Around 40% of the world’s population lives in arid and semi-arid regions where rainfall is low. These regions are facing challenges of declining water tables and increasing ground water salinity. Providing good quality drinking water for small communities in these areas is highly challenging. Although existing membrane technologies are able to produce potable quality water, issues such as high energy consumption, osmotic pressure constraint, brine management and large centralized designs make them unsuitable for application in these areas. Membrane distillation (MD), a thermal integrated membrane process, is a burgeoning technology with the potential to address and overcome these issues. As a vapour pressure operated system, MD is not restricted by saline feed solutions and therefore can achieve good quality distillate with minimal brine discharge. Furthermore, an MD system can be built as a standalone compact system suitable for small community application. The modest temperature requirement for MD operation (generally between 60°C to 80°C) enables the system to use alternative energy sources such as solar power. Despite such advantages, MD has not as yet been used widely in commercial applications. Several essential problems concerning MD process performance, namely, lower production rate, fouling propensity, energy efficiency and long term performance must be addressed.
In this study, the performance of a scaled-up modified design vacuum membrane distillation system termed ‘vacuum multi effect membrane distillation (V-MEMD)’ was evaluated. A bench scale direct contact membrane distillation (DCMD) was employed for detailed fouling analysis. The four main sections of this work incorporate: (i) V-MEMD operation; (ii) scaling development in MD; (iii) organic fouling development in MD; and (iv) pretreatment and membrane cleaning in MD. These sections present and explain critical aspects of MD performance in the context of drinking water production.
“V-MEMD operation” Firstly, in this study the beneficial features of a modified V-MEMD system were highlighted. These include the internal heating and internal condensing which reduces heat loss and makes operation possible at modest feed temperatures from 45°C to 55°C. A semi-empirical mathematical modeling in this study showed that operating at these feed temperature ranges minimized the effect of temperature polarization (TP) to a low range of between 0.96 and 0.99. The findings of the V-MEMD performance analysis indicated that feed temperature and permeate pressure were the most influential operating parameters. Lowering the permeate pressure from Pp =15.0 kPa to 10.0 kPa increased the permeate flux by almost 200%, whereby the highest permeate flux of 13.5 L m⁻² h⁻¹ (LMH) was achieved when the permeate pressure was reduced to Pp =5.0 kPa. In the V-MEMD concept, vacuum application is essential in order to create a sustainable driving force, especially for a scaled up modular unit with several membrane stages. At the same time, increased feed temperature exponentially increased the permeate flux. A small variation of feed temperature from 45.0°C to 65.0°C significantly improved the permeate flux from 3.6 LMH to 11.8 LMH.
The V-MEMD system proved to be suitable for producing 9.4 LMH of good quality permeate (more than 99.5% rejection rate) with highly saline feed water (1 M of NaC1 feed solution concentrated up to 3 M of NaC1). Only a 10-15% reduction in permeate flux was observed at high feed concentration. The modeling data revealed that high turbulent feed flow velocity of 2.2 m/s (Re = 17, 300) in the V-MEMD system effectively minimized concentration polarization (CP), but the recovery ratio reduced with increased feed flow velocity. An intermediate feed flow velocity of 1.1 m/s (Re = 6,100) was more appropriate for balancing the effect of CP and maintaining a reasonable recovery ratio.
“Scaling development in MD” In achieving near zero liquid discharge under thermal conditions, inevitably, the MD membrane would be exposed to highly concentrated sparingly soluble salts such as calcium sulphate (CaSO₄). In this study, an evaluation of CaSO₄ scaling development in MD operation was carried out, focusing on the role of hydrodynamic (flow velocity) conditions. This study found that permeate condition influenced CaSO₄ scaling development. For instance, in the V-MEMD system, the CaSO₄ crystal size in the membrane module increased from 62.68 μm to 522.28 μm, with increased permeate pressure from 10.0 kPa to 15.0 kPa. Similarly, in a DCMD configuration, a small change in the permeate velocity from 0.8 m/s to 1.1 m/s was effective in changing the scaling pattern from surface crystallization to a more dominant bulk crystallization, without the need to change the feed velocity while improving the system’s performance (i.e. increase recovery ratio, reduce pumping energy, increase permeate flux). Importantly, the findings of this study also revealed that the crystals were only loosely deposited on the membrane.
In the V-MEMD system, the loose deposition was attributed to the lack of hydraulic pressure, low feed temperature (Tf = 47.6°C), high turbulence (Re = 5665.6, 0.9 m/s) and short membrane retention time (21.6 s). Increasing the feed flow velocity from 0.3 m/s to 0.9 m/s in the V-MEMD reduced the gypsum crystal size in the membrane module from 339.03 μm to 62.68 μm. Likewise, in the DCMD configuration the high feed velocity (turbulence) played an important role in controlling the membrane surface crystallization. The Field Emission Scanning Electron Microscope (FE-SEM) analysis with EDS showed significantly higher calcium and sulphate element deposition on the membrane at low feed velocity (0.5 m/s) compared to the high flow velocity (2.2 m/s).
“Organic fouling development in MD” Organic fouling is a ubiquitous problem in membrane processes. Compared to pressure driven membrane processes, the fouling phenomenon in MD operation is unique due to the presence of thermal conditions on a hydrophobic membrane at supersaturated feed concentration levels. In depth understanding of the MD fouling phenomenon is crucial if MD is to be successfully implemented in a proto-scale. This research carried out a detailed fouling development analysis using Liquid Chromatography-Organic Carbon Detection (LC-OCD) to characterize the behavior of organic compounds under thermal MD operation. The findings of this research established that organic fouling in MD was influenced by the type of organic compound present in the feed solution, the thermal state as well as the physico-chemical condition of the feed solution. Based on the LC-OCD analysis of the feed and permeate solution and membrane foulant as well as membrane analysis (contact angle and SEM-EDS analysis), both the humic acid (HA) and bovine serum albumin (BSA) compounds showed dominant fouling tendencies while the alginic acid (AA) compound exhibited minimal fouling tendencies. The latter was due to its hydrophilic nature and negative electrostatic repulsion.
The membrane SEM-EDS analysis showed that mainly the BSA compound was deposited on the membrane surface (800.6 mg/m² organic mass per membrane area) compared to the HA compound (423.2 mg/m²). This was due to the hydrophobic nature of the BSA compound which allowed it to bond with the hydrophobic MD membrane. Meanwhile, the humic substances (HS) showed changes under MD thermal conditions. The LC-OCD analysis of the HA feed solution revealed the thermal disaggregation of the HS, forming low molecular weight–HS (LMW-HS) organics. Further, the cross-section membrane SEM-EDS line analysis showed the penetration of the LMW-HS organics through the membrane pores, resulting in partial wetting. The findings for the influence of physico-chemical state of the feed solution revealed that the addition of salinity (NaC1) contributed to higher HS disaggregation to LMW-HS organics. This resulted in severe penetration of the LMW-HS organics to the permeate side. Meanwhile, in the presence of inorganics Ca²⁺ ion that acts as a binding agent, a cake layer was formed on the membrane.
“Pretreatment and membrane cleaning in MD” Finally, a practical application of MD was presented in this study by analysing the pretreatment and membrane cleaning in MD. In the first part of this section, the performance of two chemical-free pretreatments (namely, deep-bed biofilter and a submerged membrane adsorption bioreactor system (SMABR) was evaluated in terms of organic fouling reduction. Both these pretreatment systems helped to reduce HS and LMW organics as well as assimilable organic carbon (AOC) concentrations through adsorption and biodegradation mechanisms. In the second part of this section, MD performance with natural seawater was compared to SMABR pre-treated seawater. The natural seawater, which predominantly contains HS, resulted in the formation of LMW-HS organics under MD thermal conditions and pore penetration was observed to occur through the membrane.
The biofouling potential of MD operation with SW was highlighted based on the AOC concentration of the membrane foulant and feed solution. In the meantime the SMABR pre-treated seawater feed solution containing low concentrations of HS and LMW organics, resulted in more stable permeate flux and minimal LMW-HS organics pore penetration. The findings established the suitability of chemical-free pretreatments to reduce organic fouling in MD. Additionally, the membrane cleaning by water was carried out to flush away the loose deposition of crystals in the V-MEMD system. Based on the feed solution ion mass balance, with only 2 L of DI water, most ions in the feed solution, specifically the Mg, Na and C1 ions, were removed. This finding established the effectiveness of frequent DI water flushing for the V-MEMD system.