Potable Water Savings Potential Through Rainwater Harvesting in a Brazilian Fitness Centre: A Case Study

Abstract

Water scarcity and rising urban demand pose growing challenges for sustainable water management in Brazil, where over 73 million people may face shortages by 2035. Given this scenario, rainwater utilisation has emerged as a strategic alternative for preserving water resources, helping to reduce potable water consumption and relieving demand on public supply systems. This study aimed to evaluate the potential for potable water savings through the implementation of a rainwater harvesting system in a fitness centre without a swimming pool, located in southern Brazil—a building typology rarely addressed in the literature. Water end-uses were empirically characterised using water flow measurements and questionnaires conducted in an existing facility operated by the same franchise. A daily balance simulation was performed using the Netuno computer programme (Version 4), and an economic feasibility assessment was conducted based on local costs and tariff structures. The results showed that non-potable end-uses represented 24.4% of total water consumption. The rainwater harvesting simulation indicated an ideal tank capacity of 11,000 L, enabling potable water savings of 7.04%. The economic analysis showed an implementation cost of R$13,240.72 and a consequent return on investment of fifteen months. These findings confirm the technical and economic viability of rainwater harvesting systems for fitness centres and highlight the relevance of local conditions in shaping performance and investment returns.

1. Introduction

In Brazil, water consumption is influenced by significant regional disparities in resource availability and population distribution. While the Southeast and Northeast regions account for 69% of the population, they contain only 10% of the nation’s water resources. Conversely, the Amazon Basin contains nearly 80% of surface water, despite its low population density [1,2]. These disparities are influenced by socioeconomic conditions, micrometering coverage, and the operational efficiency of local water utilities [3,4]. In 2022, the national average per capita water consumption was 149.8 L/day, while in Palhoça, the study location in the state of Santa Catarina, it reached 221.0 L/day. Only 75.9% of the local population is connected to the public water supply, and 15% of people have access to centralised sewerage services [5]. This situation underscores the urgency of reducing potable water demand and promoting more sustainable water use in buildings.

Rainwater harvesting (RWH) systems are recognised as effective alternatives to reduce potable water demand, particularly for non-potable uses such as toilet flushing, cleaning, and irrigation [6,7,8,9]. In institutional environments, non-potable uses account for a substantial share of total water consumption. Prior studies [10,11] reported that toilets, showers and irrigation account for a significant share of water consumption in public, office, and educational buildings. A meta-analysis by Teston et al. [12] indicated that non-potable uses represent 46.6% of total consumption in residential buildings, 53.9% in schools, 69.6% in office buildings, and up to 77% in universities, reinforcing the high potential for potable water savings.

Beyond reducing potable water demand, rainwater harvesting systems alleviate pressure on supply infrastructure, strengthen urban water resilience, and support climate change adaptation [13,14,15]. Public financial incentives can promote the adoption of RWH and more efficient water use [16]. Nevertheless, regulatory frameworks and public policies regarding rainwater harvesting vary significantly across national contexts, influenced by diverse institutional arrangements, technical requirements, and climatic conditions. In Brazil, regulation is fragmented and primarily based on non-mandatory standards, such as ABNT NBR 15527 [17], complemented by decentralised state and municipal initiatives [18]. In contrast, countries such as Germany and Australia have adopted mandatory and technically robust regulatory mechanisms, enabling a broader adoption of rainwater harvesting in both residential and commercial buildings [19,20].

Storage tank sizing is a critical determinant of RWH performance, as it directly influences the water-saving potential, reliability, and investment cost [21,22,23,24]. Several sizing approaches exist, with daily water balance models being widely adopted for their ability to simulate inflows, outflows, and storage dynamics under variable rainfall conditions. Among these, the Yield After Spillage (YAS) and Yield Before Spillage (YBS) algorithms are most commonly used. The YAS method satisfies demand after overflow and yields conservative savings; YBS satisfies demand before overflow and often yields higher savings. The choice of algorithm has a significant impact on both hydrological and economic evaluations, particularly for smaller storage capacities [25]. The Netuno programme, developed by Cordova and Ghisi [26], applies daily water balance simulations to assess system performance across different combinations of catchment areas, storage capacities, demand scenarios, and historical rainfall data. Geraldi and Ghisi [27] recommend using rainfall data spanning at least 10 years to ensure modelling accuracy. Good practices also include the proper definition of the first flush volume, alignment between catchment potential and water demand, and preference for continuous simulations over simplified methods.

Characterising water end-uses is a critical step for accurate demand modelling and for evaluating the feasibility of using rainwater in buildings. This can be performed using user interviews and questionnaires [28], combined with total consumption records and water flow measurements [29], or via real-time smart metering approaches [30,31]. However, methods based solely on self-reported data are prone to bias: lower-income respondents tend to overestimate consumption, whereas higher-income respondents often underestimate it [31]. Incorporating socio-demographic information alongside monitored data can thus improve end-use estimates.

A study conducted in 62 cities in southern Brazil found that potable water savings in residential buildings ranged from 34% to 92%, depending on the demand [32]. In another literature review, savings potential across different building typologies varied from 9.2% to 57.2%, with reliability levels exceeding 75% in most cases [12]. These results depend on multiple factors, particularly potable water demand, rainfall availability, number of occupants, and roof catchment area. In coastal Santa Catarina, consistent rainfall and large roof catchments create favourable technical and economic conditions for rainwater harvesting [33].

Despite extensive research on water end-uses in residential, institutional, and educational buildings, few studies have explicitly examined fitness centres without swimming pools. In these facilities, water use concentrates in showers, toilet and urinal flushing, and lavatory taps. A study conducted in a municipal sports complex by Silva et al. [34] found that showers accounted for the largest share of water use, even when there were swimming pools. Although showers dominate consumption profiles, the volume allocated to non-potable applications remains significant, particularly for flushing toilets and urinals. These characteristics highlight the considerable potential for potable water savings in fitness centres. The scarcity of empirical studies on this typology in Brazil underscores the relevance of the present case, which assesses the feasibility of rainwater harvesting in a fitness-centre context.

Pressures on water resources and regional imbalances between supply and demand underscore the need for decentralised water solutions. This study assesses the potential for potable water savings through a rainwater harvesting system designed for a fitness centre without a swimming pool in southern Brazil. The analysis combines: (i) empirical end-use characterisation; (ii) daily balance simulation using Netuno; and (iii) an economic model aligned with local tariff structures. The contribution lies in documenting water use patterns in a rarely studied building typology, estimating the non-potable demand that can be met with rainwater, and benchmarking economic performance—based on payback and internal rate of return (IRR)—against recent case studies. These comparisons clarify how context-specific costs and tariffs influence the viability of rainwater harvesting systems.

2. Materials and Methods

A systematic approach integrated technical, hydrological, behavioural, and economic parameters to evaluate the rainwater harvesting system. Figure 1 shows the methodological steps employed in this study.

2.1. The Fitness Centre

The case study involves the establishment of a fitness centre within a commercial development under construction in Palhoça, a coastal municipality in the metropolitan region of Florianópolis, southern Brazil. The city has a humid temperate climate zone, between parallels 27°36′ and 27°55′ south latitude and meridians 48°33′ and 48°45′ west longitude. Palhoça has a population of approximately 222,598 inhabitants, distributed across an area of 394.85 km², with 92.7% of households connected to the public water supply system [35].

The development, which was still under construction at the time of the study, will have a total built area of 3602 m², distributed across three levels: ground floor, upper floor, and rooftop. The ground floor will feature eight commercial rooms, ranging in size from 74.40 m² to 168.36 m², as well as a 42.76 m² access hall to the fitness centre. The upper floor, with a total area of 1069.69 m², will contain the fitness centre. The rooftop will include a 1068.75 m² car park, of which 638.45 m² will be covered by a metal-tile roof structure serving as the rainwater catchment area evaluated in this study. Figure 2 illustrates the rainwater harvesting system, which includes pre-treatment consisting of leaf screening at the inlet, diversion of the first 2 mm of rainfall (first flush), and cartridge filtration before the upper tank.

The fitness centre was designed in accordance with the construction and installation standards adopted throughout the franchise network. In addition to spaces for strength training and cardiovascular exercise, the project includes a reception area, management office, massage chair room, storage room, staff room, three potable water fountains, cleaning supply storage, male and female changing rooms, and a changing room for individuals with special needs.

2.2. Water End-Uses

The water end-uses were estimated by collecting data on water flow rates, consumption habits, and the number of employees and users. A fitness centre of the same franchise in Greater Florianópolis served as the reference site for measurements and interviews. It has a floor plan area similar to that of the fitness centre under study and was selected to obtain empirical data on usage patterns. The water appliances are identical to those planned for the new fitness centre, except for the shower and tap in the changing room for individuals with special needs, and the bowl-and-tank toilets. The water end-uses identified and assessed were lavatory taps, showers, urinal flushing, toilet flushing, kitchen taps, and cleaning activities. Water consumption related to urinal flushing, toilet flushing, and cleaning was considered non-potable, i.e., it can be supplied with non-potable water, such as rainwater.

2.2.1. Water Flow Rate

The kitchen tap flow rate was estimated using a graduated container, a scale to measure the water volume, and a stopwatch. Three fitness centre employees who regularly use the tap assisted in this procedure. Participants operated the tap as usual, and the time to fill the container was recorded. Each employee repeated the process three times, and the average of the nine recorded flow rates was used to determine the typical flow rate of the tap.

Consumption at the self-closing lavatory taps was estimated based on the volume of water used per activation. Each tap in the men’s and women’s changing rooms was fully activated, and the graduated container and scale were used to measure the volume consumed. Three samples were taken from each tap, and the average was calculated to determine typical usage.

Cleaning in the fitness centres is carried out using buckets; therefore, the volume and number of buckets used weekly were estimated. For bowl-and-tank toilets, showers, and urinals, consumption was obtained from manufacturers’ catalogues. Consumption related to drinking fountains was excluded from this study due to high data variability. Fitness centre users typically drink directly from the fountains or fill bottles of varying sizes, making it difficult to obtain representative and standardised measurements.

2.2.2. Monthly Consumption

Monthly consumption was estimated from interviews and questionnaires. Due to the large population size, it was not possible to survey all users and employees. Therefore, Equation (1), as presented by Barbetta [36], was used to determine a representative sample based on the total population and the sampling error adopted in this study.

$$n\ge {\displaystyle \frac{n_0\times N}{n_0+N}}$$

(1)

where $n_0\ge {\displaystyle \frac{1}{E_0^2}}$; $N$ is the number of people that use the building (people); $E_0$ is the desired sampling error (%); and $n$ is the sample of people interviewed (people).

The company provided data on the number of users and employees currently attending the operational fitness centre, as well as estimates for the new facility under study. Two questionnaires were developed to assess consumption habits: one for users and one for employees. The user questionnaire was made available online for one week via a link posted at the fitness centre reception desk, and receptionists actively encouraged participation. The employee questionnaire was administered in person.

Questionnaire data, combined with user and employee counts from the operational facility, supported the estimation of monthly consumption for the new fitness centre. Monthly water consumption from the lavatory taps (Equation (2)) was calculated by multiplying the number of activations per person by the average flow per activation per day. This sum was divided by the number of samples and multiplied by the weekly estimate of users attending the new facility, representing the total number of weekly activations. Finally, this value was multiplied by the volume per activation and by the number of weeks in the month. Consumption habits were analysed separately for users and employees.

$$C_{lt}={\displaystyle \frac{N_{dm}}{7}}\times V_{lt}\times \left(N_u\times {\displaystyle \frac{{\sum }_i^{n_u}{f}_{lt,i}\times N_{lt,i}}{n_u}}+N_e\times {\displaystyle \frac{{\sum }_i^{n_e}{f}_{lt,i}\times N_{lt,i}}{n_e}}\right)$$

(2)

where $C_{lt}$ is the lavatory taps monthly water consumption (L/month); $N_{dm}$ is the number of days in the month; $V_{lt}$ is the lavatory taps volume per activation (L/activation); $N_u$ and $N_e$ are the weekly numbers of users and employees (people/week); $n_u$ and $n_e$ are the number of users and employee samples; $f_{lt,i}$ is the daily frequency of lavatory taps use (uses/day.person); and $N_{lt,i}$ is the number of lavatory tap activations per use (activations/use).

Shower usage was estimated based on the time spent using the showers by users and employees. It was observed that users of the women’s changing room spent, on average, longer periods in the showers. Consequently, monthly water consumption for showers (Equation (3)) was calculated separately for women and men.

$$C_s={\displaystyle \frac{N_{dm}}{7}}\times Q_s\times \left[\begin{array}{c}\left({\displaystyle \frac{{\sum }_i^{n_{fu}}{f}_{s,i}\times t_{s,i}}{n_{fu}}}\times P_{fu}+{\displaystyle \frac{{\sum }_i^{n_{mu}}{f}_{s,i}\times t_{s,i}}{n_{mu}}}\times P_{mu}\right)\times N_{u\hspace{0.5em}}\\ +\left({\displaystyle \frac{{\sum }_i^{n_{fe}}{f}_{s,i}\times t_{s,i}}{n_{fe}}}\times P_{fe}+{\displaystyle \frac{{\sum }_i^{n_{me}}{f}_{s,i}\times t_{s,i}}{n_{me}}}\times P_{me}\right)\times N_e\end{array}\right]$$

(3)

where $C_s$ is the shower monthly water consumption (L/month); $N_{dm}$ is the number of days in the month; $Q_s$ is the shower flow rate (L/s); $n_{fu}$ and $n_{mu}$ are the number of samples of users who use the female and male changing room; $P_{fu}$ and $P_{mu}$ are the percentage of users who use the female and male changing room (%); $N_u$ and $N_e$ are the weekly numbers of users and employees (people/week); $n_{fe}$ and $n_{me}$ are the number of samples of employees who use the female and male changing room; $P_{fe}$ and $P_{me}$ are the percentage of employees who use the female and male changing room (%); $f_{s,i}$ is the daily frequency of shower use (uses/day); and $t_{s,i}$ is the shower’s utilisation time per use (s/use).

Urinal (Equation (4)) and toilet (Equation (5)) consumption was estimated similarly to lavatory taps, but considering the percentage of users and employees using the men’s changing room.

$$C_u={\displaystyle \frac{N_{dm}}{7}}\times V_u\times \left(N_u\times P_{mu}\times {\displaystyle \frac{{\sum }_i^{n_{mu}}{f}_{u,i}\times N_{u,i}}{n_{mu}}}+N_e\times P_{me}\times {\displaystyle \frac{{\sum }_i^{n_{me}}{f}_{u,i}\times N_{u,i}}{n_{me}}}\right)$$

(4)

where $C_u$ is the urinal monthly water consumption (L/month); $N_{dm}$ is the number of days in the month; $V_u$ is the urinal volume per flush (L/flush); $N_u$ and $N_e$ are the estimate of the weekly number of users and employees (people/week); $P_{mu}$ and $P_{me}$ are the percentage of users and employees who use the male changing room (%); $n_{mu}$ and $n_{me}$ are the number of samples of users and employees using the male changing room; $f_{u,i}$ is the daily frequency of urinals use (uses/day); and $N_{u,i}$ is the number of urinal activations per use.

$$C_t={\displaystyle \frac{N_{dm}}{7}}\times Q_t\times \left[\begin{array}{c}\left({\displaystyle \frac{{\sum }_i^{n_{fu}}{f}_{t,i}\times N_{t,i}}{n_{fu}}}\times P_{fu}+{\displaystyle \frac{{\sum }_i^{n_{mu}}{f}_{t,i}\times N_{t,i}}{n_{mu}}}\times P_{mu}\right)\times N_{u\hspace{0.5em}}\\ +\left({\displaystyle \frac{{\sum }_i^{n_{fe}}{f}_{t,i}\times N_{t,i}}{n_{fe}}}\times P_{fe}+{\displaystyle \frac{{\sum }_i^{n_{fm}}{f}_{t,i}\times N_{t,i}}{n_{me}}}\times P_{me}\right)\times N_e\end{array}\right]$$

(5)

where $C_t$ is the toilet monthly water consumption (L/month); $N_{dm}$ is the number of days in the month; $V_t$ is the toilet volume per flush (L/flush); $n_{fu}$ and $n_{mu}$ are the number of samples of users who use the female and male changing room; $P_{fu}$ and $P_{mu}$ are the percentage of users who use the female and male changing room (%); $N_u$ and $N_e$ are the estimate of the weekly number of users and employees (people/week); $n_{fe}$ and $n_{me}$ are the number of samples of employees who use the female and male changing room; $P_{fe}$ and $P_{me}$ are the percentage of employees who use the female and male changing room (%); and $f_{t,i}$ is the daily frequency of toilet use (uses/day).

Monthly water consumption from the kitchen tap (Equation (6)) was estimated by monitoring its use by three fitness centre employees. During interviews with employees responsible for cleaning, they were asked to estimate the number of buckets used daily for cleaning purposes. The number of working days per week was also confirmed for each employee.

$$C_{kt}={\displaystyle \frac{N_{dm}}{7}}\times V_{kt}\times N_e\times {\displaystyle \frac{{\sum }_i^{n_e}{f}_{kt,i}}{n_e}}$$

(6)

where $C_{kt}$ is the kitchen tap monthly water consumption (L/month); $N_{dm}$ is the number of days in the month; $V_{kt}$ is the water consumption per use of the kitchen tap (L/use); $N_e$ is the estimate of the weekly number of employees who will attend the fitness centre under study; $f_{kt,i}$ is the daily frequency of use (uses/days); and $n_e$ is the number of employee samples.

To estimate the distribution of end-uses in the fitness centre, the total monthly water consumption was first determined by summing all types of water consumption. Based on this total, the proportion of water allocated to each specific use was then calculated and expressed as a percentage of overall consumption.

2.3. Sizing the Rainwater Tank

The rainwater tank was sized using the Netuno computer programme, version 4, which employs deterministic behavioural modelling to simulate the performance of rainwater harvesting systems in buildings. Simulations run on a daily time step using historical rainfall data. It considers variables such as catchment area, surface runoff coefficient, total water demand, rainwater demand, and the capacities of the lower and upper tanks. The daily rainwater collected is determined based on precipitation, area, and runoff coefficient, and is then added to the rainwater already available in the tank, considering its maximum capacity. The model then assesses start-of-day availability and overflow occurrence, quantifying consumption, overflow losses, and overall reliability [26].

The model simulates different tank configurations (lower only or lower + upper), evaluates daily demand fulfilment (total, partial, unmet), and estimates potable-water savings. When an upper tank is included, the system simulates automatic pumping from the lower tank, provided that a minimum volume is maintained in the upper tank, from which the supply occurs [26].

The system design under study consists of two tanks: a lower and an upper tank. Daily precipitation data were obtained from the Meteorological Database provided by the National Institute of Meteorology [37]. The automatic station 806 was used as the primary source, while the conventional station 83,897 was used to supplement missing records. Rainfall data were collected from 2 January 2003 to 31 January 2024. Both stations are located at a latitude of −27°36′ and longitude −48°37′ in the city of São José, approximately 7.5 km from the fitness centre’s location.

The rainwater catchment area corresponds to the covered parking area, which has a metal-tile roof and measures 638.45 m². The surface runoff coefficient for corrugated metal tiles ranges from 0.80 to 0.90, as reported by Tomaz [38]. A runoff coefficient of 0.85 was adopted in this study. The total inflow demand corresponds to the volume required to supply the building’s overall potable water consumption, with the average daily consumption of the fitness centre considered as the total water demand. The rainwater demand was calculated by summing the water end-uses from urinals, toilets, and cleaning.

The volume of the upper rainwater tank was selected based on the commercially available volume closest to the average daily non-potable water demand, which was calculated as the sum of consumption from all non-potable end uses. The simulation was conducted considering a maximum volume of 15,000 L for the lower rainwater tank, with increments of 500 L to evaluate the potential for potable water savings. The stopping criterion set the ideal lower-tank size as the smallest volume for which an additional 1 m³ yielded ≤1% extra savings.

Table 1. Input data for assessing the potential for potable water savings in the Netuno programme.

Input Data for the Simulation	Value
Rainfall data	From 2 January 2003 to 31 January 2024
First-flush (mm)	2
Catchment area (m2)	638.45
Total daily water demand (L)	25,093
Rainwater demand (%)	24.4
Runoff coefficient	0.85
Capacity of the upper rainwater tank (L)	5000
Maximum capacity of the lower tank (L)	15,000
Interval between capacities (L)	500
Difference between potential water savings through the use of rainwater (%/m3)	1

shows the input data used in Netuno.

2.4. Economic Feasibility Assessment

The economic feasibility of implementing the rainwater harvesting system was also assessed using Netuno. Based on potable-water savings, the programme calculates the monthly savings generated and compares them with the initial costs (tanks, pumps, pipes, and installation), operating costs (electricity, maintenance, and water treatment), as well as water and sewerage tariffs. From these inputs, Net Present Value (Equation (7)), payback, and Internal Rate of Return (Equation (8)) were estimated.

$$NPV=-I_0+{\displaystyle \frac{{CF}_1}{{\left(1+MARR\right)}^1}}+{\displaystyle \frac{{CF}_2}{{\left(1+MARR\right)}^2}}+\dots +{\displaystyle \frac{{CF}_n}{{\left(1+MARR\right)}^n}}$$

(7)

$$0=-I_0+{\displaystyle \frac{{CF}_1}{{\left(1+IRR\right)}^1}}+{\displaystyle \frac{{CF}_2}{{\left(1+IRR\right)}^2}}+\dots +{\displaystyle \frac{{CF}_n}{{\left(1+IRR\right)}^n}}$$

(8)

where $NPV$ is the Net Present Value (R$); $I_0$ is the initial investment (R$); ${CF}_1$ is the cash flow in period 1 (R$); $MARR$ is the Minimum Acceptable Rate of Return (%); $n$ is the number of periods for analysing the investment; and $IRR$ is the Internal Rate of Return (%).

Tanks and pump costs were calculated based on three quotations from local suppliers, with the lowest price being adopted. The labour cost for implementing the rainwater harvesting system was estimated based on the study by Istchuk and Ghisi [39], who reported an average equivalent to 24% of other initial costs in eight Brazilian cities. For pipes, fittings, first-flush diverter components (including valves and piping), and cartridge filters, a 15% total investment was assumed, consistent with Ghisi and Ferreira [29].

The operating costs considered included energy consumption by the pump and biannual tank cleaning. The fitness centres operate in the free electricity market, which allows them to negotiate energy purchases directly with producers and trading agents, while still being subject to distribution costs and sector charges paid to the regional distribution company. Consequently, electricity charges are issued in two invoices: one from the Electric Company of the State of Santa Catarina and another from the energy trading agent contracted by the fitness centre’s parent company. The electricity tariff used in Netuno (R$·kWh⁻¹) was obtained by adding the value of the September 2024 bill from the distribution utility and the bill from the free-market energy supplier for the fitness centre. The total cost (R$) was then divided by the facility’s metered electricity consumption (kWh) in the same month to obtain the corresponding tariff (R$·kWh⁻¹).

Netuno uses estimates of total water consumption and potable water savings to calculate cash flow and financial savings from the rainwater harvesting system, in addition to accounting for water and sewerage tariffs. The water company of the State of Santa Catarina, responsible for supplying water to the fitness centre, applies three tariff bands for commercial buildings, in addition to a Fixed Infrastructure Availability Tariff. The bands, implemented in Netuno, correspond to different rates charged per cubic metre according to the volume of water consumed.

The analysis period was set at 20 years, consistent with previous studies [40,41]. The minimum acceptable rate of return was 0.57% per month, corresponding to the savings yield stipulated by the Central Bank of Brazil for September 2024. Monthly inflation was calculated as the average of the Broad National Consumer Price Index from October 2023 to September 2024. This corresponded to an inflation rate of 0.36% per month, with a 12-month adjustment period for costs.

Table 2. Input data for assessing the economic feasibility analysis in the Netuno programme.

Group of Variables	Variable	Value
General data	Lower tank capacity (litres)	10,000
	Inflation (% per month)	0.36
	Adjustment of water and sewerage tariffs (months)	12
	Analysis period (years)	20
	Minimum Acceptable Rate of Return (% per month)	0.57
	The month of system installation	September
Water and sewerage tariffs	Fixed tariff (R$)	43.31
	Variable tariff—1 to 10 m3 of water consumed (R$/m3)	6.37
	Variable tariff—11 to 50 m3 of water consumed (R$/m3)	17.89
	Variable tariff—51 to 999.999 m3 of water consumed (R$/m3)	22.51
	Sewerage tariff (% of water tariff)	100
Initial costs	Lower tank (R$)	4989.00
	Upper tank (R$)	2636.58
	Labour (R$)	2228.45
	Pipes, fittings and filters (R$)	1727.05
Operating costs–motor pump	Motor pump unit power (CV)	0.5
	Motor pump efficiency (%)	50%
	Flow rate (litres/hour) *	2500
	Start-up time (s)	5 s
	Unit cost (R$)—per pump (simulation assumed two pumps—one operational, one spare)	829.82
	Electricity tariff (R$/kWh)	0.675
Operating costs–maintenance	Tank cleaning—fixed cost every 6 months (R$)	1600.00

Note: * The 2500 L/h rainwater flow maintains the upper-tank operating level during clustered demands; the pump does not operate continuously.

shows the data used in Netuno.

3. Results and Discussions

3.1. Sample Analysis and Consumption Habits

A target sampling error of 12% was adopted to determine the number of questionnaires, based on the number of users and employees at the operational fitness centre. However, due to limitations in employee availability and user participation, the minimum required sample size was not achieved, resulting in sampling errors exceeding 12%.

Table 3. Representativeness of the samples of consumer habits obtained from the fitness centre in operation.

Group	Employees			Users
Gender	Total	Female	Male	Total	Female	Male
Percentage interviewed	76%	86%	70%	3.04%	3.33%	2.77%
Sample error	13%	15%	21%	10%	14%	15%

shows the percentage of the total population that responded to the questionnaire for each group, disaggregated by gender, along with the corresponding sampling errors.

Interviews with employees indicated that 41% of the staff at the current facility are women and 59% are men. Among users, the gender distribution was 48% female and 52% male. These proportions were adopted in estimating the expected composition of users and employees for the new fitness centre under study. The water consumption habits of employees and users, as reported in the questionnaires, are presented in

Table 4. Consumption habits of employees and users of the fitness centre in operation.

Appliance	Description	Quantity *
		Employees	Users
Lavatory taps	Activations	8.23 uses/day	4.55 uses/day
Female showers	Usage time	700.00 s	144.49 s
Male showers	Usage time	256.14 s	69.55 s
Urinals flushing	Activations	4.14 uses/day	1.77 uses/day
Female toilet flushing	Activations	6.67 uses/day	1.04 uses/day
Male toilet flushing	Activations	2.71 uses/day	0.34 uses/day
Kitchen tap	Uses	3.08 uses/day	-

Note: * Average daily values per employee or user.

.

In general, the results showed that both users and employees frequently used showers and toilets, with notably longer average shower durations among female employees. This pattern is consistent with Makki et al. [42], who found that men typically take shorter but more frequent showers, while women tend to have longer showers, often due to hair washing and post-practice hygiene. In their study of Australian households, the average shower lasted 7.6 min (456 s), accounting for approximately 34% of indoor water use. Notably, 18% of showers lasted more than 10 min, with 61% of these taken by women, reinforcing gender-based trends. In this study, female employees recorded average shower durations of 700 s (≈11 min 40 s), higher than residential benchmarks, likely due to post-practice routines. Recognising such behavioural differences is essential for modelling demand and sizing infrastructure appropriately.

Several factors influence water demand and consumption patterns in buildings. Studies have shown, for example, that although higher-income households tend to install more efficient water appliances, they also have higher per capita water consumption [28,31,43]. Moreover, the rational use of water in commercial and institutional buildings requires a different approach from that adopted in residential settings, due to operational and behavioural specificities. A key factor is that users typically do not pay directly for their water consumption, which reduces the incentive for conscious use. In addition, there is considerable variation in user behaviour and environmental awareness levels. These challenges are compounded by insufficient maintenance of water appliances, which compromises operational efficiency and increases the likelihood of malfunctions. Complex supply and pumping infrastructure further hinder leak detection, a problem exacerbated by the lack of sub-metering, as a single water meter often serves an extensive and heterogeneous system. Finally, water appliances in such buildings are subject to more frequent use. They are, therefore, more prone to failure compared to those in residential environments, highlighting the need for targeted water management and control strategies in these settings [44].

3.2. Water End-Uses

The following results derive from the empirical data collected as described in the methodology. According to the manufacturers’ technical data sheets for the toilets in the fitness centre, an average of 6.8 L of water is used per flush; thus, this value was adopted as the consumption per flush. For the urinals, the volume per flush, as indicated in the flush valve’s technical data sheet, is 1.2 L. The manufacturer’s catalogue specifies that the shower provides a constant flow rate between 7 and 12 L per minute; therefore, a value of 9.5 L per minute was used to estimate shower consumption. The average consumption per operation of the lavatory taps was 0.495 L. The kitchen tap flow rate was 0.089 L per second, based on the average of nine samples collected with the assistance of the fitness centre’s employees. The average usage time for this tap was 21.22 s, resulting in an average consumption of 1.89 L per use.

Table 5. Water end-uses obtained for the fitness centre.

End Use	Potability	Estimated Monthly Consumption (m3)	Share of Total Consumption (%)
Showers	Potable	501.7	66.6
Toilets flushing	Non-potable	143.6	19.1
Cleaning	Non-potable	7.2	1.0
Lavatory taps	Potable	64.8	8.6
Urinals flushing	Non-potable	32.3	4.3
Kitchen tap	Potable	3.0	0.4

presents the monthly water consumption for each end-use. Showers alone accounted for 66.6% of the total water consumption, reflecting the intensive use of changing rooms, particularly by female users, who have longer average shower durations. Non-potable uses, including toilet flushing, urinal flushing, and cleaning, represented 24.4% of total water consumption. These water end-uses can be supplied by the rainwater harvesting system, corresponding to approximately 183,178 L per month.

The data highlight the importance of end-use characterisation in fitness centres, particularly when specifying water appliances in the project according to their water flow rates. Similar patterns were observed by Silva et al. [34] in a study conducted at a public indoor swimming pool complex in Bragança, Portugal. Although the context differs from that of a fitness centre, both cases involve high foot traffic, shared facilities, and intensive use of showers and taps. In the Bragança complex, showers accounted for 34.7% of total water use, followed by swimming pools (25.6%), backwash filtration (24.5%), taps (11.8%), and toilet flushing (2.5%). Such end-uses are consistent with those identified herein, which also found showers and lavatory taps to be the primary points of consumption. Notably, Silva et al. [34] reported operational issues, including users leaving taps open, which led to excessive waste. This concern was mitigated in the fitness centre due to the use of self-closing taps. Although toilet flushing volumes were higher in Bragança (9 L per use), simple adjustments could reduce consumption by up to 33%. These findings underscore the importance of selecting suitable appliances and promoting user awareness in managing water use in institutional settings.

In addition to showers, other hygiene-related water uses—particularly toilet flushing and handwashing—contribute significantly to total water consumption in fitness centres. Although direct end-use data from gym environments remain limited, findings from residential and institutional settings help illustrate consumption patterns in these settings. Marinoski et al. [45] conducted detailed field research in low-income households in southern Brazil. They observed that showers accounted for 30% to 36% of total indoor water consumption, followed by toilets (18% to 20%) and kitchen taps (15% to 20%). Similarly, Nascimento and Sant’ana [46] analysed water end-uses in two hotels in Brasília, which, like fitness centres, experience high user turnover and shared water appliances. In that study, showers were identified as the largest consumer (55%), followed by employee-related uses (17%). These results, although from different contexts, highlight the predominance of showers and hygiene-related uses, reinforcing the importance of targeting these end uses to achieve water savings in fitness centres.

Although standard procedures—interviews, direct measurements, and water flow estimates—were applied to assess water end-uses, it is important to highlight that future studies, with broader financial and technical resources, could benefit from the use of smart-metering systems with event-disaggregation capabilities. Such systems enable the identification of specific usage patterns with high temporal resolution, distinguishing between individual appliances and user behaviours [47,48]. Their application would allow for even more precise demand profiling, capturing peak usage events, behavioural variations, and operational anomalies that are difficult to detect using manual measurements or survey-based methods. Nonetheless, the methodological approach adopted herein provides a robust foundation and offers valuable insights into the dynamics of water consumption in fitness centres.

3.3. Potential for Potable Water Savings

Figure 3 shows the results of the rainwater tank sizing performed using Netuno, considering storage capacities ranging from 0 to 14,000 L.

Figure 3a shows that potable water savings increase non-linearly as a function of tank capacity, reaching approximately 7.5% at the maximum simulated capacity. However, a noticeable reduction in the slope of the curve occurs after 10,000 L, indicating diminishing marginal gains with additional storage. The dashed vertical line represents the optimal point identified by Netuno, i.e., 11,000 L, beyond which each additional cubic metre yields less than 1% of further savings. This supports the selection of such a capacity as the ideal capacity from both technical and economic perspectives.

Figure 3b shows the average daily volumes of potable water consumption, rainwater consumption, and overflow. As storage capacity increases, rainwater use gradually rises, stabilising at around 2000 L per day, while potable water use declines modestly and plateaus at approximately 23,000 L per day. Overflow volumes are high for smaller tanks (up to 2500 L per day) but drop sharply to around 500 L per day and remain stable thereafter. These results indicate that small tanks are unable to effectively buffer rainfall variability, resulting in higher water losses, whereas larger tanks enhance system reliability and performance.

Figure 3c shows the extent to which non-potable demand is met over time, expressed as the percentage of days in which demand was fully met, partially met, or not met at all. As storage capacity increases, the frequency of days with unmet demand gradually decreases, reaching approximately 60% with an 11,000 L tank. Concurrently, the percentage of days with fully met demand rises to around 17%, while the number of days with partially met demand stabilises at approximately 22%. These results confirm that the rainwater harvesting system alone is insufficient to fully meet non-potable demand on most days in high-consumption commercial buildings, reinforcing the need for complementary measures such as water-efficient appliances or alternative supply sources.

Campisano et al. [15], in a comprehensive review of rainwater harvesting systems across various building types, emphasised that in commercial settings it is often not feasible to supply more than 50% of non-potable demand using rainwater alone, especially in regions with seasonal rainfall or limited roof catchment areas—a finding consistent with this case. Gonçalves et al. [49] reported, in an analysis of 607 water appliances across five sports facilities in southern Brazil, that 77.8% of toilets had leaks and 30.9% of taps showed dripping, highlighting the operational inefficiencies that can undermine water-saving efforts. Silva et al. [34], in a study conducted in a public sports complex with swimming pools, demonstrated that low-cost retrofit actions were able to reduce total water consumption by up to 20.4%, even without alternative sources such as rainwater. Together, these findings underscore the importance of integrating rainwater harvesting with water-efficient appliances and broader demand management strategies to maximise water savings in high-consumption commercial buildings.

Figure 3b also suggests that managing overflow may contribute to urban drainage benefits. While rainwater harvesting tanks are primarily designed to remain full to meet water demand, there is potential to mitigate surface runoff, albeit to a limited extent. National studies, such as that by Custódio and Ghisi [33], estimated a reduction ranging from 2.7% to 14.3% in peak flows across different scenarios in the urban area of Joinville (Brazil) when rainwater was used for non-potable purposes. Additionally, independent studies have demonstrated a significant reduction in runoff through residential rainwater harvesting. In Sicily, the use of rainwater harvesting reduced peak roof runoff by 30% to 68% in at least half of the monitored events, provided that domestic rainwater demand was met [50]. At the urban scale in Genoa, continuous simulation resulted in an average 33% reduction in peak flow and a 26% decrease in runoff volumes discharged [51]. On a neighbourhood scale in Seoul, equipping each building with a 10 m³ tank per 100 m² of roof area reduced total runoff volumes and peak flows by 20% and 18%, respectively [52].

The connection between flood mitigation and rainwater harvesting can also be enhanced by utilising real-time control devices for automation. These systems can release stored volumes based on rainfall forecasts and tank water levels, ensuring that predicted rainfall can be effectively harvested. Fisher-Jeffes et al. [53] demonstrated that real-time control enhanced systems can mitigate flooding by attenuating downstream flows through strategic water storage. Similarly, Liang et al. [54] showed that algorithm-based real-time control systems, using automated valves, can reduce stormwater peak flows by up to 85% during rainfall events. Although such strategies are primarily applied at the urban scale, they underscore the growing relevance of digital technologies in sustainable water management.

These findings indicate that Netuno’s daily water-balance approach provides realistic estimates relative to YBS (optimistic) and YAS (conservative) formulations. While YBS tends to overestimate savings by prioritising demand before overflow, and YAS follows a more conservative approach, Netuno simulates storage and consumption step by step, offering a more accurate reflection of operational constraints. The validation conducted by Rocha [55], based on monitored data from a real building, confirmed that Netuno’s estimates closely matched actual potable water savings when sufficient rainwater was available to meet demand. In scenarios of water scarcity, differences were observed between estimated and measured values; however, the errors remained relatively small.

Finally, based on the estimated non-potable consumption and the available catchment area (638.45 m²), simulations using Netuno indicated that the ideal capacity for the lower tank would be 11,000 L, with a potential potable water saving of 7.04%. However, due to commercial constraints, a 10,000 L tank was adopted, resulting in a slight reduction in potential savings to 6.08%. The upper-tank capacity (5000 L) was chosen as the closest standard market size, balancing structural and cost considerations, and reflecting Netuno’s threshold criterion (at most 1% additional savings per extra m³). However, the average daily non-potable demand was estimated at approximately 6.1 m³ per day. Although non-potable end uses account for less than one-quarter of the total water consumption in this case, the corresponding monthly potable water savings (approximately 45,770 L) are substantial. These results not only demonstrate the technical feasibility of rainwater harvesting but also reinforce its importance as a key strategy for enhancing water sustainability in high-consumption commercial buildings.

3.4. Economic Feasibility Assessment

The economic viability of the rainwater harvesting system was confirmed through a financial analysis module in Netuno. For the base scenario (excluding sewerage charges), a Net Present Value (NPV) of R$146,085.85, an Internal Rate of Return (IRR) of 6.99% per month (computed on monthly cash flows), and a payback period of 15 months were estimated. These outcomes reflect low upfront costs and the prevailing local tariff structure. When a future scenario including sewerage charges was considered, the financial savings increased significantly, raising the NPV to R$359,946.54 and reducing the payback period to 8 months. These results demonstrate strong economic attractiveness under the adopted assumptions.

As shown in Figure 4, the cumulative NPV increases steadily throughout the 20-year simulation period, with monthly savings consistently exceeding operational costs. Monthly returns show a cyclical pattern, with lower gains in drier months, particularly in winter. This variation is due to the seasonal variation in rainfall, as the study assumed constant water consumption throughout the year. Consequently, during periods of low precipitation, there is less rainwater available to replace mains water, temporarily reducing monthly financial gains.

Recent studies have revealed significant variations in economic performance, attributable to factors such as rainwater demand and infrastructure, tariffs, and incentives. In Portugal, Alves et al. [56] analysed a commercial building, while Matos et al. [57] examined an industrial facility that does not need water for production. The payback period for the rainwater harvesting system in the commercial building was approximately 4 years, based on the best scenario analysed [56]. For the industrial building, the payback periods were projected to be between seven and fifteen years. These periods were found to be longer in periods of drought and shorter in years with average rainfall. This variation is attributed to the rainfall regime, as well as operating and maintenance costs [57].

In southern Brazil, de Gois et al. [58] conducted a study of a shopping centre, observing a return period of between twenty and thirty-eight days. The low payback was attributed to the high water consumption for toilets and the potential use of rainwater in the conditioning systems, which generated significant monthly savings. In a study evaluating different strategies to reduce potable water consumption in a library, also located in southern Brazil, Figueroba et al. [59] reported a payback period of 73 months for the rainwater harvesting system. These more extended payback periods are primarily attributed to higher initial investment costs, which often result from the inclusion of more complex treatment infrastructure. Nevertheless, findings from other previous studies in the Greater Florianópolis region corroborate the results obtained herein.

Similar outcomes were found by Silva et al. [34] in a municipal sports complex with swimming pools. Their analysis of various retrofit scenarios showed that replacing taps and showers, as well as adjusting toilet flushing, could reduce total water consumption by up to 20.4%, with payback periods ranging from three to four months, even without the use of alternative water sources. These findings reinforce the importance of integrated strategies that combine user behaviour, appliance efficiency, and decentralised water supply solutions such as rainwater harvesting. Consistent with the case study presented herein, they indicate that significant water savings and rapid financial returns can be achieved when rainwater harvesting systems complement efficiency measures and behavioural adjustments in high-consumption facilities.

Overall, system simplicity and regular demand for non-potable water are key factors influencing the financial success of the proposed rainwater harvesting system. Incorporating smart technologies and automation may further enhance water efficiency and help mitigate the effects of seasonal variability. In urban areas with high rainfall and limited sewerage infrastructure, rainwater harvesting systems are not only technically feasible but also strategically advantageous, serving as a tool for sustainable water management in collective-use buildings such as fitness centres.

3.5. Limitations and Future Studies

Despite the comprehensive methodological framework, some limitations should be acknowledged. Extrapolating empirical data from an existing fitness centre in Florianópolis to a new facility in Palhoça may not fully capture contextual differences in user behaviour, seasonal variations, or socio-demographic characteristics. Although both locations share climatic and cultural similarities, assuming behavioural uniformity may introduce uncertainties into the estimation of water end-uses. Furthermore, although sample sizes were calculated using statistical criteria, participation rates—particularly among users—fell short of the target, resulting in sampling errors that limit the generalisability of the findings.

Another limitation is the inability to account for peak-period variations in water use, which are common in fitness centres, as well as seasonal variability in user attendance, which tends to decline during the winter months. These dynamics may significantly influence demand patterns, system performance, and the reliability of the estimated water savings. Future research should incorporate time-series analyses with higher temporal and seasonal resolution to consider such variations more accurately.

In simulation, the system was modelled in Netuno, which—while robust and widely used—does not account for positive externalities such as reduced stormwater runoff, improved local water resilience, environmental benefits, or increased property value associated with sustainable infrastructure.

Climate variability and changing rainfall patterns pose another limitation. As highlighted by Zhang et al. [60], increased rainfall intensity combined with longer dry periods may reduce the reliability of conventional rainwater harvesting systems. Although this study used a 21-year historical rainfall series, future research should incorporate climate change scenarios using regional climate models and stochastic simulations to assess long-term system resilience. Moreover, integrating real-time control technologies (e.g., smart valves and weather-based automation) could further enhance operational efficiency and enable dual functionality for both water savings and flood mitigation.

Future research should also explore the applicability of rainwater harvesting systems in other high-demand, collective-use buildings such as schools, healthcare facilities, sports arenas, and administrative complexes. Comparative studies combining technical performance, economic feasibility, and life cycle assessment can provide a more comprehensive understanding of system benefits. In particular, life cycle assessment approaches can consider embodied energy, carbon footprint, and long-term environmental trade-offs, thereby complementing economic analyses.

In summary, although this case study provides valuable empirical evidence on the application of rainwater harvesting in fitness centres, further investigations incorporating enhanced monitoring, broader typological coverage, and integrated sustainability metrics are necessary to inform evidence-based policy and design practices in the built environment.

4. Conclusions

This study demonstrated the technical and economic feasibility of rainwater harvesting systems to supply non-potable water demand in fitness centres, based on empirical end-use characterisation, Netuno simulations, and financial analysis. The findings confirmed that non-potable water end-uses, such as toilet flushing, urinal flushing, and cleaning, represent a significant opportunity for achieving potable water savings with short payback periods. However, economic outcomes are context-dependent. Variations in equipment prices, labour costs, and utility tariffs across regions can affect payback and internal rate of return (IRR). These factors should be considered when transferring the results to other locations.

Notably, the analysis revealed that rainwater harvesting alone cannot fully meet non-potable demand in high-consumption buildings, underscoring the need to combine it with complementary measures, such as water-efficient appliances, greywater reuse, and automated control technologies. Beyond direct water savings, the integration of rainwater harvesting can reduce stormwater runoff and improve urban water resilience, supporting broader sustainability goals.

Focusing on an underexplored building typology, the study adds empirical evidence to inform design practice and policy, integrating technical and economic perspectives and discussing potential environmental co-benefits. Future research should explore scaling these findings to other collective-use facilities, integrate life cycle assessment approaches, and investigate digital monitoring solutions to enhance system performance and sustainability outcomes.