Experimental and theoretical study of a lab scale permeate gap membrane distillation setup for desalination

Desalination 419 (2017) 197–210

Contents lists available at ScienceDirect

Desalination journal homepage: www.elsevier.com/locate/desal

Experimental and theoretical study of a lab scale permeate gap membrane distillation setup for desalination

MARK

Farzaneh Mahmoudia,⁎, Gholamreza Moazami Goodarzia, Saeed Dehghania, Aliakbar Akbarzadeha a

Energy CARE Group, School of Aerospace, Manufacturing and Mechanical Engineering, RMIT University, Melbourne 3083, VIC, Australia

A R T I C L E I N F O Keywords: Sustainable desalination Permeate gap membrane distillation Flat module design Specific thermal energy consumption Permeate flux rate

Abstract Membrane distillation (MD) as a novel thermally-driven process with moderate operating temperatures is a known effective technology for salt-water desalination. In this research, a lab scale plate-and-frame permeate gap membrane distillation (PGMD) module with internal heat recovery characteristic is designed. The developed PGMD module performance is experimentally investigated for fresh and saline water feed in terms of permeate water flux rate, specific thermal energy consumption (STEC) and gained output ratio (GOR). The experimental results show that for a feed sample with 130 (g/kg: ppt) concentration (nearly four times seawater salinity), increasing the feed flow rate from approximately 0.4 to 1 L/min, led to increasing the distillate flux from 3 to 5 kg/m2 h. However, increasing the feed flow rate in this range also led to approximately 40% increase in the STEC of the system. Furthermore, a single node theoretical model based on the PGMD module configuration is developed and the modelling results validated with experimental values at different feed water flow rate and salinity. The comparison shows a good agreement between the developed model results and experimental outcomes. It is also concluded, optimization of the MD module performance to improve internal heat recovery and produce higher fresh water rate would be achievable by adjusting the effective membrane surface area and feed flow rate.

1. Introduction The need for fresh water is considered to be a critical international problem and according to the World Water Council, 17% of the world population will be living in short of the fresh water supply by 2020 [5]. Consequently, the demand for alternative sustainable water sources including ground water, desalinated water and recycled water increased in recent years and as a result, the implementation of desalination plants is growing on a large scale. Fresh water can be derived from sea water by evaporation processes e.g., multi-stage flash (MSF), multi- effect distillation (MED) or membrane based processes such as reverse osmosis (RO), electro dialysis (ED) and membrane distillation (MD). Membrane distillation is a separation process which involves phase change (liquid-vapour equilibrium) across a hydrophobic, highly porous membrane. In contrast to most membrane separation processes, which are isothermal and have driving forces as trans membrane hydrostatic pressures, concentrations, electrical or chemical potentials, MD is a non-isothermal process. If a temperature difference occurs

⁎

Corresponding author. E-mail address: [email protected] (F. Mahmoudi).

http://dx.doi.org/10.1016/j.desal.2017.06.013 Received 21 March 2017; Received in revised form 11 June 2017; Accepted 11 June 2017 0011-9164/ © 2017 Elsevier B.V. All rights reserved.

across a non-wetting membrane, the created partial vapour pressure difference as a driving force, leads to water molecules evaporating at the hot side, crossing the membrane in the vapour phase and condensing at the cold side. Commercially developed RO technology is associated with high electrical energy consumption in the range of (6–12) kwh/m3 with the electricity currently being generated from non-renewable and polluting fossil fuels [3]. In contrast, MD is a thermal process using lower top temperature (80 °C or less) compared with the traditional thermal desalination processes such as MSF and MED, making it suitable for using waste heat or solar heat. Table 1 provides a comparison between the most developed current desalination technologies in terms of STEC, specific electrical energy consumption (SEEC) and operating temperature. In addition, an advantage of the MD process [4] is that aqueous solutions of salts with higher concentrations than seawater can be treated by MD, reducing discharge volumes and increasing the water recovery factor up to 95% which considerably diminishes the environmental impact of the brine disposal.

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

Nomenclature

qSTEC q̇ C q̇ cold q̇ E q̇ hot q̇ M q̇ M,C q̇ M,L ReE S SCi SEi SEo

Symbol Am Cm CpC CpE Dh Ei f GOR hfg hC hE hF hPG hM Jp KE Km Kpg L ls ṁf ṁCi ṁEi ṁEo ṁPG Pv,w Pv,sw PrE

membrane surface area (m2) membrane mass transfer coefficient (kg/(Pam2s)) condenser channel specific heat capacity (J/kgK) evaporator channel specific heat capacity (J/kgK) hydraulic diameter (m) total power input (W) friction factor (−) gained output ratio (−) specific heat of vaporization (J/kg) heat transfer coefficient at the condenser channel (W/ m2K) heat transfer coefficient at the evaporator channel (W/ m2K) heat transfer coefficient at the impermeable polymeric film (W/m2K) heat transfer coefficient at the permeate gap (W/m2K) heat transfer coefficient at the membrane (W/m2K) permeate flux (kg/m2s) evaporator channel thermal conductivity (W/mK) membrane thermal conductivity (W/mK) permeate gap thermal conductivity (W/mK) module length (m) orthogonal distance between net spacer filament (m) feed flow rate (kg/s) condenser channel inlet mass flow rate (kg/s) evaporator channel inlet mass flow rate (kg/s) evaporator channel outlet mass flow rate (kg/s) permeate output rate (kg/s) pure water vapour pressure (Pa) saltwater vapour pressure (Pa) Prandtl number in evaporator channel (−)

TCi TCo TEi TEo TMe TMp TPG um V̇ f V̇ p α β ?? ??F δm δPG ΔP ρ

MD configurations, including: a) DCMD); b) air gap membrane distillation (AGMD); c) vacuum membrane distillation (VMD); d) sweeping gas membrane distillation (SGMD); PGMD or liquid gap membrane distillation (LGMD) is a recently introduced configuration of DCMD, which the permeate is extracted from the highest module position, so that the gap between the membrane and the impermeable film fills with permeate during the operation. Some recent research works studied around different MD configuration and compared the developed MD configuration in terms of main output factors. Furthermore, the energy source of the MD process is an important issue for commercialization of this technology as a sustainable process. Membrane distillation associated with renewable energy is considered to be a highly promising process, especially for situations where lowtemperature solar, waste or other heat is available. The STEC of MD systems varies based on the module configuration, setup scale and operating condition. A wide dispersion of reported values is observed in the literature for STEC based on different MD configurations, with the STEC varying in a range of (1–9000) kWh/m3. Moreover, the energy consumption of a small scale installation is much higher than for pilot plants with higher effective membrane surface areas [14]. Concerning comparing different MD configuration, Cipollina et al. [6] also developed a lab scale plate-and-frame membrane distillation module for seawater desalination by applying PTFE membrane with the effective membrane surface area of 0.042 m2. Three different channel configurations were investigated during this research, including free air gap, permeate-gap and partial vacuum air gap. As well, this study also investigated the effect of different operating conditions including variation of hot channel inlet feed flow rate and temperature on distillate

However, the MD process is still under study and the lack of experimental data has indicated that there is a need for more comprehensive research in this field, both experimentally and mathematically. The central issues are the external energy source for MD units, lack of MD membranes and fabrication of modules for each MD configuration. There are also uncertain energetic and economic costs as well as difficulties with long-term operation and the possibility of membrane pore wetting and membrane fouling. Overall, optimization of MD plants is required in order to reach higher MD performance and to decrease energy consumption [17]. The reported values for permeate flux are relatively low and to overcome this issue, an appropriate redesign of the MD module is demanded in order to achieve mass transfer improvement and to increase the membrane surface area per module volume. In addition, with the exception of direct contact membrane distillation (DCMD), which has been more widely studied, other MD configurations have not been properly investigated, so more focus on other MD configurations is required [1]. Generally, there are four basic

Table 1 Comparison of most developed desalination technologies [15]. Technology

Plant capacity (m3/day)

STEC (kWh/m3)

SEEC (kWh/m3)

Operation temperature (typical) (°C)

MSF MED RO

4000–450,000 100–56,000 0.01–360,000

55–220 40–220 –

4–6 1.5–2.5 2.8–12

90–120 (112) 50–70 (70) ˂ 40

specific thermal energy consumption(kWh/m3) convective heat flux from the condenser channel (W/m2) condenser channel heat flux (W/m2) convective heat flux from the evaporator channel (W/m2) evaporator channel heat flux (W/m2) convective heat flux from the membrane surface (W/m2) specific conductive heat flux (W/m2) specific latent heat flux (W/m2) Reynolds number in evaporator channel (−) salinity (g/kg) condenser channel inlet feed water salinity (g/kg) evaporator channel inlet feed water salinity (g/kg) feed water salinity in the output of the evaporator channel (g/kg) temperature at the condenser inlet (°C) temperature at the condenser outlet (°C) temperature at the evaporator inlet (°C) temperature at the evaporator outlet (°C) temperature at the membrane surface in the evaporator side (°C) temperature at the membrane surface in the permeate gap side (°C) temperature in the permeate gap channel (°C) feed velocity (m/s) feed flow rate (m3/s) permeate flow rate (m3/s) Antoine equation coefficient (−) Antoine equation coefficient (−) Antoine equation coefficient (−) impermeable film thickness (m) membrane thickness (m) permeate gap thickness (m) pressure drop (Pa) feed density (kg/m3)

198

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

flux rate and STEC of the system. It is realized that at a constant feed flow rate and temperature and for a 35 ppt feed salinity, PGMD configuration has the lowest value in terms of STEC and highest value in term of distillate flux compared with vacuum and air gap configurations. It is notable that, over recent years some commercial MD modules in different configurations have been developed, Zaragoza et al. presented a comparison, especially in terms of energy analysis during several months of operation, between the most advanced commercial MD module prototype technologies in different configurations (AGMD, PGMD and VMD) and different structures, including (plate-and-frame and spiral wound). The investigated commercial MD modules are listed [12,29]:

because of the availability of excess low cost waste heat. [9]. By reviewing studies carried out around the different MD configuration and PGMD advantage in compare to other developed MD configurations, a more reliable and comprehensive understanding of the current status of this technology is achieved. In this paper, aiming to develop a sustainable desalination technology in terms of energy consumption and by considering the result of accomplished literature survey around different MD module configuration, the permeate gap membrane PGMD, as a novel sustainable MD design having internal heat recovery characteristics, is introduced. Initially, an experimental approach by designing and developing a lab scale PGMD system configuration with ability to reduce STEC of the system is studied. This study outcome provides a good estimation of the most important characteristics of a MD setup, including permeate flux, STEC, GOR and their dependency to the module design and the particular operating conditions. To predict the PGMD module performance, numerical modelling of the heat and mass transfer phenomena in this configuration is studied a single node theoretical model is developed using some simplifying assumption. Furthermore, validation of the PGMD technology numerical model is provided by comparing the experimental data with the modelling results.

• Flat sheet AGMD applying a PTFE membrane with 2.8 m membrane area from a Swedish company (SC module) • Two different MD modules from a Singaporean company 2

• •

(KeppleSeghers) licensed from Memstill including firstly a single flat sheet PGMD with a total surface area of 9 m2 (M33 module) and secondly three MD modules connected in series each of them with 3 m2 area for each membrane to increase latent heat recovery of condensation (PT5 module) Spiral wound PGMD from a German company (Solar Spring) with 10 m2 PTFE membrane area. There was an additional channel for permeate. The system was designed to maximize internal heat recovery (Oryx 150 module). A flat sheet vacuum multi effect MD (V- MEMD) system with multiple stages and 5.76 m2 total surface area of PTFE membrane was built by Aquaver (licensed by the Memsys Company). The stage low pressure allowed the feed water to boil at a reduced temperature. However, the specific energy consumption to create and maintain vacuum in the module should be considered in the assessment (WTS-40A module).

2. Experimental approach In a PGMD module, water and volatiles components evaporate at the membrane interfacial surface of the evaporator channel, diffuse through the micro-porous membrane structure, then are condense and extracted from the membrane module at the permeate channel outlet. As it is mentioned briefly, PGMD or LGMD are a modified configuration of DCMD which has been recently introduced. In this configuration, the permeate is extracted from the highest module position, so the gap between the membrane and the impermeable film fills with permeate during the operation. Fig. 1 shows the PGMD module arrangement. In PGMD by considering a third channel for produced fresh water with an impermeable film on the permeate side, the cold fluid in the condenser side separates from the permeate and therefore it could be any other liquid like saline feed water. Through introducing the permeate gap and impermeable film, there is an additional resistance to heat transfer across the membrane which leads to a reduction of the effective temperature difference, so that the permeate gap and film thickness should be minimized to decrease the thermal resistance. As mentioned, the main advantages of PGMD configuration is

A comparison of performance of these commercial modules in terms of produced distillate quality and quantity and energy efficiency was performed by considering different operating conditions including feed temperature and feed salinities. It is observed that the PGMD module in the spiral wound structure had higher latent heat of condensation recovery compared with all the single effect configurations, leading to the minimum value of 210 KWh/m3 for the STEC [29]. Winter et al. also developed and applied a 10 m2 spiral wound PGMD configuration to reduce the overall external energy demand of the system and in other words to achieve the lower value of STEC and higher value of GOR in comparison to MD systems without internal heat recovery feature. The main objective of the developed compact spiral wound PGMD module with high membrane surface area is to establish higher rate of internal heat recovery [26]. So, as described literature survey, PGMD configuration identifies as the most optimized MD configuration in terms of energy consumption which make it as an efficient configuration to develop a sustainable MD desalination system. However, in situation with the availability of abundant waste heat or low grade thermal energy source, the external energy demand of the system could not be a dominant operating parameter, although, achieving the higher quality and quantity of distillate are more critical factors. As, in a recent research conducted by Dow et al., DCMD configuration with 0.67 m2 membrane area joint to the excess waste heat of power station. It was confirmed that, by applying a MD plant with internal heat recovery characteristics, it would be possible to reduce the value of the STEC of the system, which is a more important consideration when thermal energy is supplied from an expensive heat source. However, in their study which abundant low-grade waste heat was available, the DCMD system showed better performance in terms of produced distillate quantity and the required membrane surface area. Therefore, in this trial situation, achieving high distillate flux and smaller plant size had higher priority than high STEC (or low GOR)

Fig. 1. PGMD module arrangement with internal heat recovery.

199

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

maximum 0.5% moisture, maximum 0.050% water insoluble and typical trace elements including calcium, magnesium, sulphate, copper and iron respectively with 800 ppm, 400 ppm, 1500 ppm and < 1.0 ppm values. A schematic diagram of the lab scale experimental setup is shown in Fig. 3. The feed water is pumped from a 100 L storage tank using a small 12 V DC water pump. The feed flow rate to the module controlled either by adjusting the DC pump voltage or by an in-line controlled valve. A 100-μm pore size mechanical filter was installed before the pump to protect the module and pump from unwanted solids. To adjust the inlet feed temperature to the condenser channel, the lab cooling circuit was used. The feed water gradually preheats whilst flowing through the condenser channel and using the latent heat of condensation and conduction via the PP condensing film. The condenser outlet temperature was increased by using an external electric heat source as it flows through a copper heating coil immersed in an insulated electric water tank (2.4 KW), to provide the required evaporator channel inlet temperature for the PGMD module. In the outdoor experimental setup, the thermal energy from evacuated tube solar collectors (ETSC) with a surface area of nearly 25 m2 and the estimated power of 20 KW at Renewable Energy Park - RMIT will be coupled to MD setup which provides the required system external energy demand in a sustainable way. In the evaporator channel the hot water vaporizes at the membrane surface, diffuses through the hydrophobic PTFE membrane pores and condenses on the permeate channel film. The evaporator channel outlet feed, with higher salinity than the inlet feed to the condenser channel, returns to the feed tank. In order to maintain the feed tank salinity at a constant level, the fresh water pumped from permeate tank by applying a floating ball valve and electronic scale. The produced fresh water exited from the top manifold of permeate gap. The inlet and outlet water temperatures to the condenser and evaporator channels were continuously measured using four T-type

providing internal heat recovery for cold salt feed water by pre heating through absorbing latent heat of condensation from the permeate channel. A novel optimized experimental approach was based on a lab scale plate-and-frame PGMD module with 0.12 m2 effective membrane area, which was constructed using two transparent poly acrylic sheets with 25 mm thickness for evaporator and condenser sides and the third sheet for the produced permeate water of 6 mm thickness. As is seen in Fig. 2a, two main cylindrical channels with 10 mm diameter holes were milled in each evaporator and condenser plate for the flow inlet and outlet manifold. To improve flow distribution, a set of 11* 3 mm distributor holes, was drilled in each flow channel. The hydrophobic PTFE membrane with 0.22 μm nominal pore size and (140–200) μm thickness on a PP support, provided by Membrane Solutions LLC (MSPTFE2700222B), with an effective surface area (760 × 160) mm2 was applied. The permeate channel was separated from condenser channel by an impermeable 100 μm clear PP film (Plastics (Aust) Pty Ltd), this plastic PP film had a thermal conductivity similar to that of the PTFE membrane. The permeate gap was filled by plastic net spacers as a mechanical support between membrane and condensing polymeric film and turbulence promoter (as depicted in Fig. 2b). The gap width in condenser and evaporator channel is adjusted to 1.5 mm and the gap width in permeate channel is considered 6 mm. Four rubber gasket frames, as shown in Fig. 2b, with 1.5 mm thickness used as a sealing material providing required connection between both the membrane and impermeable film with poly acrylic sheets. To make the module structure simple and make the membrane washing and replacement easier, metal G clamps used to assemble all module components as a unique configuration (as Fig. 3c). The applied salt, to make synthetic salt water solution, provided by Pyramid Salt Pty Ltd. with minimum 98.00% NaCl purity (dry basis),

Fig. 2. a) Module's Solid Works assembly design b) inside design of MD setup with mesh spacer and rubber gasket c) outside view of the manufactured setup.

200

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

Fig. 3. Schematic diagram of the PGMD experimental setup.

dissolved solid (TDS) function in ppm during all experimental runs (both digital and analogue output). Each experiment ran for at least 90 min to reach a steady state condition and the average values of the recorded data in the steady state condition were used for analysis. Fig. 4 shows the indoor lab scale MD setup.

thermocouples connected to the data acquisition system, Agilent 34970 A. Three Digital pressure sensors were used to measure the condenser channel inlet and outlet pressure and evaporator channel inlet pressure. Two waterproof portable EC/TDS/ /Salinity Meters (HI98192) applied to monitor and measure the salinity of feed and distillate by total

Fig. 4. Actual indoor lab scale PGMD experimental setup.

201

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

3. Mathematical model development In this study, in conjunction with experimental approach, the numerical modelling of the system performance by considering the heat and mass transfer balance equations and the experimental system configuration is studied. To predict the PGMD module performance, a single node theoretical model is developed using some simplifying assumptions. The following assumptions are considered:

• Steady state condition • Stagnant air inside the membrane pore • Fully filled permeate channels with pure water • No total pressure difference across the membrane, so no mass transfer by viscous flow • No heat loss by conduction to the environment • Stagnant permeate in the permeate channel and so no heat transfer by convection in the permeate gap

For the described experimental module with a total membrane surface area of (760 × 160) mm2, applying heat and mass conservation equations by a single node modelling assumption could provide a reliable estimation of the system condition. Therefore, in the single node modelling temperature, salinity and feed flow rate values evaluated only at the inlet and outlet of the module channels and these parameter variations along the module length ignored. However, for the longer module configuration by applying multi node modelling concept the variation of these parameters along the module length could be considered. To model the PGMD system, the above assumptions, heat and mass conservation laws in all channels and the five main thermal resistances between evaporator and condenser channels, as shown in Fig. 5 are considered. In addition, by defining the known variables including: TCi,TEi, ṁCi, ṁEi, SCi and SEi which are respectively temperature (°C), mass flow rate (kg/s) and salinity (g/kg) at the condenser and evaporator channels inlets, governing equations are developed. It is also required to specify the membrane properties including pore size, porosity, thickness and thermal conductivity, impermeable film thickness and thermal conductivity, bulk conditions and module geometry in order to carry out numerical modelling. The convective heat flux from the evaporator channel to the membrane surface is expressed in Eq. (1):

q̇E = hE ⎛ ⎝

TEi + TEo − TMe⎞ 2 ⎠

Fig. 5. Single node mathematical modelling of PGMD module.

number, as comprehensively described by Essalhi et al. [11]. This coefficient is a function of membrane physical properties including pore size, porosity, thickness and tortuosity besides the temperature and pressure inside the membrane pores.

Jp = Cm (Pv, swe − Pv, swp)

(1)

For the applied membrane samples with specified structure and under the system operating condition, the dominant mass transfer mechanism is defined as a combination of Knudsen and molecular diffusion (0.01 < Kn < 1) [1,2,18,22]. Figures for membrane mass transfer coefficient (Cm) also was in a range of 3 × 10− 7 < Cm < 5 × 10− 7 kg/Pa m2 s for (30–80) °C temperature range and assuming atmospheric pressure in the membrane pores, which was in close agreement with some reported values[9,16,24,25]. The partial vapour pressure at the membrane surface could be defined by Antoine equation (Eq. (5)) as a function of membrane surface temperature in the evaporator respectively (TMe) and (TMp). In Antoine equation Pv,w is vapour pressure of pure water (Pa), T is absolute temperature (K) and for water α, β and γ are taken as 23.1964, 3816.44 and − 46.13 respectively [13].

where, q̇E is the evaporator channel heat flux (W/m2), hE is the heat transfer coefficient at the evaporator channel (W/m2K). TEi and TEo are respectively the temperatures at the evaporator channel inlet and outlet and TMe is the temperature at the evaporator side of the membrane surface. The heat transfer rate from the membrane surface q̇M arises from the latent heat of the produced vapour flux and the heat transferred by conduction across both the membrane matrix and the gas-filled membrane pores [13].

q̇M = hM (TMe − TMp)

(2)

In this equation hM is the heat transfer coefficient at the membrane (W/m2K) (its inverse depicted in Fig. 5 as membrane resistance).

q̇M =

Km (TMe − TMp) + Jp hfg δm

(4)

β ⎞ ⎟ Pv, w (T) = exp ⎛⎜α − γ + T⎠ ⎝

(3)

hfg is the water vaporization enthalpy (kJ/kg), Km and δm are the membrane thermal conductivity (W/mK) and thickness (m), respectively. Jp (kg/m2s) (based on Eq. (4)) is defined as a function of partial vapour pressure in two sides of the membrane and membrane mass transfer coefficient (Cm). The membrane mass transfer coefficient could be derived by defining the main mass transfer mechanism via Knudsen

(5)

In desalination applications, for salt feed water, the reduction of hot feed water vapour pressure by the presence of salt ions should be take into account and some empirical correlation for water vapour pressure of seawater are given in the literatures. Sharqawy et al. [23] summarized the existing correlations and data with their range of validity and 202

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

permeate gap is completely filled with distillate water, the permeate gap thermal conductivity is equivalent to thermal conductivity of fresh water. Moreover, the heat is transferred from the permeate gap to condenser channel by a series combination of thermal resistance including permeate channel and impermeable polymeric film and condenser side thermal resistances. Therefore, the condenser channel heat flux (q̇C) is defined as below:

accuracy for estimation the thermos-physical properties of seawater. As a primary estimation by assuming seawater as an ideal solution, the reduced vapour pressure may be calculated by Eq. (6) in which S is feed water salinity in g/kg, as derived from Raoult's law. Therefore, in Eq. (4), the partial vapour pressure in the evaporator side Pv , swe and permeate gap sidePv , swp, may be derived from Eq. (6). By assuming (SPG = 0) in the permeate gap side, Pv , swp would be equal to Pv , w.

Pv, sw

⎜

(6)

−4S − 3.5012 × 10−7S2)

Pv, sw = Pv, w 10(−2.1609 × 10

qĊ =

̇ = qhot

TCi + TCo ⎞ 2 ⎠

δF KF

+

1 hC

⎛TPG − ⎝

TCi + TCo ⎞ 2 ⎠

Am

Pvsw, 13% salinity

SEo =

ṁ Ei SEi ṁ Eo

Pvsw, 25% salinity

23000 18000 13000 8000 3000

335 325 330 Temperature (K)

(14)

Solving above 7 main Eqs. ((1), (3), (4), (9), (11), (12) and (13)) using Matlab Equation solver, the 7 unknown variables (TMe, TMP, TPG, TCo, TEo, JP, q̇ ), as depicted in mathematical modelling schematic, may be identified. Based on the JP value and mass balance equation for saline feed water on the evaporator side, ṁEo and SEo may be evaluated by Eqs. (15) and (16). SEi, SEo and ṁPG describe respectively, the feed water salinity in the input and output of the evaporator channel and the produced distillate rate in the permeate channel.

28000

320

(12)

(13)

Am

33000

315

(11)

ṁ Ei CpE (TEo − TEi )

ṁ Eo = ṁ Ei − ṁ PG = ṁ Ei − Jp Am

Pvsw, 3% salinity

(10)

ṁ Ci CpC (TCo − TCi )

̇ = qĊ = qcold ̇ = qhot ̇ qĖ = qṀ = qPG

(8)

310

⎛TPG − ⎝

Given assumption of steady state condition, it may be concluded that:

In this equation, KPG (W/mK) and δPG (m) are the permeate gap thermal conductivity and thickness, respectively. Since it is assumed the

Partial vapour pressure (Pa)

+

(9)

305

1 hC

1 δPG 2KPG

̇ = qcold

(TMp − TPG )

Pvw

+

KF (W/mK) and δF (m) are the impermeable polymeric film thermal conductivity and thickness, respectively and TCi and TCo are the temperatures at the condenser channel inlet and outlet, respectively. Considering the energy balance correlations in both evaporator and condenser channels and the total membrane surface area (Am), Eqs. (12) and (13) could be assumed respectively for the evaporator channel and condenser channel heat flux (q̇ hot and q̇ cold).

In this equation, hPG is the heat transfer coefficient at the permeate gap (W/m2K) (its inverse depicted in Fig. 5 as permeate gap resistance), which defined as Eq. (9). TPG is the permeate temperature at the permeate gap

1

1 hF

Similarly, in this equation, hF and hC (W/m K) are the heat transfer coefficient at the impermeable polymeric film and condenser channel, respectively. The inverse is depicted in Fig. 5 as permeate gap and condenser channels resistance, as described in Eq. (11).

Fig. 7 compares these two correlations (Eqs. (6) and (7)) for different feed salinity in a range of (30–160) ppt and at a similar temperature. It is evident from this graph, two correlations approximately are in good agreement, however the reduced vapour pressure calculation based on the assumption of Rault's law (Eq. (6)) has lower accuracy compared to experimental measurement values (Eq. (7)'s data) especially for high concentrate solutions, since in the Rault's law theory, the interaction between different ions in sea water is ignored which led to high error for a non-dilute solution. So, in this theoretical modelling, Eq. (7), which is derived based on experimental measurement, is applied to describe seawater vapour pressure. Based on the assumptions made, the produced permeate is considered stagnant in the permeate channel, so the heat transfer from the gap takes place only in the form of conduction calculable by Eq. (8).

δPG 2KPG

+

2

(7)

̇ = hPG (TMp − TPG ) qPG

1 hPG

⎟

Fig. 6 compares the partial vapour pressure of salt water solution in different salinity as a function of temperature by correlation in Eq. (6). As it is obvious from this graph, increasing salt concentration in solution reduces the partial vapour pressure, which the reduction is more significant at higher temperature. Emerson and Jamieson also introduced Eq. (7), based on experimental measurement in synthetic seawater for (30–170) ppt salinity range and (100–180) °C temperatures range. The reduced vapour pressure Pv,sw based on pure water vapour pressure Pv,w may then be calculated using Eq. (7) [10].

̇ = qPG

1

qĊ =

−1

S ⎞⎞ = Pv, w ⎛1 + 0.5735 ⎛ ⎝ 1000 − S ⎠ ⎠ ⎝

340

345

203

350

(15)

(16) Fig. 6. Partial vapour pressure as a function of temperature and salinity by assuming seawater as ideal solution.

Desalination 419 (2017) 197–210

Partial vapour pressure (Pa)

F. Mahmoudi et al.

Fig. 7. Reduced partial vapour pressure by two different correlations (Eqs. (6) and (7)) at T = 80 °C.

Partial vapour pressure (Rault's law assumption) Partial vapour pressure (Experimental measurement) 47000 46500 46000 45500 45000 44500 44000 43500 43000 42500 42000

30

40

50

60

70

80

90 100 110 120 130 140 150 160 Salinity (ppt) most important characteristic values including permeate flux, STEC, and GOR are presented.

hC and hE in Eq. (1) and Eq. (10) are similarly calculated by Eq. (17) for a spacer filled channel and for a situation that spacers do not induce change in flow direction [8,20].

Nu = 0.644 Re0.5 Pr 0.333

4.1. Hydraulic study and heat transfer results

2Dh 0.5 ls

( )

(17)

Figs. 8 and 9 illustrate the effect of feed flow rate on the pressure drop and required pumping power in the condenser and evaporator channels. This experiment is done to study the hydrodynamic condition in flow channels for cold fresh water feed. As it is evident from these graphs, by increasing the feed flow rate, the pressure drop increase, so that the required pumping power which is a function of feed flow rate (V̇ f) and pressure drop (ΔP), as described by Eq. (19), will also increase. As is shown by the hydraulic tests, the max pressure drop in the MD setup for the designed flow rate (< 2 L/min) were < 6 kPa in both evaporator and condenser channels and, < 0.5 W pumping power was required for the designed system, within the experiment flow rate range. The selected feed flow rate range was chosen to have the optimum amount of feed residence time and therefore the optimum internal heat recovery, which will be explained in the next section. It is noted that, the reported values are based on fresh feed water (nearly 0% salinity feed), however the pressure drop and pumping power would increase for saline feed water.

Which in this equation, ls (m) is orthogonal distance between the net spacer filaments and Dh (m) is the hydraulic diameter and can be calculated as explained by [20]. Hence, the applied heat transfer correlations significantly effects on the whole modelling of MD system, applying correct and accurate heat transfer equations based on the system channel/spacer configuration are so essential to consider real MD system performance. Phattaranawik et al.[19] and Winter[25] proposed to consider the complex hydraulic condition in a spacer filled channel and improve the accuracy of modelling of MD system, applying the basic forms of heat transfer correlation (as Eq. (18)) and defining the exponents value and constants for Nu number by obtained experimental results based on each specific system geometry would be the more accurate approach.

d d Nu = a Reb Pr c ⎛ h ⎞ ⎝L⎠

(18)

Pumping Power = Vḟ × ΔP 4. Results and discussion

0.11 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01

f=

−ΔPDh L

2 ρum 2

which Dh is the hydraulic diameter (defined by Eq. (21)), L is the Fig. 8. Pressure drop and required pumping power in condenser channel at different fresh water flow rates.

Pumping power (W)

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

1.5

1.8

2.1

2.7 2.4 3.9 3 3.3 3.6 Fresh Water Flow Rate (L/min)

(20)

4.2

204

4.5

Pumping power (W)

Pressure drop (bar)

In this section, the influence of feed water flow rate on hydraulic system parameters including pressure drop, friction factor and the applied pumping power in condenser and evaporator flow channels is studied. Furthermore, the effect of different feed water flow rates and salinities on internal heat recovery through the system, besides the

Pressure drop (bar)

(19)

Furthermore, the friction factor in the condenser and evaporator channels may be calculated based on Darcy equation:

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

Pressure drop (bar)

1.3 1.2 1.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2

0.13 0.11 0.09 0.07 0.05 0.03 0.01

2

2.3

2.6

2.9 3.2 3.8 3.5 4.1 4.4 Fresh Water Flow Rate (L/min)

4.7

module length and um is the feed velocity. It is noted that from Fig. 10, by increasing the feed flow rate, the turbulence in the flow channel increases so the Reynolds number increases. Increasing Reynolds number leads to improved heat transfer rate in flow channels and reduction of the temperature polarization effect at the two sides of the membrane surface as well as promoting the temperature difference across the membrane surface causing the permeate flux to increase, which will be discussed subsequently. As described by Eq. (20) and shown in Fig. 10, increasing the feed flow rate also leads to lower friction factor in the module channels.

Dh =

4A c 4 WH = P 2(W + H)

Km (TMe − TMp) + Jp hfg δm

(22)

In the second part of experiments, to study the heat flux in the absence of mass flux, the membrane sheet is replaced by an impermeable PP film with the same thermal conductivity of PTFE membrane. Therefore, no vapour mass flux could diffuse through the impermeable polymeric film and as Eq. (23), heat flux only included conduction term through the impermeable polymeric films to condenser channel. In this configuration, the total heat flux through the membrane is lower than the first set of experimental run (simultaneous heat and mass transfer through the module).

qṀ = qṀ , C =

In Eq. (21), Ac is the flow cross-sectional area and P is the wetted perimeter. For the described lab scale PGMD module the effective surface area was (760 ∗ 160) mm2, the gap width was adjusted to 1.5 mm. So, the channel cross- sectional width (W) was equivalent to 160 mm and the channel cross sectional gap height (H) was 1.5 mm. In this research a set of experiments also accomplished to study the heat transfer characteristics in MD module and to validate the heat transfer part of theoretical modelling for a spacer filled channel. The experiments were conducted by considering two individual module arrangements, as is depicted in Fig. 11. At the first set of experiments, the PTFE membrane is mounted between hot saline feed channel and permeate gap and the plastic impermeable film separates coolant from permeate gap flow (as real PGMD module configuration). Thus, in this situation heat flux from evaporator channel to permeate channel (q̇ M) included either conduction across the membrane material and its gas filled pores (q̇ M,C) besides latent heat of evaporation q̇ M,L [7], as described by Eq. (22).

Friction Factor (-)

5

qṀ = qṀ , C + qṀ , L =

(21)

Friction factor (f)

Pumping power (W)

Pumping power (W)

0.15

Km (TMe − TMp) δm

(23)

These set of experiments studying the module performance in terms of heat transfer rate in absence of mass transfer phenomena, led to a better understanding of the system behaviour in terms of energy balance and also validation of the heat transfer correlations in theoretical modelling, which provides a reliable theoretical modelling by considering both heat and mass conservation equations at steady state condition. As it is shown in Fig. 11, the amount of cold fluid temperature rise (TCo-TCi) is higher when membrane mounted in module hence the released latent heat of condensation in the permeate channel transferring to the cold channel by conduction. In addition, the amount of internal heat recovery inside the cold flow channel, decreased by increasing the feed flow rate within the system operating condition. Increasing of feed flow rate, lead to the lower feed residence time in the flow channels, so there was less time for heat transfer between hot and cold channels, so the temperature rises in condenser channel was less significant [12,13,28]. Fig. 10. Friction factor and Reynolds number of condenser channel at different fresh water flow rates.

Reynolds Number (Re)

5 900 4.6 800 4.2 700 3.8 600 3.4 500 3 400 2.6 300 2.2 200 1.8 100 1.4 1 0 1.5 1.7 1.9 2.1 2.3 2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 4.1 4.3 4.5 Fresh Water Flow Rate (L/min)

205

Reynolds Number (-)

Pressure drop (bar)

Fig. 9. Pressure drop and required pumping power in evaporator channel at different fresh water flow rates.

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

TCo-TCi (°C)

Membrane 50 45 40 35 30 25 20 15 10 5 0 0

0.1

0.2

0.3

0.4 0.5 0.6 0.7 0.8 0.9 Fresh Water Flow Rate (L/min)

1

4.2. Module performance in terms of distillate rate and energy consumption In MD operation the most important characteristic values are permeate flux (JP), STEC and GOR, which are defined as: The permeate flux JP of the MD module could be defined by dividing the permeate output rate ṁPG (kg/s) to the total membrane area (Am):

ṁ PG Am

(24)

The quantity qSTEC is the amount of total energy input (Ei) to produce 1 m3 of fresh water [21]. In this equation V̇ P is the produced distillate rate (m3/s).

qSTEC =

ṁ Ei Cp (TEi − TCo ) Ei = ̇ Vp Vṗ

(25)

The GOR is an indication of how well the total energy input to the system is utilized to produce fresh water:

GOR =

ṁ PG hfg (26)

Ei

TCo-TCi (°C)

S=0

S=30000 ppm

1.1

1.2

The effect of the feed flow rate on permeate flux for a constant set of operating condition (TCi = 15 °C, TEi = 82 °C) and for a range of feed salinities from 0 to 300 ppt, is investigated and shown in Fig. 13. It is seen in this graph that by increasing the feed flow rate, the permeate flux increases. For fresh water feed (assuming 0% salinity), increasing feed flow rate from 0.14 to 1.03 L/min, led to approximately 400% increase in the permeate flux. For a feed sample with 130 ppt (nearly four times seawater salinity), doubling the feed flow rate led to a similar upward trend on produced distillate rate. That is the, Jp increases from approximately 3 to 5 kg/m2 h (nearly 70% increase). This effects may be explained by the higher turbulence in the flow channel at the higher feed flow rate, associated with higher value for the Reynolds number, which improved the heat transfer rate in flow channels and reduced the temperature polarization effect on the two sides of the membrane surface. As a result, the temperature difference across the membrane surface increased, leading to higher permeate flux. As it is mentioned the feed and distillate tank salinity recorded continuously within all experimental runs. As Zaragoza et al., results confirmed that the maximum distillate flux is obtained at maximum feed flow rate because of higher turbulence at the higher feed flow rate and the lower temperature polarization effect [29]. Fig. 13 also presents the effect of salinity on permeate flux, which shows a decrease in permeate water flux for saline water as compared to the fresh water feed case. As seen in Eq. (6) and according to the Rault's law, higher concentration of solutes in the feed causes a decrease of vapour pressure above the solution and decrease of the MD process partial vapour pressure difference across the two sides of the membrane, eventually leading to the lower permeate flux [13]. The produced distillate rate decreases significantly for high salt

The effect of feed water salinity on internal heat recovery was also investigated and the results depicted in Fig. 12. As seen in this figure, by increasing the feed water salinity from fresh water feed (nearly 0% salinity) to seawater salinity (around 30 ppt) and then up to higher value to 130 ppt, the temperature rises in condenser channel (internal heat recovery rate) decreased. This pattern could be explained by the negative effect of salinity on permeate flux rate (as describe in Section 4.2.) which led to a lower amount of released latent heat of condensation (q̇ M,L) at the higher salinity.

Jp =

Fig. 11. Fresh water temperature rise (TCo-TCi) by internal heat recovery for two different situations; a) applying impermeable polymeric film, b) utilizing membrane in MD module, test condition: TCi = 15 °C, TEi = 82 °C.

Impermable polymeric film

Fig. 12. Cold channel temperature rise (TCo-TCi) by internal heat recovery through the membrane at different feed salinity, test condition: TCi = 15 °C, TEi = 82 °C.

S=130000 ppm

50 45 40 35 30 25 20 15 10 5 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Feed Flow Rate (L/min)

206

1.1

1.2

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

Permeate Flux (kg/m 2 h)

S=0 S=200,000 ppm

S=30,000 ppm S=300,000 ppm

S=130,000 ppm

Fig. 13. Permeate flux at different feed flow rates for S = 0, 30, 130, 200 and 300 g/kg, test condition: TCi = 15 °C, TEi = 82 °C.

12 10 8 6 4 2 0 0.1

0.2

0.3

0.4

0.5 0.6 0.7 0.8 Feed Flow Rate (L/min)

0.9

1

S=30,000 ppm

1.2

permeate flux quantity and quality within an acceptable range. Necessarily, it is expected that by applying these low mesh size filtration stages, there will be higher-pressure drop along the module length, demanding higher pumping power and so higher SEEC. The effect of feed flow rate on STEC of the system for three different feed salinity values (0, 30 and 130) ppt, which are calculated based on Eq. (25), is shown in Fig. 14. The resulted values of STEC for higher salt concentration (20 and 30) ppt, because of low distillate rate (as shown in Fig. 13), which led to a high amount for STEC did not plot. The results show that by increasing the feed flow rate, the required amount of thermal energy will increase. Higher feed flow rate is equivalent to a higher value for ṁ Ei , resulting in higher demand of external energy input (Ei) to reach the constant TEi value in comparison with the lower feed flow rate. Also, as explained previously, high feed flow rate leads to shorter feed residence time in the flow channel, therefore a less efficient sensible heat recovery in condenser channel is possible [12,28]. As a result, for a similar operating condition, TCo decreases and the amount of external heat demand (STEC) to reach the designed value for TEi, increases. On the other hand, as seen in Fig. 13, by increasing the feed flow rate, the permeate flux rate increases (higher value of V̇ P in Eq. (25)). However, the overall, the effect of higher energy demand (higher value of Ei in Eq. (25)) is not completely compensated by higher permeate flux, so the STEC values increase at the higher value of feed flow rate. As is clear from this figure, at higher feed salinities, the amount of STEC is also increased because to the permeate output at higher salinities which is also confirmed by previous studies [6,27]. The values determined for distillate flux rate and STEC are in a good agreement with similar research by Cipollina et al. [6], as their setup configuration is described earlier in introduction section. In this research, by applying a PGMD module of 0.04 m2 area and with saline feed water having 35 ppt salt concentration, 0.6 L/min feed flow rate

concentration (200 − 300) ppt, which is near the saturated state of saline feed water. As seen in this graph, for a feed sample with 20 ppt salinity and at nearly 1 L/min flow rate, the distillate flux decreased to < 2 kg/m2 h. However, Dow et al. [9] reported the approximately 3 kg/m2 h permeate flux over a 3 months' operation of a membrane distillation plant pilot trial to treat the power station's waste water, using a multi layered flat sheet DCMD system with 0.67 m2 membrane area, driven by low grade waste heat (< 40 °C) from a power station. The STEC for the trial period was reported as approximately 1500 kWh/m3, which increased by the end of the trial. Dow et al., reported their desalination setup performed consistently without considerable flux lost even at high feed concentration factors. At the end of their trial, feed TDS value had reached approximately 70 ppt, which attributed to applying a series of pre filtration stages as an effective solution. A filter series including a 25 μm high capacity bag filter followed by a 5 μm cartridge filter in feed cycle and especially at the point with the highest temperature in the MD module (before the evaporator channel inlet), applied by Dow et al. to remove the precipitating minerals and also prevents scale formation on the membrane surface. These mechanical filter stages lead to high salt rejection and prevent pore wetting at the membrane surface area, which then leads to longer life span of applied polymeric membrane, less frequent membrane replacement and production of higher quality distillate. As a further stage in this study, in development of a pilot scale PGMD module desalination rig with 0.62 m2 membrane area, a 22 μm screen pore size strainer with plastic mesh basket following by a 5 μm pore size plastic bag filter will be installed. The first filter after the saline feed tank and the second at the highest temperature position of the MD rig to filter reverse solubility salts (after the electrical heater and before the evaporator channel inlet), to prevent both scale formation and pore wetting at the membrane surface and to maintain the

S=0

1.1

Fig. 14. Specific thermal energy consumption at different feed flow rates for S = 0, 30 and 130 g/kg, test condition: TCi = 15 °C, TEi = 82 °C.

S=130,000 ppm

7700

STEC (kWh/m 3)

6900 6100 5300 4500 3700 2900 2100 1300 500

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Feed Flow Rate (L/min)

207

1.1

1.2

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

and for 80 °C of TEi, a value of 8 kg/m2 h is reported for distillate rate. The reported result is comparable with the value for S = 30 ppt under test condition mentioned in this study as depicted in Fig. 13, (a value of 8 kg/m2 h was obtained at approximately 0.7 L/min). In addition, the effect of feed flow rate on produced distillate rate showed a similar trend in both researches. Comparison of STEC values at the stated test condition shows that Cipollina et al. reported a value of 6000 kWh/m3, which as shown in Fig. 14, is much higher than the value from the present study (around 2100 kWh/m3). This difference may be explained by the lower effective membrane surface area, which is nearly one third of that in the present study. Therefore, negligible sensible heat recovery and thermal integration along the system happened in this condition. Gained output ratio is an alternative representation of the STEC of the system and for analysing the thermal efficiency of the desalination systems may be used to quantify the module's capability for internal heat recovery. As it is explained by Winter et al. [26], in MD desalination systems because of heat loss by conduction through the MD module, all the thermal energy input could not be applied to the evaporation process and the GOR must be < 1. However, in ideal MD systems with optimum internal heat recovery with high surface area for heat transfer between hot and cold fluids, a higher value of GOR would be possible. As investigated in this study and illustrated in Fig. 15, by increasing feed flow rate and feed water salinity, GOR values show a downward trend. The maximum achieved value for GOR was approximately 1, which confirmed the high heat loss rate through the lab scale module and insufficient internal heat recovery by the developed PGMD module with the geometry described. A longer module flow channel with higher membrane surface area will provide more efficient sensible heat recovery leading to a higher GOR value and so develop a more thermally efficient MD system [26]. Winter et al. [26] reported GOR values of approximately 5 and 3.5 respectively for fresh water feed and sea water feed (nominally 35 ppt) at 300 kg/h feed flow rate in a compact spiral wound PGMD module. They carried out a set of similar experiments with a system having 10 m2 total membrane area, 25 °C and 80 °C, respectively at the condenser and evaporator channel inlets and feed flow rates varied from 200 to 500 kg/h with different feed salinities. For the specified fresh water feed source (S = 0 ppt), increasing the feed flow rate from 200 to 500 kg/h led to the distillate rate significantly increasing from 10 to 25 kg/h, and the specific energy consumption rising from 130 to 207 kWh/m3. However, for the higher values of feed water salinity in a range from 0 to 105 g/kg (ppt), the distillate rate decreased with salt concentration by approximately 1 kg/ h per 10 ppt of feed water salinity. These results enable finding the optimum feed flow rates at each given salinity level to provide minimum specific energy demand, which is very important parameters in MD system with internal heat recovery feature. It is also observed

S=0

S=30,000 ppm

S=130,000 ppm

S=200,000 ppm

that the effect of increasing membrane surface area on reduction of specific energy consumption of the system is more dominant in compare to increasing fresh water flux as it provides more contact time between feed and membrane and better heat transfer performance in the system. A comparison between reported values by Winter el al. and the values from this study, shows that the current lab scale arrangement with 0.1 m2 membrane surface area has an order of magnitude higher thermal energy demand than the spiral wound full scale arrangement with 10 m2 membrane surface area. Therefore, in the PGMD configuration based on internal heat recovery characteristics, it would be achievable to reduce the amount of external thermal energy demand by providing more fluid residence time in the module channels. Therefore, providing longer module channels will provide a higher heat transfer rate between cold and hot channels, which will increase the temperature rise in the condenser channel and reduce the external STEC rate. Based on the developed theoretical model described in Section 3, the influences of feed flow rate and salinity on two most important desalination system characteristic of permeate flux and also internal heat recovery rate are plotted and the results are compared with the measured values. Figs. 16 and 17 show good comparison between experimental results and theoretical values for produced fresh water rate and for internal heat recovery parameters. The modelling results are depicted in these figures as line and compared with experimental data points. Comparison is made for feed flow rate in the range 0.1 to 1.1 L/min and with three inlet feed salinity of approximately (0, 30 and 130) ppt under the same test condition. As is evident from these graphs, values obtained from numerical modelling using heat and mass balance equations and using the Matlab Equation solver for a single node model, as explained in the Mathematical model development section, are in good agreement with experimental measured values. Therefore, the developed theoretical model could be applied as a reliable tool, to design the geometrical configuration of an optimized PGMD setup based on the simulation of the system performance. The theoretical study provides a basis for developing a more efficient PGMD module by estimating the effect of system parameters including module length, feed flow rate and temperature on the main output parameters including permeate flux rate and STEC. As a further step, the developed theoretical model is used for numerical study of the PGMD pilot plant under construction, which is a new module design with 0.6 m2 membrane area to provide higher thermal efficiency and lower STEC rate. 5. Conclusion A lab scale plate-and- frame PGMD module with 0.12 m2 effective membrane area has been developed and tested. A set of experimental test has been performed to investigate the designed MD module's main

S=300,000 ppm

0.7 0.6

GOR

0.5 0.4 0.3 0.2 0.1 0 0

0.1

0.2

0.3

0.4

0.5 0.6 0.7 0.8 0.9 Feed Flow Rate(L/min)

1

208

1.1

1.2

Fig. 15. Gained output ratio at different feed flow rates for S = 0, 30, 130, 200 and 300 g/Kg, Test condition: TCi = 15 °C, TEi = 82 °C.

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

S=0% model prediction

Permeate Flux (kg/m 2 h)

12

S=3% model prediction

Fig. 16. Theoretical (line pattern) and experimental (point marker) values comparison for influence of feed flow rate and salinity on permeate flux, Test condition: TCi = 15 °C, TEi = 82 °C.

S=13% model prediction

10 8 6 4 2 0 0.1

0.2

0.3

0.4

0.5 0.6 0.7 0.8 Feed Flow Rate (L/min)

0.9

1

1.1

S=0%

S=3%

S=13%

model prediction

model prediction

model prediction

1.2

Fig. 17. Theoretical (line pattern) and experimental (point marker) values comparison for influence of feed flow rate and salinity on internal heat recovery, Test condition: TCi = 15 °C, TEi = 82 °C.

45 40

TCo-TCi (°C)

35 30 25 20 15 10 0.1

0.2

0.3

0.4

0.5 0.6 0.7 0.8 0.9 Feed Flow Rate (L/min)

1

characteristics in different operating condition including feed flow rate and salinity. Under the designed system conditions, TCi = 15 °C, TEi = 82 °C, feed flow rate in range of 0.1 to 1.1 L/min and feed salinity in a range of (0 −30) ppt the permeate flux varied from (2 − 12) kg/ m2h, specific thermal energy consumption was between 1000 and 2500 kWh/m3 and GOR was below 1. The experimental results show that lower feed flow rate provides higher residence time and lower STEC (higher GOR). Operation at lower STEC is also achievable by increasing flow channel length and providing more contact time between feed stream and membrane surface leading to higher heat recovery and lower external energy demand of the system. However, working at low feed flow rate and high membrane surface area results in lower permeate flux and higher investment cost respectively. Therefore, to develop a sustainable PGMD configuration design, from the design point of view, the module length and effective membrane surface area need to be optimized in order to improve internal heat recovery rate in the system and so reduce the external energy demand, especially in the situation where the external energy source derives from non-renewable fossil fuels. In conclusion, the research results provides an reliable technical data for a scaled up PGMD module characterization to design a more efficient and sustainable desalination system via minimizing the thermal energy demand of the system and also producing higher distillate rate.

1.1

1.2

[4] L. Camacho, L. Dumée, J. Zhang, J.-D. Li, M. Duke, J. Gomez, S. Gray, Advances in membrane distillation for water desalination and purification applications, Water 5 (2013) 94–196. [5] C. Charcosset, A review of membrane processes and renewable energies for desalination, Desalination 245 (2009) 214–231. [6] A. Cipollina, M.G. Di Sparti, A. Tamburini, G. Micale, Development of a membrane distillation module for solar energy seawater desalination, Chem. Eng. Res. Des. 90 (2012) 2101–2121. [7] C. Cojocaru, M. Khayet, Sweeping gas membrane distillation of sucrose aqueous solutions: response surface modeling and optimization, Sep. Purif. Technol. 81 (2011) 12–24. [8] A.R. Da Costa, A.G. Fane, D.E. Wiley, Spacer characterization and pressure drop modelling in spacer-filled channels for ultrafiltration, J. Membr. Sci. 87 (1994) 79–98. [9] n. Dow, s. Gray, J.-D. Li, J. Zhang, E. Ostarcevic, A. Liubinas, P. Atherton, G. Roeszler, A. Gibbs, M. Duke, Pilot trial of membrane distillation driven by low grade waste heat: membrane fouling and energy assessment, Desalination 391 (2016) 30–42. [10] W.H. Emerson, D.T. Jamieson, Some physical properties of sea water in different concentrations, Desalination 3 (1967) 207–212. [11] M. Essalhi, M. Khayet, Fundamentals of Membrane Distillation, (2015), pp. 277–316. [12] E. Guillén-Burrieza, D.-C. Alarcón-Padilla, P. Palenzuela, G. Zaragoza, Technoeconomic assessment of a pilot-scale plant for solar desalination based on existing plate and frame MD technology, Desalination 374 (2015) 70–80. [13] M. Khayet, Membranes and theoretical modeling of membrane distillation: a review, Adv. Colloid Interf. Sci. 164 (2011) 56–88. [14] M. Khayet, Solar desalination by membrane distillation: dispersion in energy consumption analysis and water production costs (a review), Desalination 308 (2013) 89–101. [15] J. Koschikowski, Water Desalination: When and Where Will it Make Sense? (2011). [16] A. Kullab, Desalination Using Membrane Distillation Experimental and Numerical Study (Doctoral Thesis), Division of Heat and Power Technology Department of Energy Technology Royal Institute of Technology, 2011. [17] G.W. Meindersma, C.M. Guijt, A.B. De Haan, Desalination and water recycling by air gap membrane distillation, Desalination 187 (2006) 291–301. [18] K. Nakoa, A. Date, A. Akbarzadeh, A research on water desalination using membrane distillation, Desalin. Water Treat. 56 (2014) 2618–2630. [19] J. Phattaranawik, R. Jiraratananon, A.G. Fane, Effects of net-type spacers on heat and mass transfer in direct contact membrane distillation and comparison with ultrafiltration studies, J. Membr. Sci. 217 (2003) 193–206. [20] J. Phattaranawik, R. Jiraratananon, A.G. Fane, C. Halim, Mass flux enhancement

References [1] A. Alkhudhiri, N. Darwish, N. Hilal, Membrane distillation: a comprehensive review, Desalination 287 (2012) 2–18. [2] A. Alklaibi, N. Lior, Heat and mass transfer resistance analysis of membrane distillation, J. Membr. Sci. 282 (2006) 362–369. [3] F. Bai, A. Akbarzadeh, R. Singh, A Novel System Of Combined Power Generation And Water Desalination Using Solar Energy, Solar09, the 47th ANZSES Annual Conference, (2009).

209

Desalination 419 (2017) 197–210

F. Mahmoudi et al.

[21]

[22] [23] [24]

[25]

using spacer filled channels in direct contact membrane distillation, J. Membr. Sci. 187 (2001) 193–201. J.A. Sanmartino, M. Khayet, M.C. García-Payo, Chapter 4 - Desalination by membrane distillation A2 - Hankins, Nicholas P, in: R. SINGH (Ed.), Emerging Membrane Technology for Sustainable Water Treatment, Elsevier, Boston, 2016. R.W. Schofield, A.G. Fane, C.J.D. Fell, Heat and mass transfer in membrane distillation, J. Membr. Sci. 33 (1987) 299–313. M.H. Sharqawy, J.H. Lienhard, S.M. Zubair, Thermophysical properties of seawater: a review of existing correlations and data, Desalin. Water Treat. 16 (2012) 354–380. j. Swaminathan, h.W. Chung, d.M. Warsinger, f.A. Almarzooqi, H.A. Arafat, V.J.H. Lienhard, Energy efficiency of permeate gap and novel conductive gap membrane distillation, J. Membr. Sci. 502 (2016) 171–178. D. Winter, Membrane Distillation a Thermodynamic, Technological and Economic

Analysis (Ph.D. Thesis), University of Kaiserslautern, 2015. [26] D. Winter, J. Koschikowski, S. Ripperger, Desalination using membrane distillation: flux enhancement by feed water deaeration on spiral-wound modules, J. Membr. Sci. 423-424 (2012) 215–224. [27] D. Winter, J. Koschikowski, M. Wieghaus, Desalination using membrane distillation: experimental studies on full scale spiral wound modules, J. Membr. Sci. 375 (2011) 104–112. [28] J. Xu, Y.B. Singh, G.L. Amy, N. Ghaffour, Effect of operating parameters and membrane characteristics on air gap membrane distillation performance for the treatment of highly saline water, J. Membr. Sci. 512 (2016) 73–82. [29] G. Zaragoza, A. Ruiz-Aguirre, E. Guillén-Burrieza, Efficiency in the use of solar thermal energy of small membrane desalination systems for decentralized water production, Appl. Energy 130 (2014) 491–499.

210