A novel internal heat recycling concept for reducing energy consumption and increasing flux through three-stage air-gap–water-gap membrane distillation system

AGMD

Air-gap membrane distillation

WGMD

Water-gap membrane distillation

(AG-WG)MD

Air-gap–water-gap membrane distillation

C_{saline water}

Salt concentration of feed saline water (%)

C_{permeate water}

Salt concentration of permeate water (%)

C_pf

Specific heat capacity of hot feed solution (J/kg·°C)

C_pmem.

Specific heat capacity of membrane material (J/kg·°C)

C_mem.

A constant referring to a mass transfer mechanism style inside the membrane (J/kg·°C)

C_{air gap}

Molecular diffusion coefficient within the air gap (J/kg·°C)

C_Pw

Specific heat of water (J/kg·°C)

ESA

Effective evaporation surface area in accordance with internal diameter of the PVDF hollow fiber membrane (m²)

Q_{hot feed, conv.}

The heat transported from the hot feed bulk solution to the membrane surface by convection (kJ/h)

Q_{hot feed, diffus.}

The heat transported from the hot feed bulk solution to the membrane surface by diffusion (kJ/h)

Q_{mem., condu.}

The heat transported through the membrane matrix via conduction (kJ/h)

Q_{mem, latent heat}

The heat transported through the membrane matrix via evaporation latent heat (kJ/h)

Q_{mem, sensible heat}

The heat transported through the membrane matrix via sensible heat (kJ/h)

Q_{film, condu.}

The heat transported through the condensate film thickness via conduction (kJ/h)

Q_{film, conv.}

The heat transported through the condensate film thickness via convection (kJ/h)

Q_{surface, condu.}

The heat transported by conduction through condensing surface thickness (kJ/h)

Q_{cold feed, conv.}

Convective heat transfer in the bulk cold fluid (kJ/h)

Q_{l.h., permeate water}

Latent heat of vapor flux (kJ/h)

Q_h.i

Specific waste heat input (MJ/kg)

h_{hot feed}

Convective heat transfer coefficient through the hot feed (W/(m²·°C))

h_mem.

Heat transfer coefficient through the membrane (W/(m²·°C))

h_{water gap}

Heat transfer coefficient within the water-gap (W/(m²·°C))

h_film

Heat transfer coefficient through the condensate film (W/(m²·°C))

h_{cold feed}

Convective heat transfer coefficient through the cold feed (W/(m²·°C))

GOR

Gained output ratio (dimensionless)

K_mem.

Membrane thermal conductivity (W/(m·°C))

K_{air gap}

Air thermal conductivity (W/(m·°C))

K_film

Thermal conductivity of condensate film (W/(m·°C))

K_surface

Thermal conductivity of condensing surface (W/(m·°C))

M_f

Feed flow rate (l/h)

M_m

Water molar mass (≈18 g/mol)

P_v,feed

Partial vapor pressure at feed side (MPa)

P_{v,air gap}

Partial vapor pressure at air-gap side (MPa)

SRF

Salt rejection factor (%)

STEC

Specific thermal energy consumption (kWh/kg)

t

Time (minute)

T_f

Hot bulk feed temperature (°C)

T_fm

Hot feed temperature at the membrane surface (°C)

T_fm,1

Temperature of the membrane surface from the air-gap side (°C)

T_pr

Temperature of the condensate film from the air-gap side (°C)

T_c,1

Temperature of the condensate film near to condensing surface (°C)

T_c,2

Temperature of the condensing surface from the cold feed side (°C)

T_c

Cold feed bulk temperature (°C)

ΔT_cross

Temperature difference between the evaporator inlet (T_evap.,inlet) and the condenser outlet (T_cond.,outlet) (°C)

ΔH_V

Evaporative latent heat (≈2,326 kJ/kg)

F_wv

Water vapor flux (kg/(m²·h))

F_{WV, air gap}

Mass vapor transported by diffusion style within the air gap (kg/(m²·h))

F_{WV, water gap}

Mass vapor transported by diffusion style within the water gap (kg/(m²·h))

δ_mem.

Membrane thickness (m)

δ_{air gap}

Thickness of air gap (m)

δ_film

Thickness of condensate film (m)

δ_surface

Thickness of condensing surface (m)

φ

Ideal gas constant (≈8.3145 J/mol·K)

W_wv

Weight of distilled water (kg)

h_universal

Universal membrane coefficient (kg/(m²·h)·mbar)

h_d

Diffusive coefficient through membrane (kg/(m² h)·mbar)

h_c

Free convective coefficient through air gap (kg/(m²·h)·mbar)

Membrane distillation (MD) is a thermally driven separation process that has emerged as a motivating alternative to the current prevalent desalination systems like reverse osmosis (RO) and multi-stage flash (MSF). Some researchers (e.g.Hickenbottom & Cath 2014; Lee et al. 2016) used MD as a supplementary process attached squarely to the RO and MSF plants due to the lesser footprint demand, low sensitivity to heavy salt concentrations, and potential working at small-scale applications. The quantity of pure distilled water obtained from seawater or brackish water using MD primarily relies upon the available temperature difference between the warm fluid and the cold fluid which affects profoundly the generated vapor pressure difference (i.e., permeation driving force). The produced vapor molecules are transmitted from the interfaces of liquid/vapor at the feed side and then dry membrane pores to be condensed and converted at the end into pure distilled water on the exterior cooling surface at the membrane permeate side (Martínez & Rodríguez-Maroto 2008; Zhao et al. 2013; Lu et al. 2017). Air-gap membrane distillation (AGMD) and water-gap membrane distillation (WGMD) are two of the most common MD processes as well as material gap (MGMD), direct contact (DCMD), and vacuum membrane distillation (VMD) (Zhang et al. 2009; Francis et al. 2013; Khalifa et al. 2014; Khalifa 2015; Sivakumar et al. 2015; Ebadi et al. 2019; Shahu & Thombre 2020). AGMD and WGMD processes are distinguished by the capability of yielding distilled water with zero salts (Cath et al. 2004; El-Bourawi et al. 2006; Khayet & Matsuura 2011) using high intensity salt concentration solution (Khayet & Matsuura 2011; Xu et al. 2016) under low operating pressure (Laganà et al. 2000; Alkhudhiri et al. 2012) and temperature (El-Bourawi et al. 2006; Alkhudhiri et al. 2012). Further, the AGMD process has two more interesting characteristics that have given it advanced standing for salt fluid desalination through pilot investigation systems (Schwantes et al. 2013; Alsaadi et al. 2015). Firstly, latent heat of condensation could be completely recovered inside the membrane module without wanting a heat exchanger/vacuum pump as in sweeping gas (SGMD)/vacuum membrane distillation (VMD) (Guijt et al. 2005; Koschikowski et al. 2009; Khayet & Matsuura 2011; Alsaadi et al. 2013,, 2015; Duong et al. 2015; Xu et al. 2016). Secondly, the thermal insulation area constituted by the air gap between the membrane and the cooling plate minimized deeply the heat loss by conduction through the membrane matrix, leading to appreciable improvement in AGMD thermal efficiency in comparison with DCMD. But some investigators (e.g. Minier-Matar et al. 2014; Wang & Chung 2015) mentioned this area imposes additional mass transfer impedance, leading to flux impoverishment as well as feed side thermal-concentration boundary layers, low membrane operation stability, membrane fouling, and wetting, finally reducing the chance of investment extensively (Alkhudhiri et al. 2012; Cipollina et al. 2012; Hawlader et al. 2012; Saffarini et al. 2012; Alsaadi et al. 2013; Minier-Matar et al. 2014; Khalifa et al. 2015).

The multi-stage system accompanied by internal heat recovery has been endorsed as an achievable technique to enrich the vapor flux and decrease the thermal energy consumption which has already been applied successfully in several experimental investigations (e.g.Hogan et al. 1991; Ding et al. 2005; Guillén-Burrieza et al. 2011; Zhao et al. 2013; Lee & Kim 2014; Shim et al. 2014; González-Bravo et al. 2015; Kim et al. 2015; Abu-Zeid & ElMasry 2020). Furthermore, Khalifa et al. (2017) reported that the water flux system of a multi-stage AGMD connected in series and parallel was maximal about 2.6–3.0 times more than the water flux system composing of only one stage. Likewise, the vapor flux module of a multi-stage AGMD was greater compared with the vapor flux produced from a one-stage system (Geng et al. 2015). Yao et al. (2012) proclaimed thermal efficiency of 80% and a performance ratio of 13.8 by virtue of a continuous-effect AGMD module designed with internal heat recovery. Pangarkar & Deshmukh (2015) improved profoundly the permeation flux of the AGMD system composed of four stages prepared specifically for a water treatment process, reaching up to 166.4 kg/(m²·h). In another work, a zero-liquid discharge was fulfilled through a successful combination between a multi-stage flat sheet AGMD module and an evaporative crystallizer (Guo et al. 2016). Lawal & Khalifa (2015) stated that the performance of an AGMD system composed of two stages (128.46 kg/(m²·h) flux) was better compared with a one-stage AGMD system (65.81 kg/(m²·h) flux) at different operating conditions. Later, Lee et al. (2019) reported an exciting value for gained output ratio (GOR) reaching up to 24.4 through a fruitful connection between a multi-stage AGMD reversal design and a system of natural/forced cooling corresponding to only 3.89 without a natural cooling system.

However, despite the important role of the multi-stage system in vapor flux enrichment and energy consumption reduction, the quantity of pure distilled water obtained from each stage is inversely proportional to the number of incorporated stages because of the significant temperature drop and feed salinity increase which affects profoundly the temperature difference across the membrane, heat and mass transport; especially when a highly saline water is used as feed. Thence, the present research work focused on achieving the following aims:

• 1.

  Upgrade the system performance of three-stage air-gap membrane distillation (AGMD) by incorporation of it with the water-gap module consecutively.

• 2.

  Carry out a comparative study between the systems of three-stage and two-stage air-gap–water-gap membrane distillation (AG-WG)MD with the system of one-stage AGMD respecting vapor flux (F_wv) and gained output ratio (GOR).

• 3.

  Analysis of heat and mass transfer mechanism theoretically for the systems of the three-stage (AG-WG)MD and AGMD.

The key idea of inserting the water-gap module into the system of the three-stage AGMD is to take advantage of water-gap characteristics (e.g. efficient internal heat recovery, low impedance to mass transfer, and less heat loss through the permeate water) in decreasing the thermal-concentration feed side boundary layer thickness, increasing the trans-membrane temperature difference, and stimulating the heat and mass transport through the integrated AGMD stages.

To implement this idea in a laboratory, a system of three-stage AGMD composed of three similar air-gap MD modules was installed and tested experimentally. After that, the obtained results from the system of three-stage AGMD were compared with the results of one-stage AGMD, two-stage (AG-WG)MD, and three-stage (AG-WG)MD systems. The last two systems are composing of one water-gap module fixed mutually at different positions before, between, and after the air-gap modules. A set of equations were used to analysis in theory the heat and mass transport through various three-stage MD systems with and without the water-gap module. The experimental investigations were performed at different feed temperatures, feed flow rates, and feed salt concentrations. The performances of different three-stage, two-stage, and one-stage systems were evaluated respecting vapor flux (F_wv), gained output ratio (GOR), thermal energy consumption (STEC), and waste heat input (Q_h.i). The membrane modules of the air gap and water gap were designed with internal heat recovery.

The heat and mass transfer process within the membrane distillation (MD) module are executed concurrently with a continual flowing of warm and cold fluids in a counter-current direction at the hot feed and the cold permeate sides. As elaborated schematically in Figures 1 and 2, the heat and mass transfer within the presently studied air-gap and water-gap membrane modules are done via convection, conduction, and evaporation.

The heat transfer across the water-gap module is similar to the heat transfer across the air-gap module excepting the heat transfer within the air-gap area which is replaced by heat transfer via evaporation and convection through the permeate water.

In view of no/weak impedance to mass transfer during the WGMD process, the vapor quantity transferred across the membrane of the water gap (F_{WV,water gap}) is much higher than that in the air gap (F_{WV,air gap}). Additionally, the heat transfer by convection across the water gap is faster by 24.17 times than the heat transfer by conduction through the air gap (i.e., the thermal conductivity of water is 0.58 W/m·K and the thermal conductivity of air is 0.024 W/m·K). This behaviour can be viewed obviously in Figure 3, which illustrates the formation of thermal-concentration feed side boundary layers during the system of three-stage AGMD with a low temperature difference across the membrane, which profoundly influences the heat and mass transport and the vapor flux correspondingly. In contrast, Figure 4 elucidates no thermal-concentration boundary layers at the feed side during the system of three-stage air-gap–water-gap (AG-WG)MD with a high temperature difference across the membrane due to the function of the water-gap module in making a tangible change in the heat and mass transport and enriching the flux.

Sketches of the three-stage AGMD and (AG-WG)MD systems installed in a vertical direction are elucidated in Figure 5(a)–5(d). The three-stage AGMD symbolized as (AG-AG-AG)MD is used as a reference system for comparison with different systems of three-stage (AG-WG)MD symbolized as (WG-AG-AG)MD, (AG-WG-AG)MD, and (AG-AG-WG)MD. The sketches of the two-stage (AG-WG)MD and one-stage AGMD systems are not included here as they are similar to current systems excepting only the number of incorporated stages. The systems consist of feeding tank, membrane modules, circulation pump MP-55RZ (Shanghai Xinxishan Industrial Limited Company, China), rotameter LZB-4 (Huanming, Yugao Industrial Automation Instrument Company, Zhejiang, China), valves, thermostatic water bath CS-501 (Tongzhou Branch of Shanghai Jinping Instrument Limited Company, China), and coolant DLSB-10 (Tianjin Xingke Instrument Limited Company, China).

The membrane modules prepared by our research team are manufactured from plexiglas and enclosed externally by thermal insulation to minimize the evaporation losses to the environment. The modules are designed to use polyvinylidene difluoride (PVDF) hollow fiber membranes with total effective area of 1.08 m² (0.36 m² each membrane module) and thickness of 150 μm, average pore size of 0.20 μm, and average porosity of 85%. The channels for evaporating and condensing in each module have 120 PVDF hollow fibers (0.18 m² interior membrane surface area) and 240 polypropylene (PP) acts as a heat exchange tube (0.18 m² interior membrane surface area). Inner/outer diameter (m × 10⁻³) of hollow fiber membranes (i.e., evaporating channels) and heat exchange tubes (i.e., condensing channels) are about 0.80/1.10 and 0.40/0.50 respectively. The total and effective length of hollow fiber membrane and heat exchange tube are around 0.77 m and 0.59 m, respectively. The gap width employed in the air-gap and water-gap modules is ≈5 mm.

As illustrated in Figure 5(a), the feed solution was pumped from the feeding tank to the evaporator inlet channels in stages 1, 2, and 3, respectively by a circulation pump in the orientation from up to down. After the hot feed solution leaving the evaporator inlet channel in stage 3 flows into the coolant for temperature decrease, it then enters the condenser inlet channel in stages 3, 2, and 1, respectively as cold feed in a counter-current flow and then goes back to the feeding tank for a new circulation. To keep the feed volume and salt concentration constant inside the feeding tank throughout the experiment, the collected pure distilled water is returned to the feeding tank.

The performance of the systems was examined at different hot feed temperatures of 45 °C, 55 °C, 65 °C, and 75 °C, feed salt concentrations of 340 ppm, 7,500 ppm, 15,000 ppm, and 225,000 ppm, and feed flow rates of 15 l/h, 20 l/h, 25 l/h, and 30 l/h (flow velocities of 0.069 m/s, 0.092 m/s, 0.115 m/s, and 0.138 m/s and Reynolds numbers of 55.09, 73.45, 91.82, and 110.18, respectively). The temperature at the cold permeate side was kept stable at 20 °C. To make the obtained data more accurate and credible, each experiment was implemented three times under the same operating conditions for one hour operation time and average values are given. Prior to initiating the experiment, the laboratory-scale of various three-stage MD systems is left running for one hour to guarantee no dissolved gases in the feed stream, the gap area filled completely with permeate water, and reaching a steady-state.

In view of the difficulty of measuring the interface temperature at the membrane surface, the temperatures through and between each membrane module were continuously watched by several temperature sensors fixed at the inlets and outlets of the evaporator and the condenser channels and mean temperature differences are reported. The temperature controller XMTD-3001 (Easey Commercial Building, Hennessy Road, Wanchai, Hongkong, China) was used to control the hot inlet feed temperature. The vapor pressures of P_v,feed and P_{v,air gap} at feed and air-gap sides were viewed continuously using a manometer.

The obtained pure distilled water was weighed by electronic balance every ten minutes to note the variation of vapor flux. The electrical conductivity of the distilled water and saline feed were checked throughout using conductivity meter DDS-11A (Shanghai Leici Xinjing Instrument Company, China) to ensure the safety of the PVDF hollow fiber membrane and the quality of the pure distilled water.

The change in the system performance was evaluated by estimation of GOR, F_wv, SRF, Q_h.i, and STEC. The values of specific heat (C_Pw) and water density (ρ_w) utilized in the determination were obtained at ambient air temperature of 25 °C and are listed in Table 1. The experimental vapor flux (F_wv) with data errors are calculated and given in Tables 2–4.

The effect of feed temperature on vapor flux (F_wv) for the three-stage (AG-WG)MD and three-stage AGMD systems is elaborated in Figure 6(a). Feed temperature was changed between 45 °C and 75 °C at 5 °C intervals while flow rate, cooling water temperature, and feed salt concentration were kept at 20 l/h, 20 °C, and 340 ppm, respectively. Exponential flux increase with temperature that occurred in different multi-stage systems is attributed to the corresponding exponential vapor pressure increase. As shown in Figure 6(a), vapor flux for different systems of three-stage (AG-WG)MD was larger than that in the three-stage AGMD system. The flux was raised by 8.49%, 10.17%, 13.83%, 7.64%; 17.92%, 18.22%, 18.58%, 13.54%; and 33.49%, 27.12%, 24.51%, 17.36% at 45 °C, 55 °C, 65 °C, and 75 °C respectively. The improved flux is ascribed to the low effective air-gap and high driving force thanks to the incorporated water-gap while in the system of three-stage AGMD the air-gap impeded vapor mass transfer and the temperature decrease per stage leading to flux reduction.

In comparison with other systems like two-stage (AG-WG)MD and one-stage AGMD, the three-stage (AG-WG)MD systems enhanced the flux around 17.59%, 12.90%, 12.31%, 13.64%, and 211.69%, 204.35%, 138.89%, 118.12% at 45 °C, 55 °C, 65 °C, and 75 °C, respectively. Similarly, the three-stage AGMD system gave extreme improvement in flux of about 32.50%, 6.79%, 8.12%, 17.55%, and 175.32%, 156.52%, 100.79%, 93.29% compared with other two-stage AGMD and one-stage AGMD systems. As for salt rejection factor (SRF), different quantities of pure distilled water were produced from different systems, and the employed PVDF porous membrane demonstrated high SRF reaching up to 99.7%.

As presented in Figure 6(b), various systems of three-stage (AG-WG)MD attained higher GOR than that of the three-stage AGMD system, reaching up to 13.71%, 15.29%, 19.45%, 9.54%; 28.17%, 28.52%, 26.22%, 17.85%; and 35.02%, 28.18%, 26.22%, 18.58%. The high GOR related to conductive heat loss reduction and trans-membrane temperature difference (ΔT_cross) increased by 4.54%, 4.42%, 4.70%, 1.53%; 7.95%, 7.96%, 6.04%, 3.57%; and 1.14%, 0.88%, 1.34%, 1.02%. On the other side, the low GOR in the case of the three-stage AGMD system was attributed to the small amount of thermal energy utilized to produce vapor relative to the energy used for warming the inlet feed stream. Additionally, poor cooling of the air gap compared with the water gap was because of the lower specific heat capacity of air by 4.23 times than that of water (i.e., C_Pw of air = 993 J/(kg·°C) and C_Pw of water = 4,200 J/(kg·°C)).

In comparison with the two-stage (AG-WG)MD and one-stage AGMD systems, GOR for the three-stage (AG-WG)MD systems was increased profoundly by 60.57%, 42.69%, 19.11%, 2.16%, and 204.33%, 172.45%, 162.78%, 136%. In like manner, the system of three-stage AGMD enhanced GOR about 31.83%, 17.81%, 2.60%, 3.54%, and 142.24%, 119.62%, 112.11%, 104.50% compared with two-stage AGMD and one-stage AGMD systems. The percentage increases in GOR for the three-stage (AG-WG)MD systems were greater compared with the three-stage AGMD system except at a temperature of 75 °C for two-stage (AG-WG)MD and two-stage AGMD systems.

The main cause behind GOR enhancement in the case of the three-stage (AG-WG)MD systems was an efficient internal heat recycling introduced by the integrated water-gap module, leading to thermal energy reduction. Vapor flux and thermal power consumption over experimental time were utilized to realize the STEC using Equation (16). As can be noticed from Figure 6(c), the various systems of three-stage (AG-WG)MD consumed less thermal energy compared with the three-stage AGMD system. The STEC values read for the (WG-AG-AG)MD, (AG-WG-AG)MD, and (AG-AG-WG)MD systems were 36.81–46.55 kWh/kg, 30.04–40.97 kWh/kg, and 25.18–39.36 kWh/kg, respectively while the corresponding value in the (AG-AG-AG)MD system was 45.39–54.77 kWh/kg. Identically, the Q_h.i values were 0.31–0.52 MJ/kg, 0.27–0.48 MJ/kg, and 0.26–0.48 MJ/kg, and the corresponding value in the case of (AG-AG-AG)MD was 0.35–0.57 MJ/kg (Figure 6(d)).

Figure 7(a) demonstrates the quantity of vapor flux resulted from the systems of three-stage (AG-WG)MD and three-stage AGMD as a function of feed flow rate at constant feed temperature of 55 °C, cooling water temperature of 20 °C, and feed salt concentration of 340 ppm. As can be noticed from Figure 7(a), the flux quantity of the various three-stage (AG-WG)MD systems was maximal compared with the three-stage AGMD system. Compared with the system of (AG-AG-AG)MD, the systems of (WG-AG-AG)MD, (AG-WG-AG)MD, and (AG-AG-WG)MD boosted vapor flux by about 5.53%, 10.17%, 8.65%, 7.74%; 22.58%, 18.22%, 12.78%, 10.10%; and 29.95%, 27.12%, 22.93%, 17.85% at 15 l/h, 20 l/h, 25 l/h, and 30 l/h, respectively. The possible reasons for high flux were assigned to an obvious temperature-concentration polarization reduction and low impedance to vapor molecules condensing instantly at the membrane/water-gap interface.

The three-stage (AG-WG)MD systems developed flux by 38.50%, 12.90%, 2.69%, 1.53%, and 223.75%, 204.35%, 127.61%, 124.32% compared with the other two-stage (AG-WG)MD and one-stage AGMD systems, respectively. Also, the three-stage AGMD systems enhanced flux remarkably by 82.35%, 6.79%, 4.72%, 4.58%, and 171.25%, 156.52%, 98.51%, 100.78% compared with the other two-stage AGMD and one-stage AGMD, respectively. The fabricated hydrophobic PVDF hollow fiber membrane used in the experiment rejected around 99.7% of salts at different feed flow rates.

Figure 7(b) advertises that the various three-stage (AG-WG)MD systems performed with better GOR than the three-stage AGMD system, reaching up to 7.79%, 15.29%, 15.17%, 18.02%; 27.90%, 28.52%, 22.52%, 26.66%; and 30.98%, 28.18%, 25.83%, 21.40%. The enhancement of GOR was associated with the following factors: firstly, changing the mass and heat transfer significantly by virtue of permeated water filling the gap; secondly, increasing the amount of heat absorbed through the cold feed stream and decreasing the conductive heat loss across the membrane; thirdly, raising the ΔT_cross by 2.05%, 4.42%, 5.62%, 8.70%; 4.11%, 7.96%, 7.87%, 13.04%; and 0.68%, 0.88%, 2.25%, 2.90%.

The three-stage (AG-WG)MD systems increased GOR by about 35.27%, 42.69%, 33.17%, 37.91%, and 197.36%, 172.45%, 175.43%, 204.69% compared with the other two-stage (AG-WG)MD and one-stage AGMD systems. Also, the GOR of the three-stage AGMD systems was enhanced appreciably by 34.96%, 17.81%, 27.83%, 34.29%, and 143.17%, 119.62%, 127.30%, 149.69% compared with the other two-stage AGMD and one-stage AGMD systems, respectively. This result could be explained by the effective role of the water-gap module in reducing the thermal-concentration boundary layer thickness at the membrane surface. Consequently, increasing the GOR economized the energy required for heating and decreased extremely the waste heat input into the system as illustrated in Figure 7(c) and 7(d).

As seen from Figure 7(c), the absence of the water-gap module negatively affected the heat recovery in the (AG-AG-AG)MD system, leading to high STEC. High recorded STEC were between 27.20 and 53.90 kWh/kg corresponding to low STEC in the (WG-AG-AG)MD, (AG-WG-AG)MD, and (AG-AG-WG)MD systems around 21.39–47.41 kWh/kg, 19.51–34.40 kWh/kg, and 19.02–31.70 kWh/kg, respectively. Likewise, the Q_h.i in the (AG-AG-AG)MD system were in the range of 0.29–0.42 MJ/kg corresponding to 0.25–0.39 MJ/kg, 0.23–0.33 MJ/kg, and 0.24–0.32 MJ/kg (Figure 7(d)).

The variation of flux with feed salt concentration for the systems of three-stage (AG-WG)MD and three-stage AGMD is characterized in Figure 8(a). Experiments were conducted at a constant temperature of 55 °C, cooling water temperature of 20 °C, and feed flow rate of 25 l/h. A remarkable flux decline with salt concentration was due to parallel vapor pressure decrease at the feed side for the investigated systems. As can be perceived from Figure 8(a), the maximal flux of the three-stage (AG-WG)MD systems was imputed to low concentration polarization influence by virtue of the incorporated water-gap module. The (WG-AG-AG)MD, (AG-WG-AG)MD, and (AG-AG-WG)MD systems improved flux by 8.65%, 3.60%, 4.33%, 9.05%; 12.78%, 10.80%, 12.55%, 15.24%; and 22.93%, 20%, 22.08%, 23.33% compared with the (AG-AG-AG)MD system at 340 ppm, 7,500 ppm, 15,000 ppm, and 22,500 ppm, respectively.

Respecting the flux comparison, three-stage (AG-WG)MD boosted the system flux by 2.69%, 3.72%, 5.24%, 10.96% and 127.61%, 138.46%, 161%, 179.31% compared with the other systems of two-stage (AG-WG)MD and one-stage AGMD. Also, the three-stage AGMD system upgraded the flux by 4.72%, 9.65%, 10%, 9.95% and 98.51%, 113.68%, 131%, 141.38% as compared with the other two-stage AGMD and one-stage AGMD. For the salt rejection factor (SRF), increased feed salt concentration from 340 ppm to 22,500 ppm resulting in a tenuous decrease in the separation efficiency of the PVDF hollow fiber membrane ranging from 99.7% to 99.5%.

As shown in Figure 8(b), increases of 15.17%, 8.39%, 10.84%, 13.67%; 22.52%, 17.27%, 21.88%, 24.13%; and 25.83%, 22.70%, 25.77%, 25.47% were accomplished in GOR for the (WG-AG-AG)MD, (AG-WG-AG)MD, (AG-AG-WG)MD systems compared with the (AG-AG-AG)MD system. This result enhanced ΔT_cross by 5.62%, 4.35%, 5.66%, 3.94%; 7.87%, 5.43%, 7.55%, 7.09%; and 2.25%, 2.17%, 2.83%, 1.57%. In comparison with the two-stage (AG-WG)MD and one-stage AGMD systems, the three-stage (AG-WG)MD systems increased GOR by 33.17%, 35.25%, 21.16%, 2.96% and 175.43%, 168.44%, 171.63%, 255.91%, respectively. Similarly, the three-stage AGMD system promoted GOR about 27.83%, 24.59%, 8.67%, 3.90% and 127.30%, 131.18%, 127.44%, 193.70% compared with the two-stage AGMD and one-stage AGMD.

In connection with the STEC in the case of the three-stage AGMD system higher values were registered compared with the three-stage (AG-WG)MD. The (AG-AG-AG)MD system recorded STEC between 36.82 and 82.43 kWh/kg corresponding to 29.14–66.59 kWh/kg, 26.40–57.68 kWh/kg, and 23.57–53.34 kWh/kg in (WG-AG-AG)MD, (AG-WG-AG)MD, and (AG-AG-WG)MD, respectively (Figure 8(c)). Likewise, the Q_h.i of the (AG-AG-AG)MD system was between 0.35 and 0.62 MJ/kg corresponding to 0.30–0.55 MJ/kg, 0.29–0.50 MJ/kg, and 0.28–0.50 MJ/kg in (WG-AG-AG), (AG-WG-AG), and (AG-AG-WG)MD (Figure 8(d)).

Under all the above-mentioned studied operating conditions (feed temperature, flow rate, and salt concentration), the GOR values obtained for the conventional multi-stage AGMD system which operated without a water-gap module are low, while the GOR values for the novel three-stage (AG-AG-WG)MD system operated with a water-gap module are high. As well, the STEC considered in this experiment refers to the heat that is continuously replenished by the heater during the circulation. Due to efficient internal heat recovery unlike the conventional MD systems, the circulating cold feed will absorb a part of the heat from the steam condensation, which results in a reduction in the heat (energy consumption) provided by the external heater during the cycle, thereby achieving a high GOR and vapor flux for the novel three-stage (AG-AG-WG)MD system.

Table 5 shows a vapor flux comparison between the novel three-stage (AG-AG-WG)MD system investigated in this study and other conventional multi-stage AGMD systems reported in the the published literature. In this study, the novel (AG-AG-WG)MD system was assessed by comparing it with conventional (AG-AG-AG)MD systems. It was found that the system of (AG-AG-WG)MD performed better compared with (AG-AG-AG)MD using the same PVDF hollow fiber membrane under the same operating parameters. However, variations in flux were observed when our novel (AG-AG-WG)MD system was compared with other experimental works because of a variance in different operating conditions and membrane characteristics as well. As listed in Table 5, these variations result from decreasing the air gap width (δ_gap = 4 mm) and increasing the pore size (0.45 μm) (Khalifa & Alawad 2018). Additionally, increase of feed temperature and flow rate, decrease of membrane thickness, coolant temperature, and feed salt concentration promoted the flux as revealed by some authors (e.g. Bahar et al. 2014; Geng et al. 2015). In conclusion, different operating conditions and membrane characteristics induced various fluxes as seen in Table 5. Therefore, if all multi-stage MD systems examined used the same membrane under the same operating parameters, our novel system will be outperformed.

In this research work, the water-gap module improved profoundly the performance of the three-stage air-gap membrane distillation (AGMD) system. The new three-stage air-gap–water-gap membrane distillation (AG-WG)MD system reduced deeply the energy consumption and increased the vapor flux due to efficient internal heat recycling introduced by the water-gap module. At optimal operating conditions involving a hot feed temperature of 45 °C, feed flow rate of 20 l/h, cooling water temperature of 20 °C, and feed salt concentration of 340 ppm, the system of three-stage (AG-WG)MD promoted the flux by 17.59% and 211.69%, and gained output ratio (GOR) by 60.57% and 204.33% compared with other systems of two-stage (AG-WG)MD and one-stage AGMD, respectively. The salt rejection factor (SRF) registered was on average about 99.5% at highest salt concentration. In the present research work emerged experimentally the effective role of the water-gap module in reducing the boundary layer thickness and improve the flux. Consequently, increasing the flux and GOR economized mightily the energy required for heating and decreased waste heat input into the system. Also, the number of stages could be increased with high flux and more operation stability.

Supported by the Program for Innovative Research Team in University of Tianjin (No. TD13-5044). Supported by the Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT) of Ministry of Education of China (Grand no. IRT17_R80). National Center for International Joint Research on Separation Membranes, Tianjin Polytechnic University, Tianjin 300387, China.

All relevant data are included in the paper or its Supplementary Information.