Process Integration and Optimization for Sustainability (2022) 6:901–920

https://doi.org/10.1007/s41660-022-00273-z

ORIGINAL RESEARCH PAPER

Energy Auditing and Conservation for Educational Buildings: a Case

Study on Princess Sumaya University for Technology
Omar Mohamed1 · Saba Fakhoury1 · Georgina Aldalou1 · Ghaliah Almasri1

Received: 16 May 2022 / Revised: 26 July 2022 / Accepted: 28 July 2022 / Published online: 12 August 2022
© The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2022

Abstract
This paper introduces shortcut and effective approaches for energy auditing and conservation for educational buildings
with application to Princess Sumaya University for Technology. The building electricity is significantly covered by a pho-
tovoltaic system, which is included in the building assessments and solutions. The short and simplified computations of the
Normalized Performance Indicator of the building have shown fair performance, but some conservation opportunities are
still possible. Two feasible energy conservation approaches are presented; the first is by optimization of the operating times
of 357 air conditioners in the building using a Genetic Algorithm optimizer. Rigorous effort was done to get an accurate
relationship between the comfortable time spans and the average temperatures to ensure the users’ convenience with consid-
ering their willingness to exceed the pre-selected optimized time span. Nineteen scenarios over 6 months have attained an
average saving of 12.5%, which leads to a payback period of 9.27 years. The second approach is by ThermaCote® building
insulation, which attains 38% cost savings with a combined payback period of 19.16 years, including external replacement.
Although the existing photovoltaic system supplies a considerable free energy proportion and reduces the billing cost, the
proposed two methods have improved the normalized performance indicator from the “fair” range in the existing situation
to the “good” range.

Keywords Energy auditing · Normalized performance indicator · Energy conservation · Genetic algorithm · Optimization ·
ThermaCote®
Abbreviations HACs ize Operating hours of a group of ACs of a uni-
ACs Air conditioners fied size
CDD Cooling degree days Htotal Summation of total operating hours
GA Genetic algorithms αi, βi, and γi Parameters of the cost functions
HDD Heating degree days
NPI Normalized performance indicators
PSUT Princess Sumaya University for Technology Introduction
Symbols
Background and Motivation
Ci Cost function
Hi Operating hours
Energy auditing has increasingly become one of the leading
services provided by energy companies to the consumers
* Omar Mohamed in order to evaluate strategic efficiency improvements and
o.mohamed@psut.edu.jo
identify considerable energy savings (Yang and Yu 2015).
Saba Fakhoury Globally, this can be also very useful from many well-known
sabaafakhoury@gmail.com
important perspectives: (1) lower cost of consumption for
Georgina Aldalou the consumers or the owners of the buildings, (2) cleaner
Georgina.aldalou@gmail.com
environment due to reduction in emissions of conventional
Ghaliah Almasri generation, and (3) improved transmission capacities in the
ghaliahalmasri@gmail.com
grid due to lowered transmitted power to the consumers.
1
Princess Sumaya University for Technology (PSUT), One of the main challenges is to have accurate data sets
Amman 11941, Jordan through understanding of the facility; identifying problems

13
Vol.:(0123456789)
902 Process Integration and Optimization for Sustainability (2022) 6:901–920

of excess energy consumption, if any; and formulating a suit- may have different users or occupants, may have different
able and logical solution. Energy auditing can be compre- floor areas, or may have different weather conditions. The
hensive, targeted, or preliminary (Beggs 2002). Applying proposed method of NPI is a unified number range that indi-
direct comprehensive energy auditing can be too detailed, cates poor, fair, or good performance of similar types of
expensive, and time consuming (Kreith and Goswami 2008). buildings but having different conditions.
Preliminary energy audits are crucial in order to identify It is better to elaborate on the background of the topic
the essential problems regarding energy consumption in the of energy auditing and conservation and their cognitive
building energy and suggest practical solutions and this can domains to recognize their multidisciplinary nature and out-
be usually done without the need to undergo detailed energy comes. Figure 2 shows the multidisciplinary tendency of the
audits (Beggs 2002). From intensive readings in this field paper topic with emphasis on some useful outcomes of this
of study, the processes of energy auditing and conservation research and its positive impact on the economy and envi-
reports have become deeply understood, which can be sum- ronment. The research in energy auditing and conservation
marized as shown in Fig. 1. This figure shows the simplified may be initiated by a center in one engineering discipline
or preliminary strategy of auditing using normalized per- with interdisciplinary collaborations with other programs,
formance indicator (NPI). The NPI is a unified approach for but the research outcomes and usefulness go beyond the
building assessment according to its type (educational, com- boarders of any one of these disciplines alone.
mercial, industrial...etc.). It avoids using the direct numeric Students and researchers in the above educational pro-
value of energy consumption, as similar types of buildings grams should upgrade their knowledge in the overlapping

Fig. 1  The proposed research

methodology

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 903

Fig. 2  the multidisciplinary

tendency of energy auditing and
conservation principles

region between those disciplines in order to attain the dual transient building energy simulations. Foroughi et al. (2021)
advantage of inexpensive methodology and the correspond- proposed an optimization model using genetic algorithms
ing reasonably acceptable results for both energy auditing (GA), which is coupled with the EnergyPlus software pack-
and energy saving solutions. It is necessary to revise the age in order to obtain the optimum window-to-wall ratio
methodologies published for the aims of building energy (WWR), aspect ratio, and window locations in commercial
auditing and conservation, which are discussed in the next buildings in the USA. Wang et al. (2017) have proposed a
subsection. MILP (mixed integer linear programming)-based approach
to schedule a group of interruptible air-conditioning loads
Related Work in order to relieve the uncertainties of some stochastic vari-
ables, such as wind power and ambient temperatures, which
The variety of some recent published articles on differ- eventually improve the wind power utilization and reduce
ent buildings confirms the interdisciplinary nature of this the operational costs. Belany et al. (2021) have dealt with
research area in demanding different disciplines of engi- the possible lighting retrofit and with directed economic
neering and artificial intelligence (Veiga et al. 2021). The analysis for the sake of increasing the energy efficiency of
literature has shown variety of auditing methods and energy buildings by LED lighting systems. The accuracy of the ret-
conservation approaches with emphasis on novel concepts rofitted lighting systems has been validated. López-Ochoa
and building characteristics. Al-Saadi (2021) have diagnosed et al. (2019) have explored the possible improvement solu-
the building’s operation deficiency via multi-phase energy tions for some educational buildings in Spain with the aid of
audits and have suggested pragmatic retrofitting strategies. some reported Technical Building Codes published by the
Rodriguez et al. (2021) have rather focused on the uncer- Spanish Ministry of Development. Twelve proposals have
tainty in audit estimations; which has been handled via the been presented for realizing nearly zero-energy educational
Morris method of sensitivity analysis. Al-janabi et al. (2019) buildings. More sophisticated audits are doable by regres-
have presented novel information concerning the compu- sion and uncertainty analytics, which have shown accurate
tational and modeling capabilities of two—widely used— performance for modeling or prediction of building energy
building performance simulators. Osorio et al. (2022) have consumption (Mottahedia et al. 2015). Conventional regres-
presented the direction of research followed by two private sion modeling for building energy estimation has been inves-
higher education institutes in Columbia for carbon neutral- tigated via three models with data classifications for per-
ity. It has been concluded that higher education institutes formance improvements (Ridwana et al. 2020). A detailed
not only can contribute to carbon neutrality but also can review of uncertainty analysis in building energy assessment
ensure sufficient cultural awareness of sustainability for their has been published (Tian et al. 2018). The review concludes
graduates. Nadimuthu and Victor (2021) have improved the with some possible future directions for a more suitable and
energy efficiency of medicine manufacturing unit via a smart rigorous uncertainty analysis of building energy. Lucchi
energy management system. Ascione et al. (2017) have pre- (2018) has critically reviewed the visualization method that
sented retrofitting of energy in educational buildings with uses infrared thermography as applications of auditing the
application to Italian universities. The energy retrofit meas- building energy consumption, where the applications have
ures have taken into account all levers of energy efficiency presented standard and detailed energy audits through detec-
in buildings, such as building envelops and photovoltaic tion of excessive heat walls and testing building insulations.
generation. Multi-objective genetic algorithms have been Energy audit for a University building has been reported in
used as an optimizer to select optimal packages of energy Leiva et al. (2018), which has been done by investigating the
retrofit measures by minimizing thermal energy demand for influences of the individual loads on the total consumption
space heating and coupled with EnergyPlus software for without proposing energy conservation solutions. Lucchi

13
904 Process Integration and Optimization for Sustainability (2022) 6:901–920

and Delera (2020) have presented an educational experience on calculating the Normalized performance indicators
of social housing in the San Siro neighborhood in Milan, for (Beggs 2002). The novelty of the contribution is that
refurbishment and improvements of buildings’ sustainability, of the method has been simple enough to be applied
users’ comfort, and energy efficiencies. The wide range of by undergraduate engineering students to their Uni-
understandable knowledge of the local situation in terms of versity building. This is uncommon in the literature, in
the economy and society of that area stimulate the selection which the audits are mostly based on complex energy
simulations and expensive software packages that are
commonly used by utilities of energy auditing. The NPI
computational model of the building has been developed
. Castrillón-Mendoza et al. (2020) have described the imple- via MATLAB software and Excel formulas that use the
mentation of a photovoltaic (PV) system at the building of monthly electricity bills of 2019 and other relevant data
Autónoma of Occidente University by investigating the sys- such as the cooling degree days (CDD) and heating
tem outcomes through the building energy consumption post degree days (HDD) of the building. It is intended to use
PV installation. It is now clearly realized that the literature the bills of 2019 although other more recent bills are
has shown a wide-range, cumulative, and appreciated effort available because that year is the latest year where the
that lead to affirmative advancements in the field of build- building was full of students and their professors before
ing energy auditing and conservation by a combination of the COVID-19 crises that motivate distance teaching
simple, multidisciplinary, and complicated methods. How- and learning. Overall, the building NPI, with the inclu-
ever, educational buildings still lack detailed and systematic sion of the installed PV system, is found to be within
studies, the scientific contribution of this paper is described the fair range. Nevertheless, the building still has some
in the next subsection. There are common bottlenecks in remaining opportunities for energy savings. The build-
implementing promising energy auditing methods in edu- ing per se is still inherently inefficient because it has
cational buildings, such as lack of data, financial barriers, been just partially supplied by the external free renew-
and traditional cultural awareness. However, the proposed able source (The PV system) of energy, but without
method in our paper is as simple as it can be conducted by genuine improvements in the building energy per se,
engineering B.Sc. level of students to evaluate the building and this motivates the 2nd contribution.
performance and suggest potential energy savings, thereby 2. Two proposed techniques for energy conservation are
avoiding detailed or expensive energy audits that involve, for suggested based on the analysis of the aforementioned
instance, computational fluid dynamics (CFD) simulations NPI computational model and those energy conservation
or infrared thermography extensive accurate diagnosis and techniques have made significant improvements in the
assessments. building energy savings over the existing energy situ-
ation of the building. One method is based on using a
The Paper Contributions GA optimizer for optimizing the ACs’ operating hours
within the working day time window and grouping the
This paper presents a simplified and validated approach for ACs according to their sizes. By studying the financial
energy auditing that can be followed by engineers and new consequences of the practical implementation of the
graduates of engineering with a thorough understanding ACs’ optimizer in the building, it has been found that
of the concept. This proposed preliminary auditing, which the system is economically viable with around 9.27
is based on NPI, is effective to diagnose the main issues years payback period of the proposed practical con-
that cause excess energy consumption in the building and it trol circuitry. The second energy conservation, which
has not been considered in many papers. To the best of the saves around 38% of the consumed energy, is by build-
authors’ knowledge and comprehension, it is highly desir- ing envelop based on ThermaCote® insulation, with an
able to present a low-cost and direct methodology for energy estimated overall payback period of 19 years, including
auditing and conservation for educational buildings with the external replacement. The improvements in the energy
available data and resources, which eliminates the neces- consumption have been confirmed by the NPI compu-
sity of having expensive and complex simulation packages tation after application of those methods, the NPI has
to simulate the energy flows, or requires major changes in shown that both methods are capable to transfer the
the building. Therefore, the main novel contributions of this NPI from the “fair” range to the “good” range, which
article are: indicates genuine improvement in the building energy
performance.
1. A simple and cost-effective method for energy audit-
ing has been conducted on the building of Princess The rest of the paper is organized as follows: “Materi-
Sumaya University for Technology. The method is based als and Methods” presents the building description and the

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 905

required weather data the monthly energy consumption data. 2021 and no significant energy consumption in the PSUT
“The Building Performance Indication via NPI” presents the building. Because some of these buildings are physically
NPI calculations, “Proposed Techniques for Improvements connected, the total number of meters in the building is
in the Building Energy Consumption” illustrates the two only four, which supply the whole University campus with
proposed techniques for energy savings with their payback electrical energy. Two of the total meters are connected to
periods for practical installation, and finally, “Conclusions” the PV system as energy supplied to the grid was present in
states the conclusions and future recommendations. some of their bills. The university has been thus treated as
one facility as some individual meters cover more than one
building. The analysis requires the energy consumed from
Materials and Methods the grid as it reflects the loads; therefore, summation of this
energy from the four meters data was done for each month
The case under study is Princess Sumaya University for as shown in Table 1.
Technology (PSUT) located at Amman in Jordan, where Energy auditing requires studying the total consumption
preliminary energy auditing has been conducted, which is of the building. Consequently, the energy produced by the
dedicated to the whole building performance assessment. PV system is needed in order to know the actual electrical
The PSUT building facility is shown in Fig. 3, which has energy consumption of the building which reflects the actual
six main buildings: King Abdullah II School of Engineer- loads. The data set for PV energy contributions has been
ing, King Hussein School of Computer Science, King Talal provided and is shown in Table 2.
School of Business Technology, King Abdullah I School The NPI is calculated by the net energy consumed by
of Graduate Studies and Scientific Research, Deanship of the building from the grid and the total gross one, which
Admission and Registration, Presidency of the University, is consumed by the building that includes the PV. Thereby,
Deanship of Students’ Affairs, and Al-Hassan Library. Some the calculated NPIs in both cases altogether have avoided
of these buildings are attached to each other, which gives a likely misleading picture for building performance and
the possibility of feeding those connected buildings from diagnosis if only one of them is considered. As a result, the
one meter. field can be opened for many alternative methods for inher-
The electrical energy bills for the year 2019 have been ent improvements of the building towards lower net energy
provided by the Financial Department at the University. The consumption rather than just supplying partially with PV or
monthly bills for the fuel and electricity have been intention- other renewable resources, which has no ingrained effect on
ally adopted for building assessment because they clarify the building consumption. The total electrical energy con-
the most recent informative data before the COVID-19 cri- sumed can be calculated using the energy produced by the
ses, which led to online teaching and learning in 2020 and PV system and the electrical energy bills simply as follows:

Fig. 3  Princess Sumaya University for Technology (PSUT)

13
906 Process Integration and Optimization for Sustainability (2022) 6:901–920

Table 1  Energy consumed— Data provided—energy consumed from grid (kWh)

from grid
Meter number 20166001920 20164004411 20164004217 20166001926 Total/month

January 13452 5223 7626 7309 26301

February 6941 4777 6698 6552 18416
March 9402 11263 5581 5642 26246
April 3922 5861 5047 5206 9783
May 4125 5271 4812 1028 15236
June 336 5110 2186 2874 10506
July 205 3129 1824 5199 10357
August 4017 6529 6224 2555 15308
September 5563 7862 2895 3452 19772
October 104 4219 6243 4573 15139
November 3280 5893 5046 5796 20015
December 5288 5261 5066 8319 23934
Total/meter 52618 70398 54201 33796 211013

Table 2  PV system generation Date/time Generated Another approach to find the tariff is by using the billed
data energy energy consumption and the total cost to take all costs into
(kWh) account, from the bills available, as follows:
1/1/2019 0:00 53537.29 (
JOD
) Total cost (JOD)
2/1/2019 0:00 54007.45
Tariff = → factor
kWh Billed energy consumption (kWh)
3/1/2019 0:00 71397.69 (3)
4/1/2019 0:00 86611.81 The tariff using this equation has been calculated. Then, the
5/1/2019 0:00 102526.48 cost of the billed energy has been calculated using both tariffs;
6/1/2019 0:00 104017.23
the factor and table, to check which one is more consistent
7/1/2019 0:00 109760.07
with the cost on the bills since the cost may be affected by a
8/1/2019 0:00 101538.43
certain block more than the other. The most accurate tariff is
9/1/2019 0:00 88804.79
chosen for the study. The energy from all sources should be
10/1/2019 0:00 76059.23
considered; therefore, the fuel energy is essential for the study.
11/1/2019 0:00 63441.44
Five fuel bills were provided that show the fuel consumption
12/1/2019 0:00 45421.31
in liters and equivalent cost. This data is shown in Table 4.
Total 957123.26
According to reference Kolb and Siegemund (2017), 1 L of
diesel fuel contains 10 kWh of energy. Therefore, the energy
consumption of the fuel was calculated as follows:
1. Find the sum of the energy consumed from the grid
(kWh) from the provided bills.
2. Find the sum of the energy delivered to the grid (kWh) Table 3  Tariff structure
from the provided bills. Tariff table (regular subscribers’ tariff)

3. Energy consumed from the PV system Consumption (kWh) Tariff

(1) (JOD/
= Energy produced by the PV − energy delivered to the grid Lower bound Upper bound kWh)

1 160 0.04
4. Total energy consumed = Energy consumed from the PV
(2) 161 300 0.09
+ energy consumed from the grid
301 500 0.109
The steps described above have been done for all months. 501 600 0.145
It is specified on the bills that the tariff for the university is 601 750 0.169
“Governmental”, this corresponds to the Regular Subscrib- 751 1000 0.19
ers Tariff, which is an increasing block tariff as shown in 1001 – 0.256
Table 3.

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 907

Table 4  The fuel bills Diesel fuel bills provided

Bill number Quantity (liter) Cost (Fils/ Fuel cost (JOD) Building(s) Date
Liter)

1 15000 567 8505 Multiple 14/1/2019

2 12000 567 6804 Multiple 11/2/2019
3 11060 607 6713.42 Multiple 7/3/2019
4 4975 617 3069.575 IT 22/4/2019
5 11734 602 7063.868 Multiple 24/11/2019
Total 54769 – 32155.863 – –

(
kWh
) heating has been also calculated and corrected to achieve
Diesel energy (kWh) = Diesel (liter) ∗ 10 (4)
Litre more accurate results since it is estimated to be approxi-
mately 32.5% of the total energy consumed, that is, the
The building energy data has been summarized as two pie
diesel fuel energy over the total energy. The building effi-
charts in Fig. 4a and b.
ciency is assessed in the following subsection.
Unfortunately, the detailed equipment utilization is not
provided for technical reasons; however, it is not quite nec-
essary for NPI calculation and/or suggesting new energy-
NPI Calculations
saving techniques. The equipment utilization is requested
in case of detailed and comprehensive energy audits, which
The process of calculating the NPI is clarified through the
is intended in our study. In our study, it has intended to use
work-flow diagram shown in Fig. 5, where step 1 is for the
simplified auditing techniques to prove the ability of engi-
Corrected space cooling energy consumption, step 2 is for
neering students to conduct such audits without expensive or
Corrected space heating energy consumption, step 3 is for
complicated computations and provision of enormous data.
the Corrected total energy consumption, and finally step 4
is for calculating the NPI. The NPI method is detailed in
the text (Beggs 2002).
The Building Performance Indication via NPI

This is the analysis technique used to evaluate the energy

NPI Input Data
consumption of the university. Based on the book followed
in Beggs (2002), if the building is air-conditioned, then the
The data inputs needed for the NPI calculation are the total
energy consumed for space cooling should be calculated,
energy consumed, degree-day data, the total AC energy
corrected, and then used to find the NPI of the building.
consumed, the hours of use in the building, the exposure
However, in this study, the energy consumed for space

Fig. 4  Data pie chart summary.

a Grid vs. PV. b Electricity vs.
diesel fuel diesel fuel
33%
78%
From the PV system

From the Grid Electricity

(a) (b)

13
908 Process Integration and Optimization for Sustainability (2022) 6:901–920

Cooling energy Heating energy

Step 1 inputs Step 2 inputs

Annual AC 75% Diesel Annual AC

Standard Actual Standard Actual
consumed
for cooling
CDD CDD
Energy
(kWh)
+ consumed
HDD HDD
for cooling

= = = =
Space cooling Cooling Space heating
Heating
energy
consumption
x weather energy
consumption x weather
coefficient
coefficient

= =
Corrected
Corrected
space cooling
x Exposure
Coefficient space heating x Exposure
Coefficient
energy energy
consumption consumption
(kWh) (kWh)

Step 4
Standard Actual hours of
hours of use of / use of the Step 3 Corrected total
the building building Electrical energy
Energy (kWh) consumption
(kWh)
=
+ =
Corrected total
energy Hours of use
consumption x coefficient
Corrected
Space heating
(kWh) Diesel Energy
energy
(kWh)
consumption
= (kWh)

Normalized
- +
Annual Energy Total floor
Consumption / area (m^2) Space heating
Corrected
Space cooling
(kWh) energy
energy
consumption
consumption
(kWh)
= (kWh)

- +
NPI
(kWh/m^2)
Space cooling Non cooling/
energy heating energy
consumption = consumption
(kWh) (kWh)

Fig. 5  Workflow diagram for NPI calculations

coefficient, and the total floor area. These requirements are (BizEE website 2022) for the base temperature 18 °C. The
briefly explained in the following subsections. average is based on the past 5 years; 2016–2020 with an aver-
age uncertainty of 0.208% for both CDD and HDD. This aver-
The Total Energy Consumed in the Building age is used as the standard CDD and HDD of the building.

It is calculated using the electrical energy from the bills Actual CDD and HDD The daily CDD and HDD for the year
and PV system as already explained, and the energy from 2019 were also downloaded from the same widely used reference
the fuel. Both energy sources are expressed in the same (BizEE website 2022), for the same base temperature of 18 °C, and
unit, kWh. The energy of the fuel is multiplied by 0.75 have an average uncertainty of 0.0657%. Then, the CDD and HDD
since this is the proportion of the fuel used for space of the weekends, Fridays and Saturdays, and the holidays based on
heating (Beggs 2002). PSUT’s academic calendar for the year 2019 have been excluded
in calculating the actual CDD and HDD for the building.
Degree days This data was downloaded from the station Amman Air-
port, JO (35.99E, 31.97N) since it was the closest to PSUT’s
The cooling and heating degree days, CDD and HDD, location which is (35.88E, 32.02N). The base was chosen to
respectively, are needed to find the standard and actual CDD be 18 °C for two reasons:
and HDD to calculate the weather coefficients.
1. The work reported in (Almuhtady et al. 2019) explains
Standard CDD and HDD The average CDD and HDD have that this base is usually used for the past 30 years based
been downloaded from the widely used website reference on Jordan’s climate.

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 909

2. According to the American Society of Heating, Refriger- these 3 parameters, the consumption was calculated through
ating and Air-Conditioning Engineers ASHRAE 55, the Eqs. (17)–(18):
indoor comfort temperature is around 22 or 23 °C, which AC consumption (kWh) Relevant days Assumed hours
is consistent with the chosen base taking the indoor heat Month
= Total AC power (kW) ∗
Month
∗
Day
sources and Jordan’s climate into consideration, refer to (5)
the reference (ASHRAE. 2017). ∑ AC consumption
Annual AC consumption (kWh) =
Month
, of the relevant months (6)
To ensure logical results, the CDD data was also downloaded
Since we are dealing with an educational building, it can
for base temperatures 15.5 °C and 20 °C, which gave much
be reasonable to estimate ACs operating hours per working
higher and lower values compared with base 18 °C, respectively.
day. The assumptions are mentioned in Table 6.

Total air conditioner AC energy consumed

Hours of Use
Air Conditioner AC Data The sizes of the air conditioners
The standard hours of use per year are 4250 (Beggs 2002) to
present in PSUT along with the number of each size were
be able to compare the NPI with the yardsticks. The actual
provided as shown in Table 5. It shows that the total AC
hours of use per year were calculated by assuming 8 working
power for PSUT 578.5 kW.
hours per day, excluding the weekends; Fridays and Satur-
days, and the holidays based on PSUT’s academic calendar
AC energy consumed for cooling and heating The total AC
of 2019. Similarly, the days of use have been calculated.
energy consumed can be estimated by either assuming a
percentage of the total energy based on a study already
Total Area
done on a similar building with a similar climate, or by
calculating an approximate based on relevant data. The sec-
The area of each floor for all buildings of PSUT was pro-
ond approach was done as follows: Using the daily actual
vided. The net internal area for each floor for all buildings
degree days DD data, the number of days per month for
has been added to find the total floor area of the building,
which the DD is more than 1.5 was calculated in order to
which is needed for the NPI. The average area for each build-
exclude days with DD of 0.1. Then, the sum of CDD days
ing was found by finding the average of all floors for each
for each month was found to know how many days cooling
building. Then, the total ground area by finding the summa-
is needed each month. Similarly, the HDD days per month
tion of the average area for all buildings. Also, the total floor
were found. The number of operating hours of ACs is influ-
area is approximately found using the following relation:
enced by many factors, such as weather temperature, users’
willingness, and comfort. However, the total operational
(
Total f loor area m2 = Total ground area m2 ∗ Number of f loors
( )
hours can be reasonably assumed to be linked with either
(7)
the CDD or a forecasted dry-bulb temperature in order to
optimally decide the most economical operating hours for
each size of the ACs. NPI Results
The number of hours per day for which the AC is on for
both cooling and heating was assumed with the best reasoning The NPI has been calculated for different cases in order to
known. find the most logical and accurate result. Figures 6 and 7
The total power of all AC sizes has been calculated; assum- show the input and output as depicted from the developed
ing 1 kW/1 ton is drawn as a rule of thumb. Therefore, using NPI excel calculator sheet, respectively.

Table 5  AC data provided. Rule of thumb: 1 ton consumes 1 kW

Number Size (ton) Power (kW) Table 6  AC hours/day assumptions

The PSUT building 96 1 96 Season Hours/day—total Hours/day—energy

ACs 90 1.5 135 energy consumed (PV consumed from grid
+ electricity bills) only (electricity bills)
160 2 320
11 2.5 27.5 Cooling [May–Oct] 6 2
Total 357 – 578.5 Heating [Nov–April] 2 2

13
910 Process Integration and Optimization for Sustainability (2022) 6:901–920

Input
Electrical Diesel AC hours- AC hours - AC Energy AC energy Standard Actual Standard Actual Exposure Standard Actual Area
(kWh) (kWh) Winter Summer (kWh)- (kWh)- CDD CDD HDD HDD Coeff. Hours hours (m^2)
Remarks - case summer winter
studied
Bills Bills Assumption Assumption Total AC Total AC Avg of 5 Daily DD - Avg of 5 Daily DD - Location Standard Without Actual
power * power * years Holidays years Holidays Standard table holidays data
hours hours

Only energy
from grid is 244786.739 547690 2 2 70577 122642 1437 732.6 1128 827.6 1 4250 1760 14958
considered
Total energy is
1134149.01 547690 2 6 319332 124956 1437 732.6 1128 827.6 1 4250 1760 14958
considered

Fig. 6  Inputs of NPI calculator

Proposed Techniques for Improvements in the building. The time windows of operation for
in the Building Energy Consumption each size-grouped air conditioners are decided in
advance with free desire of the users to switch on/
In this section, the energy conservation methods for off. The assumptions have included many operating
building consumption improvements are introduced. optimal scenarios produced by GA that have been
Although the building energy is largely covered by compared with other hypothetical scenar ios with
the PV system, especially in the summer season, it average cost savings of around 12.5 % of the total
has been found that the building is inherently still not cost per year with 9.27 years payback period. The
efficient and some energy conservation strategies must second method is by proposing a building envelop of
be applied to reduce its gross energy consumption. ThermaCote® that attains further 38% cost savings
The PV system acts like an external free source and per year with a combined payback period of 19.16
has no ingrained effect on the building consumption. years. It is noticed that the latter can be relatively
To avo i d a ny p o s s i b l e m i s l e a d i n g p i c t u re s o n longer to recover; however, this payback period is
the building efficiency, two energy conser vation combined after the applications of the PV system
principles have been proposed, one is based on GA to and the GA air-conditioning optimizer, and even
optimally schedule the operating time span for all air so, it can be comparable to other buildings with
conditioners according to the forecasted temperature other energy investment projects for other buildings
or the predicted CDD from histor ical data. The (Zhang et al. 2021). Overall, this is certainly useful
reason for selecting air conditioners’ circuit rather environmentally and economically with reasonably
than lighting or computer can be justified through acceptable payback periods for educational buildings
basic inspection of the rating of the devices in the due to their sustainability for several decades to
building, which obviously has shown that the air come. The next subsections describe the proposed
conditioners are the highest energy-consuming devices energy conservation with their investment analysis.

Output
Space Space CDD Weather HDD Weather Corrected Corrected Non- Corrected total Hours of use Normalized NPI
cooling heating Coeff Coeff space cooling space heating cooling/heating energy coeff annual energy (kWh/m^2)
energy energy energy energy energy consumption consumption
consumption consumption consumption consumption consumption (kWh) (kWh)
(kWh) (kWh) (kWh) (kWh) (kWh)
Remarks - case
studied
AC Diesel Standard Standard Space cooling Space heating Sum of TOTAL (Corrected total Standard Corrected Normalized
Consumption Consumption CDD/Actual HDD/Actual energy * CDD energy * HDD energy - non- heating & hours/Actual total energy * annual
* 0.75+ AC CDD HDD weather coeff weather coeff corrected cooling) + (non- hours hours of use energy/Area
for heating * Exposure * Exposure heated/cooling cooling & non- coeff.
energy heating)

Only energy
from grid is 70577 533409.5 1.961507 1.3629773 138437.28 727025.03 188490.239 1053952.547 2.414773 2545055.9 170.1452
considered
Total energy is
319332 535723.5 1.961507 1.3629773 626371.94 730178.96 826783.508 2183334.41 2.414773 5272256.4 352.4673
considered

Fig. 7  Outputs of NPI calculator

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 911

Fig. 8  GA flow diagram

Generate Initial Evaluate fitness Optimization Best individuals
Population function Criterion met?

Selection

Generate
New Crossover
Population

Mutation

The Genetic Algorithm Optimizer Htotal = HACs ize1 + HAC size1.5 + HAC size2 + HAC size2.5 (11)

Genetic Algorithm is a well-known meta-heuristic optimiza- where H i is the scheduled operating hours for a group
tion technique that represents a class of artificial intelligence. of ACs that have the same size; these scheduled hours
Because it was inspired by the Darwinian theorem, it follows are to be optimally found by GA to minimize the total
the mechanism of natural selection, but in a mathematical and cost C i of electricity. The coefficients α i , β i , and γ i
iterative fashion. The surviving parameters produce the best are simply determined by a curve fitting tool either in
fit which evolves towards the global optimal solution by three MATLAB or Microsoft excel. The curves are built by
main actions: reproduction, crossover, and mutation, which are the relationship of the CDD or average temperature with
applied in each generation of parameters. Every action has spe- the operating hours of a group of ACs. H total has been
cific settings in order to obtain—or accelerate—the solution. It considered equality constraint, which is the summation
has been shown that GA is a rigorous optimization technique of total operating hours for the groups of ACs pre-
that is capable of resolving many highly complex and practical determined either by the forecasted average temperature
optimization problems, such as parameter identification of coal or the CDD on a working day. It can be alternatively
mills (Guo et al. 2014) and modeling energy-efficient super- represented by the operating hours of the ACs in the
critical power plants (Mohamed et al. 2011). The simplified offices, the classrooms, and the stadiums. The inequality
schematic for GA is shown in Fig. 8. The idea is mathemati- constraints are adjusted to fit the working day hours to
cally based on the fact that there are infinitely set of operat- allow maximum flexibility for the AC user. After using
ing hours of the AC groups that could satisfy the consumers, determining the coefficients of the cost function of each
according to the daily expected average temperature, with dif- group, they have been recognized by the GA optimizer
ferent costs of consumption; nevertheless, there is an optimal as an M-File representing the objective function. Many
choice among all these sets that make the cost minimum which scenarios were tested; nonetheless, the most satisfactory
corresponds to minimum energy consumption. results are obtained with population-type double vector,
The objective function to be minimized for a unified size population size of 50, heuristic crossover, Gaussian
group of ACs is assumed to be a quadratic cost function and mutation function, and mutation rate of 0.1.
the optimization problem can be mathematically described
through Eqs. (7)–(10) Sizes, Quantities, and Rooms of Air conditioners
∑1, 1.5, 2, and 2.5
4 groups wtih sizesCi = n=1
ai + 𝛽i + Hi + 𝛾i + H 2i In the data collection phase of the audit, it has been given
(8) that there are 4 AC sizes, distributed in 4 types of rooms as
Subject to the following equality and inequality in Table 7.
constraints The power of each size is calculated using Eq. (11),
assuming that (1 ton = 1kW):
Hmin ≤ Hi ≤ Hmax (9)
Power [kW] = size [ton].quantity (12)
∑
Hi = Htotal (10)

13
912 Process Integration and Optimization for Sustainability (2022) 6:901–920

Table 7  Air conditioner data for PSUT operation of this optimal strategy in comparison to other
Room type Size [ton] Quantity Power [kW]
hypothetical cases on the same number of days where the
GA optimizer and the proposed smart system do not work.
Offices 1 96 96 A curve fitting tool has been used to develop the cost
Classroom 1.5 90 135 equations for each AC size so we have four quadratic cost
Bigger rooms, 2 160 320 equations representing the time-span of the operating
auditoriums, and
hours of each grouped size. The daily data of the cool-
corridors 1
ing degree days of Amman between May and October is
Bigger rooms, 2.5 11 27.5
auditoriums, and obtained. Considering 18 °C as the base temperature; the
corridors 2 maximum and minimum CDD are assigned to the maxi-
mum and minimum number of operating hours assumed
for PSUT which are equal to 24 and 6 hr./day, respectively.
The Bounds of Operating Hours and Practical Correlations The 0 h/day is assigned to the base temperature, i.e., where
no cooling is needed. It is logical to relate the daily oper-
In order to optimize the system, the number of operating ating hours to the forecasted daily average temperature
hours for each size of the 4 AC sizes must be bounded. The using historical data, which was calculated using the CDD
bounding criterion is decided according to the daily aver- as follows:
age temperature in summer, in which the maximum daily
average temperature corresponds to the full working-day
Average temperature = Base temperature + CDD (13)
hours and the minimum average temperature corresponds to Note that the minimum CDD is 1.5 DD even if smaller
the minimum hours, which are selected to be two hours for CDD are available because this is the minimum CDD value
the offices and classrooms. The number of operating hours to consider cooling on that day. The next subsection shows
boundaries for each room type have been assumed logically the practical automation system, cost savings associated with
depending on the occupancy trends for each room type in GA, and the payback period.
PSUT, two cases were studied as follows:

1) The first case was that all air conditioner sizes had the The Proposed Practical GA Optimizer
same lower and upper bounds which are 2–7 h/day.
2) The second case was to keep the first 3 sizes with the The Genetic Algorithm method is a proposed energy con-
same 2–7 h/day boundary and reduce the boundary of servation system that needs a reasonably acceptable budget
the 4th size to 0–3 h/day. to implement. The practical implementation of the Genetic
Algorithm method of optimizing the number of daily oper-
The lower and upper bounds are assumed according to ating hours for each size of the air-conditioning system is
the most guessed users’ activity during the daily work- done using a smart system which is responsible for all the
ing hours of the building. The credit hours of the larg- air conditioners in PSUT. This computerized smart system
est air conditioners are actually low even on the hottest should be connected to all the air conditioner switches in the
days because they are installed in auditoriums and cor- facility. Figure 9 depicts the proposed practical automation
ridors where the occupancy rate is very low, especially system for more economical operation of the ACs. In order
the auditoriums where the activities are only the invited to do this, the following components are needed:
talks and non-scheduled lectures that may happen once
every week throughout the academic year. In addition, 1. Centralized smart computer (1 item).
the two assumed cases have been extensively tested for 2. Thermostats—one for each air conditioner (357 items).
many simulation scenarios for the proposed automatic 3. Timer switches—one for each air conditioner (357
energy conservation via GA and the results were more items).
consistent when applying the constraints in case 2. It 4. D/A converters or Arduino—one for each air conditioner
also is important to mention that, the automation of the group with the same size (4 items).
proposed design of the practical control system, which
is described in the next subsection, allows flexibility of To implement this system, specialists in the field of
the user to exceed the time span decided by the GA opti- smart systems are needed. Also, a long time and a budget
mizer if needed or even switch off the AC, but the GA are needed in order to purchase these components. As
optimizer controls the ACs operation hours instructively mentioned earlier in the abstract, in some cases, the user’s
and automatically. Therefore, the savings can be economi- remote control may bypass this automation circuitry to
cally viable as average cost savings after many days of switch ON/OFF the AC in his/her room; however, such

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 913

Fig. 9  Practical foresight for the

proposed optimizer or control
strategy Mathworks
GA optimizer

D/A converters or
Arduino devices
(quantity :4)

Logic Relay
(quantity :4)

Time Delay Switch

or Timer Switch
(quantity : 357 )

T T T T Thermostat Switches
(quantity (for total Acs ): 357 )

?
2.5 Ton ACs (11 ) 2 Ton ACs (160 ) 1.5 Ton ACs (90 ) 1 Ton ACs (96 )

Total ACs : 357 working economically

cases are not dominant, and therefore, they have trivial Table 9  Savings percentage in comparison to other hypothetical cases
effects on the overall intelligent decision for this large
Assumption Total operating hours Savings
number of ACs because of the following reasons: (hours/day) percentage
(%)
1) The operating hours are decided overall by the GA opti-
1 (maximum CDD) 24 0
mizer for the large number of ACs within boundaries of
2 23 0.003
practical time spans that are rooted in the average tem-
3 22 8.14
perature and the CDD; therefore, any slight mismatches
4 21 5.85
result in near-optimal decisions, which is also economi-
5 20 9.92
cal. If the mismatch is due to switching off, then that can
6 19 18.08
be more optimal, if it is due to exceeding some few time
7 18 18.98
spans, then it is a near-optimal decision.
8 17 21.16
9 16 5.255
10 15 4.16
Table 8  The control system components, quantity for each compo- 11 14 14.97
nent, and price 12 13 13.94
Component Quantity Price [JOD/unit] 13 12 12.74
14 11 26.94
Centralized smart computer 1 2800
15 10 23.81
Thermostats 357 15x357
16 9 28.34
Timer switches 357 20 x357
17 8 13.20
D/A converters|Arduino 4 44 x4
18 7 11.66
Installation – 1000
19 (minimum CDD) 6 0
Total cost (investment) 16471
Average 12.5 %

13
914 Process Integration and Optimization for Sustainability (2022) 6:901–920

Table 10  Economic analysis summary of assumption 9

Hypothetical operating hours Operating hours decided by GA tool Economic Analysis
Size (ton) Number of operat- Energy Cost (JOD/day) Optimal number Optimal Minimum Saved energy (kWh/ Saved cost
ing hours (hours/ (kWh/ of operating hours energy cost (JOD/ day) (JOD/day)
day) day) (hours/day) (kWh) day)

1 7 672 164.68 7 672 164.68 0 0

1.5 5 675 165.42 3.99 539.86 132.30 135.13 33.11
2 2 640 156.84 2 640.32 156.92 -0.32 -0.07
2.5 2 55 13.47 3 82.5 20.21 -27.5 -6.73
Total 16 2042 500.43 16 1934.68 474.13 107.31 26.30
Saving percentage %
5.25%

2) The proposed controller has been designed and tested The Controller Economic Analysis
in the worst cases, in which all ACs are turned ON and
compared against many hypothetical cases with consid- An economic analysis is done to ensure the feasibility of
erable average cost savings. Therefore, the success of the proposed control system. Table 8 summarizes the com-
this strategy is clarified by the fact that the cost savings ponents needed, quantity of each component, and estimated
are calculated over many days of operation, not neces- prices. The prices are estimated by surfing many national
sarily per-day cost saving. and international websites for marketing embedded systems,
but the prices are mentioned in JOD. This strategy of ACs
In the suggested design, 357 timer switches are used, control is designed for cooling purposes. Thus, the assump-
one for each AC, or an advanced multiple-device program- tion for the air-conditioning usage period is from May to
mable timer switches may be used, with less number of October. Many hypothetical assumptions have been assumed
required items. The time span for each grouped size ACs (Table 9).
is calculated as follows These hypothetical cases are calculated to take advan-
tage of the average cost savings, in order to compare with
Time span for each size = No.of ACs ∗ optimal daily hours from the GA (14)
the optimal near-optimal decisions of the GA controller
It is also expected that the practical installation should and it is not recommended to mention the details of all
reveal some needs for calibrations either on the data inputs of them here; however, one sample achieved saving in
or the scheduling limits; however, the overall investment, someday is mentioned in Table 10.
cost savings, apparatus used, and payback period is compa- The GA method resulted in 12.5% average savings if
rable with the calculated one in the next subsection. used in PSUT which are obtained theoretically as shown
in Tables 9 and 8; 19 assumptions of different total oper-
ating hours have been studied to calculate the most real-
istic average savings of the GA tool which appears to be
12.5%. The two extreme cases which are the minimum
Table 11  Total cost and saved cost
and maximum operating hours which result in 0 savings
Total summer energy cost [JOD] were taken into consideration when calculating the aver-
Cost before GA “fom Saved cost age savings. The total energy costs before implementing
bills” after GA (12.5 the GA and after the GA with the corresponding savings
%) in JOD are mentioned in Table 11.
May 2698.40 337.30
Thus, the payback period is calculated using the invest-
June 1761.43 220.17
ment cost and saved cost as follows:
July 1137.30 142.16 Investment cost
August 3237.05 404.63
Payback period = = 9.27 years (15)
Saved cost
September 2737.65 342.20
October 2631.87 328.98 To conclude the analysis, the results above ensure
Total 14203.72 1775.46 that implementing this control system is feasible and

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 915

Input - After GA

Electrical Diesel AC hours- AC hours - AC Energy AC energy Standard Actual Standard Actual Exposure Standard Actual Area
(kWh) (kWh) Winter Summer (kWh)- (kWh)- CDD CDD HDD HDD Coeff. Hours hours (m^2)
Remarks - case summer winter
studied
Bills Bills Assumption Assumption Total AC Total AC Avg of 5 Daily DD - Avg of 5 Daily DD - Location Standard Without Actual
power * power * years Holidays years Holidays Standard table holidays data
hours hours

Only energy
from grid is 235964.614 547690 2 2 61754.9 122642 1437 732.6 1128 827.6 1 4250 1760 14958
considered
Total energy is
1094232.51 547690 2 6 279416 124956 1437 732.6 1128 827.6 1 4250 1760 14958
considered

Fig. 10  NPI inputs—post GA

beneficial for PSUT as it returns its initial cost approxi- The Building Insulation Method
mately 9.3 years after the installation, which is acceptable
for an educational building. This is the worst payback Proper building insulation reduces both the energy and
period due to the largest number of components used; NPI by 38% according to the product data sheet and pres-
it may range from 5 to 9 years if other approaches are entations of ThermaCote® (ThermaCote® website 2022).
chosen. Thermacote® Insulation of buildings is a high-performance
paint ceramic heat-resistant material that leads to signifi-
Improvement Reflected on the NPI cant energy savings for the building (ThermaCote® web-
site 2022). Hence the actual NPI reaches the “good” range
The NPI has been improved since the GA results in 12.5% for either case of total energy consumed by the building or
decrease in energy used for cooling. It was calculated as excluding the energy shared by the PV system, as shown in
already explained, but with the following changes since the Table 12 and Fig. 12. It has saved 301141.16 kWh for the
reduction is only in the cooling energy: assumed case.
AC energy used for cooling ∣ post GA
(16) Economic Analysis Results of the Proposed Energy
= AC energy used for cooling ∗ 0.875 Conservation Methods

The total electrical energy ∣ post GA = The total electrical energy

The results of the economic analysis are shown in Table 13.
(17)
− (0.125 ∗ AC energy used for cooling) It is stated that the ThermaCote® product is guaranteed for
life for indoor applications, and 10 years for outdoor use (Ther-
The NPI is reduced to 3.59% and is equal to 339.82
maCote® website 2022). If the total energy is considered, case
kWh/m2, which is still in the “fair” range. This is shown in
1, the payback is 2.89 years which implies the recommendation
Figs. 10 and 11.

Output - After GA
Space Space CDD Weather HDD Weather Corrected Corrected Non- Corrected total Hours of use Normalized NPI
cooling heating Coeff Coeff space cooling space heating cooling/heating energy coeff annual energy (kWh/m^2)
energy energy energy energy energy consumption consumption
consumptionconsumption consumption consumption consumption (kWh) (kWh)
Remarks - case (kWh) (kWh) (kWh) (kWh) (kWh)
studied
AC Diesel Standard Standard Space cooling Space heating Sum of TOTAL (Corrected total Standard Corrected Normalized
ConsumptionConsumptionCDD/Actual HDD/Actual energy * CDD energy * HDD energy - non- heating & hours/Actual total energy * annual
* 0.75+ AC CDD HDD weather coeff weather coeff corrected cooling) + (non- hours hours of use energy/Area
for heating * Exposure * Exposure heated/cooling cooling & non- coeff.
energy heating)

Only energy
from grid is 61754.875 533409.5 1.961507 1.3629773 121132.62 727025.03 188490.239 1036647.888 2.414773 2503269 167.3516
considered
Total energy is
279415.5 535723.5 1.961507 1.3629773 548075.45 730178.96 826783.508 2105037.917 2.414773 5083188.2 339.8274
considered

Fig. 11  NPI outputs—post GA

13
916 Process Integration and Optimization for Sustainability (2022) 6:901–920

Table 12  Insulation energy reductions

Energy and NPI analysis
Energy con- Electrical Fuel energy Total energy Post-insulation Saved energy NPI - pre insu- New NPI—post
sumption case energy (kWh) (kWh) (kWh) total energy (kWh) lation (kWh/ insulation (kWh/
(kWh) m 2) m2)

1 Total 1134149.008 547690 1681839 1042740.18 639098.82 352.46 218.52

energy consid-
ered
2 Only energy 244786.73 547690 792476.73 491335.57 301141.16 170.14 105.48
from grid
considered

is feasible. However, this is not the actual cost since the energy 2. Using the following investment cost:
covered by the PV system is for free now, so this only means that Old Thermacote cost
if the insulation was improved before installing the PV system it New ThermaCote cost = Old ThermaCote cost +
2 (18)
would have been feasible.
That is, the investment cost is 50% higher when assum-
Two scenarios were studied where only the energy from
ing the external insulation is replaced which accounts for
the grid is considered:
the worst case of maintenance. This case has shown a
lifetime of 20 years, so a payback of 19.33 years is fea-
1. Using the same investment cost and this gives a pay-
sible and saves 26451.18 JOD annually. The cost-saving
back period of 12.89 years. It is greater than 10 years;
opportunities due to the load shifting concept may exist
however, the insulation is internal and external, so it can
if the tariff depends on the time, peak, night, or day, or if
be considered feasible. However, to ensure that, mainte-
a penalty on the maximum demand exists. However, this
nance of the external coating should be considered for a
is not applicable for this study since the tariff structure
longer lifetime.

Fig. 12  Improvements before

and after insulation Pre and Post-Insulation energy consumption
2000000
1500000
1000000
500000
0
Electrical Energy Fuel Energy Total Energy Post-insulaon Saved Energy
[kWh] [kWh] [kWh] total Energy [kWh]
[kWh]

1 - Total energy considered 2- Only energy from grid considered

Pre and Post Insulation Normalized Performance

Indicators
400
300
200
100
0
NPI - pre insulaon [kWh/m2] New NPI - post insulaon [kWh/m2]

Total Energy considered Only the Energy from Grid is considered

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 917

Table 13  Insulation economic analysis

Economic analysis
Energy case Electricity cost (JOD) fuel cost (JOD) Total cost (JOD) ThermaCote® Annual saved Payback
cost (JOD) cost (JOD)

1 Total energy 277949.68 32155.9 310105.54 341020 117840.10 2.89

2 Only energy from grid 37452.50 32155.9 69608.37 341020 26451.18 12.89
511530 26451.18 19.33

Table 14  Improvement The total energy consumed is considered—improvements

summary over one year
of typical actual energy Data Initial Post GA control Post insulation Post GA
consumption of 2019 control and
insulation

NPI (kWh/m2) 352.46 339.82 218.52 210.69

Percentage reduction in NPI (%) 0 3.58 38 40.22
Final total energy (kWh) 1681839 1641922.50 1042740.18 1017991.95
Percentage reduction in energy (%) 0 2.37 38 39.47
Final total energy cost (JOD) 69608.37 67832.90 43157.19 42056.40
Percentage reduction in cost (%) 0 2.55 38 39.58

is the Regular Subscribers’ tariff which is an increas- Combined Investment Cost (JOD) = GA cost + Insulation cost
ing block tariff. Consequently, there are no cost-saving Combined Investment Cost (JOD) = 16471 + 511530 = 528001
opportunities depending on this concept.
Investment 528001
Payback = Annual saved cost
= 27551.96997 = 19.16 years
(20)
Summary of Combined Improvements The NPI calculation of both improvements is depicted
in Figs. 13 and 14 from the NPI excel calculator.
The cost reductions for the GA control and insu- Finally, the calculated payback period for the building insu-
lation improvements are summarized in Table 14, lation could be too long for residential customers. However, It
along with the reduction caused by installing both can be accepted by most educational buildings and there are
recommendations. similar payback periods in the literature for other solutions and
The payback period if both improvements are done other types of buildings. One may argue that this is too long
together can also be calculated using Table 14 as follows: to be accepted by most residential customers, but in fact, it is
Combined Annual Saved Cost (JOD) = Initial cost − Post GA&insulation cost very suitable and economically viable for educational buildings
Combined Annual Saved Cost (JOD) = 69608.37 − 42056.40 = 27551.96 where the building is very likely to be with the same owner(s)
(19)

Output - ThermaCote Insulation

Space Space CDD Weather HDD Weather Corrected Corrected Non- Corrected total Hours of use Normalized NPI
cooling heating Coeff Coeff space cooling space heating cooling/heating energy coeff annual energy (kWh/m^2)
energy energy energy energy energy consumption consumption
consumptionconsumption consumption consumption consumption (kWh) (kWh)
Remarks - case (kWh) (kWh) (kWh) (kWh) (kWh)
studied
AC Diesel Standard Standard Space cooling Space heating Sum of TOTAL (Corrected total Standard Corrected Normalized
ConsumptionConsumptionCDD/Actual HDD/Actual energy * CDD energy * HDD energy - non- heating & hours/Actual total energy * annual
* 0.75+ AC CDD HDD weather coeff weather coeff corrected cooling) + (non- hours hours of use energy/Area
for heating * Exposure * Exposure heated/cooling cooling & non- coeff.
energy heating)
Only energy
from grid is 43757.74 330713.89 1.961507 1.3629773 85831.112 450755.52 116863.9482 653450.5793 2.414773 1577934.6 105.49
considered
Total energy is
197985.84 332148.57 1.961507 1.3629773 388350.6 452710.96 512605.775 1353667.334 2.414773 3268799 218.5297
considered

Fig. 13  The inputs to the combined improved NPI computing

13
918 Process Integration and Optimization for Sustainability (2022) 6:901–920

Output - Combined Improvements

Space Space CDD Weather HDD Weather Corrected Corrected Non- Corrected total Hours of use Normalized NPI
cooling heating Coeff Coeff space cooling space heating cooling/heating energy coeff annual energy (kWh/m^2)
energy energy energy energy energy consumption consumption
consumption consumption consumption consumption consumption (kWh) (kWh)
Remarks - case (kWh) (kWh) (kWh) (kWh) (kWh)
studied
AC Diesel Standard Standard Space cooling Space heating Sum of TOTAL (Corrected total Standard Corrected Normalized
Consumption Consumption CDD/Actual HDD/Actual energy * CDD energy * HDD energy - non- heating & hours/Actual total energy * annual
* 0.75+ AC CDD HDD weather coeff weather coeff corrected cooling) + (non- hours hours of use energy/Area
for heating * Exposure * Exposure heated/cooling cooling & non- coeff.
energy heating)
From grid only -
GA then 38288.023 330713.89 1.961507 1.3629773 75102.223 450755.52 116863.9482 642721.6903 2.414773 1552026.8 103.758
insulation
Total energy -
GA then 173237.61 332148.57 1.961507 1.3629773 339806.78 452710.96 512605.775 1305123.509 2.414773 3151576.7 210.693
insulation
From grid only -
insulation then 38288.023 330713.89 1.961507 1.3629773 75102.223 450755.52 116863.9482 642721.6903 2.414773 1552026.8 103.758
GA
Total energy -
insulation then 173237.61 332148.57 1.961507 1.3629773 339806.78 452710.96 512605.775 1305123.509 2.414773 3151576.7 210.693
GA

Fig. 14  The combined Improvements as reflected in NPI

for many coming decades. The return cycle could be much lower employees, before the COVID-19 crisis that stimulates
if the PV contribution is not considered; however, it is intended online classes with no significant energy consumption.
to include the PV contribution as free energy and consider the • Although the NPI has shown fair-range performance, two
cost of electricity from the grid, which results in this payback proposed energy conservation methods are proposed,
period. As might be a subject of interest to some researchers, one on optimizing the ACs’ operation times by GA, and
the reduction in carbon emission from the contribution of the the other by building insulation with high-performance
solar PV system may be mentioned. From the website (Boston ceramic material manufactured by Thermacote.
Solar Website 2022) as a typical generalized example of the • The relevant payback periods are calculated and dis-
basic calculation, 1 kWh electricity from solar is equivalent to cussed for guaranteed and economically viable perfor-
0.846 lbs reduction in C
­ O2 emission or 0.383 kg. By referring to mance.
Table 2, 957123.26 kWh is the total annual PV generation from • These approaches have successfully reduced the cost and
the data of 2019, the year just before the corona crises, which is energy by 12.5 % in the summer and the NPI by 3.59 % in
equivalent to a carbon reduction of 366578.2 kg. comparison to hypothetical and actual cases. In addition,
building insulation developed by a recognized company
has been suggested, which has resulted in 38% savings
Conclusions and energy, NPI, and total cost. Both methods cause
approximately 40% savings.
In this paper, a simplified energy audit has been conducted • The proposed methods have driven the NPI from the
on the PSUT building, which has targeted the whole build- “fair” range to the “good” range according to interna-
ing consumption with suggested improvements to the AC tional standards.
system.
The following points present the concluding remarks The proposed methods have been evaluated based on
and findings from the data and results: the real data of the Jordan climate and the accurate energy
consumption of the building, which considers the data of
• The energy consumed by the PSUT building has been gas and power consumption in addition to the PV power
partially covered by a PV system since 2016. share. This method is much simpler than other detailed
• Energy audits of the building have been conducted via energy audits and solutions and comparable in accuracies
the NPI method, with monthly energy bills of 2019, and outcomes. Other characteristics are best assumed from
including the share of the PV system, with the result our experience of the building and search skills, such as
of fair building performance. the constraints imposed on the working hours of the ACs
• The reason for choosing the 2019 bills is that it is the and the estimated prices for the appliances of the con-
last year where it was full of students, professors, and trol strategy of ACs. However, every study needs some

13
Process Integration and Optimization for Sustainability (2022) 6:901–920 919

potential data besides some sort of logical assumptions so and its susceptibility to the ambient air temperature towards sus-
that the study can be realized. tainable electricity generation. Energy, Sustain Soc 9(1):39
Al-Saadi SN (2021) Pragmatic retrofitting strategies for improving
Future recommendations can be stated as follows: The thermal, energy, and economic performance of an institutional
Genetic Algorithm method can be extended for heating building in a cooling-dominated climate. J Build Eng 44:103326
purposes during winter alongside the current usage of American Society of Heating, Refrigerating, and Air-Conditioning
cooling purposes controlling the operating hours using dif- Engineers [ASHRAE] (2017) ANSI/ASHRAE Standard 55-2017:
Thermal environmental conditions for human occupancy. Ameri-
ferent codes of GA to account for HDD instead of CDD. can Society of Heating, Refrigerating, and Air-Conditioning Engi-
This eventually results in further billing cuts. The paper neers, Atlanta
has included the full data and information so that it can Ascione F, Bianco N, De Masi RF, Mauro GM, Vanoli GP (2017)
be additionally regarded as an educational guide to other Energy retrofit of educational buildings: transient energy simula-
tions, model calibration and multi-objective optimization towards
future researchers worldwide to conduct such simplified nearly zero energy performance. Energy Build 144:303–319
audits, subject to data availability in the building facil- Beggs C (2002) Energy management and conservation. Elsevier Ltd,
ity or facilities they aim to improve. Further explanations Amsterdam
could be published as a critical review article to discuss, Belany P, Hrabovsky P, Kolkova Z (2021) Combination of lighting
retrofit and life cycle cost analysis for energy efficiency improve-
for instance, how the proposed NPI method is simple and ment in buildings. Energy Rep 7:2470–2483
to what extent other methods are expensive and compli- BizEE Degree Days Weather Data for Energy Saving (2022) Available
cated. The implication of more accurate consumption of at https://​www.​degre​edays.​net/. Accessed August 2022
the ACs seems to be possible by smart energy meters. Boston Solar Website: How Much CO2 Do Solar Panels Save? (2022)
Available at: https://​www.​bosto​nsolar.​us/​solar-​blog-​resou​rce-​
Despite the satisfactory performance of the GA optimizer, center/​blog/​how-​much-​co2-​do-​solar-​panels-​save/. Accessed July
swarm optimizers may have faster capability for producing 2022
the solutions. More walk-through audits could be made to Castrillón-Mendoza R, Manrique-Castillo PA, Rey-Hernández JM,
test the building via infrared thermography and thermal Rey-Martínez FJ, González-Palomino G (2020) PV Energy
Performance in a Sustainable Campus. Electronics 9(11):1874.
examination of the building’s proposed insulator if it is https://​doi.​org/​10.​3390/​elect​ronic​s9111​874
approved and installed. LED lighting replacement with a Foroughi R, Asadi S, Khazaeli S (2021) On the optimization of energy
recent lighting system is another possible future feasibil- efficient fenestration for small commercial buildings in the United
ity study. Sophisticated studies can be also suggested to States. J Clean Prod 283:124604
Guo S, Wang J, Wei J, Zachariades P (2014) A new model-based
account for emission reduction from building insulation approach for power plant tube-ball mill condition monitoring and
and smart control of air conditioners. Finally, since the PV fault detection. Energy Convers Manag 80:10–19
system covers the highest proportion of energy of 78% of Kolb O, Siegemund S (2017) For DENA- German energy agency: final
the building consumption, the solar PV system may need report- study on the implementation of article 7(3) of the “direc-
tive on the deployment of alternative fuels infrastructure” –fuel
extensive studies to be conducted as optimization prob- price comparison. European Commission, Directorate General for
lems, such as maximum power point tracking (MPPT) and/ Mobility and Transport, Brussels
or Operating and Maintenance Standard Operating Pro- Kreith F, Goswami Y (2008) Energy management and conservation
cedures (O&M SOPs). However, the last point requires handbook. Editor. CRC Press, Boca Raton
Leiva E, Villacreses S, Benítez DS (2018) Electrical energy audit of
enormous data and significant amount of research work. university buildings: Case study at Universidad San Francisco de
Quito. In: In 2018 IEEE ANDESCON. IEEE, Santiago de Cali,
pp 1–6
Data Availability The datasets generated during and/or analyzed dur- López-Ochoa LM, Bobadilla-Martínez D, Las-Heras-Casas J, López-
ing the current study are available from the corresponding author on González LM (2019) Towards nearly zero-energy educational
reasonable request. buildings with the implementation of the Energy Performance
of Buildings Directive via energy rehabilitation in cold Mediter-
ranean zones: The case of Spain. Energy Rep 5:1488–1508
Declarations Lucchi E (2018) Applications of the infrared thermography in the
energy audit of buildings: a review. Renew Sust Energ Rev
Conflict of Interest The authors declare no competing interests. 82:3077–3090
Lucchi E, Delera AC (2020) Enhancing the historic public social hous-
ing through a user-centered design-driven approach. Buildings
10:159. https://​doi.​org/​10.​3390/​build​ings1​00901​59
References Mohamed O, Wang J, Guo S, Al-Duri B (2011) Mathematical mod-
eling of coal fired supercritical power plant and model parameter
Al-janabi A, Kavgic M, Mohammadzadeh A, Azzouz A (2019) Com- identification using genetic algorithms. In: Ao SI, Gelman L (eds)
parison of energyPlus and IES to model a complex university In Electrical Engineering and Applied Computing. Lecture Notes
building using three scenarios: free-floating, ideal air load system, in Electrical Engineering, 90th edn. Springer, Berlin
and detailed. J Build Eng 22(2019):262–280 Mottahedia M, Mohammadpour A, Amirib SS, Rileyb D, Asadib S
Almuhtady A, Alshwawra A, Alfaouri M, Al-Kouz W, Al-Hinti I (2015) Multi-linear regression models to predict the annual energy
(2019) Investigation of the trends of electricity demands in Jordan consumption of an office building with different shapes. Procedia
Eng 118:622–629

13
920 Process Integration and Optimization for Sustainability (2022) 6:901–920

Nadimuthu LPR, Victor K (2021) Optimization of energy-intensive Wang D, Meng K, Gao X, Coates C, Dong Z (2017) Optimal air-con-
process in ayurvedic medicine manufacturing unit—a case study. ditioning load control in distribution network with intermittent
Process Integr Optim Sustain 5:975–992. https://​doi.o​ rg/1​ 0.​1007/​ renewables. J Mod Power Syst Clean Energy 5:55–65
s41660-​021-​00194-3 Yang M, Yu X (2015) Energy efficiency: benefits for environment and
Osorio AM, Úsuga LF, Vásquez RE, Nieto-Londoño C, Rinaudo ME, society. Springer, London, pp 11–42
Martínez JA, Filho WL (2022) Towards Carbon Neutrality in Zhang C, Hu M, Laclau B, Garnesson T, Yang X, Tukker A (2021)
Higher Education Institutions: Case of Two Private Universities Energy-carbon-investment payback analysis of prefabricated
in Colombia. Sustainability 14(3):1774. https://​doi.​org/​10.​3390/​ envelope-cladding system for building energy renovation: Cases
su140​31774 in Spain, the Netherlands, and Sweden. Renew Sust Energ Rev
Ridwana I, Nassif N, Choi W (2020) Modeling of building energy con- 145:111077
sumption by integrating regression analysis and artificial neural
network with data classification. Buildings [online] 10(11):198 Publisher’s Note Springer Nature remains neutral with regard to
Rodriguez A, Smith ST, Potter B (2021) Sensitivity analysis for build- jurisdictional claims in published maps and institutional affiliations.
ing energy audit calculation methods: handling the uncertainties
in small power load estimation. Energy. 238:121511 Springer Nature or its licensor holds exclusive rights to this article under
Thermacote® Website (2022) Available at: https://​therm​acote.​com/. a publishing agreement with the author(s) or other rightsholder(s);
Accessed August 2022 author self-archiving of the accepted manuscript version of this article
Tian W, Heo Y, De Wilde P, Li Z, Yan D, Park CS, Feng X, Augen- is solely governed by the terms of such publishing agreement and
broe G (2018) A review of uncertainty analysis in building energy applicable law.
assessment. Renew Sust Energ Rev 93:285–301
Veiga RK, Veloso AC, Melo AP, Lamberts R (2021) Application of
machine learning to estimate building energy use intensities.
Energy Build 249:111219

13