Mitigating the aftermath of Malaysia's worst flood in 50 years: emergency drinking water supply with ultrafiltration

Disaster management in developing countries like Malaysia is a challenging task due to limited resources and ability to finance important economic programs (Hochrainer-Stigler et al. 2015). The massive flood that hit Peninsular Malaysia in December 2014 was unprecedented since the country achieved independence in 1957. Several thousands of households were affected, with electricity and public piped water supplies cut-off abruptly. Government appealed to private and non-government organizations (NGOs) to provide humanitarian assistance in the disaster areas.

In times of natural disaster, victims face problems arising from contaminated water supplies (Akhter et al. 2015). Ultrafiltration is a good barrier for bacteria and virus removal to ensure the safety of drinking water (Di Zio et al. 2005). Techkem Water heeded government requests for humanitarian assistance and deployed four types of mobile ultrafiltration membrane water treatment systems to flood-hit disaster areas. Each of these systems has distinctive features, as discussed in the subsequent sections.

Figure 1 shows the flood aftermath in Kelantan, Malaysia in January 2015. Thousands of houses were damaged or swept away by the flood waters, with no electricity or public piped water supply to the affected areas. The flood caused damage estimated at USD 560 million in Peninsular Malaysia's northern states. More than 200,000 flood victims were affected in 5 states with the largest number of victims reported in Kelantan. An estimated 7% of the 1.7 million population of Kelantan had been affected by this massive flood by the end of December 2014.

Humanitarian assistance from the government, private sector and NGOs was deployed to the disaster sites to mitigate the situation. The table on the right side of Figure 2 names the areas (as Malaysian states) in the first column and shows the numbers of flood victims on 25th and 26th December 2014, while the map of Peninsular Malaysia (Figure 2) shows the total number of victims up to 1,700 hours on 28th December 2014. More than 400 relief centers were established to provide temporary accommodation for flood victims.

To accommodate both site conditions and the level of demand, four types of mobile ultrafiltration membrane water treatment systems were deployed – see Table 1.

All four systems (Table 1) utilized Polyethersulfone (PES) based Multibore^® hollow-fiber ultrafiltration membranes manufactured by Inge GmbH, Germany. There are two types, with capillary diameters of 0.9 (0.9 MB) and 1.5 mm (1.5 MB) respectively. Turbidity is a good physio-chemical indication of filtrate quality (Roig et al. 2014). As shown in Table 1, filtrate turbidity was below 0.20 NTU, indicating crystal clear water, from all four systems regardless of feed water turbidity (range from 30 to 170 NTU). The systems could provide variously between 0.3 to 10.0 m³/hr of filtrate.

Ultrafiltration membrane systems have been implemented successfully at industrial-scale to supply drinking water in Kelantan, Malaysia, producing high quality filtrate with significant turbidity reduction from the surface-water feed (Chew et al. 2015). Filtrate production costs for such industrial-scale systems were about USD 0.04/m³ and turbidity was consistently below 0.30 NTU. The current global shift towards membrane technology has made membrane prices very competitive and the small footprint for such systems generates a lot of interest. The four mobile ultrafiltration systems used in this study have much smaller footprint than conventional sand/media filtration systems of similar capacities, making them highly suitable for deployment in disaster zones.

Dead-end feed mode in ultrafiltration membrane systems requires a pre-determined periodic backwash sequence set-point timer to remove the accumulated solids on the membrane surface. This is usually carried out automatically using programmable logic controllers (PLCs). All four mobile ultrafiltration membrane systems were operated manually, however, without automation. Due to the variation in feed-water characteristics (especially suspended solids concentration and turbidity), the systems were manually backwashed instead, after reaching a pre-determined pressure set-point of 1.0 bar. All operated within a filtration flux range of 70 to 90 L·m⁻²·hr⁻¹. Depending on feed-water characteristics, it took between 1 and 3 hours of continuous filtration to reach the pressure set-point. Filtrate turbidity from the systems was consistently below 0.20 NTU even after 3 hours of continuous filtration, however. Ultrafiltration is based on the cake-filtration concept to ensure that all solid particles exceeding the membrane pores size are segregated from the filtrate to ensure consistent quality. Similar consistent filtrate quality below 0.20 NTU was also reported for feed-water originating from rivers using PES-based ultrafiltration membrane systems (Chew et al. 2016). The build-up of solid particles arising from the cake-filtration mechanism causes significant pressure increases and reduces the filtration flux rate over time.

All the four mobile ultrafiltration membrane systems were equipped with constant speed feed pumps. Initially the filtration flux rate was controlled using a manual valve on the feed-pump discharge port. The valve was adjusted every thirty minutes to meet the required filtration flux. Solid particle build-up on the membrane surface caused the filtration flux to decrease over time and making periodic valve adjustment necessary. Since the pump operated at constant speed, the system power requirement was almost constant, with very little variation.

The principal selection criteria for the four systems were based on site conditions and water demand. Type 2, for instance, could produce up to 10.0 m³/hr of drinking water and was suitable for places with more than 3,000 flood victims – e.g., Manek Urai. System types 1, 3 and 4, were all suited to smaller populations, ranging from 300 to 500 people. This assessment is based on 8 hours system operation per day and a requirement of 8 liters drinking water per person per day.

Other selection criteria included accessibility by road, fuel availability (for a generator set), space, and the quantity and quality of the feed water available. Figure 3 shows the four ultrafiltration membrane systems deployed in January 2015. Site operating conditions are reported in Table 2. The specific energy requirements for the systems range between 0.48 and 0.55 kWh/m³. The type 4 system can be powered by battery, with similar energy requirements to type 3, or with a hand pump.

As most of the hardest-hit flood disaster areas were in remote areas and the access roads were in very poor condition, the ultrafiltration membrane systems needed to be light enough to be lifted by hand. While this was not true for the type 1 system, all others could be transported using 4-wheel-drive (4WD) vehicles.

All four systems had either a generator or lead acid battery to operate the pumps. Most of the flood-hit areas were cut-off from the main electricity supply for safety reasons and portable supplies were essential for operation. The generators needed between 20 and 50 liters of gasoline for 8 hours of operation, and gasoline supplies could be bought within 50 km of the operating locations.

Each ultrafiltration membrane system has its own distinct advantages and weaknesses, which were relevant to usability at the various sites. Table 3 is a comparative summary of the four systems.

The system footprint was critical as type 1 and 2 systems, which require substantial set-up areas. In order to cope with different feed-water suspended solid concentrations, two types of membrane (0.9 and 1.5 MB) were used – see Figure 4. For low turbidity waters (<100 NTU) the 0.9 MB membrane is sufficient but for higher turbidity (>100 NTU) the 1.5 MB membrane is better. The larger diameter of the 1.5 MB hollow-fiber membrane is also stronger and better able to withstand backwashing.

All four systems could be mobilized easily within 1 to 2 days. This is essential for flood disaster relief with respect to drinking water supply, when both electricity and public piped water supplies are cut-off abruptly.

Malaysia and most of South East Asia is tropical – i.e., hot and humid all year round. The monsoon season brings heavy rains to most parts of Peninsular Malaysia and is commonly associated with floods. Malaysians are accustomed to occasional floods, including flash flood, during the monsoon seasons but the downpours in December 2014 caused massive flooding in most of northern Peninsular Malaysia exceeding anything that had occurred in the previous 50 years.

The readiness of government and public to anticipate floods, and the need for mitigating measures, arises from the fact that they are used to them. It has been estimated that the 2014 floods caused losses of USD 560 million in Malaysia. Many towns had a meter or more of flood water – see Figure 5(a) – and international assistance from USA, Japan, China, Thailand and Singapore provided humanitarian needs like food and temporary shelter for victims – Figure 5(b).

One of the most essential needs during disasters is drinking water. Disruption of the public piped water supply forces flood victims to seek nearby, untreated, river water to fulfil their daily needs –see Figure 5(c). Filtrate from the ultrafiltration membrane systems was analyzed for the presence of bacteria including E. coli using the SenSafe^® on-site water quality test kit produced by Industrial Test Systems, Inc. All the filtrate samples tested were free from bacteria. On-site water quality test kits are very practical for use at disaster sites, where proper laboratory analysis facilities are lacking. Potential water sources must be analyzed prior to ultrafiltration to confirm they are free from herbicides, insecticide, etc., and Figure 5(d) shows a team of operators collecting river water samples for pesticide analysis prior to ultrafiltration water treatment.

All raw water sources used were analyzed to ensure the absence of herbicides, pesticides and/or insecticides. Ultrafiltration is a solid-liquid separation process and not suitable for solute separation, which requires the smaller pore sizes of nano-filtration or reverse osmosis. Such systems have operating pressures of 6 bar and above, compared to the 1 bar required by ultrafiltration.

Four types of mobile ultrafiltration membrane systems were deployed to flood-hit disaster sites in Malaysia in January 2015. Two types were equipped with larger membrane capillary diameters of 1.5 mm, capable of handling high suspended solid concentrations and turbidity exceeding 100 NTU. The type 3 and 4 systems were more suitable, however, for sites with limited space, because of their smaller foot-prints. All four system types yielded filtrates with turbidity below 0.20 NTU and free of bacteria. The type 2 system, which could provide up to 10.0 m³/hr of drinking water, was deployed in an area with many victims (more than 3,000 people). All of the systems were manually adjusted to operate with filtration fluxes of between 70 and 90 L·m⁻²·hr⁻¹. Periodic backwashing was conducted after 1 to 3 hours of continuous filtration, to remove solid particles from the membrane surface. Each of the systems has its own advantages and weaknesses, which must be carefully assessed to suit the disaster site conditions. More innovative and cost effective water treatment systems are likely to be needed for disaster relief operations in future.

The authors would like to express appreciation for the support from Ir. Hj. Wan Mohd Zamri Bin Wan Ismail, Air Kelantan Sdn. Bhd., Rekarunding, University of Malaya and Inge GmbH in this humanitarian effort.