulated to reduce heat loss to the atmosphere with rock wool insulation material. Due to the unavailability of sea water, artificial sea water is prepared in the laboratory by adding the required weight of laboratory-grade NaCl salt to tap water. The weighing device is calibrated, and the error is ±0.008 g. The salinity of the feed is varied from 5 g/litre to 30 g/litre and the feed flow rate (FF) is varied from 500 ml/min to 1.5 l/min.

Figure 2

Schematic of the CAGMD experimental setup.

Figure 2

Schematic of the CAGMD experimental setup.

In the present CAGMD module the total condenser area is less than the total effective area of the membrane, because the cold water tube is the innermost part of the cylindrical module, although the literature indicates that the condenser area should be more for higher driving force (Aryapratama et al. 2016). The cold water flow was set to a fixed maximum value of 2.5 lpm for all experimental runs to compensate for this reduction in the cooling area. A stabilization duration of 1 hour is considered after each change of parameter before taking readings; this ensures steady-state conditions, as well as that any disturbances or external noises are settled down. Electrical conductivity and PPM of the hot feed water and the permeate are measured with the help of an electrical conductivity meter.

The permeate flux is calculated by dividing the total mass of permeate collected, by the total collection time and total effective membrane area. The formula for flux calculation is given by Equation (1): 
formula
(1)
where ΔW (kg) is distillate weight collected over a fixed time period t (hr), and A is the effective permeation area (m2). The effective membrane area is 0.015 m2. The length of the effective membrane module is 0.27 m.

DESIGN OF EXPERIMENTS

Statistical techniques facilitate the analyzing of the system under distinct conditional boundaries with relatively fewer numbers of experimental runs and fewer resources, and extract maximum information about system behavior. DOEs also allows determination of the principle affecting parameters and establish a relationship between the response and the major influencing factors that help to predict the system behavior under any operating conditions within or beyond the limits. RSM and the Taguchi method are the ways to apply the DOE to an analysis. These methods also present a platform to assist in developing a systematic performance matrix of factors to carry out experimentation. Analysis of variance (ANOVA) is a tool that statistically analyses the effect of individual factors as well as their interaction effect on a response. It also indicates the relative significance of the linear factors and their square and two-way interactions on the response. The present study employs a Taguchi design to study the CAGMD system statistically by creating an L16 orthogonal performance matrix. A regression model is developed through RSM because RSM can effectively establish a relationship between operating parameters (linear interaction and square term) and response, and also shows the relative significance of their combined effect on response. This facility is not available in the Taguchi method (Khalifa & Lawal 2016). The factors considered are feed temperature, feed flow rate and salinity of feed solution.

The control factors affect the response and can be controlled, however, there is always external noise associated with the experiments, which is an uncontrollable factor. The primary objective is to determine the optimal control factor setting that makes the design response resistant to noise factors. The S/N ratio is a measure of the variation of response for the targeted value under different noise conditions. In this case for Taguchi design, we employ the larger, the better S/N ratio as our aim is to maximize the flux. For this case, the relation of S/N ratio to maximize the response is given by Equation (2): 
formula
(2)

Here n is the number of experiments and Y is the response in each experiment. Confirmation tests were performed to verify the stability and validity of the regression model. The whole model is optimized to determine the most suitable conditions for flux and energy.

RESULTS AND DISCUSSION

CAGMD experiments were performed based on the orthogonal design array L16 developed for the Taguchi technique. The experimental response for flux and energy is shown in Table 2.

Table 2

Taguchi L16 orthogonal array and experimental results

Run orderParameters
Results (experimentation)
Results (model)
Feed temperature T1 (°C)Feed flow rate FF (lpm)Salinity S (g/litre)Flux (kg/m2 hr)Energy (kWh)Flux (kg/m2 hr)Energy (kWh)
45 0.50 0.80 0.57 0.86 0.47 
45 0.75 10 1.11 0.76 1.31 0.90 
45 1.00 20 1.72 1.35 1.58 1.23 
45 1.50 30 1.74 1.54 1.75 1.53 
50 0.50 10 1.35 1.68 0.95 1.67 
50 0.75 1.45 2.00 1.56 1.99 
50 1.00 30 1.52 2.10 1.44 2.07 
50 1.50 20 1.68 2.40 1.58 2.46 
55 0.5 20 1.50 2.75 1.36 2.75 
10 55 0.75 30 1.62 2.80 1.58 2.82 
11 55 1.00 2.62 3.40 2.45 3.34 
12 55 1.50 10 2.66 3.60 2.46 3.48 
13 60 0.50 30 2.20 3.64 2.21 3.63 
14 60 0.75 20 2.40 4.17 2.24 4.07 
15 60 1.00 10 2.70 4.33 2.89 4.38 
16 60 1.50 3.60 4.50 3.71 4.51 
Run orderParameters
Results (experimentation)
Results (model)
Feed temperature T1 (°C)Feed flow rate FF (lpm)Salinity S (g/litre)Flux (kg/m2 hr)Energy (kWh)Flux (kg/m2 hr)Energy (kWh)
45 0.50 0.80 0.57 0.86 0.47 
45 0.75 10 1.11 0.76 1.31 0.90 
45 1.00 20 1.72 1.35 1.58 1.23 
45 1.50 30 1.74 1.54 1.75 1.53 
50 0.50 10 1.35 1.68 0.95 1.67 
50 0.75 1.45 2.00 1.56 1.99 
50 1.00 30 1.52 2.10 1.44 2.07 
50 1.50 20 1.68 2.40 1.58 2.46 
55 0.5 20 1.50 2.75 1.36 2.75 
10 55 0.75 30 1.62 2.80 1.58 2.82 
11 55 1.00 2.62 3.40 2.45 3.34 
12 55 1.50 10 2.66 3.60 2.46 3.48 
13 60 0.50 30 2.20 3.64 2.21 3.63 
14 60 0.75 20 2.40 4.17 2.24 4.07 
15 60 1.00 10 2.70 4.33 2.89 4.38 
16 60 1.50 3.60 4.50 3.71 4.51 

RESPONSE SURFACE METHODOLOGY AND ANOVA

Analysis of the results is carried out using RSM with a 95% confidence level. Regression models are developed by RSM for flux and energy and are shown individually in Equations (3) and (4): 
formula
(3)
 
formula
(4)

It is to be noted that the results of the model and the experimentation are in excellent agreement. Therefore this regression model can be used for prediction of the performance of the present CAGMD module. The results of ANOVA are also displayed for flux in Table 3 and for energy in Table 4. The interpretation of the terms from the ANOVA results demonstrates that if the P-value of factors is less than 0.05, that shows that the factor is strongly significant towards the response. From the ANOVA table, it is evident that the contribution of feed temperature and feed flow rate are the most significant for determining the system performance. F-value is another tool to determine the level of significance of each factor. For the factor to be most influential, its F-value should be high.

Table 3

Analysis of variance (ANOVA) results for flux

SourceDFAdj SSAdj MSF-valueP-value
Model 7.58 0.8431 26.77 0.000 
T1 0.31 0.31 9.92 0.020 
FF 0.15 0.14 4.60 0.020 
S 0.31 0.31 9.86 0.076 
Square terms 0.28 0.09 3.01 0.117 
2-way interaction 0.35 0.117 3.72 0.08 
Residual error 0.188 0.03   
Total 15 7.7774    
Model summary 
R-sq R-sq (adj) R-sq (pred)   
0.177 97.57% 93.93% 86.12%   
SourceDFAdj SSAdj MSF-valueP-value
Model 7.58 0.8431 26.77 0.000 
T1 0.31 0.31 9.92 0.020 
FF 0.15 0.14 4.60 0.020 
S 0.31 0.31 9.86 0.076 
Square terms 0.28 0.09 3.01 0.117 
2-way interaction 0.35 0.117 3.72 0.08 
Residual error 0.188 0.03   
Total 15 7.7774    
Model summary 
R-sq R-sq (adj) R-sq (pred)   
0.177 97.57% 93.93% 86.12%   
Table 4

ANOVA results for energy

SourceDFAdj SSAdj MSF-valueP-value
Model 23.46 2.60 204.10 0.000 
T1 4.76 4.76 372.32 0.000 
FF 0.329 0.32 25.80 0.002 
S 0.05 0.05 3.60 0.107 
Square terms 0.068 0.022 1.79 0.25 
2-way interaction 0.08 0.02 2.13 0.19 
Residual error 0.0766 0.001   
Total 15 23.54    
Model summary 
R-sq R-sq (adj) R-sq (pred)   
0.11 99.67% 99.19% 97.69%   
SourceDFAdj SSAdj MSF-valueP-value
Model 23.46 2.60 204.10 0.000 
T1 4.76 4.76 372.32 0.000 
FF 0.329 0.32 25.80 0.002 
S 0.05 0.05 3.60 0.107 
Square terms 0.068 0.022 1.79 0.25 
2-way interaction 0.08 0.02 2.13 0.19 
Residual error 0.0766 0.001   
Total 15 23.54    
Model summary 
R-sq R-sq (adj) R-sq (pred)   
0.11 99.67% 99.19% 97.69%   

The model gave R2 of 97.57% and 99.67% respectively for flux and energy analysis, which means the model has efficiently considered the effect of all the significant factors for their determination. The value of adjusted R2 shows that the model can capture 93.93% of the variation in flux and 99.19% of the variation in energy by variation in feed temperature, feed flow rate, and salinity. Also, a good value for predicted R2 shows that the model can predict the values of flux and energy for any new condition with the right level of accuracy.

ENERGY CONSIDERATION

The setup was fabricated to form an economical arrangement for studying CAGMD so that it can be directly handed over to the community. Very high salinity feed was considered, keeping in mind the saline content of the original sea water. The energy required for heating a saline feed to a required temperature will be more compared with that of a nonsaline feed. The fact behind this is that when the salt molecules are present in the feed, they absorb some energy to break their components so that they can be dissolved easily, and the temperature of the water is reduced. In the present case the tap water was first heated in a geyser and then the salt was added to it, because heating saline content in a regular electrical geyser causes the coil of the geyser to be damaged as scale forms on the geyser coil. Therefore with increase in salinity, the heating temperature of the feed was also increased, so that after salt addition the feed could be maintained at the required temperature in the supply tank. For this reason, the direct effect of salinity on raising the temperature of the feed could not be determined accurately, but the general trend of energy consumption was in accordance with the literature (Xu et al. 2016), which supports the results (Figure 3). It is to be mentioned here that the maximum portion of energy was utilized to raise the temperature of the feed water and only a minimum amount of energy was used to run auxiliaries, which include pump, sensors, and the controllers.

Figure 3

Main effects plot for energy.

Figure 3

Main effects plot for energy.

EFFECT OF FEED TEMPERATURE

It is evident from the table that the feed temperature has the strongest effect on flux and energy. This result is in accordance with the literature. Figure 4 shows the main effect plot for flux. An increase in feed temperature of 25% from 45°C to 60°C results in an increase in flux of more than 50%. The reason is the exponential relation between the vapor pressure and the feed temperature (Alkhudhiri & Hilal 2017). At low feed temperature, the vapor pressure is low, and it increases exponentially with temperature. Interaction of feed temperature and salinity results in different results for flux, as increasing the salinity reduces the flux. So at a higher temperature and low salinity, the flux had a higher value than at other salinities (Figure 5).

Figure 4

Main effects plot for flux.

Figure 4

Main effects plot for flux.

Figure 5

Effect of feed temperature (°C) and salinity (g/litre) on flux (kg/m2 hr).

Figure 5

Effect of feed temperature (°C) and salinity (g/litre) on flux (kg/m2 hr).

EFFECT OF FEED FLOW RATE

The effect of gravity was utilized to cause the flow of feed through the module. By the ANOVA results, it is clear that the feed flow rate has one of the maximum positive effects on flux. Increasing the feed flow rate results in increased turbulence and thus increased heat transfer coefficients near the membrane surface. The surface plot in Figure 6 shows the combined effect of feed flow rate and salinity. For higher values of flux, the feed flow rate should be higher with minimum salinity. There was a 40% increase in the mean flux with raising the flow rate from 0.5 lpm to 1.5 lpm. Increase in flow rate also washes away any salt deposits from the membrane surface and helps to improve vapor diffusion through the membrane. Increase in flow rate also helps to reduce temperature polarization by properly mixing feed-side bulk flow. In the present vertical cylindrical module, the feed inlet was at the bottom of the module to ensure that the module is always filled with the feed. At lower feed flow rates, the turbulence is suppressed at the top of the module by the effect of gravity, and therefore, the overall turbulence of the module is reduced. At higher flow rate, the turbulence is maintained throughout the module length, and thereby the permeation is greater for the whole length from the bottom to the top of the membrane. Therefore higher flow rates are desired for CAGMD modules.

Figure 6

Effect of feed flow rate (lpm) and salinity (g/litre) on flux (kg/m2 hr).

Figure 6

Effect of feed flow rate (lpm) and salinity (g/litre) on flux (kg/m2 hr).

EFFECT OF SALINITY

The increase in the salt content of the feed results in a decrease in the flux. Through the surface plots it is also apparent that in the case of the combined effect of salt with feed temperature or with feed flow rate, it diminishes their positive impact on the flux as the salinity values increases. The fact behind this reason is the deposition of a salt layer on the membrane surface and the forming of a concentration boundary layer which hinders the vapor diffusion through the membrane by reducing active membrane pores. In the present case of the cylindrical module, when the water vaporizes and rejects the salt, the salt travels downwards and accumulates near the bottom side. The feed entry is provided at the bottom, which helps to create turbulence near the bottom and breaks this concentration development near the bottom. From Figures 5 and 6 also, it is evident that under the influence of salinity, the effect of feed temperature on flux shows gradual variation, whereas that of the feed flow rate shows abrupt variation. This indicates that for higher salinity feed, the flow rate should always be maintained at a much higher value to counteract the negative effect of salinity and to produce higher flux.

Table 5 shows a comparative study for AGMD by considering different modules. It is obvious that the performance of the present CAGMD module is in the range of literature results and presents a viable choice with design flexibility and lower cost.

Table 5

Comparison of different AGMD modules from the literature and present study

Module configurationMembrane materialFeed flow rate (lpm)Feed/coolant temperature (°C)Air gap width (mm)Feed solutionMaximum flux (kg/m2·hr)Reference
Hollow fiber module Polypropylene 0.4 60/20 2.55 (averaged) NaCl 3.5%-w 4.7 Aryapratama et al. (2016)  
Flat-sheet superhydrophobic condenser surface PVDF 15 60/40 1.5 Saline solution 7.5 Warsinger et al. (2015)  
Finned tubular module PTFE 1.5 50/28 0.55% 10 Cheng et al. (2011)  
Double pipe AGMD PVDF 70 0.45 530 μm 4.8 Liu et al. (2016)  
Cylindrical AGMD PTFE 1.5 60/30 5 (g/l) 3.6 This study 
Module configurationMembrane materialFeed flow rate (lpm)Feed/coolant temperature (°C)Air gap width (mm)Feed solutionMaximum flux (kg/m2·hr)Reference
Hollow fiber module Polypropylene 0.4 60/20 2.55 (averaged) NaCl 3.5%-w 4.7 Aryapratama et al. (2016)  
Flat-sheet superhydrophobic condenser surface PVDF 15 60/40 1.5 Saline solution 7.5 Warsinger et al. (2015)  
Finned tubular module PTFE 1.5 50/28 0.55% 10 Cheng et al. (2011)  
Double pipe AGMD PVDF 70 0.45 530 μm 4.8 Liu et al. (2016)  
Cylindrical AGMD PTFE 1.5 60/30 5 (g/l) 3.6 This study 

RSM was used to optimize the results with the conditions to maximize flux and minimize energy (Figure 7). It is evident that maximum values of flux will be obtained with a maximum feed temperature, feed flow rate and minimum salinity. And it has been mentioned before that the energy is utilized in raising the feed temperature only and flow rate does not primarily affect the total energy consumption in the present case. Thus with the view to maximize the flux while minimizing the energy, optimum values were suggested at 45 °C, 1.5 lpm, and 5 g/litre salinity with a flux of 2.58 kg/m2 hr. Table 6 shows the values of flux and energy at optimum conditions.

Table 6

Results obtained by different methods at optimum conditions

MethodOperating parameter
Flux (kg/m2 hr)Energy (kWh)
Feed temperature T1 (°C)Feed flow rate FF (lpm)Salinity S (g/litre)
Experimentation 45 1.5 2.62 1.70 
Regression model 2.76 1.60 
Response optimizer 2.58 1.61 
MethodOperating parameter
Flux (kg/m2 hr)Energy (kWh)
Feed temperature T1 (°C)Feed flow rate FF (lpm)Salinity S (g/litre)
Experimentation 45 1.5 2.62 1.70 
Regression model 2.76 1.60 
Response optimizer 2.58 1.61 
Figure 7

Optimized plot for maximum flux under optimum conditions.

Figure 7

Optimized plot for maximum flux under optimum conditions.

From Table 6 it is evident that the regression model that gave the equations for flux and energy separately, and the response optimizer that gave the results by simultaneously maximizing the flux and minimizing the energy, are in good agreement. Also the experimentation gave similar results at optimum conditions. Higher weight was given to minimizing the energy consumption in optimization and therefore the value of temperature is low while maximizing the flux. With higher weight to maximization of flux, a very high value of flux will definitely be obtained. Therefore in the systems where high temperature feed is not available, the feed flow rate can be set to a higher value to produce higher flux with lower energy that will ultimately result in lower operating cost.

CONCLUSION

A new design of CAGMD is presented. The idea of gravity-driven feed circulation was presented for minimizing the overall cost of the system, a DOE technique was used and an experimental matrix was developed using the Taguchi method. RSM was used to develop a regression model for analysis and optimization of the CAGMD system. The effect of feed temperature, feed flow rate, and salinity on flux was analyzed. Energy consumption was considered for feed water heating, and it is stated that most of the energy was used for feed water heating, while the pump and other auxiliaries consumed a very minimal amount of energy. Therefore the use of renewable energy sources, like a solar water heater for heating the feed, is encouraged for such systems for better energy economy. Salt rejection for this system is always more than 98%, which ensures that the system can be used to make potable water out of saline feed. At optimum conditions of 45°C, 1.5 lpm, and 5 g/litre salinity, the flux was 2.58 kg/m2 hr, while the maximum flux obtained from the system was 3.6 kg/m2 hr within the operational limits considered. It is shown that when the primary aim is to maximize the flux with minimizing the energy, the combination of lower feed temperature and higher feed flow rate presents the best scheme that produces higher flux with lower energy cost, as the pump consumes only a minimal amount of energy. In the present study a very compact CAGMD system is developed that can be used for purification of sea water, saline or brackish water with minimum cost, and also presents a design flexibility for MD systems. Therefore CAGMD presents an economical and viable solution to all potable water needs.

REFERENCES

REFERENCES
Alkhudhiri
A.
Hilal
N.
2017
Air gap membrane distillation: a detailed study of high saline solution
.
Desalination
403
,
179
186
.
https://doi.org/10.1016/j.desal.2016.07.046
.
Alsalhy
Q. F.
Ibrahim
S. S.
Hashim
F. A.
2018
Experimental and theoretical investigation of air gap membrane distillation process for water desalination
.
Chemical Engineering Research and Design
130
,
95
108
.
https://doi.org/https://doi.org/10.1016/j.cherd.2017.12.013
.
Aryapratama
R.
Koo
H.
Jeong
S.
Lee
S.
2016
Performance evaluation of hollow fiber air gap membrane distillation module with multiple cooling channels
.
Desalination
385
,
58
68
.
https://doi.org/10.1016/j.desal.2016.01.005
.
Attia
H.
Alexander
S.
Wright
C. J.
Hilal
N.
2017
Superhydrophobic electrospun membrane for heavy metals removal by air gap membrane distillation (AGMD)
.
Desalination
420
,
318
329
.
https://doi.org/http://dx.doi.org/10.1016/j.desal.2017.07.022
.
Cheng
L.-H.
Lin
Y.-H.
Chen
J.
2011
Enhanced air gap membrane desalination by novel finned tubular membrane modules
.
Journal of Membrane Science
378
,
398
406
.
https://doi:10.1016/j.memsci.2011.05.030
.
Deshmukh
A.
Boo
C.
Karanikola
V.
Lin
S.
Straub
A. P.
Tong
T.
Warsinger
D. M.
Elimelech
M.
2018
Membrane distillation at the water–energy nexus: limits, opportunities, and challenges
.
Energy and Environmental Science
11
,
1177
1196
.
https://doi.org/10.1039/c8ee00291f
.
Feng
S.
Zhong
Z.
Wang
Y.
Xing
W.
Drioli
E.
2018
Progress and perspectives in PTFE membrane: preparation, modification, and applications
.
Journal of Membrane Science
549
,
332
349
.
https://doi.org/https://doi.org/10.1016/j.memsci.2017.12.032
.
Francis
L.
Ghaffour
N.
Alsaadi
A. A.
Amy
G. L.
2013
Material gap membrane distillation: a new design for water vapor flux enhancement
.
Journal of Membrane Science
448
,
240
247
.
https://doi.org/10.1016/j.memsci.2013.08.013
.
Gao
L.
Zhang
J.
Gray
S.
Li
J.-D.
2019
Influence of PGMD module design on the water productivity and energy efficiency in desalination
.
Desalination
452
,
29
39
.
https://doi.org/10.1016/j.desal.2018.10.005
.
Goh
P. S.
Lau
W. J.
Othman
M. H. D.
Ismail
A. F.
2018
Membrane fouling in desalination and its mitigation strategies
.
Desalination
425
,
130
155
.
https://doi.org/https://doi.org/10.1016/j.desal.2017.10.018
.
Grossi
L. B.
Alvim
C. B.
Alvares
C. M. S.
Martins
M. F.
Amaral
M. C. S.
2020
Purifying surface water contaminated with industrial failure using direct contact membrane distillation
.
Separation and Purification Technology
233
,
116052
.
https://doi.org/10.1016/j.seppur.2019.116052
.
He
Q.
Li
P.
Geng
H.
Zhang
C.
Wang
J.
Chang
H.
2014
Modeling and optimization of air gap membrane distillation system for desalination
.
Desalination
354
,
68
75
.
https://doi.org/https://doi.org/10.1016/j.desal.2014.09.022
.
Khalifa
A. E.
Alawad
S. M.
2018
Air gap and water gap multistage membrane distillation for water desalination
.
Desalination
437
,
175
183
.
https://doi.org/https://doi.org/10.1016/j.desal.2018.03.012
.
Khalifa
A. E.
Lawal
D. U.
2016
Application of response surface and Taguchi optimization techniques to air gap membrane distillation for water desalination – a comparative study
.
Desalination and Water Treatment
57
,
28513
28530
.
https://dx.doi.org/10.1080/19443994.2016.1189850
.
Lee
J.
Alsaadi
A. S.
Ghaffour
N.
2019
Multi-stage air gap membrane distillation reversal for hot impaired quality water treatment: concept and simulation study
.
Desalination
450
,
1
11
.
https://doi.org/10.1016/J.DESAL.2018.10.020
.
Liu
Z.
Gao
Q.
Lu
X.
Zhao
L.
Wu
S.
Ma
Z.
Zhang
H.
2016
Study on the performance of double-pipe air gap membrane distillation module
.
Desalination
396
,
48
56
.
https://doi.org/10.1016/j.desal.2016.04.025
.
Lokare
O. R.
Tavakkoli
S.
Khanna
V.
Vidic
R. D.
2018
Importance of feed recirculation for the overall energy consumption in membrane distillation systems
.
Desalination
428
,
250
254
.
https://doi.org/https://doi.org/10.1016/j.desal.2017.11.037
.
Lu
K. J.
Chen
Y.
Chung
T.-S.
2019
Design of omniphobic interfaces for membrane distillation – a review
.
Water Research
162
,
64
77
.
doi: https://doi.org/10.1016/j.watres.2019.06.056
.
Shahu
V. T.
Thombre
S. B.
2019
Air gap membrane distillation: a review
.
Journal of Renewable and Sustainable Energy
11
,
45901
.
https://doi.org/10.1063/1.5063766
.
Swaminathan
J.
Chung
H. W.
Warsinger
D. M.
AlMarzooqi
F. A.
Arafat
H. A.
Lienhard V
J. H.
2016
Energy efficiency of permeate gap and novel conductive gap membrane distillation
.
Journal of Membrane Science
502
,
171
178
.
https://doi.org/10.1016/j.memsci.2015.12.017
.
Swaminathan
J.
Chung
H. W.
Warsinger
D. M.
Lienhard V
J. H.
2018
Energy efficiency of membrane distillation up to high salinity: evaluating critical system size and optimal membrane thickness
.
Applied Energy
211
,
715
734
.
https://doi.org/https://doi.org/10.1016/j.apenergy.2017.11.043
.
Warsinger
D. E. M.
Swaminathan
J.
Maswadeh
L. A.
Lienhard V
J. H.
2015
Superhydrophobic condenser surfaces for air gap membrane distillation
.
Journal of Membrane Science
492
,
578
587
.
https://doi.org/10.1016/j.memsci.2015.05.067
.
Woo
Y. C.
Tijing
L. D.
Shim
W.-G.
Choi
J.-S.
Kim
S.-H.
He
T.
Drioli
E.
Shon
H. K.
2016
Water desalination using graphene-enhanced electrospun nanofiber membrane via air gap membrane distillation
.
Journal of Membrane Science
520
,
99
110
.
https://doi.org/10.1016/j.memsci.2016.07.049
.
Xu
J.
Singh
Y. B.
Amy
G. L.
Ghaffour
N.
2016
Effect of operating parameters and membrane characteristics on air gap membrane distillation performance for the treatment of highly saline water
.
Journal of Membrane Science
512
,
73
82
.
https://doi.org/10.1016/j.memsci.2016.04.010
.