Improving the hydrological performance of a rainwater for drinking (RFD) system at a rural hospital in Vietnam

A rainwater harvesting system (RWHS) collects rainwater for potable and nonpotable water use, and these systems have been used since historical periods when centralized water supply systems were not well established. The RWHS has been expanded to areas of ecosystem management, urban flood control, nonpoint source pollution control, integrated basin management, green infrastructure, and so on. Researchers worldwide have investigated whether rainwater collected from roofs is suitable for drinking, and their results show improper water quality, particularly regarding microbial quality. Vietnam has abundant rainfall and has traditionally used the RWHS for drinking water (Thuy et al. 2019). Rainwater is simpler and easier for individuals to use to produce drinking water compared to river water, which requires treatment, and groundwater, which is contaminated by arsenic. Filtration and disinfection systems can make rainwater meet drinking water standards even though collected rainwater might contain pollutants from the atmosphere or roofs.

The rainwater for drinking (RFD) system, consisting of catchment, storage, and treatment components, was built in the Ly Nhan Health Care Facility (HCF) to produce drinking water. The water quality from the RFD system met all requirements by the drinking water standards of Vietnam, QCVN 01-1:2018 (Domestic water standard) and QCVN 06-1:2010/BYT (Drinking water standard). In this study, we focus on only the quantity of rainwater and not the water quality, which will be analyzed in future research. The performance of most RFD systems has been evaluated with certain fixed conditions, which result in a fragmented understanding of the performance and cannot provide suggestions to improve the operational efficiency. This type of conservative and passive performance evaluation is not reliable due to a lack of rainfall data (Ghisi et al. 2007) and could cause much overflow during the rainy season. No matter how well the catchment area and tank size of the system are designed, the system might be useless during unpredictable changes in rainfall if its operating performance is fixed. Performance improvement is critical for sustainably utilizing the RFD system, and this system needs to be evaluated hydrologically.

Hydrological evaluation methods have been reported in terms of reliability based on mass balance (Kim & Yoo 2009; Ward et al. 2010; Mun & Han 2012; Temesgen et al. 2015; Dao et al. 2017; Nguyen & Han 2017; Guo & Guo 2018). Reliability is the most suitable nondimensional indicator for quantitative analysis and is divided into two types, volumetric reliability and time reliability (Liaw & Tsai 2004; McMahon et al. 2006; Unaini et al. 2017), which are also considered as efficiency (Fewkes 1999; Palla et al. 2011) and security (Umapathi et al. 2019). Kim & Yoo (2009) evaluated three cases of an RWHS for nonpotable purposes in Korea using sensitivity analysis and presented the number of rainfall days and the amount of water consumption. Guo & Guo (2018) analyzed the water supply reliability of an RWHS in a humid and arid area and found that the relationship between reliability and the controlling factors is nonlinear. Those results presented the capacity of rainwater tanks considering water demand but not methods for improving operational performance. Therefore, this study aims to hydrologically evaluate RWHS and suggest operation methods to ameliorate performance under varying conditions.

Ly Nhan HCF is a part of the Ly Nhan District Preventative Medical Center in Ha Nam Province and is 90 km from Hanoi, Vietnam (Figure 1). Ly Nhan is located in the Red River region, and most of its groundwater is contaminated by arsenic (Berg et al. 2001; Nguyen et al. 2009). Ha Nam province has a tropical monsoon climate with 1,900 mm of annual rainfall, which is concentrated from May to September (People's Committee of Ha Nam Province 2016).

Seoul National University (SNU) cooperated with the World Health Organization representative office in Vietnam (WHO Vietnam) and the Vietnam Health Environment Management Agency (VIHEMA) to build an RWHS called the RFD system. In the first survey for understanding the current situation, they used to boil the water for the patients, which incurred energy and labor expenses. To install the system, which supplies 300 L of water per day, part of the roof was used for the catchment, four 4-ton tanks, and water treatment facilities, and drinking fountains were set in the building (Figure 2). This system is expected to enhance water, sanitation, and hygiene as a pilot implementation of the Water, Sanitation, and Hygiene-Facility Improvement Tool (WASH-FIT). Based on the lessons from this RFD system at Ly Nhan HCF, other RFD systems may be distributed to other medical centers in rural areas to supply safe, clean water.

The RFD system in Ly Nhan HCF consists of five parts, as shown in Figure 3: catchment, first flush tank, rainwater harvesting tank, rainwater treatment room, and drinking water fountains. Rainwater harvested from the roof flows to a downpipe through gutters. The downpipe connected to the first flush tank has a valve to control the flow of rainwater. When the storage tank is full, the valve will be opened to direct overflow away. A first flush tank removes the first flush of rainwater with contaminants because pollutants on a roof are mostly swept away by the initial rain. Four stainless steel tanks (4 m³ each) store the rainwater and have a drain valve at the bottom of the tank to drain sediments from the tanks. The four tanks-in-series improve water quality by successively removing sediments. To improve water quality, fiber filters remove residual sediment, and ozone is applied to remove odor and color in the ozone contact tanks at the water treatment room. The facilities in the water treatment room are equipped on the basis of requests by Vietnam agencies although the ozone contact tanks are unusual parts of the RWHS. The drinking fountains are distributed to several buildings and equipped with taps, nanofilters, and UV lamps to guarantee the best drinking water.

Daily rainfall data for Ly Nhan District were not obtained and were replaced by that for Hanoi in 2005 as an alternative (Figure 4). The amount of rainfall in Hanoi is normally between 1,200 and 2,300 mm/year, and the average annual rainfall is 1,665 mm based on records from 2002 to 2018 (General Statistics Office of Vietnam 2020). In 2005, the annual precipitation reached 1,355 mm at Lang station in Hanoi, which was the only daily rainfall we were able to obtain.

Understanding rainfall is critical to determine the optimum design for the RWHS. A well-designed RWHS allows people to use rainwater continuously all year in terms of supply.

An RFD system with low NWD and high RUR is better because it ensures a more stable supply and maximizes rainwater use with less overflow, which flows without use. After evaluating the RFD system at Ly Nhan HCF for the initial plan of 300 L/day of water demand, 400 m² of the catchment area, and 16 m³ of tanks, other scenarios with different water demands were simulated to suggest methods for achieving lower NWD and higher RUR.

The RFD system at Ly Nhan HCF obtains 22.2% annual RUR and zero NWD with a catchment area of 400 m², a tank volume of 16 m³, and a daily water consumption of 300 L when starting with fully filled tanks (Figure 6). NWD changed to 4 days in the case of starting with empty tanks, which means that the RFD system would be vulnerable to insufficient rainfall initially unless operation starts with full tanks or in the rainy season. An empty tank cannot supply water before it is filled with rainwater from rain events. Full tanks at the start work the same as pump priming, the process of introducing fluid into a pump to operate it.

As shown in Figure 6, the volume of stored rainwater changes daily with rainfall events and cannot store excessive rainwater, which affects the RUR value. The annual RUR is an indicator of where annual precipitation is evenly distributed throughout all 12 months. Monthly RUR, on the other hand, is more useful for understanding the RFD operation and to take countermeasures in areas where rainfall is concentrated within a few months. The RFD system at Ly Nhan HCF shows a broad range of RUR from a minimum of 5% to a maximum of 1,722% (Figure 7). The stored rainwater at the end of a previous month makes RUR high because more water is available than that collected in the month. Frequent overflow in the rainy season becomes the main factor for decreasing RUR; thus, complementary operation between the rainy season and the dry season is required to increase both monthly and annual RUR.

The two results above are based on given conditions. Therefore, sensitivity analysis is essential not only for evaluation but also high-performance operation under variable conditions. Primarily increasing water demand, decreasing precipitation, and broken systems threaten normal operation (Figure 8). The RUR and NWD should be calculated with variable external factors, rainfall, and water demand to determine countermeasures in terms of operation under various conditions. Two operation indicators of catchment area and tank volume were evaluated under varying annual precipitation of 1,200, 1,500, and 1,800 mm and water demand of 0.1, 0.3, and 0.5 m³/day.

To investigate the NWD and RUR related to the catchment area under variable water demands, tank volume was fixed at 16 m³, and the daily rainfall data in 2005 from Hanoi were applied. A water demand of 0.1 m³/day requires just 200 m² of the catchment area to achieve zero NWD, but a water demand of 0.5 m³/day requires more than 1,000 m² (Figure 9). Increasing catchment area without additional tanks can reduce NWD and RUR at the same time, but the larger the area is, the less effect it has on them. Considering the expenditure of expanding the catchment, 400–600 m² is a suitable range.

Figure 10 presents the effect of increasing tank volume, which decreases NWD and increases RUR. There is no change in RUR by tank volume when water demand is as low as 0.1 m³/day, but substantial changes in RUR occur when water demand is as high as 0.5 m³/day. As shown in Figures 8 and 9, 16 m³ of tank volume for 400 m² of the catchment area is suitable for the current water demand, which requires an additional 200 m² of the catchment area or two tanks at 4 m³ to decrease NWD from 58 to 26 when water demand increases by 0.5 m³/day. The RFD system should expand its catchment area or tank volume to achieve zero NWD, and the best method is determined by comparing cost and effect.

Figures 11 and 12 show how NWD and RUR change depending on the amount of annual rainfall. Those rainfall data are derived by multiplying certain rates to determine the desired rainfall, which is assumed to have the same pattern of rainfall being concentrated in a few months during the rainy season. The performance indicators were analyzed by assuming a water demand of 0.3 m³/day. The lowest rainfall represented extreme drought based on the annual rainfall data in Nam Dhin bordering Ly Nhan. As a result, 2 days of NWD occur when rainfall decreases by 1,000 mm, but it can become zero by increasing the catchment area by 10 m² or the tank volume by 1 m³. The current design is determined to be a nearly sufficient stable water supply even in a drought year.

The RFD system at Ly Nhan HCF was confirmed to have enough capacity to supply drinking water over 1 year. The system was also simulated under various conditions changing water demand and annual rainfall. Several methods for improving the hydrological performance of the system are suggested from the NWD and RUR perspectives. First, the tanks should be filled with water at the start of operation to achieve zero NWD. Most RWHS managers overlook utilizing other water resources because the RWHS is considered to use only rainfall in areas that suffer from water scarcity due to lack of clean water. Unless the RFD system is installed in the rainy season, a lack of rainwater is clearly predictable in the early period after establishment. Second, the catchment area determines the total amount of inflow to the system. A sufficient amount of large areas for collecting rainwater can overcome the limit of low precipitation if the costs are acceptable considering the cost and effect. Third, augmenting tank volume decreases NWD by storing the amount of water required during continuous days without rain. Finally, decreasing water use during drought is not recommended because the RFD system is designed for the amount of drinking water needed for humans, and this amount of water should be secured every day.

RUR indicates whether the system is being operated efficiently without wasting water, so the person in charge of operation and management needs to monitor this value closely and not use the yearly value. RUR is the ratio of total inflow and total water supply; in other words, RUR can be improved when total inflow is low and water supply increases. The rainfall in the dry season is used efficiently because of low precipitation, while that in the rainy season is wasted through overflow because water demand is fixed in the case of Ly Nhan. Therefore, the RUR could be improved seasonally, especially during monsoons (Figure 13). First, another water supply line must be connected before rainwater treatment to produce nonpotable water. Washing dishes, laundry, and cleaning consume more water than drinking, and their water source can be replaced with rainwater during monsoon, which causes substantial overflow. Second, rainwater can be stored and used for drinking in the dry season using a foldable plastic bag that can be stored when not in use eliminating the need for a permanent installation space.

We can determine the optimal catchment area, tank volume, and water consumption by considering the cost. The cost is calculated based on the expenses of the RFD system at Ly Nhan HCF, which are 1,000 USD for 100 m² of roofing and 385 USD for a 4 m³ tank. Table 1 shows the cost of various conditions and methods to respond to the change of increasing water demand, decreasing rainfall, or both. Even under conditions that have the same effect, the cost decreases by increasing the number of tanks. If additional roof areas are available, then the cost will decrease, but if not, then additional tanks are prioritized. When the RFD system needs to be revised due to changes in rainfall or water use patterns, installing a tank is suggested first before increasing the catchment area.

While RFD systems are very simple and easy to use, most of them are designed inaccurately due to a lack of hydrological consideration. Evaluation of the system performance is not useful and reliable because it focuses on the design and not the operation and management. An RWHS consisting of just a catchment and tank cannot satisfy the current water needs and still totally depends on the rainwater itself for quality and quantity. The new RFD system developed in this study is able to supply clean and safe drinking water, and its performance is evaluated by sensitivity analysis simulating different situations for appropriate operation during unfavorable conditions.

On the basis of the sensitivity analysis, we suggest the method of improving performance to enable zero NWD and increased RUR. It is fundamentally possible to enhance the performance by changing the catchment area and tank volume; in particular, extending the catchment area decreases both NWD and RUR, and increasing the tank volume facilitates a low NWD and high RUR during drought and high water demand. In this study, tank installation is recommended prior to building a new catchment area when water demand is higher than the amount of collected rainwater. Another water supply line is also advantageous for reducing overflow waste and preparing for unexpected obstacles.

Every RFD system should be carefully designed using sensitivity analysis for operation under any given situation. Moreover, monthly monitoring of RUR and NWD is important for performance improvement and reflects the value of operation. Daily records of rainfall, water use pattern, and volume of water in the tank should be recorded to support better operation of the RFD system considering seasonal management. We advise that various RFD systems should be evaluated using sensitivity analysis to develop the appropriate RFD system guidelines reflecting the characteristics of each region and seasonal management.

This research was supported by Science and Technology Support Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (MSIT) (NRF-2018K1A3A9A04000025) and the Institute of Construction and Environmental Engineering at SNU. The authors wish to express their gratitude for the support. This paper was supported by the KOICA/WFK Scholarship funded by the Korea International Cooperation Agency (2020-00201).

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