An improved technique for reducing water wastage from micro-RO-membrane-based water purification systems: An experimental study

Point-of-use (POU) water purification for home-use drinking water purification has grown significantly in parts of the world including India, in the past two decades. Several technologies, including ultra-violet (UV) radiation, chlorine-based disinfection, filtration using activated carbon, and media made of polymeric fibers, have been used in these home-use water purification systems. However, small reverse osmosis (RO)-based water purification systems (micro-scale RO purifiers) have emerged as the most-favored technology, with or without some of the other technologies mentioned. One of the major advantages of RO-based systems is the discernable ‘improvement’ in the taste of water due to the reduction of dissolved ions in water. This taste change assures the user that the water is purified or ‘improved’ compared to other technologies, such as UV-based disinfection, which do not produce a discernable taste improvement in the product water. It is also known that RO removes pathogenic bacteria, viruses, other micro-organisms, dissolved organics, pesticides, and dissolved ions from drinking water. Some of the ions removed such as arsenic, lead, mercury, nitrates, and fluorides are known to be detrimental to human health and are occasionally present at dangerous levels in groundwater sources. Other ions removed, such as iron and aluminum, are not desirable in drinking water due to health-related or esthetic reasons. Some ions removed such as sodium, potassium, calcium, magnesium, and chlorides may be required by the human body, but these are also available from other inexpensive food sources. In addition, the amount of these ions, available from drinking water, may not be sufficient to meet our nutritional needs, hence their removal is often tolerable, given the huge benefits arising from the removal of the toxic ions and pathogenic micro-organisms. Finally, drinking water is consumed to meet our need for hydration and not for nutrition. The above benefits have made RO technology the clear front-runner for home-use and POU water purification systems.

However, the home-use RO technology has a clear negative, which is the considerable volume of water wasted as the ‘reject’. This problem is enhanced by consumers and manufacturers choosing to operate the water purifiers at a recovery of 10–40%. This is done to prevent scale formation and hence irreversible membrane fouling. Membrane damage due to scaling requires the replacement of the most expensive component of the water purification system, which is the RO membrane element. Hence, a technology for automated membrane cleaning that allows the water purifier to be operated at ∼75% pure water recovery, without significantly compromising the membrane life, could lead to significant water saving at the national and international levels. This study results from an ongoing program, which is developing technologies that will enable the operation of home-use RO purification systems at near ∼75% recovery of product water.

The following aspects of home-use RO-based water purifiers which make them distinct from other continuously operated, larger RO-based water purification systems, must be kept in mind, to appreciate this work. First, home-use RO-based water purification systems are not continuously operated, since the typical requirement of a family, for drinking water, is met within an hour of operation of the purification system. Hence, a typical purification system is on stand-by mode for ∼23 h/day. Some larger systems could be on stand-by mode for ∼8 h between cycles of operation. Second, the power consumption of a typical RO-based home-use water purification system is quite low (∼30 W), in comparison to other home appliances, such as modern TVs (40–65 W), ceiling fans (75 W), and room lights (60–10 W). Hence, a small increase in the same, while it is in use, would not impact the users of the technology.

Osmotic backwash in continuous RO membrane filtration processes has been proposed and applied previously (Sagiv & Semiat 2005; Avraham et al. 2006; Qin et al. 2010). Various studies of the osmotic backwash of RO membranes and UF membranes in large-scale desalination systems have been reported (Nam et al. 2012; Ramon et al. 2013; Gilabert-Oriol et al. 2014; Kim 2014; Park et al. 2014; Jiang et al. 2015; Gao et al. 2016). Several studies have reported the use of osmotic backwash for maintaining flux in forward osmosis (FO) systems (Blandin et al. 2016; Gwak & Hong 2017; Motsa et al. 2017), which are typically used for energy recovery from wastewater. Even in these systems with high organic fouling potential, osmotic backwashing showed positive results. Warsinger et al. (2018) reported a study of osmotic backwash on a batch RO system, wherein they found that, if the residence time of the high-concentration feed, is kept below the ‘induction time’ for crystallization, scale formation can be effectively prevented. Dana et al. (2019) reported a study of osmotic backwashing of RO and NF membranes used for brackish and wastewater treatment and found that sulphate-based solutions have a better ability to maintain the permeate flux by cleaning the CaPO₄ scale. Yaranal et al. (2020) studied pressure-assisted osmotic (PAO) backwash for RO membrane modules, in which backwashing was achieved by pressurizing the permeate, in addition to permeate flow, due to the osmotic potential. Lee et al. (2021) compared ‘salt’ cleaning and osmotic backwash in lab-scale nano-filtration membranes and found that due to the ion exchange of the Ca ions with Na ions, the integrity of the fouling layer is weakened. Simple physical cleaning using feed water and/or permeate has been reported in several studies. Oh et al. (2009) studied scale formation on RO membranes due to the precipitation of calcium salts. The study found that while surface nucleation dominates, bulk crystallisation is also responsible for scale formation. Ma et al. (2013) reported a study of backwashing of fouled UF membranes, using both UF permeate and RO permeate, and found that better results were achieved when backwash was performed with RO permeate. Zhang et al. (2015) reported a study of cleaning membranes used for membrane distillation. Chemical-based cleaning has been reported in several studies (Ferrer et al. 2015; Lee et al. 2017a, 2017b), in which sodium hypochlorite has been used, occasionally in combination with HCl. Some papers (Al-Ghamdi et al. 2019; Alpatova et al. 2020) have investigated a unique backwashing approach using an aqueous solution of CO₂, wherein the gas nucleates as the pressure drops, forming bubbles, which dislodge the foulants on the membrane surface. More recent studies have explored the use of ultra-sonication (Hube et al. 2023), ‘water-hammer’ type effects induced by pressure variations (Aslam et al. 2022), and other novel chemical means for the prevention of membrane fouling (Khan et al. 2023; Qiao et al. 2023).

During the study, reported in this paper, which involved the measurement of the total dissolved solids (TDS) of the backwash-effluent, it was observed that the gravity and osmotically driven backwash-effluent showed a dual-minima in the effluent TDS (Figures 6(a) and (b) and 7(a) and (b)). Such an observation has not been reported earlier, to the best of our knowledge. These observations were studied under multiple operating conditions, to understand the causes for the dual-minima. The above observations can be used to optimize the gravity and osmotically driven backwash system. The membrane element is ‘cleaned’ using a gravity and osmotically driven backwash using the permeate produced by the system. This study examines the effect of the permeate volume used for the backwash, on the extent of residual salt removal from the membrane module.

The estimate of recovery from the above validated the recovery estimate from the flow rates. A level control element (13) in the permeate collection tank is used to cut-off power to the power supply (14), which shuts off the pump and the solenoid valve, when the permeate tank is filled to capacity (∼25 L), thereby mimicking the batch-operating mode of most POU and home-use drinking water purification systems. The closure of the solenoid valve (6) prevents water from draining out of the membrane cartridge (8). At this point in time, the small volume of water, entrained in the RO casing, is discharged through the reject line (11). This discharge-water from the reject line after shut-down was collected at regular time intervals, and its volume and TDS were measured. These values are reported and discussed in the subsequent sections. All material parts used above were procured individually, which are available as spare parts in the Indian market.

The experimental control apparatus, shown in Figure 1, was operated to fill the storage tank (12), which had a capacity of ∼25 L. This is termed as one operating cycle. One operating cycle lasted for ∼1 h at the beginning and increased to 4 h after ∼30 cycles. The post-shut-down discharge-water was collected from the reject port (11), for measurements. The system was allowed to be on stand-by for at least 1 h, during which the permeate storage tank was drained. Typically, the operating cycles of the ‘control’ experiment (Figure 1) and the experimental system (Figure 2) were performed alternately. The operating cycle for the experimental system was followed by the ‘backwash’, which lasted longer. The typical ‘cycle time’ for the experimental system was shorter than the above and increased from ∼1 h, at the start to ∼2 h after ∼30 cycles. A life-time experiment lasted for ∼40 cycles.

Backwash-discharge data, similar to the data shown in Figure 5, was collected from the ‘experimental’ system. The backwash-discharge data consisted of the time required to collect incremental volume samples of 20 mL, and the TDS in the same. Figure 6(a) shows the measured TDS variation of the backwash-discharge with the volume of this discharge, from the reject line (point 11 in Figure 2). The maximum reject-TDS for these experimental runs were similar. Figure 6(b) shows the measured TDS variation of the ‘backwash-discharge’ with time, from the point of shut-off, for the same systems. From Figure 6(a), it can be observed that the backwash-discharge TDS variation showed a primary minimum, which occurs after ∼100 mL of ‘discharge’. From Figure 6(b), it can be observed, that the primary minimum in backwash-discharge TDS occurs after ∼10 min of shut-off. This dual-minima in the backwash-discharge-water TDS is of interest and has not been reported earlier, to the best of our knowledge.

The possible causes for the above observations will be discussed in following sections. Figure 6(c) shows the estimated mass of dissolved salts removed from the RO membrane casing, based on Equation (3), during the backwash experiment. Also shown in Figure 6(c) is the mass of dissolved salts removed from the RO membrane casing, which was estimated based on data from a ‘control’ experiment having similar maximum reject-TDS value. It can be observed that while the ‘control’ experiment removes about 68 mg of dissolved TDS from the casing, the ‘experimental’ backwash system is able to remove around 160 mg of dissolved salts.

The data shown in the previous section suggest the following. First, the experimental system for backwashing, shown in Figure 2, is able to enhance the life of the RO membrane system, when it is operated at a high recovery of ∼75%. The TDS data from these ‘life-time’ experiments demonstrate that the membrane integrity and ability to perform are not compromised by the proposed process of backwash. The above enhancement in life is achieved while using less than 2.5% of the product water for membrane cleaning. The proposed backwashing system is capable of removing 2.5 times higher mass of dissolved salts from the RO membrane element, compared to the ‘control’ experiment. This mass of salts would have remained within the RO membrane element in the absence of the proposed backwash. The relatively longer time required for the backwash is not an issue with the home-use RO-based water purification systems, as these systems are typically operated for ∼1 h/day, spending the remaining 23 h in ‘stand-by’.

In the ‘control’ experiments, the discharge volume of ∼150 mL was observed, post-shut-off, which occurs due to the following. The high TDS of water on the feed-reject side of the membrane drains out under the pressure which the membrane was under, at the point of shut-off. This pressure dissipates very quickly post-shut-off, which causes the rapid decline in the discharge flow rate. Since the throttle valve position under the high-recovery operating condition, is almost closed, the discharge occurs slowly, over a period of minutes. The reject water at the discharge end of the casing is at the highest reject-TDS, while the water at the feed end of the feed-reject side of the membrane is at the feed TDS. Hence, in the control experiments, the TDS of discharge-water can be expected to vary from the maximum reject-TDS to the feed TDS. This is what is experimentally observed and is shown in Figure 5(a) and (b).

In summary, this experimental study demonstrates a technology which can be used to cause a significant reduction in water wastage from home-use water purification systems, with minimal added parts and cost. It further shows how the volume and time requirements for a complete washout of the RO membrane system can be estimated. It also demonstrates how a significantly higher mass of dissolved salts can be removed from the RO membrane casing by a complete hydrostatic plus osmotic (HPO) backwash. Most importantly, this work demonstrates the risk of retaining a high TDS zone on the membrane surface, when the system goes into stand-by, in the absence of a complete backwash. The observed dual-minima in the backwash-effluent TDS is due to the delay in the reverse flow of the permeate into the membrane core, which is required to trigger forward-osmotically driven backwash. In future research, a mathematical model based on the above will be studied in detail.

The authors wish to thank BITS-Pilani, Hyderabad Campus, for supporting this study.

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

J.C., the corresponding author is a ‘Technical Expert’ for the Water Quality India Association, which represents a group of companies that sell water purification products in South Asia.