sustainability

Article
Scenario Aggregation-Based Grid-Connected
Photovoltaic Plant Design
Ahmad Abuelrub 1, * ID
, Osama Saadeh 2 and Hussein M. K. Al-Masri 3 ID

1 Department of Electrical Engineering, Jordan University of Science and Technology, Irbid 22110, Jordan
2 Department of Energy Engineering, German Jordanian University, Amman 11180, Jordan;
[email protected]
3 Department of Electrical Power Engineering, Yarmouk University, Irbid 21163, Jordan; [email protected]
* Correspondence: [email protected]; Tel.: +962-792-907-358




Received: 14 March 2018; Accepted: 12 April 2018; Published: 20 April 2018


Abstract: As the global population continues to increase and living standards in developing countries
continue to improve, the demand for energy is surging. This is also coupled with technological
advances, which are leading to the increased electrification of transportation, manufacturing,
and home appliance. Classical fossil fuel-based energy generation is unsustainable and a significant
cause of air pollution. Therefore, clean, local, and sustainable sources of energy, such as solar
energy, have recently been receiving more attention. In this paper, a complete design approach for
grid-connected photovoltaic (PV) plants is developed. Jordan University of Science and Technology
(JUST) is presented as a case study. The design is formulated as an optimization problem to find
the optimal PV plant size needed to minimize the system cost and meet the design constraints.
The uncertainties of solar radiation and temperature are considered using appropriate stochastic
models. The optimization problem is solved using the Scenario Aggregation technique. The results
show the financial and environmental feasibility of the proposed system. The optimal PV plant
size which meets the annual demand of JUST is found and the financial and environmental benefits
derived from executing the proposed system are emphasized.

Keywords: photovoltaic plant; grid-connected; sustainable energy; CO2 emissions reduction; scenario
aggregation; system uncertainties

1. Introduction
Energy stored in fossil fuels that is released by human activity causes environmental pollution.
Furthermore, fossil fuel energy resources are exhaustible, and alternatives will be needed in the
near future. On the other hand, renewable energy resources are clean and cannot be depleted [1].
Photovoltaic energy is one of the most sustainable forms of energy, as it harvests energy from the
sun. It is silent, ecological, local, and economically feasible. So, market expansion in PV technologies
has rapidly increased due to reduced initial capital investment in addition to the urgency of realizing
energy access in remote areas, as this technology does not require transmission lines. In essence,
PV technology promotes sustainable local energy generation.
The published literature is rich with research projects studying PV energy. In Reference [2],
a stochastic method is involved in order to size an off-grid PV system with energy storage. This includes
Markov chain and beta probability density function (pdf). The stochastic results are compared
with a deterministic method. The stochastic solution provides more reliable and realistic results.
In Reference [3], an off-grid PV/diesel battery is discussed and designed for a village in Russia as
a case study. The levelized cost of energy (LCOE) and the CO2 emissions are calculated to indicate the
economical and the environmental feasibility of the proposed configuration. Results show savings

Sustainability 2018, 10, 1275; doi:10.3390/su10041275 www.mdpi.com/journal/sustainability

Sustainability 2018, 10, 1275 2 of 13

in both fuel and oil cost (7.8%) and emissions (51.0%) compared with a conventional fossil fuel plant.
The authors of Reference [4] evaluated the investment in a multi-site photovoltaic plant as a profitability
analysis with an objective to maximize the net return. In a techno-economic study [5], numerous PV
configurations are investigated. The LCOE is calculated, which includes capital cost, as well as
operation and maintenance costs over the project lifetime. Also, the economic impacts of using batteries
or not are discussed. In Reference [6], a stochastic programming model for the optimal scheduling
of a distributed energy system with multiple energy devices including renewables, considering
economic and environmental aspects, is presented. The results give the optimized operation strategies
which reduce the expected energy costs and CO2 emissions. In Reference [7], the authors discuss the
feasibility of a grid-connected rooftop/building integrated photovoltaic system for the electrification of
consumption with incorporating feed-in tariffs/net metering process regulation in New Delhi, India.
Jordan’s area is about 89,200 km2 , with 80% of its 9.5 million inhabitants living predominantly in
urban centers in the northwest of the country in areas that amount to approximately 10% of the total
available land. As 96% of all energy in Jordan is imported, the country is highly affected by the cost
of energy imports, which have been a major burden on its economy. Volatile fuel prices are expected
to cost Jordan $3.21 billion annually, representing 19.5% of the GDP. This has a financial impact on
the country, as well as a political impact. In particular, Jordan’s potential for renewable energy is
significant, but its generation represents less than 5% of current electrical energy production.
To encourage the deployment of renewable energy systems, the government has approved
a net-metering scheme, as shown in Figure 1 below. This scheme employs a net-meter, which in
essence is a bidirectional kWh counter. Any excess energy generated by the system that is not
consumed locally at the load is injected into the grid, and at the end of a specific contract period,
the net consumption is billed. In Jordan, the contract period is annual, which allows consumers to
generate excess energy during the high radiation summer months that can later be retrieved during
low generation winter months.

Figure 1. Grid-connected PV plant using the net-metering system.

This is especially true for large universities such as Jordan University of Science and Technology
(JUST), which are considered large energy consumers. JUST consumed 33.4 GWh in 2016, at a cost of
$12.25 million, which represents 12% of the university’s annual budget. This is a huge burden that
affects potential development and expansion of the university.
Universities and institutes of higher education are expected to be community role models
and lead change in their local surroundings. This is especially true for developing countries
were universities house unique intellect and expertise. The United Nations Decade of Education
for Sustainable Development (2005–2014) resulted in more attention being directed toward higher
education institutions and the sustainability of their operations. In addition students, alumni, and
Sustainability 2018, 10, 1275 3 of 13

governments are pressuring universities to lead local, national, and international efforts toward
increased integration of renewable energy.
In this paper, a full optimized PV system design for JUST is presented. In Section 2, detailed
mathematical modeling of the proposed system is presented. Section 3 provides a complete PV plant
sizing procedure which is implemented as a stochastic optimization problem solved using a scenario
aggregation technique. The case study of the PV plant at JUST is considered and a detailed design of
the PV plant, which includes its main components, is illustrated in Section 4. In Section 5, a detailed
financial and environmental benefit analysis of the suggested PV plant is presented. Finally, Section 6
draws conclusions and shows the financial and environmental feasibility of the proposed system.

2. Mathematical Modeling of the PV System

Mathematical modeling is considered a crucial step in predicting the behavior of a system using
a list of inputs, a list of outputs, and the algorithm needed to find the optimal output function. In other
words, mathematical modeling is used to evaluate the system performance for various inputs and
conditions. Therefore, it is an essential process especially for large utility scale systems, such as
photovoltaic renewable energy systems, which have high initial investment. Detailed mathematical
modeling for the components of the proposed PV system is presented in this section.

2.1. PV Plant Output Power

The electrical power generated from one PV panel (PDC ) depends on the global solar radiation
and the panel cell temperature and can be calculated using (1) [8]:

GA
PDC = PSTC × × [1 + ( TC − TSTC )CT ] (1)
GSTC

where PSTC is the power of the PV module under STC conditions in Watts. GA is the total in-plane
irradiance on the surface of the PV module in Watt/m2 . GSTC is the STC irradiance (1000 Watt/m2 ).
TC is the temperature of the PV cell. TSTC is the STC temperature (25 ◦ C); CT is the manufacturer’s
temperature power coefficient. The cell temperature of the PV system operation can be estimated
using (2) [9,10]:
TNOCT − 20
 
TC = Ta + Ga (2)
800
where Ta is the ambient temperature in ◦ C. TNOCT is the PV cell temperature at an irradiance of
800 Watt/m2 and an air temperature of 20 ◦ C.

2.2. Uncertainty in Solar Radiation

The solar radiation is random in its nature; therefore, a stochastic model is needed to model its
uncertainty. In the published literature, Beta distribution is widely used to model solar radiation
uncertainty, as in References [11,12]. In this paper, the Beta distribution for each time step is used to
model the solar irradiance. The probability density function of the Beta distribution is given in (3):

Γ ( α i + β i ) S I α i −1 (1 − S I ) β i −1
( )
Γ ( αi ) Γ ( β i )
0 ≤ SI ≤ 1
f b (S I ) = (3)
0 Otherwise

where fb (SI ) is the Beta distribution function of SI . αi, , βi are Beta distribution parameters for time step
i. The parameters of the Beta distribution can be estimated from the historical solar radiation data
using (4) and (5): !
µsi (1 + µsi )
β i = (1 − µsi ) −1 (4)
σsi2
Sustainability 2018, 10, 1275 4 of 13

µsi β i
αi = (5)
1 − µsi
where µsi and σsi are the mean and the standard deviation of the solar radiation data for time step
i, respectively.

2.3. Uncertainty in Temperature

The daily temperature variation over a long period is usually modeled using Normal probability
distribution [13]. The Normal distribution is defined by two parameters: the mean value µt and the
standard deviation σt . Temperature data is used to create the Normal distribution for each time step in
order to model the uncertainty in ambient temperature. The PDF of the Normal distribution is given
in (6):
(l −µti )2
1 −
2σ2
f ( Ta ) = √ e ti (6)
σti 2π
where Ta is the ambient temperature, µti is the mean value of Ta for time step i, and σti is the
standard deviation.

2.4. PV Plant Area

The panels of the PV array should be installed at an optimal distance (dopt ), minimizing the
required land area of the PV plant and taking into account the prohibition of shading between rows of
the PV array. As a matter of fact, to ensure zero shading, dopt is calculated at the time when the sunrise
angle is at the minimum (γs ) [14]. Note that dopt depends on the module angle of dip (β), the sunrise
angle, and the dimensions of the PV panel. It is calculated in the project’s location at noon of the winter
solstice to be 2.6 m. Then, the optimal total area required to install the PV plant can be calculated
using (7):
A PVPlant = ANcol cos β + W ( NPV − Ncol )dopt (7)

where NPV is the number of PV panels. Ncol is the number of columns in a rectangular PV array. Finally,
A and W are the area and the width of the PV panel used in this project.

3. PV Plant Sizing Procedure

In this section, a mathematical formulation of the proposed system will be presented for
optimization. The objective function measures the total cost of the proposed system, which is the
capital cost of the installed PV plants. The optimization problem constraints are as follows: the energy
generated from the PV plant is equal to load requirements, and the total PV plant area is less than the
dedicated land area at JUST. It is assumed that load will be fully supplied by the PV plant, which will
result in zero energy consumption from the grid. The optimization problem is proposed as a stochastic
optimization problem to find the optimal size of PV plant needed to minimize the cost of the system
and to meet the problem constraints.
The total system cost is a function of the PV plant size, since as the installed capacity increases the
total system capital cost increases. As shown in Equation (1), the total energy generated from the PV
plant can be expressed as in (8):

S
EPV = ∑SPV × SSTC
Ii
× [1 + ( TCi − TSTC ) TCP ] × ti (8)
i
Sustainability 2018, 10, 1275 5 of 13

where EPV is the total energy generated from the PV plant in MWh, SPV is the plant size in MW, i is the
index of the time step, and ti is the time step in hours. The problem constraints mentioned above are
shown in (9) and (10), respectively.

8760
S
EPV = ∑ SPV × SSTC
Ii
× [1 + ( TCi − TSTC ) TCP ] × ti = Dyr (9)
i =1

A PVs ≤ Amax (10)

where Dyr is the total energy demand for one year, ti is one hour, and Amax is the available area in m2 .
Since the problem has stochastic variables, a stochastic optimization technique must be applied.
In the published literature, different methodologies of stochastic optimization have been suggested to
solve problems with uncertainties. In Reference [15], a stochastic optimal sizing of a rural mini-grid
composed by a photovoltaic (PV) plant, lithium battery storage, a diesel generator, and a fuel tank is
proposed. The problem is formulated as an optimization problem to minimize the net present cost of the
mini-grid components. A particle swarm optimization (PSO) procedure integrated with Monte-Carlo
simulation is used to solve the optimization problem. PSO is used in Reference [16] to find the optimal
hybrid PV/wind configuration. However, in this paper, a simpler scenario aggregation approach for
stochastic optimization, which is presented in Reference [17], is used to solve the optimization problem.
The main advantage of this method is that it is able to capture system uncertainties in addition to being
simple to implement. The mathematical model for the stochastic optimization is shown in (11):

min ∑ ps f (x, ss ) (11)

s.t. x ∈ ∩s Cs

where scenario s is assigned with probability ps . The optimal solution of this problem is denoted by x*.
In the optimization problem of the proposed system represented by (8)–(10), the scenario s involves
solar radiation SIi and ambient temperature TCi . The optimal solution x* is the optimal size of the PV
plant SPV . The scenarios are assumed to be uniformly distributed, therefore ps is the reciprocal of the
total simulated scenarios (ns ) as in (12):
1
ps = (12)
ns
Then, the optimal solution of the stochastic optimization problem given in (11) can be
approximated over ns simulated scenarios by (13), as shown in the scenario aggregation
methodology [17].
xˆ∗ = ∑ ps x s (13)
s∈S

where xs is the optimal solution for scenario s. The convergence is verified by limiting the coefficient of
variation (COV) of the optimal vector of solutions x*, as shown in (14):

σs
COV = ≤ COVmin (14)
µs

where σs is the standard deviation and µs is the mean of the vector of solutions x*. Solving the
optimization problem gives the optimal size of the PV plant needed to minimize the system cost (i.e.,
PV plant size in (8)) and meet the problem constraints defined in (9) and (10). The PV plant sizing
procedure can be summarized by the flow chart shown in Figure 2.
Sustainability 2018, 10, 1275 6 of 13

Figure 2. PV plant sizing procedure flow chart.

4. Case Study and PV Plant Design

4.1. Case Study

In this case study, the plan of a grid-connected PV plant in JUST is presented. It is worth noting
that licensing for renewable energy projects in Jordan is done through the Energy and Minerals
Regulatory Commission (EMRS). The latest regulations governing large net-metering project only
allow installing a PV system to meet demand, and no oversizing is allowed. A license is only granted
after a thorough review of annual consumption analysis, and at the end of the billing year any excess
generation is forfeit. An overview of JUST is shown in the following discussion.

4.1.1. About JUST

JUST is a comprehensive, state-supported university located on the outskirts of Irbid. It was
established in 1986. Today, JUST has more than 1000 full-time faculty members, with 24,000
undergraduate and 1800 graduate students, in contrast to 2300 students in the 1986/1987 academic
year. This increase has led to increased construction and the establishment of more facilities to conform
Sustainability 2018, 10, 1275 7 of 13

to JUST’s ambitious strategy to extend all possible means to ensure distinctive graduates. This, in turn,
has driven the energy demand of JUST to higher levels.

4.1.2. Energy Demand

Due to the rapid increase in the JUST community, more energy resource planning is needed
to meet the increasing demand. However, energy resource planning requires careful load study
to ensure that the installed capacity of the suggested power system is sufficient to meet the load
demand. Therefore, a thorough investigation of JUST’s energy demand during the last five years was
conducted and the results are summarized in Table 1. This table shows the monthly electricity demand
of JUST during the last five years. The main contributors to electricity consumption at JUST are the air
conditioning, lighting, and water pumping systems, in addition to other minor applications.

Table 1. Energy consumption of Jordan University of Science and Technology (JUST) for the last five
years in GWh.

Month 2012 2013 2014 2015 2016

Jan 2.574 2.394 2.237 2.145 2.338
Feb 2.396 2.464 2.277 2.139 2.329
Mar 2.339 2.305 2.474 2.485 2.527
Apr 2.833 2.744 2.685 2.644 2.292
May 2.801 2.603 2.884 2.863 2.910
Jun 2.626 3.401 3.008 3.103 2.899
Jul 3.484 3.546 3.621 2.882 3.310
Aug 2.848 2.756 2.832 3.384 3.788
Sep 3.114 3.249 2.660 3.094 2.968
Oct 2.601 2.308 2.762 2.426 2.848
Nov 2.482 2.904 2.765 2.304 2.695
Dec 2.415 2.496 2.304 2.450 2.521
Total 32.513 33.171 32.508 31.919 33.425

As can be seen in the table, the energy demand decreased from September 2014 through April 2015.
This is predominately due to the implementation of an energy efficiency program at the university.
Building operations implemented a temperature control strategy for all air conditioning systems and
installed motion sensors in bathrooms and hallways for lighting control. The increase in consumption
afterward is due to the commission of a new building at the university.

4.1.3. Solar Radiation and Temperature

Jordan has, in general, a hot, dry climate characterized by long, hot, dry summers and short,
cool winters. The summer season, from late April to October, is hot and dry, with high temperatures
averaging around 32 ◦ C. The winter, lasting from November to March, is relatively cool, with high
temperatures averaging around 13 ◦ C [18]. The solar radiation at JUST’s location (32.4950437, 35.989037)
is abundant for most of the year, whereas cloudy days amount to less than 10% of the year. One year of
solar radiation and ambient temperature data averaged over a one-hour time interval were obtained
from the weather station at JUST. The data are shown in Figure 3 below.
Sustainability 2018, 10, 1275 8 of 13

Figure 3. (a) Solar radiation average for one year; (b) average temperature for four seasons.

4.2. Simulation Results

The uncertainties in solar radiation and ambient temperature were modeled using Beta and
Normal probability distribution functions. These PDFs were built using one year of historical data
(2016) that were collected from JUST’s weather station. The mean and standard deviation for each time
step were calculated. Then, Beta and Normal distributions were created for each time segment of each
day as described in Section 2. The available area for the PV plant at JUST is 250,000 m2 .
The parameters of the optimization procedure presented in Section 3 were chosen such that
ns = 5000, and COV min = 0.05. After running the PV plant sizing procedure, illustrated in Figure 4,
on MATLAB with these selected parameters, the optimal size of the PV plant was found to be 19.64 MW.
However, this installed DC capacity does not include electrical losses in the transformer, inverters,
or the cables, which are presented in the next subsection.
The accuracy of the solution obtained from the PV plant sizing procedure depends on the number
of the sampled scenarios (ns ) and the coefficient of variation (COV min ). As the COV min gets smaller,
the number of the required samples increases and the accuracy of the optimal solution increases.
Figure 4 shows the relationship between COV min and optimal PV plant size.

Figure 4. The relationship between coefficient of variation (COV) and optimal PV plant size.

4.3. PV Plant Design

JUST is supplied from the national electric grid by Irbid District Electricity Company (IDECO)
through three dedicated 33 kV feeders, namely: Techno1, Techno2, and Hassan. A grid impact study
(GIS) was carried out by IDECO to assist the suitability of the connecting grid to handle connecting
the PV system with a net-metering scheme. It was found that, due to reverse current limits, all three
feeders must be utilized to connect the PV system. The connection scheme recommendation was
8 MWp on Techno1, 8 MWp on Techno2, and 4 MWp on Hassan. Therefore, it was decided to divide the
plant into five equal zones of 3.928 MWp , with each zone supplying a separate circuit to the associated
33 kV bus. This required five transformers rated at 5 MW. As transformer efficiency is best at 80%
loading, and transformer loss calculations are not easily computed. This is due to the fact that the
Sustainability 2018, 10, 1275 9 of 13

rated PV generation only occurs for a short time during daytime generation. This dictates that the
worse-case transformer efficiency should be used for calculations. According to PVSYST, one of the
leading PV system simulation software, the major transformer losses that affect PV system design
include iron losses, ohmic losses, and night disconnect. This may be estimated to approximately 1% of
Sustainability 2018, 10, x FOR PEER REVIEW 9 of 13
transformer rating. In our case, this results in 1.3% of the PV plant size.
According
rated PVto utility standards,
generation only occursMV for apower cable
short time losses
during should
daytime not exceed
generation. This1% of power.
dictates that theThe GIS
worse-case
also requires a wide transformer
range of powerefficiency should
factor be control,
(PF) used for calculations. According to PVSYST,
which is implemented one of the Top of
by the inverters.
leading inverters,
the line on-grid PV system that simulation
have asoftware, the major
wide range of PFtransformer losses an
set points, have thatefficiency
affect PV of
system
98.5%. design
Therefore,
include iron losses, ohmic losses, and night disconnect. This may be estimated to approximately 1%
the plant should be oversized by 3.8% to meet load demand. The new system size is 20.39 MWp ,
of transformer rating. In our case, this results in 1.3% of the PV plant size.
with a new zone size of
According to4.08 MW
utility p.
standards, MV power cable losses should not exceed 1% of power. The GIS
Jenko
alsoSolar “JKM265P-60”
requires a wide range of PV modules
power factor were chosen
(PF) control, for this
which project. Jenko
is implemented by thehas been constantly
inverters. Top
ranked over
of thethe
linepast few years
on-grid by Bloomberg
inverters, as a Tier
that have a wide range1 of
PVPFmodule manufacturer.
set points, The Jenko
have an efficiency PV panel
of 98.5%.
parametersTherefore,
are given the plant should
in Table 2. be oversized by 3.8% to meet load demand. The new system size is 20.39
MWp, with a new zone size of 4.08 MWp.
Jenko Solar “JKM265P-60”Table
PV modules
2. Jenkowere chosen for
PV module this project. Jenko has been constantly
ratings.
ranked over the past few years by Bloomberg as a Tier 1 PV module manufacturer. The Jenko PV
panel parameters are given in Table 2.
Rated power 256 W
Max power voltage 31.4 V
Table 2. Jenko PV module ratings.
Max power current 8.44 A
Rated power 256 W
Max power voltage 31.4 V
The SMA Solar Technology “STP
Max60-10”
powerinverter
current is chosen
8.44 Afor this design. SMA is chosen as it is
the lead PV inverter technology in the Jordanian market. This inverter requires a DC input voltage in
The SMA Solar
the range of 565–1000 Vdc Technology
with 60 kW “STP 60-10”
rated inverter
power. As iseach
chosen for this
of the design.
zones SMA is chosen
is identical. Thus,as for
it is a total
the lead PV inverter technology in the Jordanian market. This inverter requires a DC input voltage in
generation capacity of 20.39 MWp , 68 inverters are required per zone. Each inverter requires modules
the range of 565–1000 Vdc with 60 kW rated power. As each of the zones is identical. Thus, for a total
arrangedgeneration
in seriescapacity
and parallel combinations to meet inverter voltage and power demands. Each
of 20.39 MWp, 68 inverters are required per zone. Each inverter requires modules
PV stringarranged
was chosen to consist of 24
in series and parallel modules to
combinations with
meeta inverter
string voltage of 753.6
voltage and powerV, which ideally
demands. Each PVlays in
the middle of was
string the chosen
inverter voltage
to consist of range. Each
24 modules string
with provides
a string a peak
voltage of power
753.6 V, whichof 6144lays
ideally Wpin, and
the a DC
combinermiddle of the inverter
box combines voltageper
10 strings range. Each string
inverter. The DCprovides
to ACa peak
ratiopower
at the of 6144 Wpis
inverter , and a DC
approximately
1.024 to 1.combiner box combines
The complete plant 10 strings
has per inverter.
340 inverters and The DC toPV
81,600 ACmodules
ratio at the
forinverter
20.89 is
MW approximately
p and 20.4 MWac
1.024 to 1. The complete plant has 340 inverters and 81,600 PV modules for 20.89 MWp and 20.4 MWac
planet capacity. Figure 5 shows the layout of a three-inverter section, detailing PV string connection
planet capacity. Figure 5 shows the layout of a three-inverter section, detailing PV string connection
up to theup transformer connection.
to the transformer TheThe
connection. restrest
ofof
thethesystem
system online diagram
online diagram is repeatable
is repeatable as theas the overall
overall
plant is scalable.
plant is scalable.

Figure 5. PV plant connection diagram.

Figure 5. PV plant connection diagram.
Sustainability 2018, 10, 1275 10 of 13

5. Financial and Environmental Benefits

5.1. Financial Analysis

JUST electricity is billed based on a bracket scheme, as shown in Table 3.

Table 3. JUST electricity billing brackets.

Consumption Bracket (kWh) Rate ($)

1–160 0.059
161–300 0.129
301–500 0.153
501–600 0.204
601–750 0.238
751–1000 0.268
above 1000 0.375

As JUST consumes much more than 1000 kWh each month, most of the energy consumed is
priced at the highest billing bracket. The average kWh cost for 2016 was calculated to be $0.366,
with an annual electrical bill of $12.23 million USD. The latest government PV tenders have been
awarded a price of $986 per kWp . Therefore, the designed system will cost approximately $21 million.
The payback period of this project will include the capital cost in addition to the operational and
maintenance costs. These calculations are illustrated in the following discussion.

Payback Period Analysis

The payback period analysis includes the investment cost of the PV plant and the operational
and maintenance (O&M) costs. The (O&M) costs used in this discussion are obtained from similar PV
plant projects which are currently operational in Jordan. These costs are summarized in Table 4.

Table 4. Summary of operational and maintenance (O&M) costs.

Type of Service Cost

PV modules cleaning $400/MWp /month
Landscape maintenance $700/MWp /month
Generic maintenance 3% of the capital cost/year
Inverters replacement 20% of the capital cost/nine years

PV modules’ cleaning is necessary to enhance the PV plant yield. This is especially important
in the summer months, as dust buildup affects performance. Landscape maintenance is needed to
keep landscape healthy and facilitate the maintenance of the plant; this includes weeding, which is
important during the spring, when weed height actually introduces shading to the lower part of the
panel. Generic maintenance refers to the periodic preventative maintenance needed to keep the plant
working satisfactorily. Finally, a typical inverter lifetime is eight to 10 years, and the PV plant project
lifetime is 25 years. Therefore, it is assumed that the inverters need to be replaced at years 9 and 18 of
the project lifespan.
In order to determine the payback period of this project, the net present value of the project must
be calculated. The net present value includes the plant capital cost (initial cost) and the total O&M costs
reflected into the present year. The present worth of the total O&M costs can be calculated using (15):

N
O & Mnwc = ∑DFj ∗ O & Mj (15)
j =1
Sustainability 2018, 10, 1275 11 of 13

where O&Mnwc is the present worth cost of the total O&M costs, DFj is the discount factor for year
j, and O&Mj is the O&M cost for year j. It is worth mentioning that the inverter replacement cost is
calculated for years 9 and 18 only. The discount factor for year j is a function of the nominal interest
rate i and the inflation rate f and can be calculated using (16):
 j
1+ f
DF = (16)
1+i

In this paper, the nominal interest rate is 7% and the inflation rate is 3%. Using (16) and (17),
the net present cost of the O&M costs is found to be equal to $20.7 million. Hence, the net present cost
is equal to $41.7 million USD.
The expected PV plant generation in is shown in Table 5. The generation will decrease with time,
as expected, due to the degradation phenomena of the PV modules.

Table 5. Expected lifetime PV plant yield.

Year Generation kWh Yield in m$ Year Generation kWh Yield m$

1 32,760,000 11.99 14 27,518,400 10.07
2 31,777,200 11.63 15 27,190,800 9.95
3 31,122,000 11.39 16 26,863,200 9.83
4 30,794,400 11.27 17 26,535,600 9.71
5 30,466,800 11.15 18 26,208,000 9.59
6 30,139,200 11.03 19 25,880,400 9.47
7 29,811,600 10.91 20 25,552,800 9.35
8 29,484,000 10.79 21 25,225,200 9.23
9 29,156,400 10.67 22 24,897,600 9.11
10 28,828,800 10.55 23 24,570,000 8.99
11 28,501,200 10.43 24 24,242,400 8.87
12 28,173,600 10.31 25 23,914,800 8.75
13 27,846,000 10.19

Using the same nominal interest rate used above the estimated payback period of the PV plant is
around four years and five months.

5.2. Environmental Benefits

According to the US Environmental Protection Agency (EPA), using eGRID, the US annual
non-base load CO2 output emission rate, considered an accurate “Emission Factor” for electrical
consumption, can be calculated using the following rate:

Electrical Consumption: 7.03 × 10−4 metric tons CO2 /kWh

Therefore, the annual CO2 emissions due to electric consumption at JUST is equal to:

33,425,000 × 7.03 × 10−4 = 23,497.775 metric tons or 25.9 k t US equivalent.

As with any other higher education institution, not all students and academic staff are on campus
on a daily basis. Thus, for a complete assessment of the impact of installing the PV plant, the Full-Time
Equivalent (FTE) of students and employees was determined to normalize the campus CO2 reduction.
A survey of the number of students registered per semester in 2016, in addition to course schedule,
employee work hours, and academic staff course load for the different semesters were all considered
in the FTE calculation, and the FTE for the university community (students, faculty, and staff) was
calculated to be 16110 FTEs. Therefore, the installed PV plant reduces CO2 emissions by 1.6 t per FTE.
The maximum amount of CO2 a person should produce per year in order to halt climate change is
2.0 t CO2 . The world average is 5 t, the average in Europe is 6.7 t, the average in the US is 16.5 t, and the
Sustainability 2018, 10, 1275 12 of 13

average in Jordan is 3.4. These are country averages that include CO2 emission from work, public
operations, and home. The CO2 emission reduction due to the JUST PV system brings the universities’
community average to below the 2.0 t limit.
So, installing this PV system not only saves a great deal of financial resources, but also has
a tremendous environmental impact, with huge CO2 reduction.

6. Conclusions
This paper aims to meet JUST’s electrical energy demand by considering PV plant installation
connected to the electrical grid using a net-metering technique. This was achieved by formulating
an optimization problem in which the objective is to find the optimal PV plant size needed to meet
JUST’s electricity demand and minimize the system cost. A detailed mathematical model of the PV
plant was presented. The system uncertainties inherent in solar radiation and ambient temperature
were modeled using Beta and Normal distributions. This stochastic optimization problem was solved
using the Scenario Aggregation procedure. The optimal size of the proposed PV plant was found to
be 19.64 MW. When considering MV losses, as well as commercial components, a 20.98 MW system
was proposed. The proposed PV plant was found to be financially and environmentally efficient,
with a payback period of four years and five months as well as a considerable reduction in CO2
emissions. The project was divided into two phases. The 5 MW Phase I has been completed and
is operating according to design. Phase II is under tender and work should commence in the next
few months.

Acknowledgments: We gratefully thank Jordan University of Science and Technology (JUST) for providing the
data used for this study.
Author Contributions: H.A. derived the mathematical modeling; A.A. conceived the PV plant sizing procedure
and performed the optimization; O.S. performed the plant design and the environmental benefits analysis; A.A.
performed financial analysis.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Bostan, I.; Gheorghe, A.V.; Dulgheru, V.; Sobor, I.; Bostan, V.; Sochirean, A. Resilient Energy Systems:
Renewables: Wind, Solar, Hydro; Springer Science & Business Media: Berlin, Germany, 2012.
2. Cabral, C.V.T.; Filho, D.O.; Diniz, A.S.A.C.; Martins, J.H.; Toledo, O.M.; Machado Neto, L.V.B. A stochastic
method for stand-alone photovoltaic system sizing. Sol. Energy 2010, 84, 1628–1636.
3. Bortolini, M.; Gamberi, M.; Graziani, A.; Pilati, F. Economic and environmental bi-objective design of
an off-grid photovoltaic–battery–diesel generator hybrid energy system. Energy Convers. Manag. 2015, 106,
1024–1038. [CrossRef]
4. Bendato, I.; Cassettari, L.; Mosca, R.; Williams, E.; Mosca, M. A stochastic methodology to evaluate the
optimal multi-site investment solution for photovoltaic plants. J. Clean. Product. 2017, 151, 526–536.
[CrossRef]
5. De Prada-Gil, M.; Domínguez-García, J.L.; Trilla, L.; Gomis-Bellmunt, O. Technical and economic comparison
of various electrical collection grid configurations for large photovoltaic power plants. IET Renew.
Power Gener. 2016, 11, 226–236. [CrossRef]
6. Di Somma, M.; Graditi, G.; Heydarian-Forushani, E.; Shafie-khah, M.; Siano, P. Stochastic optimal
scheduling of distributed energy resources with renewables considering economic and environmental aspects.
Renew. Energy 2018, 116, 272–287. [CrossRef]
7. Tomar, V.; Tiwari, G. Techno-economic evaluation of grid connected PV system for households with feed in
tariff and time of day tariff regulation in New Delhi–A sustainable approach. Renew. Sustain. Energy Rev.
2017, 70, 822–835. [CrossRef]
8. Thomson, M.; Infield, D. Impact of widespread photovoltaics generation on distribution systems. IET Renew.
Power Gener. 2007, 1, 33–40. [CrossRef]
Sustainability 2018, 10, 1275 13 of 13

9. Skoplaki, E.; Boudouvis, A.; Palyvos, J. A simple correlation for the operating temperature of photovoltaic
modules of arbitrary mounting. Sol. Energy Mater. Sol. Cells 2008, 92, 1393–1402. [CrossRef]
10. Meral, M.E.; Dinçer, F. A review of the factors affecting operation and efficiency of photovoltaic based
electricity generation systems. Renew. Sustain. Energy Rev. 2011, 15, 2176–2184. [CrossRef]
11. Graham, V.; Hollands, K. A method to generate synthetic hourly solar radiation globally. Sol. Energy 1990,
44, 333–341. [CrossRef]
12. Ettoumi, F.Y.; Mefti, A.; Adane, A.; Bouroubi, M.Y. Statistical analysis of solar measurements in Algeria using
beta distributions. Renew. Energy 2002, 26, 47–67. [CrossRef]
13. Klein Tank, A.; Können, G. Trends in indices of daily temperature and precipitation extremes in Europe,
1946–99. J. Clim. 2003, 16, 3665–3680. [CrossRef]
14. Mertens, K. Photovoltaics: Fundamentals, Technology and Practice; John Wiley & Sons: Hoboken, NJ, USA, 2013.
15. Fioriti, D.; Giglioli, R.; Poli, D.; Lutzemberger, G.; Vanni, A.; Salza, P. Optimal sizing of a mini-grid in
developing countries, taking into account the operation of an electrochemical storage and a fuel tank.
In Proceedings of the 6th International Conference on Clean Electrical Power (ICCEP), Santa Margherita
Ligure, Italy, 27–29 June 2017.
16. Askarzadeh, A.; dos Santos Coelho, L. A novel framework for optimization of a grid independent hybrid
renewable energy system: A case study of Iran. Solar Energy 2015, 112, 383–396. [CrossRef]
17. Wets, R.J. The aggregation principle in scenario analysis and stochastic optimization. In Algorithms and Model
Formulations in Mathematical Programming; Springer: Berlin, Germany, 1989.
18. Philip, G. Water, Life and Civilization: Climate, Environment and Society in the Jordan Valley|Water, Life and
Civilization: Climate, Environment and Society in the Jordan Valley; International Hydrology Series; Mithen, S.,
Black, E., Eds.; Cambridge University Press: Cambridge, UK, 2012.

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).