Evaluation of rainwater harvesting system in university buildings for non-potable water demand

Abstract

Urban water demand for potable water is significantly increasing due to population growth. Harvesting rainwater can reduce the pressure on the supply main and groundwater usage for non-potable water usage. In this study, the SWMM LID modelling framework was used to investigate the rainwater harvesting potential of 5 academic buildings on the university campus to compensate for daily non-potable water usage such as toilet flushing, lawn irrigation and car or bus washing. The water saving efficiency (WSE) of the RWH system for two different storage scenarios: (i) rain barrel only (50000 L), and (ii) rain barrel and underground (UG) water tank (140000) was estimated by simulating a hydraulic model using historical daily rainfall data (1982–2021). The results suggest that the WSE of non-potable water demand is greatly influenced by storage capacity, number of users, and area of usage. For toilet flushing, 29% WSE is obtained for 100 toilet users daily. WSE varies from 34.3 to 100% for irrigation depending on storage volume of 50,000 L and 140,000 L, respectively. Scenario 1 achieves 51.3–100% WSE for car and bus washing water demand for 10 buses and 10 cars, respectively. Scenario 2 fulfils the water demand for 28 cars or 14 buses. The results also showed that a cost savings of $23 for scenario 1 and $63.41 for scenario 2 could be achieved if $0.45 per 1000 L water price is considered with pumping cost, but a higher return can be achieved if drinking water requirement can be fulfilled with harvested rainwater.

Background

Water stress and scarcity are among the major growing concerns worldwide, leading to more scientific research to tackle the issue. Relying on underground freshwater resources only harms the environment¹. Widespread urbanization has become one of the prime causes of water shortage in many municipalities around the world. Changes in precipitation patterns and increasing temperatures also significantly affect the water supply efficiency. A combination of climate change impacts with uncontrolled urbanization in future might create a significant imbalance between water demand and supply, resulting in severe water scarcity in developed urban areas². Residential rainwater system analysis in Australia indicated a reduction of 2–14% in water savings under future climatic conditions³. Freshwater resources are constantly disappearing as infrastructure development increases to accommodate the growing population. Most of the land surfaces in urban areas are covered with impervious layers due to the construction of residential and non-residential structures, roads, and other concreted areas, increasing the possibility of excessive runoff and flooding⁴. This urban development directly contributes to high soil impermeability, resulting in more frequent flood events⁵. Runoff velocity increases as there exists a uniform slope throughout with greater flow volumes. This runoff carrying significant pollutants from road surfaces deteriorates water quality in receiving water bodies. Water bodies are often contaminated by transition metals in South Asian countries such as Bangladesh, India, Pakistan, and Nepal due to the overproduction of various industrial commodities and inadequate sewage treatment methods⁶. Adverse impacts on aquatic diversity and threats to human health, including carcinogenic effects, are consequences of a range of pollutants coming from urban stormwater runoff⁷. Low impact development (LID) is an alternative design approach to the traditional stormwater management system, which offers significant control over the flood discharge and peak flow⁸. One of the most common LID practices includes rain barrels that adopt the utilization of rainwater storage tanks to promote rainwater harvesting and delay flood peaks in urban watersheds⁹. The requirement for a rainwater harvesting (RWH) system has been engendered due to the increasing pressure on water supply due to excessive population growth in urban areas as well as diminishing sources of fresh water. As a sustainable solution, many countries are adopting RWH to reduce pressure on fresh water supply¹⁰. RWH system consists of collecting rainwater from impervious areas, storage, treatment in case of potable use, and supply from the storage tank for potential usage purposes¹¹¹². has carried out an spreadsheet based analysis of RWH potential large roofs under different climate scenarios considering two underground water tanks in Australia. The study emphasized on optimization of rainwater tank size with varying roof sizes with the developed daily water balance model. In an economic analysis of RWH system in Portugal, a 10% discount rate was found enough to make investment in RWH financially viable¹³. Another study was performed in Dhaka, the capital city of Bangladesh, where RWH potential of commercial buildings were analyzed with roof area ranging from 315 to 776 sqm and rainwater tank capacity varying from 100 to 600 cubic meter. Shorter payback period of 2.25–3.75 years for RWH system were found due to higher cost of water and energy in the industrial sector considering usage profile of 30 L per capita per day¹⁴.

Two types of RWH systems, passive and active, can be installed based on the different needs. Passive rainwater harvesting includes receiving the rainwater and allowing it to flow over the land to a detention pond where water will have a chance to infiltrate underground. The overland flow will help water the existing plants, reducing the irrigation water demands. Active RWH system captures the rooftop runoff, stores it and keeps it available for fulfilling potable and non-potable water demands such as agriculture, toilet flushing, laundry, car washing, etc¹⁵.

In this study, the rainwater harvesting potential of buildings on a university campus using the rain barrel option in the Stormwater Management Model (SWMM) is analyzed to predict the utilization of rainwater in various non-potable water usage, including car washing, irrigation and toilet flush. Rainwater harvesting potential is measured based on the water demand scenario of each water usage. Water supply efficiency (WSE) and reliability percentage are calculated considering two rainwater storage options: rain barrels only and rain barrels with underground (UG) water tanks.

In the rainy season, the SUB University campus gets flooded as there is an enormous amount of water coming from the high hilly area beside the campus. It is not due to the rainwater falling on the campus area but runoff from adjoining lands for some drainage issues. The first consideration was to apply low-impact development (LID) technology to reduce runoff. Before investigating the effect of LID technologies on manipulating stormwater runoff, assessing the impact of rainwater harvesting systems to subside groundwater usage was mandatory. Moreover, the campus has an inadequate drainage network to manage stormwater runoff. This prompted the development of a hydraulic model to analyze drainage network capacity in various rainfall events and also opened the opportunity to assess the volume of rainwater that can be harvested from the university buildings. As the university has a lawn for green areas as an aesthetic environment with gardening around the campus, which requires irrigation water and buses for transportation, it is a requirement that the rainwater usage includes irrigation water demand and car washing demand. Besides, toilet flushing as an inevitable non-potable water demand is also contained in the study as it is incorporated in most of the potential studies on rainwater harvesting.

Materials and methods

Study watershed

Southern University Bangladesh is a private university located in Chattogram, Bangladesh. The university’s main campus is the study area for assessing the hydrologic performance of different low-impact development modules. The campus is comprised of five academic buildings with library and office facilities. The total roof areas of the buildings are 376.8 sqm, 204.8 sqm, 352.3 sqm, 391.6 sqm, and 303.7 sqm for B1, B2, B3, B4, and B5, respectively.

Input data Preparation

A detailed layout of the study area was collected from the university authority, where seven different land use categories are identified, as shown in Fig. 1. Academic buildings cover 31% of the total area. 33% of the area comprises green areas with a grass-covered lawn, a small field, gardens in front of buildings, and grass cover in the south-side slope-protected area. The rest of the area includes a brick soling road with masonry structures, concrete surface, and earth. The combined sewer network of the study area is presented in Fig. 2 below. Only three significant channels serve stormwater and sewer water discharging purposes. The north-side drainage channel is an underground conduit with a pit discharging water only from the B1 building. The open channel between buildings B2 and B3 is partly covered with a concrete slab for the road, and the rest is open. The drainage channel from building B5 to B4 is an underground circular pipe which serves as a sewage pipe. The drainage channel on the west side of building B4 is an open channel ending at the southeast outlet.

Fig. 1.

Drainage network of the study area (Layout created by using AutoCAD version 2007, available at: https://www.autodesk.com/products/autocad/overview).

Fig. 2.

Variation of annual rainfall and annual rainy days from 1982 to 2021.

Rainfall data

Daily rainfall data records from 1982 to 2021 were obtained from the Bangladesh Meteorological Department near Ambagan station, Chattogram. This station is approximately 7.4 km from the study area. The total rainfall (mm) in each year is presented in Fig. 2.

Rain barrel and UG water tank placement in the study area

Two arrangements are considered for placing storage tanks for rainwater capturing. Figure 3a represents the plastic rainwater tanks placed besides each university campus building.

Fig. 3.

(a) Placement of rain barrel for RWH (Layout created by using AutoCAD version 2007, available at: https://www.autodesk.com/products/autocad/overview). (b) Placement of rain barrel and UG water tank for RWH (Layout created by using AutoCAD version 2007, available at: https://www.autodesk.com/products/autocad/overview).

For building B1, two 3000-litre barrels are considered as the building is a one-story academic building with classrooms, and its plinth level is 0.5 m below the level of the parking area on the right. Thus, the roof area of this building is comparatively lower than the other buildings, making it unsuitable for water tanks larger than 3000 L, which have a convenient height for capturing water from the roof. Building B2 is a two-story library building with an admission office on the ground floor and a reading room on the rooftop. For this building, a 5000-litre tank is placed. Building B3 is a one-storied laboratory building for which a 5000-litre water tank is considered. Buildings B4 and B5 are five-story structures, with B4 featuring a steel sheet roof and B5 a concrete roof. Two 10,000-litre tanks were considered for increasing rainwater capture potential.

In Fig. 3b, the rainwater tanks for building B2 and B3 are kept the same, but three underground water tanks are considered for building B1, B4, and B5. For building B1, an underground water tank with a length of 6.1 m, breadth of 3 m, and depth of 2.1 m is considered with a water tank capacity of 30,000 L. Two 50,000 L underground water tanks are evaluated for each building B4 and B5, whose length is 9.1 m, breadth 3.7 m, and depth 1.8 m. The water tank sizes are conceptualized based on available space beside those buildings.

Usage profile of harvested rainwater

Most rainwater harvesting systems analyze the efficiency of water usage for non-potable uses such as irrigation, toilet flushing, etc^16,17.In this study, three different water use scenarios based on the volume of rainwater harvested from each building are considered, including lawn irrigation, car washing and toilet flushing. This analysis will provide the advantage of individual buildings’ RWH system performance. For toilet flushing, a flush tank of 6 L per flush is considered with a usage rate of 2 times per person per day with a monthly operation of 30. The irrigation plan for the green area is assumed to be 15 days a month with a water demand of 1250.5 L per day, which comes from gardening water requirement of 15 L per square meter per week, according to BNBC 2020.

Given that the university provides buses and cars for the transportation of students, faculty, and staff, using rainwater for car washing is also an expected practice.

Here, Monthly washing days are considered only two days. The volume of water required for buses and cars is 365.5 L and 188.5 L, respectively¹⁸. The actual usage profile may differ and can be assessed if metered, even if these are standard numbers used for analysis. A 10% loss as a first flush in harvested rainwater is considered in analysis for removal of excess contaminants and maintenance purpose.

Simulation model

In this study, the Stormwater Management Model (SWMM) is used to evaluate the potential of rainwater harvesting in the Southern University Bangladesh campus. SWMM simulates low impact development (LID) scenarios by simulating rainfall-runoff process¹⁹. To properly evaluate the stormwater runoff scenario, it is necessary to evaluate the exact landuse type of the study area. It will help in investigating existing drainage network and the other LID practices in future. The study area is divided into 76 sub catchments based on the type of land cover. Subcatchments S1 to S5 represent the academic buildings which are considered for the application of RWH system. SWMM uses the nonlinear reservoir model to estimate the amount of runoff generated using the Manning equation²⁰. There are two different approaches to placing rain barrels as a LID control. LID controls can be assigned within a sub-catchment to receive runoff generated from the impervious areas, and another is to define a single LID control as a whole sub catchment where receives runoff from both precipitation and another impervious surface²¹. The first approach is adopted in this study. Two LID scenarios are considered for assessment of rainwater harvesting potential: (i) Rain barrels only and (ii) rain barrels with UG water tank. The deployment of rain barrels and UG water tanks in the sub catchments is presented in Table 1. The model simulation is carried out for the years 2007–2021 as these years represent a similar precipitation trend for 14 years. The year-wise simulation included the dry months, defined as periods when rainfall was absent or the amount was insufficient to meet the rainwater utilization demand. Furthermore, due to the inter-annual variability of monthly rainfall, the results for irrigation water demand are presented considering a range of four to ten dry months per year.

Table 1.

Rain barrel and UG water tank specification in sub-catchments.

Sub-catchments	Rain barrel only		Rain barrel and UG water tank
	Reservoir type	The volume of rain barrel or UG water tank, litres	Reservoir type	The volume of rain barrel or UG water tank, litres
S1	Rain barrel	6000	UG water tank	30,000
S2	Rain barrel	5000	Rain barrel	5000
S3	Rain barrel	5000	Rain barrel	5000
S4	Rain barrel	20,000	UG water tank	50,000
S5	Rain barrel	20,000	UG water tank	50,000

The WSE (%) is defined as the percentage of the ratio of water demand fulfilled by harvested rainwater for each year divided by the total water demand for a particular use in that year.

Where Y_t is the water demand fulfilled by rainwater from the reservoirs in litres, and D_t is the water demand in a year.

Reliability is defined as the ratio of a total number of days’ water demand of a particular use fulfilled and the total number of days. It gives an idea about the percentage of time the system will meet the water demand (Fig. 4).

Fig. 4.

SWMM model of the study area (Model created by using EPA SWMM 5.2 software²².

Results and discussion

Variations in rainfall

An analysis of annual rainfall and the number of rainy days from 1982 to 2021 reveals a continuous upward trend in rainfall after 2007, with the number of rainy days consistently exceeding 80, except in 2016, when the number of rainy days was approximately 60.

²³ represented the rainfall type based on the annual rainfall intensity expressed as “heavy” to “scanty”. Heavy rainfall represents rainfall intensity greater than 2000 mm/yr, whereas rainfall less than 500 mm/yr is termed as scanty rainfall. Rainfall intensity between 1000 and 2000 mm/yr is moderate rainfall and rainfall between 500 and 1000 mm/yr denotes less rainfall type.

Analysis of the rainfall data set from 1982 to 2021 shows that after 2006 there exists a somewhat continuously fluctuated rainfall pattern which is all above the threshold of 2000 mm indicating heavy rainfall during this period except a moderate rainfall in 2016.

Another way is to classify the seasonal rainfall as excess, normal or deficient. This procedure is based on the long-period average (LPA) and coefficient of variation (CV). Rainfall of a year is termed “excess” when the total rainfall depth is more significant than (LPA + CV), and if the rainfall depth is less than (LPA – CV), then it is called “deficient” rainfall. Rainfall between (LPA + CV) and (LPA – CV) is termed “normal” rainfall for that year²³. From the previous analysis of rainfall, as heavy rainfall was observed from 2007, this classification regarding LPA and CV was carried out for the rainfall data from 2007 to 2021 (Figs. 5 and 6).

Fig. 5.

Rainfall classification expressed as “heavy” to “scanty”.

Fig. 6.

Variation of annual rainfall within the interval set by long period average and coefficient of variation (CV).

There is no direct pattern observed in rainfall types based on LPA and CV. Only five years represents excess rainfall which has total rainfall depth over maximum value of (LPA + CV) equal to 3254.77 mm and four years with moderate rainfall over 2695.75 mm but less than 3254.77 mm.

Water saving efficiency of harvested rainwater

This study illustrates the potential of rainwater harvesting for each individual building so that it becomes easier to understand the water-saving efficiency and reliability of each building during the planning and implementation of the rainwater harvesting system.

Rainwater harvesting potential of rain barrels only

This Fig. 7 illustrates the impacts of toilet flushing on the WSE of the RWH system of each of the buildings on the campus, with the number of toilet users varying from 10 to 100 persons per day. As the number of persons in a university is variable, analysis is based on maximum 100 persons per day for each buildings. Moreover, the university has eight departments with four departments have week days on Friday and other four has week days on Saturday. So number of weekdays in a month were not considered for analysis. Based on the size of the roof area for rainwater harvesting, it can be observed that buildings B4 and B5, representing the S4 and S5 subcatchments in the SWMM model of the campus area, have higher WSE with respect to other buildings due to the placement of large capacity (20000 L rain barrel) UG water tanks beside those buildings. 12 L flush tanks are considered for toilet bowls with the usage of 6 L water per flush. Daily usage of 2 times per person per day is considered in the calculation of WSE. Around 42% WSE can be achieved for an average of 10 toilet users per day, which reduces to around 4% for an average of 100 toilet users per day in building B4 and B5.

Fig. 7.

Annual water supply efficiency for toilet flushing in each sub-catchment.

For other buildings, WSE percentage is relatively low, around 12–1%, due to the unavailability of space to place larger capacity water tanks. Figure 8 illustrates the WSE of each subcatchment based on the number of dry months when irrigation water is required for the lawn area, represented by subcatchments S6 and S7 in the SWMM model with a combined area of 583.5 sqm. The water demand for each irrigation day is 1250.5 L. The water requirement for irrigation in a year is 75,030 L for four dry months and 187,575 L for ten dry months. Here, monthly irrigation days considered is every alternate day, which is 15 days a month. It is obvious that with the increase in a number of dry months, WSE decreases up to 2.4%. With the application of a 20,000 L rain barrel in S4 and S5, WSE reaches a maximum of 24% for four dry months and a minimum of 9.6% for ten dry months. For a water tank capacity of 5000 L applied in both B2 and B3, the maximum and minimum WSE ranges from 6 to 2.4%. WSE will increase with reduced irrigation water requirement if the sprinkler system increases the water application efficiency. Buses and cars are mandatory transportation mediums in the university, which require washing due to their daily usage.

Fig. 8.

Annual water supply efficiency for irrigation of lawn in each sub-catchment.

With 365.5 L of washing water requirement for buses and 188.5 L for cars, it is seen from Fig. 9a that WSE values were obtained for the sample buildings, with Building B4 and B5 exhibiting a WSE of 40%. In comparison, a lower WSE, ranging from 20 to 25%, was recorded for Buildings B1, B2, and B3. These results were determined based on a scenario of washing five cars per month distributed across two distinct washing days. For washing five buses 2 times a month, it reduces to 10–21% respectively. As the university currently owns three cars and three buses, the WSE for car and bus washing are close to 40%, and 20% can be achieved easily with a 5000-litre rain barrel capacity.

Fig. 9.

(a) Annual water supply efficiency for car washing in each sub-catchment. (b) Annual water supply efficiency for bus washing in each sub-catchment.

Rainwater harvesting potential of rain barrels and UG water tank

With a UG water tank of 55,000 L, it can be observed from Fig. 10 that the toilet flushing requirement can be fulfilled with a WSE of around 57% for daily usage of 2 times for 20 persons for academic building S4 and S5, with a WSE of 36% for academic building S1. This result is considerably flat for buildings S2 and S3 due to their small roof area, reducing the potential of more runoff for storage. As expected in all buildings, the WSE gradually reduces as the number of toilet users increases. As the number of users is not usually going to be constant every day in each building, the WSE can be higher in each building with a reduction in a number of users and times of use. In Fig. 11, WSE for lawn irrigation for a water storage capacity of 55,000 L UG water tank applied in subcatchments S4 and S5 with a 10% loss for first flush is perceived to vary from almost 26% for ten dry months to 66% for four dry months. If rainwater harvesting is only applied with a 35,000 L UG water tank for sub catchment S1 representing academic building B1, then this building can harvest 31,500 L of rainwater per year with WSE of 42% for four dry months and WSE of 16.8% for ten dry months. WSE of Academic buildings B2 and B3 vary from 2.4 to 6%. Possibility of reaching 100% WSE can be achieved from academic buildings B1, B4 and B5 only. Here the number of cars is increased as the capacity of the UG water tank applied is higher than the rain barrels applied in the previous scenario (Fig. 12).

Fig. 10.

Annual water supply efficiency for toilet flushing in each sub-catchment.

Fig. 11.

Annual water supply efficiency for irrigation of lawn in each sub-catchment for different number of dry months.

Fig. 12.

(a) Annual water supply efficiency for car washing in each sub-catchment. (b) Annual water supply efficiency for bus washing in each sub-catchment.

In the case of car washing, almost 99.5% WSE can be achieved for seven cars with two washing days per month by harvesting rainwater from sub catchment S1 representing academic building B1, while three buses can be washed fully with the same amount of water with a WSE of 100%. Almost 50% and more of the WSE is achieved with the help of rainwater harvested from academic buildings B1, B4 and B5, for 15 cars, while this WSE percentage reduces to approximately 23–38%, respectively, for 15 buses. For academic buildings B4 and B5, if five cars and five buses are considered, the water demand for washing two times a month can be fulfilled.

Reliability of the rainwater harvesting system

Reliability percentage is measured by taking the percentage of days water demand is fulfilled by harvested rainwater for each particular use¹¹. The university can accommodate a maximum water storage capacity of almost 50,000 L with an underground water tank applied in each sub catchments S4 and S5. Figure 13a illustrates the number of days water demand for toilet flushing can be fulfilled in a year based on the average number of toilet users per day, and the related time-based reliability is presented in the adjacent Fig. 13b. From the Fig. 13a and b, it can be observed that with a water storage capacity of 50,000 L and ten toilet users per day, 100% rliability can be achieved. However, with increasing toilet users, serving days reduces to 75 days for 50 toilet users giving a reliability percentage of 21%. Suppose the volume of rainwater harvested can be increased up to 140,000 L by implementing the rainwater barrel and UG water tank with 10% loss, 29% reliability can be achieved for toilet flushing water demand for 100 toilet users per day which comes down to 12% for scenario one where only rain barrels are used with water storage volume of 56,000 L. Figure 14 portrays the reliability of storage volume based on the number of dry months in a year. 60–30% reliability can be achieved by implication of a 50,000 L storage tank if dry months range from 4 to 8 months respectively, for a water demand of 1250.5 L per day for 15 days a month where the irrigable land surface is 583.5 sqm. Applying the full rainwater storage capacity (140,000 L) across all academic buildings results in an optimal reliability of nearly 100% when encountering seven dry months annually. This reliability, however, decreases to 67% when the number of consecutive dry months extends to ten. From Fig. 15, the reliability percentage of harvested rainwater for washing combinations of cars and buses can be predicted. While the number of cars and buses are both 5, there is a possibility of reaching 68% reliability with a storage volume of 50,000 L. For up to 15 cars and buses, reliability can be increased from 23 to 63% by applying the rainwater storage volume of 50,000 L to 140,000 L, respectively. More cars and buses decline the reliability below 50%.

Fig. 13.

(a) Number of toilet users vs. water demand fulfilled in days. (b) RWH system reliability concerning storage volume for toilet flushing.

Fig. 14.

RWH system reliability concerning storage volume for lawn irrigation.

Fig. 15.

RWH system reliability with respect to storage volume for washing cars and buses.

Cost analysis

The performance of rain barrels varies based on rainfall intensity, from deficient to excess rainfall. From 1982 to 2021, there was somewhat heavy rainfall from 2007, but this scenario changes for rainfall classification according to LPA. Based on the total inflow dynamics to the Rain Barrel LID unit, as documented in Fig. 16 across the 2007–2021 simulation, it is concluded that a twofold increase in the final storage volume was not achieved. Therefore, there is no possibility for the rain barrels and UG water tanks to be filled fully with rainwater two times a year. Considering this fact, WSE, reliability and finally cost savings are calculated considering one time filling of the storage volume for each building rainwater barrels and UG water tanks (Fig. 17).

Fig. 16.

Total inflow volume compared to single fill volume and double fill volume in rain barrel simulation from 2007 to 2021.

Fig. 17.

Cost savings per year related to storage volume for each academic building.

From the analysis, $2.45 and $14.32 can be saved in a year by harvesting water in building B1 with a storage volumes of 6000 L and 35,000 L, respectively. For buildings B2 and B3, cost savings are $2.05 as the feasibility of placing large rain barrels or UG water tanks near those buildings is limited. Storage volume of 20,000 L with rain barrels and 50,000 L with UG water tanks can be installed near building B4 and B5, which increases the yearly cost savings to $8.18 and $22.50, respectively. Though the cost savings are not that significant, the non-potable water demand fulfilled by this rainwater will indeed reduce pressure on fresh water demand in the supply chain. The cost-benefit analysis is not performed as the university has limited feasibility for larger storage volumes resulting in higher payback periods.

The university is currently sourcing drinking water (9500 L per month costing $227.28 per month) from outside vendors. If the harvested rainwater can be treated for drinking, cost savings per year will be $1205.75 for scenario one and $2727.30 for scenario 2.

Conclusions

This study analyzed the rainwater harvesting system from the rooftop of five academic buildings for non-potable water usage purposes including toilet flushing, lawn irrigation, and car washing. The water demand scenarios were based on two types of storage systems: rain barrel only and a combination of the rain barrel and UG water tank. A hydraulic model was developed in SWMM to quantify the amount of rainwater inflow in rainwater storage reservoirs.

The rainfall analysis specifies that there is a heavy rainfall trend defining rainfall greater than 2000 mm observed for the last 15 years. However, based on the LPA, the rainfall classification is irregular. The university campus’s hydraulic model was simulated for 2007–2021 to evaluate the amount of rainwater captured. The rain barrel only scenario results indicate that WSE for storage volume of 50,000 L could be obtained around 10–100% for toilet flushing demand if the number of toilet users varies from 100 to 10 respectively. WSE for a rain barrel and UG water tank scenario (140000 L storage volume) reaches around 97% for an average of 30 toilet users per day and reduces to around 29% for an average of 100 toilet users per day. This WSE is computed considering 6 L per flush toilet tank which can be improved using dual flush toilet tank where per flush water volume for urinal will be 3–4.5 L. For irrigation purposes, WSE vastly depends on the number of dry months. Analysis of the rainfall pattern shows an average of 6–7 dry months in the years 2007–2021, for that WSE obtained for law irrigation are around 40–34.3%, respectively, for a storage volume of 50,000 L. For scenario 2, where rain barrels and UG water tanks are used, WSE can be increased up to 96–100% for lawn irrigation if the storage volume is increased to 140,000 L. In scenario 1, WSE for car washing is obtained around 100% for ten cars with a washing water demand of 188.5 L. For ten buses, 51.3% WSE is obtained considering 365.5 L of water demand for bus washing. As with scenario 2 with higher storage volume, around 100% water demand can be fulfilled for washing 28 cars or 14 buses. The reliability percentage for scenario 2 with a storage volume of 140,000 L varies from 97 to 29% for 30–100 toilet users, respectively. These percentages change to 35% and 10% for 50,000 L storage capacity. With 6–7 dry months in a year, the average reliability obtained for lawn irrigation is around 37%. Approximately 95% reliability is achieved for washing a combination of 10 cars and buses for scenario 2. For scenario 1, only 68% reliability is achieved for a combination of 5 cars and buses. The annual cost saving of the RWH system is quantitatively estimated at US$23. This estimation is based on the simulated RW storage volume of 50,000 L and a singular annual system replenishment cycle. The calculation incorporates a municipal water price of $0.45 per 1000 L, including the associated pumping cost from the supply main. If the storage volume is increased to 140,000 L as in scenario 2, cost savings per year will be $63.41. Though the return in harvesting rainwater for non-potable water usage is insignificant, this cost savings can be increased to a greater amount of $1205.75–$2727.30 for harvesting volumes of 50,000 L and 140,000 L respectively for drinking water. Inclusion of rain barrel and UG water tank has a moderate impact on the cost savings but there is significant drawback on payback period of the investment.

Rainwater harvesting system analysis will yield more accurate results if future rainfall events due to climate change can be known. Moreover, storage volume and longer dry seasons play a crucial role in RWH system, as with less rainfall events with limited storage volume will only satisfy a small portion of the daily non-potable water demand. Evaluating RWH system performance after installation in one building in future will also help in decision making for establishment of other RW storage tanks leading to sustainable use of water in the university campus. The results provided here give insights of non-potable water usage with RW harvesting reducing pressure on mains water supply. These results will be beneficial for other educational as well as industrial infrastructures for implementing sustainable solutions to water usage and management.

Author contributions

Data collection and model study and drafting of the research work: Mohammad Ayanul Huq ChowdhuryConceptualization, supervision and review of the draft: Aysha Akter.

Funding

Appreciation goes to the Department of Civil Engineering, Chittagong University of Engineering & Technology (CUET) for the financial support to conduct a PhD study.

Data availability

Data will be available on request: aysha_akter@cuet.ac.bd; aysha_akter@yahoo.com.

Declarations

Competing interests

The authors declare no competing interests.