PV-Battery-Diesel Microgrid Layout Design
Based on Stochastic Optimization
M. Hijjo*, F. Felgner*, and G. Frey*
* Saarland University, Chair of Automation and Energy Systems, D-66123 Saarbrücken (Germany)
E-mails: {mohammed.hijjo, felix.felgner, georg.frey}@aut.uni-saarland.de
Abstract—This work presents a layout design method for
a grid-connected PV-Battery-Diesel microgrid considering
the effect of the applied power management strategy among
the system’s components. The proposed microgrid is
dedicated to those areas in the world which suffer from
prolonged grid outages and depend mainly on diesel
generators to cover the periods of power deficit.
Conventionally, such hybrid microgrids are designed
assuming two prime impractical hypotheses: namely, a full
utilization of the generated power from renewable
resources, and no losses due to power conversion among
DC/AC buses; both hypotheses lead to specifically optimistic
and encouraging results from an economic point of view. In
addition, many approaches do not include the energy
management scheme during optimization. By contrast, the
layout herein is optimized taking into account a developed
rule-based energy management (EM) scheme and realistic
power conversion efficiencies, which together play the key
role in determining the realistic percentage of utilized
renewable energy. The optimization algorithm employs a
Genetic Algorithm (GA), which is one of the metaheuristic
stochastic optimization techniques that can provide a
solution to an optimization problem with less computational
effort than iterative ones. The simulations are carried out
using a typical load profile of a hospital in Gaza Strip over a
one-year period.
Index Terms—PV-Battery-Diesel, Microgrid, Energy
Management, Layout Design, Genetic Algorithms (GAs).
I. INTRODUCTION
The technological revolution in the renewable energy
sector has led the world to look for new opportunities to
handle wide aspects of energy supply—especially
offering efficient and affordable solutions—in different
societies around the world. An innovative concept which
utilizes a cluster of loads and micro power sources is
called microgrid [1]. It is formally defined by the U.S.
Department of Energy as “a group of interconnected
loads and distributed energy resources (DERs) with
clearly defined electrical boundaries that acts as a single
controllable entity with respect to the grid and can
connect and disconnect from the grid to enable it to
operate in both grid-connected or island modes” [2].
This concept is promising, especially to the regions
around the world which have an inadequate grid
infrastructure or/and an intermittent power from the main
grid and mainly depend on backup diesel generators to
supply their essential demand. Such examples are widely
spread in the Middle East, in Africa and in India with
978-1-5090-4682-9/17/$31.00 ©2017 IEEE
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on October 24,2022 at 18:02:01 UTC from IEEE Xplore. Restrictions apply.
various degrees of power lack [3]. Nevertheless, the
situation becomes even more complicated when
considering the expected extra cost of fuel transportation
or a breakdown of the supply chain, which may seriously
affect people’s life [4].
From this perspective, the deployment of renewable
energy resources (RES) is expected to be a good
alternative of energy at the long term. Moreover, the
cooperation between RES and diesel generators can lead
to significant synergies compared with systems
depending only on diesel fuel in case of outages. The
most appropriate hybrid system for such situations is PV-
Battery-Diesel, especially where the grid cannot meet the
load demand at every time.
A huge work has been done in this field of research,
starting from arbitrarily choosing the system components
in case of small loads up to optimizing the needed
capacity of each component according to specific and
predetermined operation strategy of the system [5]-[7].
Furthermore, the energy management scheme regarding
battery operation is often not considered in optimization
[5], [9], [10].
This work represents an easy-to-implement method to
design medium-scale, scalable, and self-sufficient
microgrids which are ready to be interconnected with the
national grid aiming at healing the weakened grid as far
as possible. The main contribution of this work is the fact
that the decision parameters are not only defining the size
of PV generation, battery bank and diesel generator but
also the switching criteria between battery and diesel
generator in case of insufficient PV generation: namely,
the limits of the state of charge (SoC) at which the battery
and diesel generator are switched on/off. Besides that, we
consider the losses of power conversion and the real
utilization factor of RES after performing the control
strategy.
The paper is organized as follows: Section II describes
the applied microgrid topology referring to the most
common operational constraints or/and hypotheses. In
section III, we describe the main power management
strategy which is developed to dispatch the power
efficiently among the components of the system and the
load as well. Section IV presents the optimization
framework to select the most appropriate components in
correspondence to the optimization goal. Section V
demonstrates a case study of a hospital complex in Gaza
30
 Strip and the corresponding results. Finally, a discussion τ
and a conclusion are provided in Section VI. EB τ
³
EB 0  PB (t )dt
0
(5)

II. SYSTEM DESCRIPTION where PB is the battery power rate (positive in charging
A simple microgrid system is assumed to carry out and negative in discharging), PBmax is the maximum limit
this work. It incorporates PV array, generic electric of the battery power, SoC is the actual battery state of
storage system (ESS) and a diesel generator set (GenSet), charge, EB is the actual energy stored in battery, CB is the
all together representing the proposed backup supply for nominal capacity of the battery.
the critical load, which all are connected to the main grid
D. Diesel Generator:
through the point of common coupling (PCC) as depicted
in Figure 1. The diesel generator is modeled as a linear or quadratic
function which fits the fuel consumption chart given by
EMS the manufacturers [12]. The total fuel cost results from
the integration of the fuel consumption over the time.
PCC AC Equation (6) and (7) represent the total fuel cost (Fc) and
DC
the generation limits of the generator set.
Main Grid PL PD PB PPV

³ aP
2
Fc ( P)  bP  c dt (6)
Critical Diesel Battery PV
Loads GenSet Storage Array min max
PGen d PGen d PGen (7)
AC zoneG(Grid/Diesel) DC zoneG(PV/Battery) Observably, the power generation efficiency of the diesel
generator is affected by its loading factor (i.e., generated
Fig. 1. Grid-connected PV-Battery-Diesel microgrid (EMS: energy power w.r.t nominal capacity). A typical efficiency chart
management system). for 250 kW diesel asset is illustrated in Figure 2, where
electrical energy produced per liter (kWh/L) is
Following is an abstract model of each system’s
proportionally increased with the loading factor.
component and/or its main operational constraints.
A. Main Grid:
The main grid can have two states: ON (available) and
OFF (unavailable) and can supply the load adequately
whenever it is in ON state. However, a frequent
interruption is imposed because of insufficient national
energy resources or/and restriction on the logistics, which
requires another standby supply. The grid outages can be
planned in advance or take place unexpectedly due to
unpredictable faults.
B. PV Array:
The output power of the photovoltaic array is the total Fig. 2. Diesel generator efficiency characteristics.
sum of the generated power of each panel at standard test
conditions considering the actual irradiance in the field Notably, the diesel price per energy unit is getting
and ambient temperature as well. The applied model for cheaper at higher generation levels. However, it is
each PV panel adopts equation (1) as follows: recommended to leave a safety margin before the highest
max
GC possible generation limit PGen as a compromise of
PPV PSTC >1  k Tc  TSTC @ (1)
GSTC lifetime and instantaneous efficiency, here it’s chosen at
Further details of the applied model can be found in [11]. Best
85% loading factor and denoted by PGen .
C. Battery Bank: E. Bi-Directional Power Inverter:
The battery bank is also composed of a group of A bi-directional power inverter is required to convert
smaller storage units to give the desired need of energy the power form between AC and DC buses. It is modeled
capacity (kWh). Equations (2) to (5) describe the
considering a constant efficiency factor K C as in Equ. (8)
operational constraints of each single battery (A single
time-slot is considered here as ' 10 min.) Pout Kc Pin (8)
PB d PBmax (2)
III. POWER FLOW MANAGEMENT STRATEGY
EB τ
SoC τ (3) The energy management system (EMS) is the core
CB
controller which routes the power from different
SoCmin d SoC τ d SoCmax (4) resources to supply the load in an efficient manner.

31
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on October 24,2022 at 18:02:01 UTC from IEEE Xplore. Restrictions apply.
 Therefore, an instantaneous power balance must be allowable SoCstp) is more than the residual of diesel
achieved as depicted in equation (9): generator. An example of the interactions among the
N PV PPV  N B PB  PD  PG PL  PLoss
microgrid components according to the applied rule-
(9)
based operation policy is explicitly illustrated in Figure 4.
where NPV and NB are the total number of photovoltaic
panels and storage units, respectively, PD is the generated
power by the diesel generator, PG is the power drawn
from the grid utility, PL is the instantaneous load demand
and PLoss is the total unused power, i.e., the lost power
due to power conversion, plus the unutilized solar power.
A flow chart diagram of the developed operation policy
in this work is shown in Figure 3.
SoC ≤ SoCmin

Diesel GenSET
Battery Bank Master Source
Master Source
PL ≤ [0.85×PD (max) = PD(Best)]
Diesel GenSet (OFF) Use the residual capacity up to
PD(Best) to charge the battery
(PL ≥ η.PPV) Discharging
(PL < η.PPV) Charging Keep (PD) as close as
possible to PD(Best)

SoC ≥ SoCstp

Grid is (OFF) Grid is (ON)

Fig. 4. Illustration of a one-day operation of the microgrid.

Grid (ON)
[ Supply unmet load ]
[ Diesel GenSet OFF ]
[ Charge the battery by the residual - up to SoCstp ]
Fig. 3. Rule-based operation policy for the proposed microgrid.
This operation policy adopts a master/slave operation
based on a pre-defined priority, i.e., the highest priority is
given to the renewable energy, then to the grid utility,
next to the battery storage system and, finally, to the
diesel generator. Besides, discharging process of the
battery cannot occur with the operation of diesel
generator concurrently. On the other hand, the charging
process of the battery can occur at times while the diesel
generator is ON.
As described in Figure 3, PPV has the highest priority
to supply the load; if it is insufficient, the grid takes the
responsibility to supply the residual if the grid state is
ON. In case of insufficient renewable supply and power
outage, the battery is imposed to supply the residual load
as long as its SoC is above a certain threshold SoCmin.
Finally, the diesel generator is operated to supply the
unmet load in case that SoC is less than SoCmin. To keep a
higher fuel efficiency of the diesel generator, its residual
power charges the battery up to a certain limit SoCstp if its
Best
loading factor is less than PGen . Such a criterion can be
met by solving equation (10):
PD τ
­ Best
min ®PGen , netD τ 
1
K'
> ½
SoCstp  SoC τ ¾ @ (10)
¯ ¿
where netD τ is the instantaneous unmet load by PV
generation, i.e., NPVPPV . Hence, the output power of the
diesel generator is kept at its best rate as long as the
remaining charge of the battery (to reach its maximum
The top figure shows the load profile (blue), available PV
generation (green), and the available power from grid
(yallow), where these signals represent the main inputs of
the EMS to direct the power to the load in an efficient
way. The next figure shows the power extracted/injected
to the battery (green) and the residual amount of PV
generation which could not be utilized neither to supply
the load nor to charge the battery (red). The third figure
shows the time-slots in which diesel generator is called to
supply the load and how much power it supplies (in total)
to cover the unmet demand and charge the battery as
well. Obviously, it operates almost always on its best rate
஻௘௦௧
(ܲீ௘௡ ) except in the last period of operation where the
total charge needed to reach SoCstp is less than the
஻௘௦௧
residual of its generation capacity up to ܲீ௘௡ . The third
figure presents also the power supplied from the grid at
times when it is available. Lastly, the bottom figure
illustrates the SoC and how it varies according to the
interactions among the load and the power resources. The
simulation of this day uses SoCstp = 90% and SoCmin =
40%, and it can be seen that these limits are not violated
while the grid and the diesel asset are ON. However, the
small margin up SoC = 100% is reserved to be covered
in the time of plenty PV generation, hence, SoC can reach
its maximum limit (100%) and the unutilized power can
be minimized.
IV. OPTIMIZATION ALGORITHM
Basically, it is crucial to find a cheaper and cleaner
supply than diesel in order to cover the essential load of a
facility at a cheaper price of energy over a long term.
Besides, a compromise should be found between the
running cost and the investment capital.

32
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on October 24,2022 at 18:02:01 UTC from IEEE Xplore. Restrictions apply.
 The optimization algorithm employs a Genetic
Algorithm (GA), which is one of the metaheuristic
stochastic optimization techniques that can provide a
solution to an optimization problem with less
computational effort than iterative ones [13].
The objective function to be optimized is the total sum
of the investment and running cost over the annual cost of
the system, where the running cost results mainly from
the purchased energy from the grid and the fuel
consumption by the diesel generator, and the investment
cost depends mainly on the instalment plan or the
installation cost of the components (PV panels, Battery
bank, Diesel Generator and the inverter), plus the
maintenance and replacement cost of some components
such as batteries. Equation (11) represents the
mathematical formulation of the cost function, equations
(12) and (13) correspond to the detailed mathematical
formulation of the investment and running costs ( Cinv ,
Crun ), respectively.
Ctot Cinv  Crun
Cinv Best
N PV C PV  N B C Bat  PDG C DG
Crun C f Fc  C g E g
N PV and N B are the total number of PV panels and
஻௘௦௧
storage units, ܲீ௘௡ is the maximum capacity of diesel
generator, ‫ܥ‬௉௏ , ‫ܥ‬஻௔௧ , ‫ܥ‬஽ீ are the prices of a single PV
panel, a single storage unit and the corresponding diesel
generator, respectively. The parameters Cg, Eg and Cf are
the energy price per kWh from the grid, the total energy
purchased from grid and fuel price per liter considering
transportation and logistics.
A pseudo-code of the applied GA, including a specific
decision on the appropriateness of each candidate 5-tuple
C of decision variables (Step 4), is depicted in Figure 5.
Each appropriate candidate is evaluated over a one-year
load and weather profile to find out the best fitness
function (Equation 14).
1. Inputs: Weather data, load profile and grid powering states.
2. Candidate of decision variables:
C := (ܰ௉௏ , ܰ஻ , DG, ܵ‫ܥ݋‬௠௜௡ , ܵ‫ܥ݋‬௦௧௣ )
3. Initialization: randomly seeded possible solutions.
4. Main Control Strategy: Is candidate C appropriate for
energy management scheme? If yes, evaluate the
penalty function. Otherwise, exclude C.
5. Selection: Select the best candidate solution among the
present generation before step in the next generation.
6. Crossover and mutation: The new possible candidate
solution is generated from the parents which survived.
7. Apply the main control strategy again (STEP 4)
8. Termination: after exceeding the time budget or
generation limit or satisfying the minimum criteria.
9. Output: the values correspond to the best/final solution.
Fig. 5. Pseudo-code of the optimization algorithm using GA.
Traditionally, a multi-objective optimization problem
is reformulated into a single-objective optimization
problem using weighted factors and aggregation [14].
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on October 24,2022 at 18:02:01 UTC from IEEE Xplore. Restrictions apply.
Here, a weighted sum of the lost or unutilized power is
combined with the cost function in order to be
minimized. This adds a new constraint on the problem
besides the operational constraints of each component.
The final multi-objective penalty function 3 can be
formulated as in equation (14).
§ ·
3 : Ctot  ¨ w
¨ ¦P ¸
Loss' ¸ (14)
© t ¹
w is a weighting factor that represents the significance of
this additional constraint. The applied searching method
is heuristic, however, time-efficient and fully exhaustive
as the so-called brute-force.
The search space of the problem is obviously very large.
Considering NPV solar panels and NB battery storage units
with NSoC different levels of SoCmin and SoCstp (which we
declared in advance) to choose from, the number of
iterations would reach (NPV×NB×NSoC×NSoC) of iterations,
each requiring k evaluations of eq. (1) to (14) if a brute-
(11) force method is chosen to check all possible
(12) combinations (as done, e.g., in [5]); k is the number of
total time slots among the year (here k := 6×8760 =
(13) 52560). Therefore, the GA is chosen as it enables the
exploration of a wide range of allotted elements in a
much shorter time. In our case, less than 100 generations
were examined until they converged to a solution (here:
just a few minutes of computation time).
V. CASE STUDY
Al-Shifa’ Hospital in Gaza Strip is taken as a case
study where the grid outage is planned in advance [4]. A
fraction of the load profile is chosen as a critical load in
the building complex to carry out the simulation where
the maximum peak during one year is 71.2 kW and
minimum load is 25.3 kW with a mean consumption of
42.8 kW and 10.4 kW of standard deviation [8].
This statistical information is described by the load
duration curve (LDC) in Figure 6 because it is a powerful
means to describe a load profile over a relatively long
period.
Fig. 6. Load duration curve (LDC) of the load over one year.
A planned grid outages due to 70% energy deficit in Gaza
Strip is applied [15]. It is scheduled by the distribution
company in which the grid is ON for 6 hours and OFF for
12 hours. Generally, the public are notified in advance by
such a schedule according to the current deficit and
available supplies.
33
 A deep cycle lead-acid battery is used as storage unit. The resulted optimized components are listed in Table II.
Besides, the yearly PV production of a single 250W p PV
panel is calculated using the weather data from the TABLE II. OPTIMIZED MICROGRID CONFIGURATION
metrological database METEONORM [16]. The used Parameter Nomenclature Value(s)
storage unit is a Trojan battery model IND17-6V adopted ܰ஻‫כ‬ Number of storage units 140 units
from [17]. It has a 5.55 kWh capacity at 20-h discharging ‫כ‬
ܰ௉௏ Number of PV panels 922 panels
rate and a relatively long lifetime under a low SoC, i.e. ୫ୟ୶ሺ‫כ‬ሻ
ܲ஽ீ Diesel generator capacity 100 kW
2200 cycles at 40% SoC with a nominal price of 1400 $. ‫כ‬
ܵ‫ܥ݋‬௠௜௡ Minimum allowable state of charge 40 %
Its maximum (dis)charging is considered 1/5 of the ‫כ‬
ܵ‫ܥ݋‬௦௧௣ Stop charging threshold from GenSet 89 %
capacity within 1 h to keep healthy operating conditions.
Considering one cycle per day, its annual capital cost is: The annual purchased energy and fuel savings in case of
1430 ($) × [365 (cycles/yr) / 2200 (cycles)] = 237 ($/yr). the proposed system over a one year simulation is listed
In addition, a 250Wp PV panel is considered the kernel of below in Table III. The results are compared with the
the PV array which has a price of 1.9$/Wp including the baseline configuration, which depends merely on diesel
installation, corresponding power electronics, and generator as a backup resource in addition to the
converters for voltage regulation and adaption among interrupting grid.
power bus bars. Assuming a 20-years warranty over the
TABLE III. ANNUAL PURCHASED ENERGY AND FUEL SAVINGS
project lifetime, the annual price of each panel is:
[1.9 ($/Wp) × 250 (Wp)] / 20 (yr) = 24 ($/yr). Diesel &
Diesel &
Parameter Grid-Connected
Four diesel generator candidates to choose from are Grid
Microgrid
assumed in this work by fitting their corresponding fuel Purchased energy from grid (MWh) 124,98 116,12
consumption data [12]. All parameters used in simulation Gross purchasing from grid ($×103) 22,45 20,90
with the corresponding nomenclature and values are Total fuel consumption (L×103) 80,98 3,31
listed below in Table I. Operation Hours (Hr) 5840 153
TABLE I. SIMULATION PARAMETERS Total fuel costs ($×103) 149,8 6,12
Parameter Nomenclature Value The supplied energy from each individual resource in
ܰ஻‫כ‬ Number of storage units [0, 600] both configurations is investigated and listed in Table IV.
‫כ‬
ܰ௉௏ Number of PV panels [0, 1000] Besides, the annual and utilized energy from all resources
‫כ‬
ܵ‫ܥ݋‬௠௜௡ Minimum allowable state of charge [40, 60] % in both configurations is presented as well.
‫כ‬
ܵ‫ܥ݋‬௦௧௣ Stop charging threshold from GenSet [70, 90] %
CB Single storage unit capacity 5.55 kWh TABLE IV. ANNUAL PRODUCED AND UTILIZED ENERGY
SoCint Initial state of charge 75 %
Diesel &
SoCmax Stop charging threshold from PV 100 % Diesel &
Annual Energy (MWh) Grid-Connected
ηb (Dis)charging efficiencies 95 % Grid
Microgrid
ηc Inverter efficiency 95 % Gross purchased energy from grid 124,98 116,12
ܲ஻௠௔௫ Battery (dis)charging power limit േ‫ܥ‬஻ ȀሺͷŠሻ Net load supplied from grid 124,98 96,87
௠௔௫
ܲீ௘௡ Diesel generator rated power 100 kW Gross produced energy from diesel 249,9 11,50
Cf Diesel fuel cost per liter 1.85 $/L Net load supplied from diesel 249,9 10,225
Cg Utility grid energy price per kWh 0.18 $/kWh Total produced energy from PV – 404,16
DG* Set of available diesel generators 75,100,175,250 kW
Net load supplied from PV – 267,78
ο Time slot (1/6) h Total energy of the load 374,88
(*) parameter to be optimized.
The simulation is carried out using MATLAB Obviously, the gross and net supply either from grid or
optimization toolbox and converged under these diesel in the old configuration are the same. By contrast,
circumstances to a final penalty value around 0.22 as with the microgrid, the net supply from any of the
depicted below in Figure 7. This value expresses mainly resources is less than the gross produced or purchased
the levelized energy price per kWh and the inherent energy in the system. The main reason of that is the lost
increasing due to the non-utilized PV energy as declared energy due to power conversion among the system’s
previously in equation (15). components. In addition to that, some non-utilized energy
is lost in case of the microgrid because the battery is fully
charged and the load is satisfied at the same time.
In addition to the previous indices, the net share of PV is
another important index which accurately indicates the
real benefit of deploying the RES in the system and gives
a better realization of the presented outcomes. It can be
described by two values, the RES utilization factor (UF)
and penetration level (PL) which both are defined in
equations (15) and (16) as follows:
Fig. 7. Convergence of penalty function over GA generations.

34
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on October 24,2022 at 18:02:01 UTC from IEEE Xplore. Restrictions apply.
 Net energy supplied by RES
UF u100%
Total RES production
Net energy supplied by RES
PL u100%
Total energy demand
The values UF and PL can be calculated by dividing the
value of the sixth row by the values in the fifth and last
row respectively.
UF = (267.78 / 404.16)×100% ≈ 66.25%
PL = (267.78 / 374.88)×100% ≈ 71.43%
As depicted below in Figure 8, a pie chart illustrates the
net share of each resource in both situations (Grid-Diesel
and Grid-Microgrid). Meanwhile, the penetration level
(PL) of PV array in the new configuration is marked with
a green color.
(a) (b)
Fig. 8. The net share of each power source: (a) Baseline (b) Microgrid.
VI. SUMMARY AND CONCLUSIONS
This work presents a straight-forward method to
provide an optimal layout design of a grid-connected
microgrid including PV array, ESS and Diesel asset. The
method applies a Genetic Algorithm, and it is appropriate
for buildings subject to frequent grid outages where, in
times of grid outage, the load is supplied merely by diesel
generators, which consume a huge amount of fuel.
However, with the microgrid, the annual payments of
installation and replacement count 75×103 US-$, but the
long-term annual savings will be 145×103 US-$, which
means that the net annual saving will be about 70×10 3
US-$, that is half the fuel cost in the old configuration. In
addition, a very reasonable utilization of RES can be
achieved with microgrid (UF = 66.25%). Besides, a high
RES penetration level can be achieved (PL = 71.43%).
Furthermore, it is noticed that the optimal selection of
components depends primarily on the adopted operation
policy in addition to the gross energy produced by RES
rather than on the initial prices of the components only.
Installing such RES in microgrids can be done gradually,
and expectedly, the savings will increase progressively.
Prospective work may include more sophisticated
operation policies and measure how much they can affect
the penetration level of RES or/and decrease the losses
due to power conversion.
Authorized licensed use limited to: Institut Teknologi Sepuluh Nopember. Downloaded on October 24,2022 at 18:02:01 UTC from IEEE Xplore. Restrictions apply.
REFERENCES
(15)
[1] R. Lasseter, “MicroGrids” Power Engineering Society
Winter Meeting, IEEE 2002.
(16) [2] Microgrid Exchange Group. “DOE Microgrid Workshop
Report” Aug. 2011 [Online] retrieved in Nov. 2016:
http://energy.gov/oe/downloads/microgrid-workshop-
report-august-2011.
[3] M. Hijjo, F. Felgner, J. Meiers and G. Frey, “Energy
Management for Islanded Buildings Integrating
Renewables and Diesel Generators,” 2016 IEEE PES
PowerAfrica Conference, Livingstone, Zambia, 2016, pp.
62-66.
[4] M. Hijjo, P. Bauer, F. Felgner, G. Frey, “Energy
Management Systems for Hospitals in Gaza-strip,”
Proceedings of the IEEE Global Humanitarian Technology
Conference (GHTC2015), pp. 18-25, Seattle/WA, USA,
Oct. 2015
[5] L. N. An, T. Quoc-Tuan, B. Seddik and N. Van-Linh,
“Optimal Sizing of a Grid-Connected Microgrid,” 2015
IEEE International Conference on Industrial Technology
(ICIT), Seville, 2015, pp. 2869-2874.
[6] R. Dufo-Lopez, and JL. Bernal-Agustin, “Design and
Control Strategies of PV-Diesel Systems Using Genetic
Algorithms,” Solar Energy Vol. 79, Issue 1, pp. 33-46,
2005.
[7] R. Atia and N. Yamada, “Optimization of a PV-wind-
diesel system using a hybrid genetic algorithm,” 2012
IEEE Electrical Power and Energy Conference, London,
Oct. 2012, pp. 80-85.
[8] M. Mushtaha and G. Krost, “Sizing a self-sustaining wind-
Diesel power supply by Particle Swarm Optimization,”
2011 IEEE Symposium on Computational Intelligence
Applications In Smart Grid (CIASG), Paris, 2011, pp. 1-7.
[9] S. X. Chen, H. B. Gooi and M. Q. Wang, “Sizing of
Energy Storage for Microgrids,” in IEEE Transactions on
Smart Grid, vol. 3, no. 1, pp. 142-151, March 2012.
[10] M. Aiad, A. Badran, “Optimal Selection of Hybrid
PV/Wind Systems for Jordanian Conditions,” Global
Conference on Renewables and Energy Efficiency for
Desert Regions GCREEDER, Sep. 2013 Amman, Jordan.
[11] J. Duffie, W. Beckman, “Solar Engineering of Thermal
Processes,” 4th Edition, Wiley, May 2013
[12] Approximate Diesel Fuel Consumption Chart. retrieved in
Nov. 2016; www.dieselserviceandsupply.com
[13] M. Mitchell, “An Introduction to Genetic Algorithms,”
MIT press Cambridge, MA, USA 1996.
[14] P. Kayal, C.K. Chanda, “Optimal mix of solar and wind
distributed generations considering performance
improvement of electrical distribution network,” Journal of
Renewable Energy, vol. 75, pp. 173-186, March 2015
[15] M. Hijjo, F. Felgner, G. Frey, “Energy Management
Scheme for Buildings Subject to Planned Grid Outages,”
Journal of Engineering Research and Technology, vol. 3,
no. 3, pp. 58-65, 2016.
[16] Meteonorm: Global metrological database.
[17] X. E. Yu, P. Malysz, S. Sirouspour and A. Emadi,
“Optimal microgrid component sizing using mixed integer
linear programming,” 2014 IEEE Transportation
Electrification Conference and Expo (ITEC), Dearborn,
MI, USA, June 2014
35