**3. Results**

#### *3.1. Benchmark Analysis*

Table 3 contains values of the electricity performance metrics of all the categories in the campus, and they are cited in Table 3 from the highest to lowest intensity. Average EUIs were calculated by first calculating the average in the three years of each building. Then, we sum up the EUIs within the category, and we dived them by the number of facilities of each category. The section of others is excluded from the benchmark study because it includes three buildings (14,22,25) that represent, respectively, a warehouse, a parking garage, and a nursery. These buildings have a weak EC, do not have an impact on the campus EC, and do not fit into any of the major categories. The EUI averages represent benchmark values of the cited categories in the Mediterranean climate.

Figures 3 and 4 summarize as percentages the total EC by categories and the sum total GFA by categories. The research and science category have the biggest share by 47% of the EC, even though it accounts for only 27% of the gross floor area (GFA), followed by the teaching and seminary category that accounts 17% of the EC and 26% of the GFA, the library category that accounts 15% of the EC and 7% of the GFA, the administration office category that accounts 15% of the EC and 27% of the GFA, by the sport facilities category that account 5% of the EC and 6% of the GFA, and finally the restaurant category that accounts 1% and 2% of the GFA. These distribution shows that there is no direct relation relationship between EC and GFA because each category has its own characteristics.

**Figure 4.** Gross floor area (GFA) proportion of all the categories.

Figure 5 summarizes the energy consumption evolution by building category from 2016 to 2018. The evolution of the total sum campus EC had a minimum value in 2016, then had peaked in 2017, and had medium value in 2018. Figure 5 reveals that all the categories followed the overall trend, except restaurant and sport facilities, where both categories account combined 6% of the total EC. Research building EUI varies from 32.5–230.3 kWh/m2. Furthermore, its facilities include spaces like academic offices, computer rooms, and laboratories, and those spaces are characterized by a longer period of operation and a large number of computers, laboratory freezers, and other electric equipment. However, the majority of research buildings have a value superior to 80 kWh/m2; the highest intensity value—230.3 kWh/ m2—was scored by the solar energy center building (31). One of the reasons behind this high consumption is that a lot of research takes place in the center, and the researchers

and students working on solar chemistry and water detoxification use several compressors with high energy consumption that cannot be powered small solar field installed on the roof of the building. The technology of information and communication center building (30) has the second highest value 188.3 kWh/m2, and the lowest value was scored by the engineering school building (9).

**Figure 5.** Energy consumption evolution by building category.

Library building EUI varies from 62.4–105.3 kWh/m2. This category is the second most energy intensive. Their locals include spaces like reading rooms, computer rooms, and common spaces and

are characterized by a centralized air conditioning system, longer operating time, a high number of occupants (especially during the exam period), and a substantial number of computers and laptops.

Sport facilities are the third biggest consumer of energy by GFA, and their EUI varies from 28.02–69.9 kWh/m2. It contains spaces like swimming pool, a covered multitask hall, and gym rooms.

Restaurant buildings had the fourth highest energy intensity, and their EUI varies from 35.13–41.34 kWh/m2. They are characterized by longer operating time and di fferent electric equipment.

Teaching and seminary rooms had the fifth highest energy intensity, and their EUI varies from 13.59–59.2 kWh/m2. This category includes spaces like regular classrooms, computer rooms, and room theaters, and it is characterized by a high number of occupants. Four buildings out of eight have similar EUI values, which are close to average energy use index of this category. The lowest value is scored by building (4), a seminary building that has a low operating frequency, while the highest value is scored by building (3), which is an exception in this category because of its infrastructure that includes a water pumping system to ge<sup>t</sup> rid of the used water for the whole campus.

Administration o ffices are the least energy intensive category, and their EUI varies from 23.3–32.5 kWh/m2. They include o ffices, meeting rooms, and common spaces and are characterized by a low number of occupant and shorter operating time.

Figure 6 outlines the scatter plot of the average EUI during last three years. In the function of the average EC, we can observe that only five buildings (31,30,29,24,32) have an EUI superior to the EUI median of universities (MU) in the Mediterranean climate given by the 2003 CBECS data [9]. Twenty-five buildings from our portfolio have an energy intensity inferior than the M.U, and 19 out of those 25 buildings are three time less intense than the k–12 schools in hot and humid zones [9]. In addition, the scatter plot demonstrates that four buildings (30,29,13,26) have an EC greater than 7.0 × 10<sup>5</sup> kWh. These facilities belong to the research and library category. However, two of those big consumers have EUI values bellow the MU, and the category that has shown more harmony in their sample of buildings is the restaurant category, which include two properties that has almost identical values in their EC, GFA, and EUI.

**Figure 6.** Scatter plot of energy use intensity (EUI) as a function of energy consumption (EC) by buildings category.

Table 4 outlines the classification of energy performance of all the buildings within each category. This classification is a useful tool for managers because it reports insights into which of the building

should be prioritized in term of investments to achieve efficiency. Managers an divide the portfolio of every category into three groups (poor practice, usual practice, and best practice).

#### *3.2. Case Study*

Figure 7 summarizes the campus yearly energy consumption from 2011 to 2018. The overall energy consumption has been varying in the range of 7.93–8.8 GWh, and it has a known growth of 9% since 2011. However, the patterns do not represent a clear trend through the calendar years; this is one of the reasons we decided to proceed with the academic years since we have the data of the campus monthly EC (Table 1).

Campus energy consumption data on a monthly basis (Table 1) shows that the evolution over the years is generally the same, which means that it plunges and peaks in the same period of the year. The peaks usually happen during the months of May and June. In this period, campus buildings make substantial use of HVAC systems, and some buildings like the library start operating for a longer period because of the exam period. The down trend starts in October after the weather begins to be cooler, and in the beginning of July, it plunges to hit the lowest values in August—during this period, the campus is practically empty, and the majority of the university buildings are non-operational because of the summer vacation.

#### *3.3. Correlation Analysis and Regression Model*

Data presented in Table 5 sums up the inputs and the outputs used for the correlation analysis and regression model which corresponds to weather parameters (CDD and HDD), the number of occupants (the number of students, professors, and administration staff), and the EC of the campus.

Figure 8 outlines the scatter plots of EC sum total of the campus in function of the following variables: CDD and HDD, number of occupants, number of students, and number of staff. In the case of the correlation of outdoor temperature with EC, EC (Kwh) = 1204.8 CDD&HCC (◦C) + 7·10 6, its correlation coefficient of 0.72 indicates a positive correlation. According to the scatter plot of the number of occupants, its correlation coefficient of 0.62, which also indicates a positive correlation (EC (Kwh) = 1073.9.7 N—8·10 6). However, the remaining scatter plots represented a weak correlation, especially for the number of students. In addition, the number of occupants is the main factor behind the excessive consumption.

**Figure 8.** Scatter plots with their correspondent coefficient of correlation.

Figure 9 outlines the linear regression model of the total EC as a function of CDD and HDD. Both variables are statistically related because the correlation coefficient is 0.719. Figure 10 shows the energy thresholds of good practice of the University of Almeria.

**Figure 9.** Linear regression model of EC in function of CDD and HDD.

**Figure 10.** Scatter plot of the total EC in function of academic year
