Rainwater collected volume estimation

Rainfall analysis is developed based on records corresponding to 59 years (1951 to 2010) retrieved from 8 climatological stations which are close to the study area (^{SMN, 2010}), as is shown in Table 1 and Figure 2.

Figure 2 Climatological stations close to the UAM Azcapotzalco. 

The building studied is five stories tall, comprising Laboratories, auditoriums, a greenhouse, an esplanade and a potential collecting area of 1974 m2. Average volume of water to be captured in one month period can be estimated based on the information on the historical rainfall average of each month and the catchment area according to Equation (1).

where, Vi,j is the volume corresponding to the month i, for the collecting area j (m³), Pi It is the average historical precipitation of the month i (m), Aj is the area j for the rain harvesting (m²), Ct is the runoff coefficient which represents the fraction of water that can be captured related to the rainwater flowrate. ^{Aparicio (1992)} suggests for rooftops values ranging from 0.75 to 0.95, but in the case of the present study it is estimated that few loses can be attain, mainly by evaporation on the same rooftop and for considering not to capture the first rains of the season, instead, use this rain to clean the colleting area. Based on the previous considerations, a runoff coefficient of 0.9 is applied.

Determination of water volume consumption

During two weeks, samplings were conducted to account for the number of persons entering to the toilets and the cleaning activities performed by the maintenance personnel, occurring from 8:00 to 17:00 h. The data collected allowed for estimating the amount of water demanded by the building, to identify and to plot variation curves for the demand for this building: Group 1 (Monday, Wednesday and Friday) comprised by students attending lectures held on others buildings, but using the toilets in the building W. Group 2 (Tuesdays and Thursdays) corresponding to students conducting research activities in the laboratories of the building W. Group 3 (Saturday) with a small probability of incurrence comprised only by students conducting research activities in the laboratories and working personnel for maintenance activities. It is estimated that no Sundays there is no need for water usage, except on rare occasions and for this reason is considered irrelevant.

Water consumption was classified according to the gender; and a factor was used for each sanitary furniture employed. For men this was 0.75 in the toilet, 0.25 in the urinal and 2.0 in the sink, meaning that for each person who enters, 75% is prone to use the toilet, 25% the urinal and activating 2 times the sink water dispenser. For women, 1.0 is assigned to the toilet and 2.0 to the sink. All sinks consume 0.25 l, all urinals 0.5 l and all toilets 5.0 l per flush accordingly to the manufacturer's figures (^{Helvex, 2000}). In addition to the water consumption in the bathrooms, a volume of 0.36 m3/day for cleaning purposes is employed accordingly to what is reported back by the cleaning staff.

Water consumption per hour can be estimated according to the Equation (2):

where Vh,p  is the volume consumed per hour as a function to the number of people p (m³), Np is the number of people in an hour, fl is the sink factor, Ql is the volume of water consumed in a lavatory service (m³), fi is the toilet factor, Qi is the volume of water consumed in a toilet service (m³), fm is the urinals factor, Qm is the volume of water consumed in a urinal service (m³).

Collection-consumption balance

With information related to average collection per month and average demand per day, a daily balance can be obtained, allowing to identify the times when the demand is greater and an external endowment is therefore required; or times when the collection is greater and there is a surplus of water that can be stored. This is called potential volume of rainwater to be stored for a period of time (day, week, month or other), and is calculated with a transit model applied to the tank as is shown in Equation (3):

where VTn is the volume stored in the tank by end of day n (or the day before n-1) (m³), VCAn is the volume of rainwater collected during the day n (m³) which results from dividing Vi,j (Equation (1)) by the number of days of the corresponding month, VCOn is the water consumption corresponding to the day n resulting from adding 24 hours of hourly consumption Vh,p  (Equation (2)) (m³).

Modeling the operation of storage tanks

Once the acquisition-demand dynamic has been obtained, three tanks capable to store and to meet the demands for water consumption in the W building during the periods of scarcity are proposed. To simulate each of the tanks operational performance in situations of filling and spillage (Figure 3), equation four is applied:

where Van is the volume stored during day n (m³); Vcn is the average volume of rainwater collected on day n (m³); Van-1 is the volume of water stored by end of day before n (m³); Don is the water consumption during the day n (m³), and Ta is the maximum storage capacity for the tank (m³).

Figure 3 Collection-storage diagram. 

Volume of water to be collected

In Mexico City, the dry season occurs from November to April, representing only 8.4% of the accumulated rainfall, while the rainy season occurs from May to October, accumulating the remaining 91.6%. This precipitation is modulated to a large extent by the passage of clouds coming from the east, providing a flow of moisture from the Gulf of Mexico (^{Magaña, López, & Vázquez, 2013}). The volume of rainwater that can be collected in each month can be is obtained by applying the Equation (1) and is shown in Figure 4.

Figure 4 Volume of water to be collected per month. 

Volume of water consumption

Consumption varies throughout the day and the week, depending on the activity that is occurring. For obvious reasons, it is different from the home time variation of Mexico City published by ^{Conagua (2015)}, where the peak consumption occurs from 8:00 to 9:00. For the holidays season (typically in May and September) the influx is represented by group 3, that is to say, its value in water consumption is the one used by the cleaning staff. On the other hand, during the schooling year, the days of greatest consumption are those of group 1 (Figure 5) with a maximum of 0.24 m³ during midday, while for group 2 (Figure 6) the highest consumption occurs after 13:00 hrs., which amounts to 0.17 m³. It turns out that this is the hour at which students leave lectures in the W building. Both groups consider 0.32 m³ of water usage for cleaning tasks.

Figure 5 Variation curve of consumption group 1. 

Figure 6 Variation curve of consumption group 2. 

Collection-consumption balance

During the dry season, April is the only month that can offer more water than that necessary (Figure 7), this means that during this season there is a deficit of about 40%, however, this is not comparable in magnitude with the surplus obtained during the months corresponding to the rainy season, for which during July is around 90% with respect to the consumption. During the rainy season, not only there is an increase in the frequency of rainfall, but in the intensity as well; practically twice as much waterfalls every time it rains (^{Rodriguez, 2017}). All this information shows the high potential for water collection concerning consumption throughout the year.

Figure 7 Collection-Consumption relation. 

By the end of a year, after collecting rainwater and satisfying the consumption of the building, a surplus of 1,259 m³ can be used to supply the demand in other buildings of the UAM Azc or to be employed in other necessities (see Figure 8), or send it back in the groundwater aquifer through the absorption well, thus helping to mitigate groundwater aquifers overexploitation in Mexico City, for this is the main source of water supply (^{Canteiro, Olea, Escolero, & Zambrano, 2019}).

Figure 8 Volume surplus throughout the year. 

Storage tanks operation

All tanks evaluated have a corresponding storage volume of 5 m³, 10 m³ y 20 m³, respectively, for which it is considered that their operation begins on the first day of the year. It turns out that for all cases, during the first year of operation, the system requires 43 m³ of an external endowment for January to March. However, starting from the second year, this external endowment is no longer required. Once the tank has reached its maximum capacity in March, the volume of rainwater collected does not decrease in the following months (see Figure 9). Based in the previous information, purchasing a tank of 5 m³ in capacity will be enough to store water in the building implying lower investment cost.

Figure 9 Storage tanks operation during two years. 

Historical operation simulation

When simulating the system behavior, daily average precipitation values are replaced for records of each day corresponding to the year 2015 (for which records are complete). The result shows what would have been the performance of the tank if the system would had been implemented since January 1 (Figure 10). It is important to clarify that the year 2015 presented an average precipitation of 680 mm, which is 19% lower than the historical average (837 mm). The tank with the required capacity to store the water is the one with volume of 5 m³. It would have been required 63 m³ of an external endowment until March. Once filled, a 100% of consumption is satisfied (see Figure 10).

Figure 10 Simulation results for the tank based on data available for the year 2015. 

In the literature, other studies with similar methodologies have been reported, however, conditions regarding to precipitation, collecting or consumption are different. ^{López, Cruz and Rojas (2018)} conducted their study in Mexico City hosted by a logistics company. These authors conclude that water harvesting can meet 100% the demand of water consumption. They explain that this is possible because of the correspondence between the collecting area and the number of users is also significant. ^{Palacio (2010)} employed a similar methodology in an educational institution in Caldas, Antioquia, Colombia. This author also considers costs for treatment to make it drinkable, concluding that during nine months of the year, toilets and cleaning activities are covered 100%, however, for the remaining months only 90% is covered. ^{Karan and Jakhar (2006)} employed a similar methodology applied to two educational institutions in New Delhi, where it is only possible to cover 65% of the water demand during five months considered by the study. The present study and that conducted by ^{López et al. (2018)} both in Mexico City, consider an average annual rainfall of 837 y 730 mm respectively, whereas the study conducted by ^{Palacio (2010)} reports the highest average annual precipitation of 2,938 mm. On the other hand, ^{Karan and Jakhar (2006)} reported in their study an average of 693 mm, most likely to be present during the summer.