*2.4. Method*

This subsection describes the definition of experimentation parameters, as well as the procedure followed to perform and monitor the rain simulation for each roof prototype.

#### 2.4.1. Definition of Experimental Conditions

The conditions to simulate precipitations were established according to the geo-climatic characteristics (altitude, temperature, radiation, and geographic location) of Valle del Cauca, Colombia, more precisely in the city of Santiago de Cali. We used the official data record from the weather station of the Universidad del Valle—located at the latitude (3◦22 39.66"), longitude (76◦32 05.26") coordinates of the WGS 84 system—because it is the closest station to the study area with official data [15].

A review of rainfall records was necessary to define the quantity and characteristics of precipitations in the study zone. Data from 1966 to 2015 were considered to define the most relevant precipitation cycles or scenarios. Based on these, three main types of precipitation were considered.

Typical rainfall: This consists of a cycle of several consecutive days of rain, the intensity of which corresponds to the wet season of the periodic cycles presented in the geographical study area. The criteria used for the selection were:


IFD (intensity–frequency–duration) curves were not considered in this rainfall scenario due to the fact that the capacity of the spray system per roof prototype was 1.2 L/min. Rainfall for the five days were simulated as follows: precipitation of 4 mm in a single period of 2.4 min (2 min and 24 s); 12 mm precipitations in two periods of 3.6 min (3 min and 36 s); 50 mm precipitation in three periods, the first two of 9 min (15 mm) and the last of 12 min (20 mm); 51 mm precipitation in three periods, the first two of 9 min (15 mm) and the last of 12.6 min (21 mm); and finally, 1 mm precipitation over 36 s. The time between each period was thirty minutes for each measurement, and the data employed to estimate typical conditions were taken from 23 to 29 October 1987 (See Appendix A—Figure A1a).

Intense rainfall: This was considered under the variables of intensity, frequency, and duration. It was selected the one with historical data higher than 30 mm/h that would have occurred in those atypical events and under the effects of La Niña phenomenon. It was also important as a requirement to have an IFD curve. The selected parameters were the total precipitation (49.70 mm) and a duration of 1 h (intensity: 49.7 mm/h).

Considering the need to simulate intense rain in the closest way to the IFD curve and taking into account the capacity and restriction of the irrigation system, this precipitation was simulated in four periods seeking to represent ten-minute intervals of distribution of the IFD curve. For this reason, the 17.9 mm precipitation that took a time of 10.74 min was joined with the second 10 mm precipitation, thus generating continuous precipitation of 16.74 min (16 min and 44 s). The second period corresponded to a 10 mm precipitation in a period of 6 min, the third to an 8 mm precipitation in 4.8 min (4 min and 48 s), and, finally, the fourth to a 3.8 mm precipitation in of 2.28 min (2 min and 16 s). Periods two, three, and four started every 10 min. Recorded data for intense rainfall conditions were taken from 2 June 1995 (see Appendix A—Figure A1b).

Saturation rainfall: This consisted of finding the number of precipitation millimeters required for simulating over the area of the roofs (average condition). In such a way, precipitation generated an effect close to the ground saturation. The way to find this was to drop the total capacity of the sprinkler system on the roof, taking a rain gauge as a measurement reference. Then, when the evacuation flow was similar in volume to the water entering to the system, the reference volume of the rain gauge was taken, giving a value of 10 mm over 6 min.
