Centralized or Decentralized Rainwater Harvesting Systems: A Case Study

: World population growth, climate changes, urbanization, and industrialization have all had a negative impact on natural resources, including water resources. Excessive exploitation and pollution have caused more and more regions to have problems with access to fresh water. Rainwater is perceived as a valuable alternative source of water that is most often used in a hybrid system supplementing tap water. Considering the possibilities of designing a rainwater harvesting system as a decentralized or central system, this research was undertaken to determine the hydraulic and ﬁnancial e ﬃ ciency of these two systems. The research was carried out for a single-family housing estate located in Poland. For this research, a simulation model was applied to determine the e ﬃ ciency of water saving and the life cycle cost indicator. In variants where rainwater was only used to ﬂush toilets, the water saving e ﬃ ciency was 80% and 79% for the decentralized and centralized rainwater harvesting system (RWHS), respectively. The use of rainwater for toilet ﬂushing and watering the garden resulted in a signiﬁcant reduction in e ﬃ ciency to 57% (the decentralized system) and 54% (the centralized system). On the other hand, the results of the life cycle cost (LCC) analysis showed that in spite of reducing tap water consumption, both the centralized and the decentralized rainwater harvesting system were not ﬁnancially viable solutions for the housing estate, and only coﬁnancing investments at the level of 25% to 50% resulted in a signiﬁcant improvement in ﬁnancial e ﬃ ciency.


Introduction
Over the last century, there has been a rapid development of cities and urbanization of areas that were previously of a natural or rural nature. Europe has evolved into a continent where about 73% of the population lives in urban areas [1], and only North America (82%) and South America (80%) belong to more urbanized continents.
An increase in the population that is concentrated in small spaces and the development of industry and communication, as well as the pursuit of raising living standards have caused a significant depletion of natural resources, including surface and underground water resources [2]. These factors, as well as economic and political problems related to water, are the main reasons for the lack of water security in the world [3]. Climate changes are also observed, which are increasingly affecting all aspects of the lives of people, including the availability of potable water. Therefore, optimal management of available water resources is necessary to deal with all these problems [4]. Sustainable urban water management has become a goal of strategic planning for water utilities [5]. Access to an adequate quality of water, including the quality supplied from a water supply network, is a basic need for human development. Failure to meet the required potable water quality standards threatens the safety of water system users and reduces the reliability of the potable water [6,7]. water availability in coming years and eventually lead to a deterioration of the quality of life for future generations [49]. Currently, the average annual resources per one Pole are estimated at about 1600 m 3 , which accounts for only 35% of the resources per statistical inhabitant of Europe.
Previous publications on the use of rainwater in Poland for various buildings and various technical parameters of installations have concerned RWHSs which have been installed as decentralized systems [50,51]. There is no scientific research related to centralized systems. In addition, in other countries, this type of rainwater harvesting system is very rarely included in the scientific studies of various researchers.
Considering the above, this research was undertaken to assess the potential of reducing potable water consumption in a housing estate by using rainwater in decentralized and centralized systems, which is a novelty not only in Poland, but worldwide. Analyses of hydraulic, technical, and financial conditions for four variants of the rainwater harvesting system were performed using the example of a single-family housing estate located in Poland. The rainwater was intended for toilet flushing and watering the garden.

Case Study
The case study, for this research, concerns the housing estate "Słoneczna Aleja" (Sunny Avenue Estates) in Warsaw, Poland ( Figure 1). The estate consists of 22 single-family houses made in skeleton technology as energy-efficient semi-detached buildings. The usable area of a single building is 135 m 2 . The horizontal projection surface of the roof is 85 m 2 . The undeveloped plot area is 235 m 2 .
Rainwater harvesting systems were considered as an additional installation for the buildings. For this purpose, technical, hydraulic and financial analyses of such an investment were performed. The calculations were carried out for the following investment options: • Variant 0, no rainwater harvesting system and the water supply for the building only from the water supply system; • Variant 1A, an individual (decentralized) system for collecting, storing, and using rainwater to flush toilets, supplementing water shortages from the water supply network; • Variant 1B, an individual (decentralized) system for collecting, storing, and using rainwater for toilet flushing and irrigating the area, supplementing water shortages from the water supply network; • Variant 2A, a centralized (joint) system for collecting and storing rainwater from all the buildings and its use for toilet flushing with additional water shortages from the water supply network; • Variant 2B, a centralized (joint) system for collecting and storing rainwater from all the buildings and its use for toilet flushing and irrigating the area, supplementing water shortages from the water supply network.

Case Study
The case study, for this research, concerns the housing estate "Słoneczna Aleja" (Sunny Avenue Estates) in Warsaw, Poland ( Figure 1). The estate consists of 22 single-family houses made in skeleton technology as energy-efficient semi-detached buildings. The usable area of a single building is 135 m 2 . The horizontal projection surface of the roof is 85 m 2 . The undeveloped plot area is 235 m 2 .  The diagram of the rainwater harvesting system in Variants 1A and 1B and Variants 2A and 2B is shown in Figure 2.
The research assumed, on the one hand, that the demand for rainwater for toilet flushing was constant throughout the entire period of the analysis. This assumption was justified as the demand time series generated by the use of toilets did not show excessive daily differences [52]. On the other hand, the use of rainwater for irrigation of the area was periodic and, in Polish conditions in accordance with design guidelines [53], was carried out for 15 days during the spring and autumn months, from April 15 to September 15. The calculation parameters for buildings, plots of land, and RWHSs are presented in Table 1. The calculations assume the statistical number of people in the household for Poland [54].

•
Variant 2B, a centralized (joint) system for collecting and storing rainwater from all the buildings and its use for toilet flushing and irrigating the area, supplementing water shortages from the water supply network.
The diagram of the rainwater harvesting system in Variants 1A and 1B and Variants 2A and 2B is shown in Figure 2. and Variant 2B. 1, precipitation; 2, building; 3, rainwater supply to the tank; 4, individual rainwater tank; 5, rainwater supply from the individual tank to the building; 6, excess rainwater discharge from the individual tank; 7, rainwater drainage system; 8, water supply network; 9, supply of tap water (supplementary) to the tank; 10, supply of tap water to the building; 11, sewage network; 12, sewage drainage from the building; 13, central rainwater tank; 14, rainwater distribution network; 15, supply of rainwater from the network to the building; and 16, excess discharge of rainwater from the central tank. and Variant 2B. 1, precipitation; 2, building; 3, rainwater supply to the tank; 4, individual rainwater tank; 5, rainwater supply from the individual tank to the building; 6, excess rainwater discharge from the individual tank; 7, rainwater drainage system; 8, water supply network; 9, supply of tap water (supplementary) to the tank; 10, supply of tap water to the building; 11, sewage network; 12, sewage drainage from the building; 13, central rainwater tank; 14, rainwater distribution network; 15, supply of rainwater from the network to the building; and 16, excess discharge of rainwater from the central tank. The simulation tests were carried out using real daily precipitation and temperature data, which were recorded at the Warsaw Okęcie meteorological station. The data from 2000 to 2013 were applied for this research. The average annual precipitation sums for this period are shown in Figure 3. In the period analyzed, the median annual precipitation sums for Warsaw was 581 mm and it did not differ significantly from the precipitation values from other long-term periods. The simulation tests were carried out using real daily precipitation and temperature data, which were recorded at the Warsaw Okęcie meteorological station. The data from 2000 to 2013 were applied for this research. The average annual precipitation sums for this period are shown in Figure 3. In the period analyzed, the median annual precipitation sums for Warsaw was 581 mm and it did not differ significantly from the precipitation values from other long-term periods.

Hydraulic Analysis
The hydraulic analysis of the rainwater harvesting system in the variants adopted was based on a simulation model based on daily mass water balance developed by Słyś [55], which can be generally described by Formula (1): where Vi is the volume of rain water retained in the tank at the end of the i-day; Ii is the volume of rain water flowing from the roof to the tank on the i-th day, m 3 ; Vi−1 is the volume of rainwater remaining in the tank from the previous day, m 3 ; Ui is the volume of rainwater collected on the i-th

Hydraulic Analysis
The hydraulic analysis of the rainwater harvesting system in the variants adopted was based on a simulation model based on daily mass water balance developed by Słyś [55], which can be generally described by Formula (1): where V i is the volume of rain water retained in the tank at the end of the i-day; I i is the volume of rain water flowing from the roof to the tank on the i-th day, m 3 ; V i−1 is the volume of rainwater remaining in the tank from the previous day, m 3 ; U i is the volume of rainwater collected on the i-th day from the reservoir, m 3 ; and Z i is the volume of excess rainwater discharged from the tank to the sewage system on the i-th day, m 3 . The inflow of rainwater to reservoir I i depends on the roof surface, A, and the amount of precipitation, H, occurring on a given day. The size of the inflow is determined from Formula (2): where ϕ is the runoff coefficient, A is the surface area, m 2 ; and H i is the daily precipitation, m. The calculation also takes into account the type of precipitation (rain or snow) and atmospheric air temperature. During periods of snowfall its quantity is automatically converted to the amount of liquid precipitation. If the following two conditions are met: (i) the presence of snowfall and (ii) the air temperature below 0 • C, the amount of precipitation is not included on a given day in the balance of water stored in the retention tank. This precipitation is taken into account on the following days when the air temperature is above 0 • C. This assumption in the calculation model considers the occurrence of solid precipitation (snow) and negative temperatures that limit snow melting and water supply in the tank.
Due to the low pollution of rainwater coming from the roofs of buildings, the simulation model did not take into account the impact of these pollutants on the functioning of the system. This assumption is similar to the assumptions of other such computational models [56,57]. This model is very similar to other models such as the one developed by Fewkes [52]. The model algorithm is based on the YAS (yield after spillage) operating rule. The general principle of the YAS operation is as follows: The volume of rainwater supplied to the tank in the current time interval is added to the volume of Resources 2020, 9, 5 7 of 18 rainwater remaining in the tank from the previous time interval, and the excess of water is removed by the overflow. The initial storage volume (V i−1 ) for the second time step is the volume of water at the end of the first step (V i ). The daily time interval was used in these studies.
On the basis of the results obtained from the RWHS simulation tests, the efficiency of tap water saving, E f , was determined from Formula (3) [55].
where U i is the volume of rainwater collected on the i-th day from the reservoir, m 3 ; J i is the number of system users on the i-th day, person; and d di is the daily water demand for a specific purpose on the i-th day, m 3 /person/day.

Financial Analysis
The test results based on the simulation model of rainwater harvesting systems (RWHS) were used as input to assess the financial efficiency of the investment regarding the possibility of using such systems in the area of the housing estate.
The financial analysis of the rainwater harvesting system was carried out using the life cycle cost (LCC) methodology. The value of the LCC indicator, which was determined from Formula (4), consists of the following: (i) INV k investments incurred for the purchase and installation of a given system, (ii) operating costs expected throughout its lifetime, OC k , and (iii) costs of disposal, MC k [58]. Cash flows occurring in subsequent years are discounted, which allows for a comparison of their amounts appearing at different periods of time. The determination of the share of individual components in the total costs of the project is very important, both from the point of view of the investor and the system operator. The LCC enables cost optimization of investment projects [59] and the making of balanced decisions [60]. The guidelines contained in [61] studies provide that wherever the life span of an analyzed system goes beyond the foreseeable future, its residual value at the completion of exploitation do not need be defined in quantifiable terms. Consequently, the MC k costs were not taken care of in the quantitative analysis, similar to studies conducted by other authors [62].
where T is the duration of the LCC analysis, years; r is the fixed discount rate; and t is another year of using the installation. The operating costs of the building water supply system can be calculated from Formula (5).
where WC kt is the cost of purchasing tap water in the k and year t option, €; SC kt is the sewage disposal costs in option k and year t, €; and RC kt is the costs of rainwater drainage to the sewage system in the k and year option t, €.
In the case of buildings without a RWHS, the value of the WC kt parameter can be calculated from Formula (6).
where D t is the water demand for specific purposes, m 3 /year, and cw t is the purchase price of tap water in year t, €/m 3 . However, for buildings additionally supplied with water from the RWHS system, the WC kt value is described by Formula (7).
Resources 2020, 9, 5 8 of 18 where UW tk is the volume of rainwater used in the building for specified purposes, m 3 /year. The demand for water for a specific purpose can be determined from Formula (8).
where J is the number of occupants, person and d dt is the average daily water demand for a specific purpose, m 3 /person/day. The costs of wastewater disposal are determined from Formula (9).
where B t is the volume of sewage discharged from the building, m 3 /year and cb t is the sewage disposal price, €/m 3 . However, in the case of using water for toilet flushing and washing, it is assumed that B t = D t , whereas in the case of using rainwater for works outside the facility (watering and utility works), the value of B t is calculated from Formula (10).
where E t is the volume of tap water discharged outside the sewage system, m 3 /year. The cost of rainwater drainage RC kt can be calculated from Formula (11).
where R t is the volume of rainwater discharged from the building to the drainage system, m 3 /year and cr t is the price for draining rainwater, €/m 3 . The discount rate is a very important parameter affecting the final results. The discount rate was assumed to be 5%, as it was used in calculations by Roebuck et al. [46], Morales-Pinzon et al. [63] and Rahman et al. [28].
Initial investments for the INV k variants were determined based on the information obtained from device manufacturers and contractors. The detailed summary of data for calculations is presented in Table 2.

Hydraulic Analysis
Using the simulation model, hydraulic calculations of rainwater harvesting system operation for various tank capacities were performed for the adopted values of the calculation parameters. The results of the hydraulic calculations for Variant 1A, Variant 2A, Variant 1B, and Variant 2B are shown in Figure 4.
The calculations carried out show a significant relationship that occurs between the tank capacity and the volume of rainwater used in buildings and the amount of tap water and the amount of rainwater discharge outside the system. A comparison of the decentralized system with the centralized system, in the case of the variant where rainwater was used only for toilet flushing (Variant 1A and Variant 2A), no significant differences were observed in the amount of rainwater used for this purpose. For the smallest tank volumes in the systems analyzed, the difference in the amount of rainwater used was about 40 m 3 per year, whereas for the largest tank volumes it was just 8 m 3 . The situation was similar for the case where rainwater was intended for toilet flushing and watering the garden. The differences between the systems were slightly larger, but they were still not significant. Greater water saving opportunities were noticed for the rainwater harvesting system installed individually for each building. Therefore, the question about the rational and financially justified capacity of the rainwater storage tank becomes relevant and this question is answered on the basis of an analysis of the system effectiveness which is determined on the basis of Formula (3). The results of calculating the effectiveness of the E f system are shown in Figure 5.
Resources 2020, 9, 5 9 of 18 the basis of an analysis of the system effectiveness which is determined on the basis of Formula (3). The results of calculating the effectiveness of the Ef system are shown in Figure 5. According to the results of the calculations, it can be clearly observed that above a certain level of tank capacity, a significant increase does not determine a relevant growth in system efficiency. For a difference in the efficiency level of no more than 1%, it can be assumed that the tank capacity would be optimal in terms of the use of rainwater. From the point of view of system efficiency, the most favorable capacities should be the following: The tank with capacity of V = 4 m 3 for Variant 1A, the tank with capacity of V = 5 m 3 for Variant 1B, the tank with capacity of V = 60 m 3 for Variant 2A, and the tank with capacity V = 75 m 3 for Variant 2B. The average performance indicator over a multiannual period is 80%, 79%, 57%, and 54% for Variants 1A, 2A, 1B, and 2B, respectively. The low level of efficiency for the last two variants is mainly due to the high demand for water, which is not covered by rainwater. Because there are such small differences in water savings that can be obtained by replacing it with rainwater, it is the financial analysis that should show which of the analyzed solutions for the rainwater harvesting systems (the decentralized or the centralized ones) is most beneficial for the housing estate considered. According to the results of the calculations, it can be clearly observed that above a certain level of tank capacity, a significant increase does not determine a relevant growth in system efficiency. For a difference in the efficiency level of no more than 1%, it can be assumed that the tank capacity would be optimal in terms of the use of rainwater. From the point of view of system efficiency, the most favorable capacities should be the following: The tank with capacity of V = 4 m 3 for Variant 1A, the tank with capacity of V = 5 m 3 for Variant 1B, the tank with capacity of V = 60 m 3 for Variant 2A, and the tank with capacity V = 75 m 3 for Variant 2B. The average performance indicator over a multiannual period is 80%, 79%, 57%, and 54% for Variants 1A, 2A, 1B, and 2B, respectively. The low level of efficiency for the last two variants is mainly due to the high demand for water, which is not covered by rainwater. Because there are such small differences in water savings that can be obtained by replacing it with rainwater, it is the financial analysis that should show which of the analyzed solutions for the rainwater harvesting systems (the decentralized or the centralized ones) is most beneficial for the housing estate considered.

Financial Efficiency
Calculations of the financial effectiveness of the investment for the adopted variants of the RWHS system were based on the results of hydraulic analyzes. Variant 0 was adopted as the comparative variant, in which no such system was used, and the buildings were supplied with water only from the water supply network. The LCC methodology was used in the analyzes. The results of calculations carried out according to Formulas (4)-(11) for the adopted variants of various reservoir retention capacities are presented in Figure 6.
The test results for various system configurations show that the amount of the LCC costs, and thus the cost-effectiveness of the use of individual variants, is significantly influenced by the capacity of the central tank or the capacity of individual tanks. In both Variant 1 and Variant 2, as the tank capacity increases, the share of INV investment costs in the total LCC costs increases.
On the basis of the results obtained from the life cycle cost analysis, it was found that none of the considered options involving the use of a RWHS is financially more favorable than Option 0, both in the decentralized and centralized system. Therefore, for such defined installation parameters from a financial point of view, none of them is justified. This result is primarily influenced by capital expenditure, which, depending on the system variant, accounted for 60% to 70% of the LCC value for Variant 1A and 41% to 52% for Variant 1B for the decentralized system; and from 58% to 67% for Variant 2A and 39% to 48% for Variant 2B for the rainwater harvesting system in a centralized system. The unfavorable financial result of the centralized and decentralized rainwater harvesting system is also influenced by the operating costs resulting from the purchase of water from the water supply network in the 30-year period analyzed. Savings resulting from replacing tap water with rainwater do not compensate for the costs necessary to carry out a RWHS.

Financial Efficiency
Calculations of the financial effectiveness of the investment for the adopted variants of the RWHS system were based on the results of hydraulic analyzes. Variant 0 was adopted as the comparative variant, in which no such system was used, and the buildings were supplied with water only from the water supply network. The LCC methodology was used in the analyzes. The results of calculations carried out according to Formulas (4)- (11) for the adopted variants of various reservoir retention capacities are presented in Figure 6.
The test results for various system configurations show that the amount of the LCC costs, and thus the cost-effectiveness of the use of individual variants, is significantly influenced by the capacity of the central tank or the capacity of individual tanks. In both Variant 1 and Variant 2, as the tank capacity increases, the share of INV investment costs in the total LCC costs increases.
On the basis of the results obtained from the life cycle cost analysis, it was found that none of the considered options involving the use of a RWHS is financially more favorable than Option 0, both in the decentralized and centralized system. Therefore, for such defined installation parameters from a financial point of view, none of them is justified. This result is primarily influenced by capital expenditure, which, depending on the system variant, accounted for 60% to 70% of the LCC value for Variant 1A and 41% to 52% for Variant 1B for the decentralized system; and from 58% to 67% for Variant 2A and 39% to 48% for Variant 2B for the rainwater harvesting system in a centralized system. The unfavorable financial result of the centralized and decentralized rainwater harvesting system is also influenced by the operating costs resulting from the purchase of water from the water supply network in the 30-year period analyzed. Savings resulting from replacing tap water with rainwater do not compensate for the costs necessary to carry out a RWHS. Generally, it can be stated that the centralized rainwater harvesting system for the housing estate analyzed is financially more advantageous than the decentralized rainwater harvesting system, although the differences in the LCC costs were not significant. For the variant where rainwater was intended for toilet flushing, the LCC indicator for the smallest tank volumes (1 m 3 and 15 m 3 ) was only about 3% lower for the centralized rainwater harvesting system. Similarly, this difference between the LCC costs for the centralized and decentralized systems for the largest tank volumes (5 m 3 and 75 m 3 ) was only 5% (Figure 6a,c). Considering the use of rainwater for toilet flushing and watering the garden (Variant 2A and Variant 2B), it was noticed that the difference in the level of LCC for the tank capacity of 1 m 3 and 15 m 3 was smaller than 1% (Figure 6b,d), and in the case of the largest tanks, this difference was at a similar level as in the situation when RWHS was intended only for toilet flushing.
In order to assess the impact of variable investment conditions on the financial effects, a sensitivity analysis was performed. The analysis involved examining the impact of future changes in the financial model parameters on the level of investment profitability. For this investment, the research was carried out according to three scenarios of changes in the model components within ±25% of the costs determined for the base year. Considering the increase in unit prices of water, sewage, and energy in recent years, and the fact that the calculations had already taken into account the annual increase in unit prices, it was found that a larger decrease or increase in the value of operating costs was unlikely. The following change scenarios were adopted in the research: • Scenario A, the change in the value of TOC operating costs resulting from changes in the prices of tap water and sanitary sewage discharged from the building to the sewage network; • Scenario B, the change in the value of OCkt operating costs resulting from changes in electricity prices; • Scenario C, the change in the value of OCkt operating costs resulting from changes in prices for rainwater drainage to the drainage network. Generally, it can be stated that the centralized rainwater harvesting system for the housing estate analyzed is financially more advantageous than the decentralized rainwater harvesting system, although the differences in the LCC costs were not significant. For the variant where rainwater was intended for toilet flushing, the LCC indicator for the smallest tank volumes (1 m 3 and 15 m 3 ) was only about 3% lower for the centralized rainwater harvesting system. Similarly, this difference between the LCC costs for the centralized and decentralized systems for the largest tank volumes (5 m 3 and 75 m 3 ) was only 5% (Figure 6a,c). Considering the use of rainwater for toilet flushing and watering the garden (Variant 2A and Variant 2B), it was noticed that the difference in the level of LCC for the tank capacity of 1 m 3 and 15 m 3 was smaller than 1% (Figure 6b,d), and in the case of the largest tanks, this difference was at a similar level as in the situation when RWHS was intended only for toilet flushing.
In order to assess the impact of variable investment conditions on the financial effects, a sensitivity analysis was performed. The analysis involved examining the impact of future changes in the financial model parameters on the level of investment profitability. For this investment, the research was carried out according to three scenarios of changes in the model components within ±25% of the costs determined for the base year. Considering the increase in unit prices of water, sewage, and energy in recent years, and the fact that the calculations had already taken into account the annual increase in unit prices, it was found that a larger decrease or increase in the value of operating costs was unlikely. The following change scenarios were adopted in the research: • Scenario A, the change in the value of TOC operating costs resulting from changes in the prices of tap water and sanitary sewage discharged from the building to the sewage network; • Scenario B, the change in the value of OC kt operating costs resulting from changes in electricity prices; • Scenario C, the change in the value of OC kt operating costs resulting from changes in prices for rainwater drainage to the drainage network.
Graphical interpretations of the results from the test outcomes are presented in Figures 7 and 8, and show that for all variants and the entire range of design parameters, assuming the use of rainwater, no significant improvements were observed in the efficiency of variants as compared to the traditional water system solution (Variant 0). When analyzing the impact of lowering operating costs on the value of LCC costs, it was noticed that the largest decrease in the LCC ratio in relation to its base value was in Scenario A for variants where the rainwater harvesting system was intended for toilet flushing and garden watering (Figure 7b,d). Reduction of the cost of purchasing water and wastewater disposal by 25%, in this case, reduce the total costs in the range of 10% (5 m 3 tank) to 12.5% (1 m 3 tank) for the decentralized system and 11% (75 m 3 tank) to 13% (15 m 3 tank) for the centralized system.
Resources 2020, 9, 5 12 of 18 Graphical interpretations of the results from the test outcomes are presented in Figures 7 and 8, and show that for all variants and the entire range of design parameters, assuming the use of rainwater, no significant improvements were observed in the efficiency of variants as compared to the traditional water system solution (Variant 0). When analyzing the impact of lowering operating costs on the value of LCC costs, it was noticed that the largest decrease in the LCC ratio in relation to its base value was in Scenario A for variants where the rainwater harvesting system was intended for toilet flushing and garden watering (Figure 7b,d). Reduction of the cost of purchasing water and wastewater disposal by 25%, in this case, reduce the total costs in the range of 10% (5 m 3 tank) to 12.5% (1 m 3 tank) for the decentralized system and 11% (75 m 3 tank) to 13% (15 m 3 tank) for the centralized system. Similarly, when operating costs increased by 25%, the greatest impact was observed for Scenario A and the use of RWHS for toilet flushing and for watering the garden, both in a decentralized and centralized systems (Figure 8b,d). The lowest impact was visible for the changes in costs according to Scenario B, and similar to the variants with RWHS these costs resulted only from the discharge of excess rainwater to the sewage system. Similarly, when operating costs increased by 25%, the greatest impact was observed for Scenario A and the use of RWHS for toilet flushing and for watering the garden, both in a decentralized and centralized systems (Figure 8b,d). The lowest impact was visible for the changes in costs according to Scenario B, and similar to the variants with RWHS these costs resulted only from the discharge of excess rainwater to the sewage system. Taking into consideration the results based on the sensitivity analysis and the high share of INV investments in the total LCC costs, this research was carried out to determine the level of cofinancing needed to implement the systems and allow the investments to be profitable. Currently, in Poland, there are very few examples of cities where residents can receive such funding. The results of this research in this area are shown in Figure 9. In the case of the decentralized system and the use of rainwater only for toilet flushing (Variant 1A), 25% cofinancing does not affect the profitability of the investment, therefore, the application of RWHS is more favorable than Variant 0 (Figure 9a). In the case of Variant 1B, where rainwater was used in addition to watering the garden, reducing the INV by 25% results in a slight reduction in the LCC costs below the total costs of Variant 0, but only for the tank volumes of 1 m 3 , 2 m 3 , and 3 m 3 (Figure 9b). For Variant 1A and Variant 1B, the 50% cofinancing significantly reduces the LCC in relation to its base value. Depending on the tank capacity, this reduction ranges from 30% to 35% of the LCC costs for Variant 1A and from 20% to 26% for Variant 1B.
In the case of the centralized rainwater harvesting system and Variant 2A, the subsidies for initial investments of 25% of their value only for a 15 m 3 capacity tank would reduce the total investments slightly below the level of these costs for Variant 0 (Figure 9c). In addition, for Variant 2B, the LCC cost differences between Variant 0 and the solutions from RWHS were insignificant. As in the decentralized system, a visible improvement in financial conditions of investments is only observed at a 50% reduction of the INV. The LCC costs were reduced from 29% to 33% for Variant 2A and from 19% to 24% for Variant 2B as compared with the base value of these costs. Taking into consideration the results based on the sensitivity analysis and the high share of INV investments in the total LCC costs, this research was carried out to determine the level of cofinancing needed to implement the systems and allow the investments to be profitable. Currently, in Poland, there are very few examples of cities where residents can receive such funding. The results of this research in this area are shown in Figure 9. In the case of the decentralized system and the use of rainwater only for toilet flushing (Variant 1A), 25% cofinancing does not affect the profitability of the investment, therefore, the application of RWHS is more favorable than Variant 0 (Figure 9a). In the case of Variant 1B, where rainwater was used in addition to watering the garden, reducing the INV by 25% results in a slight reduction in the LCC costs below the total costs of Variant 0, but only for the tank volumes of 1 m 3 , 2 m 3 , and 3 m 3 (Figure 9b). For Variant 1A and Variant 1B, the 50% cofinancing significantly reduces the LCC in relation to its base value. Depending on the tank capacity, this reduction ranges from 30% to 35% of the LCC costs for Variant 1A and from 20% to 26% for Variant 1B.

Conclusions
The aim of this study was to compare centralized and decentralized rainwater harvesting systems in terms of their hydraulic and financial efficiency. The analysis was carried out for a real estate development of single-family houses located in Poland.
The obtained research results showed that the efficiency of the RWHS depended mainly on the demand for non-potable water and the amount of rainwater flowing into the system, and to a lesser extent on the system layout (centralized or decentralized). In variants, where rainwater was only used to flush toilets, the water saving efficiency was 80% and 79% for decentralized and centralized RWHSs, respectively. The use of rainwater for toilet flushing and watering the garden resulted in a significant reduction in efficiency to 57% (the decentralized system) and 54% (the centralized system).
The results of the life cycle cost analysis clearly showed that in spite of the reduction of reducing tap water consumption, neither the centralized nor the decentralized rainwater harvesting system was a financially viable solution for the housing estate in question. This was also confirmed by the sensitivity analysis that was conducted. Such unfavorable outcomes resulted from the amount of investments which in some variants constituted even 70% of the total LCC costs. Therefore, the cofinancing of such investments significantly improved their financial indicators. Depending on the tank capacity, the reduction of investments by 50% resulted in a maximum reduction in LCC costs from 29% to 33% for the centralized rainwater harvesting system and from 30% to 35% for the decentralized one.
On the basis of the results of this research an important conclusion, which is also appropriate for other countries, can be drawn which is that in order to introduce sustainable management strategies and protection of water resources with a noticeable impact on the environment, it is necessary to raise public awareness, and above all, financial support, to encourage potential investors to implement decentralized or centralized alternative water installations. In the case of the centralized rainwater harvesting system and Variant 2A, the subsidies for initial investments of 25% of their value only for a 15 m 3 capacity tank would reduce the total investments slightly below the level of these costs for Variant 0 (Figure 9c). In addition, for Variant 2B, the LCC cost differences between Variant 0 and the solutions from RWHS were insignificant. As in the decentralized system, a visible improvement in financial conditions of investments is only observed at a 50% reduction of the INV. The LCC costs were reduced from 29% to 33% for Variant 2A and from 19% to 24% for Variant 2B as compared with the base value of these costs.

Conclusions
The aim of this study was to compare centralized and decentralized rainwater harvesting systems in terms of their hydraulic and financial efficiency. The analysis was carried out for a real estate development of single-family houses located in Poland.
The obtained research results showed that the efficiency of the RWHS depended mainly on the demand for non-potable water and the amount of rainwater flowing into the system, and to a lesser extent on the system layout (centralized or decentralized). In variants, where rainwater was only used to flush toilets, the water saving efficiency was 80% and 79% for decentralized and centralized RWHSs, respectively. The use of rainwater for toilet flushing and watering the garden resulted in a significant reduction in efficiency to 57% (the decentralized system) and 54% (the centralized system).
The results of the life cycle cost analysis clearly showed that in spite of the reduction of reducing tap water consumption, neither the centralized nor the decentralized rainwater harvesting system was a financially viable solution for the housing estate in question. This was also confirmed by the sensitivity analysis that was conducted. Such unfavorable outcomes resulted from the amount of investments which in some variants constituted even 70% of the total LCC costs. Therefore, the cofinancing of such investments significantly improved their financial indicators. Depending on the tank capacity, the reduction of investments by 50% resulted in a maximum reduction in LCC costs from 29% to 33% for the centralized rainwater harvesting system and from 30% to 35% for the decentralized one.
On the basis of the results of this research an important conclusion, which is also appropriate for other countries, can be drawn which is that in order to introduce sustainable management strategies and protection of water resources with a noticeable impact on the environment, it is necessary to raise public awareness, and above all, financial support, to encourage potential investors to implement decentralized or centralized alternative water installations.