Sustainability Assessment of Harvesting Rainwater and Air-Conditioning Condensate Water in Multi-Family Residential Buildings under Various Conditions in Israel—A Simulation Study

Abstract

The environmental impacts and water savings of different configurations of non-potable domestic water use (toilet flushing and laundry), sourced from rainwater harvesting (RWH) and air-conditioning condensate water (ACWH), in multi-family buildings in Israel are examined. Two building types differing in specific roof areas, and three climatic sub-regions were modeled. RWH satisfied 23 and 46% of the water demand for toilet flushing and laundry in high-rise and low-rise buildings, respectively. Air conditioning is used almost daily during Israel’s hot and dry summers. Hence, the combined RWH-ACWH system saved 42 and 64% in high- and low-rise buildings, respectively. Displacing desalinated seawater, a significant water source in Israel, with alternative water sources lowered the environmental impacts with an increase in storage, up to a certain volume, beyond which impacts started rising. The same infrastructure is used during winter for RWH and for ACWH during summer; thus, combining the two exhibits significant water savings, with marginal extra costs while lowering the environmental impacts.

1. Introduction

The current study explores the use of two alternative water sources, which are otherwise often lost or become inaccessible: rooftop rainfall runoff and air conditioning condensate water. In urban environments, where soil infiltration is limited due to coverage by impermeable surfaces, rainwater most likely drains into the drainage system and ends up in the sea, while air-conditioning condensation water is likely to end up on impermeable surfaces and evaporate. In other words, these are two water sources (for non-potable uses in this study, but potentially potable under different conditions) that, in most cases, would not benefit a natural ecosystem or replenish surface—or ground—water reservoirs. In the following, the literature regarding rainwater and air-conditioning condensate water harvesting is reviewed.

1.1. Water Resource Management

Pressure on already stressed freshwater resources continues to increase due to climate change, urbanization, industry, and agriculture, which are all rising, to supply demands of a growing global population, with an expected increase of 20–30% of global water demand by 2050 [1]. The problem of dwindling water availability has been approached by manipulating the environment for too long. Rivers were diverted, dams were erected, flooding valleys, and long pipelines were laid to meet growing water demand without seriously considering the environmental consequences. This is exacerbated by deteriorating water quality, due to anthropogenic activities. Hence, to ensure long-term water security, water resource management must shift towards a more sustainable approach.

The role of water reuse and using additional non-traditional water sources should be central in today’s water management policies. However, to avoid burden shifting, these solutions should be scrutinized through the lens of life cycle environmental analysis while always considering possible loss of utility elsewhere.

1.2. Rainwater Harvesting

Rainwater harvesting for non-potable domestic use is potentially advantageous not only in terms of freshwater saving but also in reducing urban runoff, mitigation of impacts of urban stormwater on receiving water bodies, potable water (PW) treatment, PW supply, and the environmental impacts related to all these services [2,3].

Rainwater harvesting (RWH) efficacy strongly depends on climate factors, especially on the quantity and seasonal distribution of rainfall. In a typical large building in most regions of the USA, less than 10% of non-potable water (NPW) demand can be met by RWH. Only in the Gulf Coast and the Pacific Northwest does this this figure increase—up to 15% in most areas, with a local peak of 42% on the Washington coast [4]. In a multi-story student accommodation building in southern India, RWH may meet up to 42% of PW demand during the monsoon season, but contribute hardly any water during the long dry season (October–May, [5]). In Slovakia, with an average rainfall of 34–59 mm/month, RWH may lead to over 60% reduction in potable water use for toilet flushing, laundry, and cleaning, and up to 64% of household water-related expenses [6]. In western Malaysia, where rainfall is abundant year-round (>2000 mm/y and at least 130 mm/month), RWH may supply over 95% of water for toilet flushing and irrigation [7]. RWH efficiency also varies greatly with design parameters such as roof area to number of occupants ratio and storage tank size, as well as local attributes such as occupants’ water demand. This variability was demonstrated by Muklada et al. [8], who have shown that for a (flat) roof area of 200 m², the water saving efficiency of RWH for toilet flushing during the rainy season (October–May) in Israel (Mediterranean climate) ranged from as much as 80% for 4 residents (single-family home) to ~10% for 64 residents (multi-apartment building). Feloni and Nastos [9] report that in single-family homes in two small Greek islands, having a very similar climate to Israel, RWH could supply 30 to 100% of the water demand for toilet flushing, changing almost linearly with normalized roof area (13.3 to 47 m²/resident). In this case, storage tank size had negligible effect on the system reliability. In a multi-family home (6.3 m²/resident normalized roof area) in the coastal zone of Israel (annual avg. rainfall 507 mm), 18% of the annual water demand for toilet flushing can be supplied by RWH with a 50 m³ storage tank [10].

Domestic RWH solutions have been investigated from a life-cycle perspective with varying results. Crettaz [11] found that using rainwater for non-potable domestic uses is environmentally advantageous only when the production and distribution of mains water entail extremely high energy demand. RWH in a single-family residence in Malaysia proved to be some 12% lower in global warming potential compared with municipal water supply. That being said, the global warming potential attributed to RWH is highly dependent on the local electricity mix. Zepon-Tarpani et al. [12] showed a 3-fold difference when using grid electricity in Brazil (591 kgCO₂eq/1000 m³) and South Africa (1878 kgCO₂eq/1000 m³), which is generated by hydropower (63%) and coal (92%), respectively.

Common to several studies [11,12,13,14] is the conclusion that the type and size of the rainwater storage tank is a critical factor in the environmental performance of RWH systems. The tank is a main part of the required infrastructure, which may contribute significant impacts due to its value chain (raw materials and production processes). Its placement (above the roof, under the roof, or at ground level) dictates the pumping arrangement and its energy demand, which is another major factor in the environmental impacts of the system [15]. On a neighborhood scale, considering urban runoff reduction, soil permeability, and the runoff treatment may also be determining factors in the life-cycle impacts of RWH [16]. It is worth noting that Martins Vaz et al. [17] in a recent review paper state that LCA of RWH needs a more comprehensive contextual approach, specifically ones considering the impact at the urban scale.

1.3. Air-Conditioner Condensate Water Harvesting and Combined Systems

Depending on ambient temperature and humidity, air-conditioner condensation water (ACW) may replace varying fractions of domestic water demand. In the US, ACW may supply up to 10% of non-potable demand for a large apartment building in most of the country, and up to 26% in the humid and warm southeastern states [4].

In the aspiration for more sustainable urban water management, ACWH (air conditioner water harvesting) and RWH need not be regarded as competing methods of supplying water demand. Since both are relatively clean water sources and are often active during different seasons, combining the two may have added benefits. Moreover, their combination would not incur much added infrastructure for collection and treatment and use the same infrastructure for distribution; hence, they will not add much to the environmental and financial burdens.

Ghimire et al. [18] found that RWH is environmentally preferable to ACWH in 4-story buildings in two locations in the US: Washington DC and San Francisco. However, in taller buildings (19 stories) in San Francisco, ACWH has lower environmental impacts than RWH. This is because a larger volume of water is harvested from air conditioners than from rainfall, due to the warmer and drier climate and a smaller normalized roof area (roof area per occupant). A combined RWH + ACWH system was environmentally equivalent to the better-performing system in all buildings and locations tested.

Israel’s climate is characterized by a very warm and dry (no precipitation) summer (June to September), during which air conditioning is extensively used in most living spaces. Thus, if only rainwater was to be harvested, the installed infrastructure for collection, storage, and supply of the harvested rainwater would be used only about six m/y. For a better payback of the installed RWH systems, both financially and environmentally, it makes good sense to use the same infrastructure during the dry season for reclaiming readily available air-conditioning condensate water. The fact that this practice hardly requires additional investment of funds and material resources means that it will reduce the overall impact of the system through the benefits of additional PW savings.

Environmental impacts and water-saving benefits of using different configurations of harvested RW and ACW for toilet flushing and laundry in a multi-family building in an urban Israeli setting are studied here comparatively, with the current situation, i.e., PW is the sole water source for all domestic uses, being the baseline scenario. It is worth noting that desalinated seawater supplies 70% of the municipal PW in Israel [19] and is expected to increase in the future. Thus, desalinated seawater use, being the marginal water source, will be reduced due to PW saving.

In summary, the sustainability of RWH systems has been studied extensively; much less attention has been given to ACWH and even less to combined RWH + ACWH systems. The current study addresses this knowledge gap by bringing three new angles, for sustainability assessment of alternative water sources: (1) using the same infrastructure for two different NPW sources, (2) performing the study in climate zones where RWH is applicable only during winter and ACWH only during summer, ranging from Mediterranean (completely dry summers) to Arid, and (3) desalinated seawater being the source of potable water.

2. Goal and Scope

The life cycle of a water harvesting system begins with the environmental impacts of resource extraction, product manufacturing, and installation. The use phase is characterized by environmental credits, achieved through reduced PW consumption. The end-of-life phase again incurs environmental costs due to the disposal and treatment of waste.

As aforementioned, the goal of this study is to compare the environmental impacts/benefits and water saving (toilet flushing and laundry) of different configurations of domestic use of RW and ACW in a multi-family building in an urban Israeli setting. The baseline scenario for this comparative analysis is the current situation, where PW is the sole water source for all domestic uses. Desalinated seawater, being the marginal water source, will be reduced due to PW saving.

2.1. System Boundaries

The analyzed system includes all sub-systems related to RW and ACW harvesting, treatment, storage, and supply for toilet flushing and laundry (washing machine) in the building from which they were harvested (Figure 1). The use phase of the water and its end-of-life (including sewage collection and treatment) are excluded, as these are not affected by the source of the supplied water if it conforms with appropriate quality. In the figure, RW and ACW are harvested, treated, stored, and used for toilet flushing and laundry (top raw brown rectangles). Desalination, conveyance, and storage and supply to domestic uses (bottom raw brown rectangles), are reduced (marked by dashed arrows) as a result of the above practice.

Due to the intermittent nature of RW and ACW, supply of PW to the household for toilet flushing and laundry must always remain available. This means that municipal water supply infrastructure must maintain its full capacity, and there are no potential savings in reducing its embodied emissions, e.g., by constructing smaller seawater desalination plants or reducing pipeline diameters, as can be done in the case of greywater reuse [20].

Urban drainage is excluded because the decrease in runoff caused by the RWH from a single roof is not significant enough to lead to any decrease in drainage infrastructure. This would become significant only when a critical mass of RWH is implemented throughout the city.

2.2. Functional Unit

The functional unit of the assessment is the supply of water for toilet flushing and laundry to 40 households (133 residents; for details, see Section 3 below) within a single residential building for one year (m³/y).

3. Methods

3.1. Infrastructure

The model refers to an urban apartment building of 40 households, built either 10 or 5 stories high (referred to as “hi-rise” and “low-rise”, respectively), with a roof catchment area of 840 and 1680 m², respectively (Table 2). RW is harvested from the building’s flat concrete roof during winter. ACW is harvested from all apartments during the hot, dry summer.

The infrastructure for harvesting both types of waters includes:

• Sloping PVC pipelines from three corners of the building, collecting water from the end of the vertical gutters or AC drainage pipes, into a storage tank.
• Underground HDPE storage tank, located at one corner of the building (hence no transverse drainage required). Excess water overflow is directed to the municipal drainage system.
• A simple floating ball chamber converter, responsible for the daily first flush. Seasonal first flush is assumed to be handled through a manual valve, directing the harvested water to the municipal drainage. Both are negligible in the overall inventory and were not inventoried.
• Harvested rainwater filter (see Treatment section).
• Distribution pumps for supplying the harvested water to building apartments.
• Distribution pipes (PE100, PN16), from the storage tanks’ outlet to all apartments.

3.2. Water Supply and Consumption

The 40 households per building are modelled as identical dwellings, each inhabited by 3.32 occupants, which is the Israeli average for 2017 [21], summing 133 occupants per building. Domestic water consumption is modeled as 138 L/(c·d) (L/(capita·day)), according to the Israel Water Authority [22]. Non-potable consumption (toilet flushing and laundry) constitutes 35% of the overall domestic water usage, i.e., 48 L/(c·d), which amounts to 6.33 m³/(bld·d) (m³/(building·d)). All PW is modeled as desalinated seawater, using inventory data as reported in Opher and Friedler [20].

3.3. Scenarios

Three climatic sub-regions were examined, represented by three monitoring stations of the Israel Meteorological Service (IMS,

Table 1. Characteristics of meteorological stations used for data collection.

	North	Center	South
Climate type	Temperate Mediterranean	Mediterranean (inner coastal/coastal)	Arid
Meteorological station location	Yehiam/Eilon	Raanana East/Beit Dagan	Beer Sheva
Coordinates (deg. N, deg. E)	32.997, 35.217 33.064, 35.221	32.182, 34.882 32.008, 34.814	31.253, 34.800
Altitude (m above sea level)	365/300	48/31	279
Average precipitation (mm/y)	767	525	196
Average number of rainy days (d/y)	79	56	39

): northern Israel (represented by Yehiam for precipitation data and by Eilon station for temperatures and humidity), central Israel (represented by Raanana East for precipitation data and by Beit Dagan station for temperatures and humidity), and the arid area in southern Israel, represented by the Beer Sheva meteorological station.

3.4. Rainwater Harvesting

Daily precipitation from each of the three climatic sub-regions was acquired from the aforementioned IMS stations from October 2010 to May 2017 (i.e., seven rainy seasons). Precipitation statistics are shown in Figure A1 (Appendix A). Average annual harvest potential was calculated for each location, based on a modeled timeline of rainwater storage and consumption (see example in Figure A2).

The rainwater harvesting model is a mass balance model, using a yield after spillage algorithm as recommended by [23]. A detailed description of the model can be found in [8,10]. The first seasonal roof-runoff event of the rainy season, known to be much more polluted than those following, is fully diverted away from the harvesting tank and was incorporated in the model. A daily first flush, intended to divert the first portion of runoff, which washes off most pollutants accumulated on the roof surface since the preceding rain event, was also modeled. Further, initial losses were also considered for each rainfall event (Table 2). Potential harvested volumes range from 145 m³/y for high-rise buildings in the south to 1112 m³/y for low-rise buildings in the north (Table 3).

Table 2. Rainwater harvesting model parameters.

Parameter	Value	Details	Source
Initial losses	2.3 mm/event If preceding day was dry.	Concrete roof, 1% slope	[8]
Seasonal first flush	1st rain event to create runoff is not harvested.
Daily first flush	0.5 mm		[9,10]
Runoff coefficient	1		[24]
Rainy season	Sept–May	273 d/y with potential RWH
Rooftop catchment area	840 m2 1680 m2	High-rise building Low-rise building

Table 2. Rainwater harvesting model parameters.

Parameter	Value	Details	Source
Initial losses	2.3 mm/event If preceding day was dry.	Concrete roof, 1% slope	[8]
Seasonal first flush	1st rain event to create runoff is not harvested.
Daily first flush	0.5 mm		[9,10]
Runoff coefficient	1		[24]
Rainy season	Sept–May	273 d/y with potential RWH
Rooftop catchment area	840 m2 1680 m2	High-rise building Low-rise building

Table 3. Potential RW harvest volumes and proportion of potable water consumption that can be displaced by harvested RW (standard deviations in parentheses).

			North	Center	South
High-rise building	Potential harvest Portion of annual NPW demand	m3/y %	556 (120) 24 (5)	413 (82) 18 (4)	145 (47) 6 (2)
Low-rise building	Potential harvest Portion of annual NPW demand	m3/y %	1,112 (241) 47 (10)	828 (161) 35 (7)	290 (93) 12 (4)

Table 3. Potential RW harvest volumes and proportion of potable water consumption that can be displaced by harvested RW (standard deviations in parentheses).

			North	Center	South
High-rise building	Potential harvest Portion of annual NPW demand	m3/y %	556 (120) 24 (5)	413 (82) 18 (4)	145 (47) 6 (2)
Low-rise building	Potential harvest Portion of annual NPW demand	m3/y %	1,112 (241) 47 (10)	828 (161) 35 (7)	290 (93) 12 (4)

3.5. Air Conditioning Water Harvesting

The use of air conditioning (AC) in the model depends on temperature and humidity data. As aforementioned, data for 2010 to 2017 were acquired from the IMS, for the three climatic sub-regions modeled. AC was assumed to operate in all households whenever the maximum daily temperature was above 25 °C and the average daily relative humidity was over 50%. The duration of AC operation was set according to temperature thresholds (

Table 4. Daily AC operation duration as a function of temperature.

Max. Daily Temp. Threshold (°C)	AC Operation Time (h/d)	Condenser Operation Time (70%) (h/d)
25	6	4.2
28	8	5.6
30	10	7.0
32	12	8.4
34	16	11.2
36	18	12.6

). Maximum operation time is 18 h/d, assuming the apartment is empty for 6 h/d during work/school hours. The AC condenser was assumed to be working 70% of the operation time. Cooling rate data are based on Grossman G. (Prof. Mechanical Engineering, Technion, cooling systems specialist; personal communication, 2017;

Table 5. Cooling rate data per modelled apartment.

Apartment area	175	m2	ICBS [21], 2015–2016 average
Net closed-space apartment area	171	m2	Minus 2% balcony area
Apartment area cooled at any given moment	85.5	m2	Assumption: 50% of net apartment area
Cooling rate	3.42	toncooling
	41,040	BTU/h
	12.03	kJ/s

).

Condensation water yield per h of AC operation was calculated, using a standard psychrometric chart, per category of maximum temperature (

Table 6. Hourly and daily “production” of condensate water by AC in central Israel.

Starting Point (Relative Humidity 66%) 1			Water Removed from Apartment at Endpoint (temp. 25 °C, Relative Humidity: 50%) 1
Ambient Temp.	Moisture Content	Enthalpy at Saturation
(°C)	(kg/kgdry_air)	(kJ/kgdry_air)	(kg H2O/h)	(kg H2O/d)
28	0.0158	60.5	25.1	105
30	0.0176	67.0	25.4	142
32	0.0198	75.5	25.3	177
34	0.0222	83.0	26.2	220
36	0.0250	92.0	26.8	300
40	0.0322	112.0	28.8	485

¹ From psychrometric chart of normal temperatures at sea level altitude.

), and the maximum daily relative humidity, averaged over the summer months (June to September) (66% in northern and central Israel, 61% in the south). For example, if on a certain day the maximum temp. in northern Israel was 33 °C and the maximum relative humidity was over 50%, then the yield of ACW was modeled according to 16 h of AC operation, cooling the apartment from 33 °C to 25 °C and drying the air from 66% (which is the average maximum daily humidity in summer) down to 50%. The model’s sensitivity to using average daytime temperature, rather than the maximum, is examined in the sensitivity analysis section.

3.6. Water Quality

As aforementioned, a diverter diverts the complete first rain event of the season and the first 0.5 mm of rain in all other events. This should significantly decrease the contaminants load in the harvested RW and reduce filtering requirements. Further decrease can be achieved by designing the rainwater tanks as “sedimentation basins”.

Total suspended solids (TSS), as measured in harvested RW in an Israeli urban environment, range from nearly 0 to 279 mg/L with an average of 29 mg/L [25]. Thanks to the first flush diverter, TSS concentration was assumed not to exceed the 75th percentile reported by Friedler et al. [25], which is 33 mg/L.

ACW is basically distilled water, but it might be contaminated by dust, heavy metals, or bacteria. Still, it is presumably cleaner than roof-harvested rainwater, so the same treatment should suffice for the intended NP use.

3.7. Water Treatment

The quality of untreated roof-harvested rainwater, as measured by Friedler et al. [25], was very high, with all parameters complying with Israel PW quality regulations, except turbidity, fecal coliforms (FC), and Cd (26 ± 33 NTU, 4.9 ± 9.4 cfu/100 mL, and 18 ± 24 μg/L, respectively). In this case, since the harvested rainwater is intended for toilet flushing and laundry, its quality does not have to comply with PW standards. However, turbidity should be low, mainly due to aesthetic aspects, and FC should be removed in order to minimize health risk, while Cd is not expected to pose any risk through the intended use. Turbidity can be efficiently removed by filtration (which may also remove some Cd) and FC by disinfection.

Harvested ACW is expected to contain some airborne contaminants, having relatively low turbidity and potentially slightly acidic pH, as well as occasional contamination by bacteria or fungi, originating from the condenser unit [26,27,28].

The treatment of both alternative water sources in the model includes filtration through a commercial stormwater filter, and disinfection by UV irradiation. Data for the stormwater filter were acquired from Contech Engineered Solutions for their StormFilter [29]. The 27” cartridge of StormFilter ZPG™ media is a proprietary blend of zeolite (90%), perlite, and granular activated carbon (GAC). As such, the filter is capable of also removing some heavy metals (e.g., Cd) and dissolved organic matter, if present in the water. Installation of the filter was assumed to be of the CatchBasin configuration, which includes a shallow underground structure and a high flow bypass. Installation components were neglected, except for the required excavation, which was assumed to be of 1 m³/cartridge. Design parameters for specific modelled conditions are based on mass loading and were acquired from ContechES and adjusted to modelled conditions (

Table 7. Model parameters for RWH filtration, based on ContechES StormFilter [29].

Building Type	Filter Characteristics	Region
		North	Center	South
High-rise	Number of filter cartridges	1	1	1
	Cartridge lifetime (y)	1.1	1.7	5.8
Low-rise	Number of filter cartridges	2	1	1
	Cartridge lifetime (y)	1.1	0.9	2.9

). Used filter cartridges were assumed to be transported to a landfill, since no data was found about regeneration or reclamation of used filter media. Inventory data for the filter can be found in

Table A1. Inventory data for one RWH filter unit.

Filter Cartridge Media
Zeolite	61.0	kg	27” ZPG filter cartridge, dry weight. Lifetime depends on pollutant loads (Table 7).	ContechES [29]
Perlite	3.5	kg
GAC	3.5	kg
Cartridge housing
PE	9.0	kg	Lifetime: 25 y	ContechES
Transport
Transoceanic freight ship	652	ton·km	Import of cartridge materials 9600 km	The National Coal Supply Corporation Ltd., Tel Aviv, Israel
Road transport	2040	kg·km	Fresh dry cartridge (68 kg) to site (30 km)	ContechES [29]
Road transport	4080	kg·km	Wet cartridge (136 kg) to treatment facility (30 km)	ContechES [29]
Installation
Hydraulic excavation	1	m3	Lifetime: 50 y	Assumption based on ContechES StormFilter brochure

.

3.8. Storage

ACWH does not require large storage capacity because its average “production” is between 2.5 (±1.4; 2SD) m³/(bld·d) (north) and 4.1 (±3.2; 2SD) m³/(bld·d) (south), which is less than the daily NPW consumption (6.3 m³/(bld·d)). Storage requirements are, therefore, dictated by RWH, of which larger volumes often accumulate over the course of a few days.

Storage size requirements for harvested RW are a function of the quantity of runoff generated on the roof surface and NPW consumption by all households. Runoff volume depends on rainfall and roof area, while water consumption is a function of the number of consumers (i.e., residents). The number of consumers, and therefore total water consumption, is set to be identical in all scenarios, so the roof size and the geographic location are the two model parameters that, together with the varying storage size, affect rainfall use efficiency (RUE—proportion of rainfall fallen on the roof that was harvested and used; for details see [8]) and volumetric reliability (aka water saving efficiency—WSE).

Storage tanks are inventoried as PE vessels installed underground, alongside the outer perimeter of the building. Five different storage sizes are modeled, varying between 10 and 150 m³, each capacity consisting of a certain combination of commercial tanks of different sizes, available in the local market (ROTONIV Technologies Ltd.; Or Akiva, Israel; www.rotoniv.com). Material resources, tank transport to site, and excavation for tank installation were modeled using approximated values acquired through linear interpolations of manufacturer’s data (

Table A2. Model parameters for harvested water storage tanks.

Variable	Modelled Values	Source
Tank volume (m3)	10, 22, 44, 88, 150
Tank weight per m3 stored (HDPE)	y = 42.01v + 3.53 v–volume in m3	Linear interpolation of manufacturer data (R2 = 1.00)
Excavation for tank installation per m3 stored	y = 1.76v + 0.49 v–volume in m3	Linear interpolation of manufacturer data (R2 = 1.00)

).

3.9. Harvested Water Supply

Households practicing NPW use have a dual distribution system—one for PW and one for NPW. Supply of water for NP uses is modeled by PE pipelines (their length varying between scenarios, according to the type of building) and a pump. PW supply infrastructure is not modeled since it remains the same for PW users as well as for users of both potable and NP water.

3.10. LCA Analysis

The system was modeled in SimaPro and inventoried using Ecoinvent version 2.2 and local data as described in Opher and Friedler [20]. Impact assessment was performed with ILCD 2011 Midpoint+ version 1.0.10, using all impact categories (climate change, ozone depletion, human toxicity (cancer, non-cancer effects), particulate matter, ionizing radiation (human health, environmental), photochemical ozone formation, acidification, eutrophication (terrestrial, freshwater, marine), freshwater ecotoxicity, land use, water resource depletion, and mineral, fossil, and renewable resource depletion).

LCA results are discussed as relative values in the context of a comparison between the defined scenarios. References to the actual magnitude of the environmental impacts are avoided since no external comparison was attempted.

4. Results and Discussion

4.1. RW Harvesting

RWH efficiency depends on climatic region through rainfall patterns; building type (high rise/low rise) through size of harvesting basin; and on storage size, which affects RUE. Following, each dependency is discussed separately.

4.2. Geography/Climate

In Israel, the number of rainy days and total rainfall decreases with decreasing latitude, moving from Mediterranean climate in the north to arid climate in the south. The rainy season, from October to March–April, is characterized by intermittent rain events, which are often quite heavy and may last several days, and dry periods of one or more days between them. With such an uneven precipitation pattern, larger storage volumes are required for capturing all potential harvest than if the same rainwater volume would have been uniformly distributed across the rainy season.

4.3. Building Type

The potential annual RW harvest of a low-rise building is double that of a high-rise, since its catchment area (i.e., roof) is twice as large, but water consumption is equal for both types of buildings, as they house the same number of residents. This necessarily means that a low-rise building requires a larger storage volume to achieve the same RUE but, at the same time, can reach a higher WSE.

4.4. Storage Size

Since rainwater availability depends on seasonal precipitation patterns and daily rainfall varies greatly (e.g., 0 to >100 mm/d in northern Israel), the size of storage available for rainwater harvesting is a crucial factor for RUE. The larger the storage, the higher the yield. The two red lines in Figure 2 describe the increasing annual volume of harvested rainwater for buildings in central Israel. For example, a 44 m³ storage volume captures 84% of total potential RW harvest when installed in a high-rise building and 58% in a low-rise, replacing 13 and 17% of the annual NPW consumption, respectively. It is worth noting that for high-rise buildings, increasing the storage size above 44 m³ hardly affects the annual harvested RW volume, while for low-rise buildings, increasing the storage volume continuously rises the harvested RW volume, but at a diminishing rate.

4.5. ACW Harvesting

Potential ACW harvest per building depends on the difference in temperature and relative humidity between the starting point and the desired end point of the cooled space, which differs between the three climatic sub-regions tested, as described in the Inventory section. It is not affected by building type, as both types modeled accommodate an equal number of apartments of the same size.

As aforementioned, ACWH requires smaller storage due to its almost uniform distribution of small daily quantities. In all three geographical areas, the minimal storage tank tested (10 m³) is enough for harvesting 100% of the daily condensate water produced in the building (orange line; Figure 2).

The daily yield of AC condensation water increases with increasing ambient temperatures, as more cooling is required to achieve comfortable living conditions. Average daily yields, calculated by the models are 2.5, 3.6, and 4.1 m³/(bld·d), for the north, center, and south regions, respectively. In all three cases, ACW volume is smaller than the daily NPW demand (6.3 m³/(bld·d)) and is therefore fully harvested, even with the smallest storage size modeled. Harvested ACW supplied 19, 27, and 31% of NPW consumption in the northern, central-coastal, and southern-desert climates, respectively. These correspond to 430, 613, and 707 m³/(bld·y) reductions in desalinated water demand.

4.6. Water Saving (Toilet Flushing and Laundry)

RWH and ACWH put together replace a growing quantity of NPW as storage size increases. With 150 m³ storage in the north, this may be as much as 1470 or 990 m³/(bld·y), in the case of a low-rise or a high-rise building, respectively. Both building types in all climatic sub-regions, except the southern high-rise, may still benefit slightly from yet larger storage tanks (Figure 3). In terms of using available water (Figure 4), ACW is fully harvested in all regions, with the smallest storage tank examined (10 m³). RWH, on the other hand, increases as storage tank sizes increase. Full utilization (i.e., 100% of potential harvest) is achieved with 88 m³ storage in the southern high-rise and with 150 m³ in the southern low-rise and central high-rise. In the other scenarios, the largest storage tank (150 m³) does not achieve full harvesting potential, capturing 99% in the central low-rise and northern high-rise and 95% in the northern low-rise.

4.7. Life Cycle Impact Assessment

The system’s impacts on global warming potential, under modeled conditions and stated assumptions, vary between 3.7 and 7.6 tCO₂eq/y. The northern and central low-rise buildings, harvesting both RW and ACW, exhibit the lowest impacts, and scenarios with large storage volumes in the southern region present the highest ones. These values should be regarded as a gross estimate since there is inherent uncertainty in the data, the modeling parameters, and the assumptions made. Uncertainty analysis was not carried out, but its effect on the conclusions of the study was kept to a minimum by excluding from the system boundaries any activities that may not be relevant to the comparison and by restricting the discussion to an internal comparison, within the modeled system only.

It is worth noting that adding ACWH to an existing RWH system adds negligible environmental impacts while enhancing water availability, as ACWH uses the same infrastructure with marginal additional construction. Hence, adding ACWH actually reduces the specific impact (impact per m³ supplied) of the combined system.

The dominant contributor to all environmental impacts of the studied system, except one, is seawater desalination in the supply chain of municipal water, which is supplied to households for the examined uses when harvesting is not practiced or harvested waters do not fully satisfy demand. The high impact of seawater desalination is mostly due to its high energy demand (~3.5 kWh/m³ produced water), supplied by the local electricity grid, which is based on fossil fuels (40% coal, 50% natural gas, 10% renewables; extrapolated for year 2020 from data by [30]. In all impact categories apart from ozone depletion, harvesting both RW and ACW results in impacts lower than the baseline scenario (no harvesting), across the whole range of climatic conditions and building types—from the least plentiful scenario of a high-rise building in the south to the relative abundance of a low-rise building in the northern region (Figure 5).

Figure 5 presents relative impacts in the various categories as storage size increases for the two extreme scenarios of smallest and largest harvest potential (southern high-rise and northern low-rise buildings). Figure 6 shows the same for the two mid-range scenarios (RWH and combined harvesting in a central high-rise building).

ACWH does not affect storage size since it is captured fully even with the smallest tank modeled. However, the balance between life cycle environmental cost and benefit is affected by the additional ACWH, as it is fully utilized thanks to its relatively uniform temporal distribution during summer, resulting in a significant contribution of NPW. Harvested ACW displaces 19, 26, or 31% (in the northern, central, or southern regions, respectively) of the annual water consumption for the NP uses considered here. The ACWH effect on the life cycle environmental impacts is apparent through a comparison between the two charts in Figure 6. A reduction in impact intensity thanks to implementing water harvesting is much more pronounced and consistent when both RWH and ACWH are practiced (bottom chart) than if RWH is practiced alone (top chart).

Injection molding, which is the HDPE storage tank production process, is a close second or third contributor to the overall system impact in most impact categories, and the main contributor to ozone depletion. Within the injection molding process, the use of organic solvents is responsible for the destructive effect on stratospheric ozone. Since the overall system’s impact on ozone depletion, unlike all other categories, is affected more by the production of the tanks than by desalination, its impact in this category increases more steeply with storage size (Figure 5). As far as ozone depletion is concerned, the smaller the storage tank, the better. It is worth noting that if the storage tank had been constructed of steel or concrete, its ozone depletion would have been much smaller, with greater impact in other categories.

Whereas storage tank size is a major contributor to ozone depletion, it is only a minor one to climate change. For example, for a low-rise building in central Israel with a storage volume of 44 m³, seawater desalination contributes 87% of the climate change impact, while all other processes contribute 13% altogether (i.e., HDPE 3%, injection molding of the tank 3%, land transport of the materials 3%, low-voltage electricity 2%, PVC 1%, and the remaining processes 1%). Considering this, the above combined RWH + ACWH system itself (excluding CO₂eq emission by desalination) emits ~500 kgCO₂eq/1000 m³, which is comparable with the values reported by [13,15].

4.8. Optimal Storage Size

On one hand, the larger the storage tank, the more PW may be replaced by reclaimed water. On the other hand, larger storage comes with an increase in environmental impacts, emanating from increased energy requirements and quantities of raw materials to produce the tanks, fuels for their transport and installation, and emissions linked to these processes. In the upper plot in Figure 5, above 10 m³ storage, all impacts increase in a near-linear curve. This is because no significant volume of harvested RW is added with increasing storage, and the increasing impacts of storage tanks are practically the only affecting parameter. Of course, impacts decrease gradually at first, but due to the low sampling resolution of small storage sizes, it is not evident in the results. Storage size should be determined so that its added benefits through reduced PW demand are not overshadowed by its negative impacts.

Determining the optimal storage size is a multi-criteria problem, as the system should maximize water saving and minimize environmental impacts (a multi-criteria problem in itself). In reality, there may be additional criteria, such as minimizing monetary costs. A multi-criteria analysis is outside the scope of this study, but results for the first two criteria are examined separately.

Figure 7 presents optimal storage sizes for each of the six scenarios of RWH + ACWH by three different methods. The first is maximizing PW replacement. Since harvested water quantities approach a plateau with growing storage size (Figure 3), a threshold of 2% reduction in PW demand is used (i.e., the first step of increasing storage size that results in <2% decrease). This results in optimal tank sizes between 22 m³ for southern high-rise buildings and over 150 m³ for northern low-rise buildings.

Storage sizes satisfying the criteria of minimal environmental impacts, averaged over all examined impact categories with equal weights to all, are between 14 and 105 m³ (Figure 7). The smallest tank of 14 m³ is required for a high-rise building in the southern region, where conditions are driest, while a low-rise building in the northern region, where rainfall is most abundant, would benefit most from 105 m³ storage.

The blue bars in Figure 7 represent storage sizes determined by the minimum of monthly NPW demand and monthly average harvest (after [19]. In all three climatic conditions in this study, the potential average collection per month is lower than the NPW demand; therefore, the optimal storage size in all scenarios, according to this method, is the monthly volume of water collected: 69 to 118 m³ for a southern high-rise and a northern low-rise building, respectively.

4.9. Sensitivity Analysis

The model’s sensitivity to the duration of AC operation was examined by comparing harvested ACW quantities when the duration is determined by average daytime temperature rather than by maximum over 24 h. Average daytime temperature was calculated using 2010–2017 IMS data of 3 h intervals from 8:00 to 20:00. This was on average 15% lower than the daily maximum temperature, with averages of 22, 23, and 23 °C, compared with 26, 27, and 28 °C for north, center, and south, respectively. The trigger for using the AC on a particular day remains unchanged—maximum daily temperature of >25 °C; however, the duration of its operation is determined by daytime average, according to Table 4, with one exception: when max. temp. is >25 °C and average daytime temp. is <25 °C, daily AC operation was modeled as 4 h.

Figure 8 shows the reduction in ACW harvest caused by changing the model parameter determining daily AC usage from the daily maximum ambient temperature (checkered orange bars) to the average daytime temperature (solid orange bars). Using the average daytime temperature resulted in a 35, 32, and 50% reduction in ACWH in the northern, central, and southern climates, respectively.

The harvesting model is very sensitive to changes in available ACW quantities because it constitutes a significant portion of the total water harvested. The difference in magnitude of its effect in the different climatic sub-regions has to do with the quantities of ACW relative to precipitation, which are much higher in the south, where rainfall is scarce and temperatures are high. There is also a larger difference in day/night temperatures in the south, characteristic of the desert climate, meaning that the lower morning and evening temperatures lower the daytime average by much more in the south than in the other two locations, resulting in a larger difference between the daily maximum and the average temperatures.

5. Conclusions

Rainwater harvesting is environmentally cost-effective only if the operational phase of its life cycle results in an offset of municipal water supply, which balances the resources invested during the rest of its life cycle. It is most beneficial and manageable where precipitation is abundant, but is, of course, more needed where it is scarce. In this study, all three climatic sub-regions are characterized by 5–6 completely dry summer months, during which temperatures rise and air conditioning is used almost daily. These sub-regions vary in the amount of rainfall and summer temperatures: annual rainfall decreases, while maximum daily temperatures increase, as one moves from north to south along the 400 km length of the state of Israel.

In terms of water saving, the model shows that RWH alone, with optimal storage capacity (calculated by maximum water saving), may replace up to 23% of water demand for toilet flushing and laundry (540 m³/(bld·y)) in a high-rise building as modeled, and twice that much in a low-rise building, as the catchment area is doubled (having the same number of dwellers). The combined RWH and ACWH system saves up to 42% (970 m³/(bld·y)) and 64% (1470 m³/(bld·y)) in a high- and low-rise building, respectively.

LCA shows a trend common to all examined impact categories of generally lower impacts when alternative water sources displace the desalinated water supplied through the municipal grid. As storage size increases, impacts tend to decrease up to a certain point, referred to as the optimal environmental storage size, after which further increase in tank size causes impacts to rise. The optimal storage size for RWH and ACWH depends on building typology (i.e., the ratio of roof area to number of residents) and precipitation patterns. The range of storage volumes examined in this study (10 to 150 m³) proved adequate for all scenarios except the low-rise northern one, which has the largest potential water harvest and is not fully utilized using a 150 m³ storage tank.

Harvesting air conditioning condensate water may not necessarily be justifiable on its own, due to the relatively small quantity of offset PW supply compared with the resources necessary for its capture. However, in a setting where the infrastructure for collection, storage, treatment, and distribution is already in place, as is the case where rainwater is being harvested, then it may be a significant added benefit, at hardly any extra costs while lowering environmental impacts.