Energies

Type of the Paper (Article)
Modelling and Energy Management of an Off-grid Distributed

Energy System: A typical community scenario in South Africa
Adewale Obaro 1, *, Josiah Munda 1 and Adedayo Yusuff 2
1
Department of Electrical Engineering, Tshwane University of Technology;
2
Department of Electrical and Mining Engineering, University of South Africa;
* Correspondence: [email protected]; Tshwane University of Technology, Private Bag
X680, Pretoria, 0001, South Africa.
Abstract
Conventional power systems have been heavily dependent on fossil fuel to meet the increasing
energy demand due to exponential population growth and diverse technological advancements.
This paper presents an optimal energy model and power management of an off-grid Distributed
Energy System (DES) capable of providing sustainable and economic power supply to electrical
loads. The paper models and co-optimizes multi-energy generations as a central objective for
reliable and economic power supply to electrical loads while simultaneously satisfying a set of
system and operational parameters. In addition, Mixed Integer Nonlinear Programming (MINLP)
optimization technique is exploited to maximize power system generation between interconnected
energy sources and dynamic electrical load with highest reliability, minimum operational and
emission costs. Due to frequent battery cycling operation in the DES, Rainflow algorithm is
applied to the optimization result, to estimate the Depth of Discharge (DOD) and subsequently
Citation: To be added by editorial
count the number of cycles. The validity and performance of the power management strategy is
staff during production.
evaluated with an aggregate load demand scenarios of sixty households as a benchmark in a
Academic Editor: Firstname Last- MATLAB program. The simulation results indicate the capability and effectiveness of optimal DES
name
model through an enhanced MINLP optimization program in terms of significant operational
Received: date costs and emission reduction of the Diesel Generator (DG). Specifically, the deployment of DES
Accepted: date minimizes the daily operational cost by 71.53%. The results further indicate a drastic reduction in
Published: date CO2 emissions, with 22.76% reduction for the residential community load scenario in contrast to
Publisher’s Note: MDPI stays
the exclusive DG system. This study provides a framework on the economic feasibility and
neutral with regard to jurisdictional effective planning of Green Energy Systems (GESs) with efficient optimization techniques
claims in published maps and capability for further development.
institutional affiliations.
Keywords: Distributed Energy System; Photovoltaic System; Energy Storage System; Wind
Turbine; Greenhouse Gas Emission; Mixed Integer Nonlinear Programming.
Copyright: © 2022 by the authors.
Submitted for possible open access
publication under the terms and
conditions of the Creative Commons
Attribution (CC BY) license
(https://creativecommons.org/license
s/by/4.0/).
Energies 2022, 15, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/energies

Energies 2022, 15, x FOR PEER REVIEW 2 of 46
1. Introduction
Generally, continuous energy shortage and global exponential population growth
have resulted in enormous atmospheric Greenhouse Gas (GHG) emission, rapid fossil
fuel reserve depletion, environmental degradation and high operational costs related to
electrical power generation [1]. In most cases, power sectors, diesel powered energy
consumers, firms and investors do not take responsibility for releasing these enormous
GHG emissions to the atmosphere. To mitigate these critical and prevailing energy
challenges, multiple Renewable Energy Sources (RESs) can be interconnected together
forming power distribution systems for powering electrical load demand, otherwise
known as Distributed Energy System (DESs) [2]. The DESs are modular Multi-Energy
Sources (MESs) which can be installed at distribution systems to provide immediate
electrical power demand, improve power reliability, minimize operational cost, and
power losses as well as diversify energy sources. These RESs are clean, ubiquitous,
sustainable, self-replenishing and cost efficient when effectively harnessed and
efficiently optimized in meeting energy needs [3]. Nevertheless, the uncertainty and
dynamic characteristics of the intermittent Renewable Energy Resources (RERs)
especially solar radiations and wind speeds result in irregular power generation,
significant electrical power discrepancy and subsequently electrical power mismatch.
These further complicate the modelling and pose serious techno-economic limitations to
the maximum utilization and reliability of the MESs in the DESs [4]. Consequently,
Energy Storage Systems (ESS) such as Fuel Cells (FC), batteries, Hydrogen Tanks (HTs),
Super-Capacitors (SCs), flywheels, Compressed Air (CA) and Molten Salt (MS) can be
integrated as backup mechanisms to regulate power exchange, improve operational
efficiency and energy utilization for economic, sustainable and reliable operation of the
DESs [5], [6]. In addition, an optimal design, control strategy and efficient energy
optimization mechanism is required to mitigate energy mismatch between generation
and consumption inherent with intermittent and irregular power generation behavior as
well as improve energy utilization efficiency in DESs [7], [8]. Although, oversizing of
system’s components in DESs can significantly minimize reliability challenges, this often
results in unnecessary high capital and replacement costs. Hence, it is imperative to
design DESs for sustainable and economic operation capable of meeting the continuous
energy shortage through efficient energy management techniques [9], [10], [11].
Related Literature
Recently, an extensive investigation has been conducted on the design and
optimization of hybrid renewable energy systems (HRESs), Hybrid Energy Systems
(HESs) and DES involving the inclusion of either turbine, diesel generator or grid system
with the aim of providing continuous power supply. More importantly, the emission
evaluation released has been neglected in objective function formulation, especially in
analyzing the economic implication of these systems. Moreover, several techniques have
been widely explored to manage and increase energy supply of Hybrid Renewable
Energy Systems (HRESs). A Stochastic risk-dependent multi-objective assessment model
for an optimal energy exchange in a Micro-Grid (MG) system consisting of photovoltaic
system, wind turbine, energy storage and flexible loads was analyzed in [12] to address
inherent volatility and unpredictability of MG. Lin et al [13] investigated a stochastic
control of distributed energy resources (DERs) in which decentralized controllers were
used to reduce the expected cost of balancing load profile. An intelligent and self-suited
multi-agent system (MAS) based energy-coordination for decentralized control of the
hybrid electrical system was presented in [14]. The MAS comprises of manager agent
and service agents which supervise the production and various load agents respectively.
The control strategy dictates the interruption, service shift, increase and reduction of the
power demand of various flexible consumption sources on the system. The optimal
energy management of a stand-alone hybrid micro-grid which seeks to enhance energy
utilization efficiency, minimize fuel emission costs using Dynamic Programming (DP)
method was expounded in [15]. The simulation results obtained shows a reasonable
reduction in total cost of the system. In [16], the State of Charge (SOC) of the Battery
Energy Storage System (BESS) which was incorporated in the optimal control of MGs
with alternative current tie-lines, depleted to zero. This indicates an abnormal
implementation of control strategy on the battery bank, as it can drastically shortens the
lifespan of the BESS. The author in [17] presented a multi-objective technique for
modelling an isolated HRES using a Backtracking Search Algorithm (BSA) with
inclusion of a dump load. The proposed HRES model was oversized which could incur
additional capital cost and fails to take into consideration operational cost of discharging
the dump load. The meta-heuristic Firefly Algorithm (FA) was applied to regulate load
frequency of an MG design linked to a photovoltaic and thermal generators at
Maximum Power Point Tracking (MPPT) [18]. The results show that the FA-based
controllers demonstrated better improvement in terms of different indices and settling
times when compared with Genetic Algorithm (GA), under various disturbances and
parameters. In the presented system, the performance of FA is inversely proportional to
distance between the fireflies. This means that the performance of FA reduces as the
distance between the fireflies increases and thus, poses a major potential setback to this
technique. In [19], the authors introduced an efficient control network for a wind-
hydrokinetic pump-back hydropower plant (PHP) integrated with the conventional
power system to optimize energy generation of the hydro reservoir with stringent
ecological water flow using Mixed Integer Linear Programming (MILP). In the proposed
system, water regulations adversely impact downstream riverine ecosystem, aquatic
biodiversity, human settlement, and revenue generation. In addition, a Mixed Integer
Nonlinear Programming (MINLP) estimation model for Demand Response Programs
(DRPs) integrated energy hubs was analyzed in [20]. Simulation results obtained
demonstrate the viability of model presented in powering an energy hub load demand.
Furthermore, a related work presented on the Economic Emission Load Dispatch (EELD)
aimed at minimizing operating cost and emission levels of thermal generator power sys-
tems employing Symbiotic Organism Search (SOS) optimization algorithm was investi-
gated [21]. The effectiveness of the studied system in the SOS simulation demonstrated a
reduced emission values and fuel cost in relation to other existing heuristic techniques. A
solution network based on Stud Krill Herd (SKH) algorithm for optimal power flow of
conventional power systems was also evaluated in [22]. The authors in [23] and [24] also
applied Particle Swarm Optimization (PSO) method to address a multi-energy schedul-
ing problems of diverse paused loads in standalone power systems and annual costs re-
duction in radial distribution networks respectively. Although, the simulation results ob-
tained showed effective performance of the PSO algorithm, the model presented only fo-
cused on the capital cost of the system while the convergence speed was not investigated.
In an effort to reduce active power losses and improve voltage profile across a network,
[25] presented an optimal sizing and location of Distributed Generators using Strawberry
Plant Propagation Algorithm (SPPA). The results showed that the SPPA achieved a rea-
sonable voltage profile improvement and power loss minimization. Additionally, a grid-
connected and an off-grid hybrid energy systems were investigated for the electrification
and optimization of selected areas using HOMER simulation software in [26] and [27] re-
spectively. A comparative study on HOMER and in-house algorithm has shown that the
in-house model is far more flexible, efficient and economically viable than the HOMER
model, which is suscptible to inaccurate design as it failed to produce results for power
shortage in some scenarios [28]. From literature review, it is evident that existing works
failed to appropriate the amount of emission and emission costs associated with fossil
fuel consumption in various models of DESs presented [1] – [28]. Non-inclusion of both
operational and emission costs can exuberate investment and replacement costs as well
as ultimately compromise the overall energy cost. To overcome the aforementioned limi-
tations, an optimal energy model with an efficient power management technique of an
off-grid DES is proposed. The proposed DES is a cluster of hybrid power generation mix-
tures capable of providing clean, economical, and sustainable energy solution to salvage
the growing energy scarcity especially in remote and isolated areas where a conventional
power system is inaccessible and economically non-viable. The DES comprises of a pho-
tovoltaic (PV) system, Wind Turbine (WT), Micro-Hydro Power System (MHPS), Battery
Storage System (BSS), Diesel Generator (DG) interlinked via an energy network which ul-
timately powers a dynamic electrical load demand. In this DES, inclusion of the multi-
RESs is aimed at minimizing operational cost through the maximization of RESs energy
generations. Due to its remarkable capability in solving complex, nonlinear optimization
problem and efficient performance in numerical optimization, the MINLP optimization
technique is exploited to manage power exchange between interconnected energy
sources and dynamic electrical load under applicable multi-power exchange limits and
system parameters while Rainflow Counting Algorithm (RCA) is further exploited to
monitor the reliability performance status of the BSS during the sampling duration. Effec-
tiveness of the proposed DES is evaluated using a typical community load demand sce-
nario in the optimization simulation. The following are some highlights of this paper's
contributions and originality.
Novelty and Contributions
1. A comprehensive optimal capacity and efficient multi-objective optimization
solution of a DES which prioritizes utilization of renewable power generation
sources and incorporates both operational and emission costs in rural areas is
presented.
2. A MINLP optimization mechanism has been applied in solving the multi-objective
cost function of meeting load demand with highest reliability, minimum
operational and emission costs, while simultaneously maintaining an optimal
power balance and system constraints in the DES.
3. To determine the degradation assessment of the BSS model in the DES, Rainflow
counting algorithm is implemented for the estimation of charge/discharge cycle
capacity taking into consideration the dynamics of the state of charge.
4. Typical community load demand scenarios in Pretoria, South Africa have been
utilized in validating the proposed model.
5. Economic and emission cost analyses have been conducted to assess the overall
operational benefits of the proposed DES network.
6.
The proposed DES network has been implemented using realistic RERs dataset
accessible through solar energy and NASA meteorology databases.
The sections of the current paper are structured as follows. Section 2 gives a description of the proposed DES model
and computational modelling of the system’s components. In Section 3, optimization of models with necessary system
constraints and parameters are described. The case study and simulation parameters are presented in section 4.
Simulation outcomes are discussed in section 5. Section 6 presents the conclusion of the paper and recommendations
for future work.
2. Overview of CO2 Emission

The CO2 emission is a major factor responsible for the rising environmental GHG
emissions. For instance, Figure 1 illustrates a two decade overall per capita CO 2 emission
progression in South Africa commencing from 1991 to 2021 and expected value till 2022.
Out of this figure, a great proportion originated from fossil fuel combustion in the same
period as depicted in Figure 2 [29]. From the figures 1 and 2, it is evident that per capita
CO2 emission in South Africa is expected to reach 8.90 metric tons in 2022 with an
increase in fossil fuel combustions of about 477 million tons in the same year.
Per capita (Annual) Co2 Emission in South Africa

10
9
Co2 Emission (Metric tons)
8
7
6
5
4
3
2
1
0
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
Year (Two Decades)
Figure 1: Per capita CO2 emission in South Africa.

Co2 emission from fossil fuel combustion in South

Africa
600
CO2 Emission (Million tons)
500
400
300
200
100
0
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
Year (Two decades)
Figure 2: Fossil fuel consumption CO2 emission in South Africa.
2.1 System Modelling and Description

Figure 3 presents a schematic layout of the proposed DES comprising of PV system, WT, BSS interlinked with a
dynamic AC load profiles based on the need of energy consumers through a power controller, with directional arrows
representing power flow. The integration of a battery system provides for energy mismatch and load variations
recovery through injection of excess and compensation for net powers. Therefore, the dual-purpose BSS acts as a sink
to conserve surplus power for future energy consumption or need when power generation exceeds load demand and
as a source to inject net power when load demand exceeds power generation and thus, maintaining an effective power
balance operation. This is made possible through the aid of a bidirectional power flow converter (which regulates the
charging and discharging mechanisms) connected to the BSS. The total power of the RES subsystems is prioritized to
completely power the dynamic daily load demand profile and hence, unmet load demand is zero at any instant.
Conversely, energy from the BSS and/or a DG may be dispersed to meet a required net power in an event that the
dynamic load exceeds total generated power with the aid of an automatic switch. A switch is included on the DG side
as a measure to automatically connect or disconnect from the DES based on the available power to cover the load
demand. Modelling of the power electronic converters and controllers has been ignored for network system
simplicity. The mathematical modelling of the proposed DES components is presented in section 2.2.
Figure 3: A network structure of the proposed distributed energy system.
2.2 Mathematical modelling of the proposed DES components

In this section, the mathematical model of individual components is formulated to
analyze the performance operation and economic impact of the proposed DES.
2.2.1 Photovoltaic system

The power generation of a Photovoltaic (PV) system relies on the hourly solar radiation, ambient temperature and
thus, plays a crucial role in the design of an efficient PV system. Hence, the solar irradiance dependent power of the
PV module is formulated by using Eq.(1) [30]:
P pv ( τ )=P R , pv
[ ][
R
R ref
1+ N T ( T c −T ref ) ] ( 1 )
where P R , pv , P pv ( τ ) and τ represent rated power of the PV panel, generated power by the PV system and discrete
time horizon set to 1 hour in this study, respectively. Tref, R and Rref represent the reference temperature (taken as
25oC), solar radiation and reference solar radiation (1000 W/m 2) respectively. NT represents panel temperature
coefficient and equates to -0.0037 per degree Celsius for single and multi-crystalline silicon cells [31]. The temperature
of the PV panel is computed using Eq. (2):
T C =T a +
[ ]R
800
( NOCT −20 )( 2 )
where Ta, NOCT and R represent ambient temperature ( oC), normal operating cell temperature (oC) usually specified
by the cell manufacturer and solar radiation respectively. The overall power produced by the PV system is thus,
estimated as follows:
Nτ
P PV ,Total ( τ )=N pv ,cell × ∑ P pv ( τ ) ( 3 )
τ=1
where NPV represents total number of panels utilized. Figure 4 presents the daily ambient temperature and solar
radiation data obtained from NASA meteorological site Pretoria, South Africa located at a longitude 28.23 oE and
latitude 25.75oS and considered for the study. The equivalent network of a PV generator can be found in [32] while the
PV cell specifications utilized in this work are presented in Table 1.
Figure 4: A real-time daily solar radiation and ambient temperature data on a typical summer day in Pretoria.
2.2.2 Wind Generator Model

A wind turbine (WT) transforms wind’s mechanical power into productive electrical power. The mechanical power of
the WG mainly depends on the speed of the wind, which can be mathematically expressed as follows [33]:
1 3
PWG ( τ ) = A S U w C P γ ( 4 )
2
where A S∧U wrepresent the swept area of the rotor blade ( 1.735 km ) and wind speed (m/s) of the WG respectively,
2
γ represents the air density( 1.25 kg /m3 ). The characteristic non-dimensional power coefficient ( C P ) of the WT relies
on both the blade pitch angle( β ) as well as the ratio of the tip speed( λ ) . The tip speed ratio is expressed as a fraction
of blade speed tip to the wind speed according as follows:
Rblade
λ=ψ blade ( 5)
Uw
where Rblade ∧ψ blade represent the radius and rotational speed of the blades taken as 23.3 m and 3.14 rad /s
respectively. The power coefficient, CP can thus be expressed as:
C P =( 0.44−0.0167 β ) sin
[ ( πλ−3 π )
15−0.3 β ]
−0.0184 β ( λ+3 ) ( 6 )
The threshold power coefficient is approximately equal to 0.539 . However, it is crucial to note that the coefficient
value may vary based on turbine types, mechanical and aerodynamic losses. Figure 5 depicts the average wind speed
data collected from WASA.
A typical daily wind speed in Pretoria

15
Wind speed
10
Wind speed [ m/s ]
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [ h ]
Figure 5. A typical 24h daily wind speed in Pretoria.
Hence, the overall electrical output power of the WT is computed according to Eq.(7):
{
1 3
U A C γ U cin ≤U w ≤ U r
2 w S P
PW ( τ ) = Rated
( 7)
PWT U r ≤U w ≤ U co
0 Otherwise
where PWT
Rated
represents the rated electrical output power of WT; U w , U r , U cin∧U co represent
the real-time, rated, cut-in and cut-out speeds individually. The characteristic curve of the wind
speed versus the corresponding mechanical power of the WG under current study is presented in
Figure 6.
1.2 Chart Title

Rated output power at rated
Figure 6: Characteristic
speed
curve of wind speed 1
against mechanical output
power for the studied WT
Cut-out
Wind Output Power (p.u)
[33]. 0.8
speed
The number of WTs
required to power a daily 0.6
load demand is calculated
using Eq.(8):
0.4
av
PL
NWT = av
(8 )
P WT 0.2 Cut-in speed
av av
where P ∧P L WG represent
average power of load 0
demand and average 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
power of the WT Wind Speed (m/s)
respectively.
2.2.3 Micro-hydro power system Model

The micro-hydro power system (MHPS) comprises a small hydro-turbine installed in a dam or reservoir which
converts mechanical power in moving water streams to electrical power through turbine rotation, without the use of
fossil fuel. However, the turbine output power viability depends on water flow rate, surface area, head drop and
seasonal variability. These variations in water flow rate or level affect the generation capacity and thus produce
variable power supply. Therefore, the turbine can be installed along waterfall top or artificial reservoir with good
topography and high-water flow rate such as streams, and rivers with few pipes leading to a small generator housing.
In addition, the reservoir is designed to allow flexible operation by regulating water flow regimes in response to
energy demand especially during periods of reduced water flow [19].
In this study, the MHPS is designed to complement other renewable energy generator (PV and WT) especially, when
the solar irradiance and/or wind speed is at minimum. Therefore, the theoretical electrical power generated by the
MHS is computed using Eq. (9) [34]:
P MHPS ( τ )=Qr × Η w × ηconv × ρw × α g ( 9 )
Q Η η α ρ
where r , w , conv, g and w denote the water flow rate in m 3/s, total water head in meters which depends on
available water, overall energy conversion efficiency (per unit) of the turbine, gravitational acceleration (9.81 m/s2) and
water density (typically 1000 kg/m3) respectively. The generated electrical power can then be conveyed to consumers.
2.2.4 Diesel Generator Model

A diesel generator (DG) is a synchronous electro-mechanical machine which can

supply uninterrupted electrical power to load demand using fossil fuel by converting
kinetic energy in moving parts to alternating current (AC) power. In the current study,
the DG is utilized as a complementary energy generation source to provide
uninterruptible and stable power to electrical load demand in the event of power
generation deficiency, using a strategy known as Load following (LF). The active
operation costs are considered in the mathematical formulation of DG’s fuel
consumption so that the primary objective of optimizing the fuel cost would not be
compromised. Therefore, DG fuel consumption cost required to power the controllable
electrical load demand is formulated as a second-order function of the power generated
according to Eq.(10) [35]:
Nτ
FC DG ( P DG ( τ ) )=∑ ( nDG + mDG P DG ( τ ) + p DG P2DG ( τ )) (10 )
τ =1
where ni , m i , pi represent cost coefficients of DG (specified by the manufacturer) taken as 0.0246, 0.0815, 0.433 (kWh/l)
respectively [1], [36]. P DG ( τ )represents the real power supplied by the DG, measured in kW.
2.2.4.1 Evaluation of Fossil Fuel Emissions

Although DG can power electrical load reliably, its exhaust releases greenhouse gas
(GHG) during operation. According to Global Emissions Center for Climate and Energy
Solutions (C2ES), the world emits million metric tons of GHG with approximately 76%
carbon emission originating from fossil fuels consumption (energy generation) [29]. As a
result, it is imperative to measure atmospheric GHG emissions of the DG and its cost
implication as this information assists in determining or compensating the mitigating
effects and proffers suitable control strategies. The emission factor of DG GHG consists
mainly of oxides of Sulphur and carbon-dioxide discharged into the surrounding during
the combustion of fossil fuel. In the current study, the hourly DG GHG emission
released per unit of diesel fuel is computed as a product of diesel fuel consumption by
the overall emission factor as in Eq.(11) [37], [38]:
EC GHG ( τ )=PDG ( τ ) × ( EF SO + EFCO ) ∈ ( kg ) ( 11)

2 2
where P DG ( τ )is the hourly power generation by the DG, EF SO2∧EF CO2 represent individual emission factor of SO 2
and CO2 considered in South Africa as 0.4 (kg/L) and 0.951 (kg/L) respectively [39], [40], [41], [42].
2.2.5 Electrical Power Converter

A power converter is used as a bi-directional converter to convert DC to AC power. During the charge and discharge
operation, the bidirectional converter converts BSS output from DC-AC and vice versa. The capacity of the converter
depends on its efficiency and computed as follows [43]:
P co
ηConv = ( 12 )
Pn z1 + Pco
( z P
1+ 1 co
Pn )
where z 1=
(
1 10
−
1
99 η10 η100
−9 =
) (
1 1−η10 η100
11 η 10 η100 )
1
z 2=
η100 −η10−1
In Eq.(12), Pco and Pcn denote the output and nominal powers of the converter respectively; η10 ¿ η100 denote the
efficiencies of the converter at 10% and 100% respectively, specified by the manufacturer.
2.2.6 Battery Storage System (BSS) Model

The BSS is a complementary energy backup incorporated to mitigate potential
disruptive frequency deviation and dwindling power generation in DES due to the
irregular characteristics of renewable energy resources (for instance the wind speed,
solar radiation), season and time of the day as well as uncertainty in net power
requirements. In this study, the investment cost is minimized through the optimal sizing
of the BSS (in kWh). The capacity of the BSS (kWh) is estimated as:
D SS × E L
C BSS = ( 13 )
DOD ×ηtemp
Where D SS , E L , DOD∧ηtemp represent the days of self-sufficiency, average daily
energy demand ( kWh/day ) , depth of discharge and temperature correction factor of the
BSS respectively.
The instantaneous state of charge is a dynamic linear function which depends on its nominal capacity and net energy
balance, evaluated as follows [6], [36]:
( )
Nτ
ηD Nτ
SoC BSS ( τ +1 )=SoC BSS ( τ ) + ηC ∑ Pinj ( τ )− ∑ P ( τ ) x δ Δ τ ( 14 )
τ =1 E Nom τ =1 dis
where SoC BSS ( τ )∧SoC BSS ( τ +1 ) is the predetermined and current SOC respectively; ηC ∧ηC are the charging and
discharging efficiencies of the BSS expressed as 90% and 60% respectively; Pinj ( τ )∧Pdis ( τ )is the hourly surplus and
net active powers injected into and discharged from the BSS during charge and discharge operations respectively;
E Nomis the nominal energy capacity; ∆ τ represents sampling time horizon; the positive and negative sign conventions
indicate charging and discharging mechanisms respectively.
Equation (14) implies that the battery energy level (SOC) at any instant depends on the initial energy level and the
amount of power conserved into or discharged out of the BSS during the current time horizon. Thus, the current SOC
becomes:
Nτ
δ Δ t η D Nτ
SoC BSS ( τ )=SoC BSS ( 0 ) +η C ∑ Pinj ( τ )− ∑ P ( τ ) (15 )
τ=1 E Nom τ=1 dis
The injected and discharged power indicate the charging and discharging
mechanisms of the BSS, which is computed based on the net load demand according to
Eq.(16) and Eq.(17) respectively:
(
Pinj ( τ )=
PW ( τ )+ P PV ( τ ) + P MHPS ( τ )−P D ( τ )
μ Conv )
if PW ( τ ) + P PV ( τ ) + P MHPS ( τ )−P D ( τ ) ≥0 ( 16 )
Pdis ( τ )=
( P D ( τ ) −( PW ( τ ) + P PV ( τ ) + P MHPS ( τ ))
μConv )if PW ( τ ) + P PV ( τ ) + PMHPS ( τ ) −P D ( τ )< 0 ( 17 )
where PWT ( τ ) , P PV ( τ ) , P BG ( τ )∧P MHP ( τ ) are power flows from the WT, PV, and MHPS respectively. μConv represents
the efficiency of the AC converter bus. The following criterion is used to calculate the minimal state of charge:
min max
SoC BSS =SoC BSS ( 1−DOD )( 18 )
where DOD represents depth of discharge of the BSS, indicated as a percentage, it can also be considered as extent of
discharge (EOD) or intensity of discharge (IOD). The design parameters considered for the lead-acid BSS under the
current study are given in Table 1.
2.2.6.1 Total Battery Cost

The overall operational cost of the battery defined as the sum total of electrical energy
cost injected into the battery and approximated degradation cost of the battery and it is
expressed as follows [44]:
Nτ
C Total = ∫ Pinj ( t ) C e ( τ ) dτ +C D ( 19 )
BSS BSS
τ=1
where Pinj ( t ) ,C e ( τ ) , τ represent the net power injected into the battery, electrical energy cost, charging time. Note
that C e ( τ ) depends on the current energy cost, Pinj ( t ) may either be negative or positive indicating discharging or
charging operation respectively.
The degradation cost of BSS is a function of its lifespan degradation due to its charge cycle. It can be expressed as
follows:
∆L
CD = × C BSS ( 20 )
BSS
L
where ∆ L∧Lrepresent the change in charge cycle lifetime degradation and complete lifespan of the battery (BSS)
provided that the charge cycle is regularly used until the end of lifespan (EOL) of the BSS respectively. C BSSrepresents
the investment cost of the BSS. The BSS lifespan can be maximized through the minimization of degradation cost,
CD .
BSS
3 Mixed Integer Non-Linear Programming Optimization Energy Model

The energy optimization of the DES network model presented in section 2 is investigated using MINLP technique.
The MINLP co-optimizes energy production and minimizes energy utilization through maximization of the
renewable energy sources in the DES while simultaneously maintaining applicable multi-parametric power
constraints and system parameters.
3.1 MINLP Optimization Algorithm

In this study, a multi-objective operational cost function with the aim of minimizing utilization of DG and GHG
emission costs through the maximization of RESs power generation in the DES is formulated as a MINLP
optimization problem with dependent discrete and continuous control variables. The MINLP is a robust and
computationally efficient optimization method capable of prioritizing RESs in meeting load profile with fast
convergence speed. In this paper, the MINLP is implemented in MATLAB program using “Intlinprog” solver. The
combination of the continuous and binary control variables forms the mixed integer nonlinear constraints. The
MNILP algorithm is expressed in a generalized canonical form as follows [45]:
{
Req x=qeq
Minimize J ( x ) ,
Subject ¿ Rineq x ≤ qineq ( 21 )
¿
x
lb≤ x ≤ ub
x i ∈ Dsw
where the nonlinear objective cost function is denoted by J(x).

The equality coefficient matrices are denoted by Req x and q eq.
The inequality coefficient matrices are denoted by Rineq x and q ineq.
The lower and upper nonlinear control variables are respectively denoted as
lb and ub.
Discrete binary control variable is denoted by D sw .
3.2 Objective Problem Formulation
The current study aims to minimize daily operational costs, consisting of fuel consumption and emission treatment
costs incurred by DG through the maximization of renewable power generation while providing reliable power
supply to the load demand in the DES. The emission cost relates to the monetized environmental emission associated
with the hourly output power produced by the DG. Daily energy demand is supplied primarily from the WT, PV and
MHPS generators while excess power is conserved into the BSS for subsequent energy requirement. The DG is only
engaged to power or complement net power demand in the event of power production deficiency. The daily
operation cost of WT, PV system and BSS is assumed zero in this paper. Thus, the multi-objective function is
formulated in Eq. (22) as follows:
{ }
Nτ Nτ Nτ
Ψ min =Min C D ∑ FC DG (P DG ( τ )). D sw + ∑ EC GHG ( τ ) × C d−∑ P R ( C R ) ; ∀ R ∈ WT , PV , BSS ( 22 )
τ=1 τ=1 τ=1
where C D ∧D sw represent diesel cost price in $/l and discrete state of the DG respectively; C d represents the
decontaminant cost by DG, taken as 0.0091$/kg; FC DG∧EC GHG represent the fuel consumption and emission costs
released by the DG respectively. The multi-objective cost function is then subjected to the following operational and
decision constraints to evaluate its feasibility:
(i) Power Equality limits:
By applying Kirchhoff’s current law (KCL) to the radial network, the total power generation injected from the DES
must be equal to the total power outflowed to the electrical load demand, losses and BSS at any time horizon τ . Thus,
the equality constraint is expressed through the power balance equation as follows:
Nτ Nτ Nτ Nτ Nτ Nτ
∑ PWT ( τ ) +¿ ∑ P Pv ( τ ) +∑ P MHPS ( τ )=∑ P L ( τ )+∑ P Loss ( τ ) +∑ P BSS ( τ ) ; ( 24 ) ¿

τ=1 τ =1 τ=1 τ=1 τ=1 τ=1
where s is a binary decision variable controlling the charging or discharging operation modes of the BSS and P Losses ( τ )
represents the power losses during distribution to consumers . However, the radius of consumption to the generation
considered in this study is less than 500m. Hence, the power losses are negligible and thus, ignored. It is worthy to
note that the unmet energy equates to zero as the combined generating units are capable to meet the load demand
completely.
(ii) Power generation inequality constraints
Technically, output power produced by each controllable power source cannot exceed a predefined minimum and
maximum boundary limit at any specific operation horizon of the DES. Hence, the continuous hourly output power
from each power generation source k should be kept within the lower and upper permissible limits expressed as in
Eq. (25):
Pk ≤ Pk ( τ ) ≤ P k ; ∀ k ∈ { MHPS ,WT , PV , DG } ( 25 )
min max
min rated
where Pk ∧Pk is the respective minimum and maximum set points of WT, PV, MHPS and DG. “k” represents
control variables.
(iii) Design variable constraint, Node power constraint

The electrical power produced by each RES must be less or equal to the power dispensed to the load demand and
power conserved into the BSSS as represented in Eq.(26):
P M ( τ ) ≤ P L ( τ ) + PBSS ( τ ) ; ∀ M ∈ {WT , PV , MHPS } ( 26 )
(iv) Absolute power generation constraint

The power generated by each power generating units in the DES must be equal to or exceed zero as defined in Eq.(27):
P N ( τ ) ≥0 ; ∀ N ∈ { MHPS , WT , PV , DG } ( 27 )
(iv) State of charge (SOC) inequality boundary:
The activity of the BSS, known as state of charge or state of energy is constrained within an acceptable minimum and
maximum limits to preserve its lifespan and obtain an optimal feasible solution, defined according to Eq. (28):
min max
SoC BSS ≤ SoC BSS ( τ ) ≤ SoC BSS ( 28 )
(v) Binary Switch state control variable

The ON/OFF operation of the DG is determined by a binary state variable, defined as follows:
Dsw = {1 , 0 } ( 29 )
where D sw indicates the ON or OFF state variable of the DG.
(vi) Charge and discharge constraints:
The charge and discharge powers into and out of the BSS shall not exceed the rated power capacity of the BSSS.
{
P inj ( τ ) P min Rated
BSS ≤ Pinj ( τ ) ≤ PBSS
PBSS ( τ ) = P ( τ ) P Rated ≥ P ( τ ) ≥ Pmin ( 30 )
dis BSS dis BSS
0 Elsewhere
0 ≤ PBSS ( τ ) ≤ P BSS ; ∀ k ∈ { charging∧discharging modes } ( 31 )

max
where PBSS ( τ ) is the controllable injected and discharged power during charge and discharge mechanisms. The
Pinj /P dis ( τ )are the controllable injected and discharged powers representing the charging and discharging activities
respectively.
(vi) Charge and discharge power flow constraint

At any specific horizon, the BSS is permitted to either operate in the charging or discharging mode. Thus, the power
injection and power discharge into and out of the BSS during the charging and discharging operation mechanism occur
independently, and not simultaneously as in Eq.(32):
PBSS ,∈¿ (t ) × P BSS ,out

( t ) =0 ;∀ PBSS ,∈¿ ,P (t )≥ 0 (32 ) ¿ ¿
BSS ,out
3.3. Renewable Energy Factor

To determine the viability and economic impact of the proposed DES, it is crucial to calculate the amount of electrical
power which originates from the renewable power sources. This can be referred to as Renewable Energy Factor (REF).
The REF is therefore expressed as the proportion of total power supplied by a combination of WT, PV generator and
MHPS to the load demand over a specific period, expressed in percentage as follows:
REF ( % )=¿
where PWT ( τ ) , P PV ( τ ) , P MHP ( τ )∧P BSS ( τ )represent the overall daily electrical power supplied by the WT, PV, MHPS,
and BSS respectively. P LS ( τ ) is the total electrical load served (kWh/day).
3.4 Estimation of BSS Degradation Using Rainflow Algorithm

In a DES, accurate cycle estimation of the battery is crucial because it ensures proactive monitoring of its usage
capacity and reliable health status during operation [46], [47]. Therefore, the current study considers the impact of
frequent charge and discharge dynamics on capacity degradation of the BSS under RCA. The RCA is a cycle counting
evaluation method capable of determining the performance reliability of BSS [48]. Generally, a battery should be
replaced once its capacity diminishes by a specific proportion as the nominal reliability of the battery cannot be
guaranteed [49]. In this study, the RCA is implemented to estimate daily cyclic degradation of the battery system
during the cycle charge-discharge operation modes. The RCA uses the SOC data as input to estimate how many cycles
are present at each sampling time. The computed charge capacity loss is analyzed to estimate the optimal
performance of the battery technology. A flowchart of the MINLP incorporated with the RCA is presented in Figure 7:
Figure 7: Flow chart of the proposed MINLP optimization technique for the energy management of the DES.
3.5. Modelling of Electrical Load demand

A load demand modelling which ensures sustainable power flow to electrical load demand in an economic and
reliable manner, is essential for effective operational decision making and optimal planning in the DES. In this paper,
typical daily load profiles measured for different households through South African electric power utility (ESKOM)
prepaid meters were used to model time-varying load profile for sixty households representing an aggregate number
of houses in a typical South African community using Normal Distribution (ND) technique. The ND is a probability
density function (PDF) suitable to provide a fitting for the collected residential load profiles, given as in Eq.(34) [17]:
[ ] ( 34 )
2
− (ld ( τ ) −μ )
1 2
f ( ld ( τ ) )=
2σ
e
√2 π σ 2
The average load profile at each hour is calculated by the probability of every possible events during a particular time
frame, which is expressed in Eq.(35) as follows:
N ld
P = ∑ Pmax
τ
ld ld x P L ( ld ( τ ) ) ( 35 )
st=1
where N ld denotes the maximum state number of the load profile; σ ∧μ represent the hourly standard deviation and
mean of the load profiles. Pld ( ld ast ) represents possible outcome of load demand determined for every state within a
τ
sample space as follows:

ld
maxst
P L ( ld ( τ ) ) = ∫ f ( ld ( τ ) ) dl ( 36 )
minldst
4. Model Parameters
Table 1 presents the technical and economic simulation parameters comprising of system configuration and
operational settings used in the simulations.
Table 1: Simulation technical specifications.
Parameters Symbol Value Unit
Simulation duration Nτ 24 hours

Sampling time τ 30 minutes
Wind turbine generator parameters
Rated power of turbine PWT ,Rated 400 kW
Swept area of the rotor blade AS 1.735 km
2
3
Air density γ 1.25 kg /m
Rotational speed of blade ψ blade 3.14 rad/s
Radius of blade Rblade 13.3 m
Pitch angle of blade β Degree (o)
Rated speed Ur m
Cut-in speed U¿ m
Cut-out speed U out m
Battery storage system parameters
Nominal capacity of battery E Nom 80 kWh
Nominal voltage V 52 V
Current Capacity Ah 1600 Ah
Depth of discharge DOD 80 %
Charge/Discharge efficiencies ηC /η D 90 / 60 Dimensionless
Minimum state of charge min 30 %
SoC BSS
Initial state of charge SoC BSS ( 0 ) 40 %

Maximum state of charge max 95 %
SoC BSS
Capital cost (excluding installation cost) Cs 4316.13 kWh

Photovoltaic cell parameters
Rated power of the cell P R , pv 20 kW
Reference solar radiation Rref 1000 W/m2
o
Reference temperature Tref 25 C
Temperature coefficient NT -3.7 x 10-3 o
C-1
o
Normal operating cell temperature NOCT 20 C
Sum of PV panels interlinked in series Ns 12 Dimensionless
Sum of PV cells interlinked in parallel Np 8 Dimensionless
Diesel Generator parameters
Cost Coefficients a, b, c 0.01840, 0.2088, 0.433 L/kWh
Rated power capacity Rated 1000 kW
P DG
Power factor pf 0.85 Dimensionless

Diesel cost (price) CD 1.12 USD/L
Emission factors EF SO / EFCO
2 2
0.4 / 0.951 (kg/L)
Micro-hydro power parameters
Water flow rate Qr m3/s
Total or net water head Ηw 10 m
Overall energy conversion efficiency of turbine η conv 60 %
Water density ρw 1000 kg/m3
Gravitational acceleration αg 9.81 m/s2
4.1 Description of Load Profile Scenario

This section presents an outline of a dynamic residential daily load scenario used in the simulation and analysis of the
efficiency of MINLP optimization method presented in section 3. Typical electricity consumptions of three residential
households in Mooikloof Ridge, a suburb of Pretoria, South Africa are considered as scenarios in this study. Each
selected household comprises of various physical and contextual characteristics with respect to design, size, age, and
thermal comfort leading to varying weekly and weekend daily demand load profiles as shown in Figure 8. The
households are regular residential buildings most of which possess basic electrical appliances such as lights,
computers, fans, refrigerators, TVs and washing machines. The weekdays and weekend daily load profiles are taken
directly from Impact meters installed at each of the three residential building by Impact meter services, an authorized
electricity distribution company approved by ESKOM. The daily load profile is divided into 24 hours with an interval
of 30 minutes sampling time. It should be noted that the load profiles are similar in both weekdays and weekends
except with a higher energy consumption proportion during weekends. These daily load profiles have similar energy
demand characteristics with a relatively low and constant demand between 0:00 – 5:00 due to inactivity of most
residents, slight surge and drop in demand between 5:000 – 9:00 resulting from activities of most residents as they
prepare for daily activities and work, maintain relative constant values between 9:00 – 17:00. A progressive and
prolonged increase in electricity demand is observed between 17:00 – 22:30 due to a peak activity of energy usage by
electricity consumers powering ON most electrical gadgets such as cooking appliances, thus leading to a significant
spike in load demand at this time range. The measured daily energy consumption profiles are then incorporated to
the statistical model presented in Section 3.4 to obtain an aggregate daily load profile for sixty residential households
(buildings), representing a typical dynamic load profile for a remote community as shown in Figure 9. In this Figure,
the daily electricity consumption ranges between 38.45 ~ 418.06kWh. The load demand is relatively constant early in
the morning indicating resting activity, when only basic loads are required, steadily rise when residents prepare for
work activities and progressively drop when the residents leave the house for work until it reaches and maintain a
uniform load demand throughout the daytime. As residents return from work, the load demand rapidly increases
again as various electrical appliances are powered ON. Due to the focus of this paper, the community daily load
profile is then considered as a load demand case study to evaluate the impacts of the DES and compared to an
exclusive DG supply of load demand, intended to optimally service the energy demand regardless of the amount of
renewable power available, economic implication and greenhouse emission factor considerations. The daily operation
mechanism of the studied DES is examined through the implementation of the MINLP optimization simulation by
computing the power supplied by each power generation sources to assess their individual performance during the
24h duration. The MINLP minimizes the daily operational cost of the system in terms of fuel consumption through
the maximization of RESs power generation while simultaneously ensuring reliable supply or satisfaction of the
dynamic load demand. In addition, the optimum operational cost and greenhouse gas emission cost of both scenarios
(DES configuration and DG sole system) of electricity production under investigation is obtained and analyzed.
Efficacy of the proposed DES model at minimizing operational costs, conserving energy, and emission reduction is
demonstrated using the numerical results obtained in the next section to assess the economic feasibility and adequacy
of the efficient MINLP optimization technique.
7
Household 1 [ Weekday ]
Household 2 [ Weekday ]
6 Household 3 [ Weekday ]
Household 1 [ Weekend ]
Household 2 [ Weekend ]
5 Household 3 [ Weekend ]
Load demand [ kWh ]
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [ h ]
Corresponding color names: Blue, Orange, Purple, Teal, Green, Magenta, Red
Figure 8: Daily household electricity consumption variation curve for three different
households daily load profiles during weekdays and weekends in Pretoria.
400 Aggregate Load Demand [ P L ]
350
300
Load Profile [ kW ]
250
200
150
100
50
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [ h ]
Figure 9. An aggregate daily energy demand profiles of a typical remote community.
5. Simulation results and discussion

A discussion and analyses of the simulation results obtained for the proposed DES network using an MINLP
technique is presented in this segment. The simulation is performed on a computer fitted with Intel Core i5 processor,
8GB RAM, 500GB hard drive capacity. The DES comprises of five energy sources which include WG, PV, MHPS, DG
and BSS. The MINLP aims to regulate optimal energy flow from energy generation sources in the DES to electrical
load demand as well as minimize daily operational and emission costs while satisfying various system parameters
and operational limits. Efficacy of the MINLP method for the DES is evaluated in powering the load demand
elucidated in section 4 and contrasted with an exclusive usage of DG system in powering the electricity demand
profile.
5.1 Analysis of daily renewable based multi-energy generation for the DES model
Figure 10 displays the daily power generation analysis of the various RES power mixtures in the DES which include
WT, PV, MHPS generators. The PV turbine progressively produces electrical power (indicated by colour yellow)
between 5:45 – 17:30 until it attains its peak power generation of concisely 332kWh at 11:30 and reaches its lowest
generation of approximately 12kWh at 17:30. During the PV power generation horizon, sufficient power which
optimally covers the load demand is produced by the multi-power sources while surplus energy is conserved into the
BSS according to Eq.(16). The sufficient power generation of the PV system is greatly influenced by the intensity of
solar irradiance and ambient temperature. Similarly, the wind energy generation results from the wind speed and the
rated capacity of the wind generator. Hence, the BSS operates in charging state and maintaining an optimal energy
balance operation. Similarly, the colour brown in Figure 10 illustrates the WT power generation. Due to the
availability of sufficient wind speed, the WT generates energy throughout the day. A relatively average wind energy
of nearly 150kWh is produced between 0:00 – 5:00 which complements the MHPS power generation, with its lowest
power generation between 5:00 – 12:00 while its optimal generation of 310kWh is attained between 16:30 and 18:30 at
the interval of maximum load demand. In Figure 10, the purple colour represents the power generation by the MHPS
generator. Importantly, the MHPS did not service the load demand during most hours of the day due to the optimum
power generation capability illustrated or displayed by both the WT and PV turbines. The MHPS is engaged to
contribute power generation in the event of inadequacy by the PV and WT in servicing the load demand optimally.
Generally, the electrical power produced by the renewable energy sources is inherently linked to the renewable
energy resources in the selected area.
350
P PW
P PV
300
P MHS
Electric output power [ kW ]
250
200
150
100
50
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
Time [ h ]
Figure 10: Power production of renewable based multi-energy sources.

5.2 Evaluation of Diesel power generation and its operational efficiency mode
Figure 11 presents the highlights of diesel power generation as well as its operational efficiency for singly operated
DG and the proposed DES model to power the load profile. An auto-starting mechanism is adopted for the integrated
DG as a complementary energy generation source using a net load following dispatch strategy. Using the MINLP
technique, the DG operates in a well-controlled and economical mode powering electrical load demand through a net
load monitoring approach, with at least its minimum operational efficiency during the periods of energy insufficiency
or deficiency in order not to compromise its overall efficiency. A typical minimum operating efficiency of the DG
proposed in [1] is adopted as a benchmark for reliable lifespan and economic operation, indicated by purple colour
while the maximum operating efficiency is indicated by the red colour provided that the DG operates at its highest
power capability at any sampling time. Between 0:00 – 18:30, the DG supplies zero power to the load demand due to
the sufficient power generation by the complementary combination of PV, WT, MHPS power sources as well as the
BSS. The enormous RERs available for the selected area and optimal capacity of each power source are obviously
responsible for the sufficient energy generation during the sampling time. Moreover, the DG is operated at its
minimum operational efficiency, suppling electrical power to the net energy demand between 18:30 – 21:30 during
which the load profile exceeds total power generation of the hybrid power sources and stored energy. The minimum
operational efficiency ensures stable, reliable, and economic operation mode of the DG. From 21:30 – 24:00, DG
supplies zero power to the load demand due to a decrease in load demand and power generation by the WT as well
as conserved energy into the BSS. In any case where the DG does not supply any electrical power to the load, the
operating efficiency becomes zero.
1000 100
Rated DG Power Output at Maximum Operational Efficiency
900 DG Output Power Flow [ Exclusive System ] 90
DG Supplied Power [ Proposed DES Using MILP ]
Diesel Generator Output Power [ kW ]
800 Minimum DG Output Power at Minimum Operational Efficiency 80 DG Operational Efficiency [%]
700 70
600 60
500 50
400 40
300 30
200 20
100 10
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [ h ]
Figure 11: Analysis of power generation by DG power supply and operational efficiency.
5.3 Analysis of the State of Energy of the Battery Storage System

The activity of the BSS is presented as a reflection of the state of energy in Figure 12. The highest and the lowest level
of charge are set at 90% and 30% respectively for economic and technical reasons. Due to the relatively constant
energy demand in the morning (between 0:00 – 5:00) and sufficient power production by the MHPS and WT, the state
of energy maintains a relatively constant level during this period. As the energy demand increases between 5:00 – 8:00
while the wind power generation becomes low and the MHPS produces zero power, net energy demand is obtained
from the BSS and the state of energy drastically decreases until adequate electrical power is produced by the PV
system at 9:00. Then, the state of charge of the BSS starts rising significantly or considerably due to the conservation of
net energy produced until the peak state of charge is attained at 12:00 during which lowest amount of wind power
and highest PV energy are produced. Coincidentally, the intensity of solar radiation and wind speed obviously
determines the amount of power during the 24 – h duration. Thus, the BSS operates in discharging mode according to
Eq.(17) by discharging its conserved energy to compensate the net load demand in any event where the total power
generated from the RESs in the DES proves insufficient to optimally service the load profile completely. It is
concluded that the capacity of the BSS determines that amount of stored surplus power from the multi-energy sources
and subsequently, the electrical power that can be drawn from the BSS to cover the net energy demand.
Figure 12: State of Energy of the BSS for the DES model.
5.3.1 Degradation assessment of State of Energy using Rainflow Algorithm

In this study, the RCA estimates the useful life period between the load points. The RCA carefully tracks the daily
utilization of the BSS operation and its corresponding degradation lifespan. This is demonstrated by the number of
cycles extracted from state of energy (SOE) profile as shown in Figure 13. Thus, it is imperative to compute the overall
capacity loss over every half cycle during the 24h duration. To achieve the stated objective, the SOC profile obtained
from the implementation of MINLP technique is used as input to the RCA. This defines analogous half and full cycles
which then pairs local SOC minima and maxima limits. Overall, five full/complete cycles (black, cyan, magenta, blue
and green) and four half cycles (red, black, magenta, and cyan) are detected by the algorithm representing
degradation effect on the BSS. This shows good health status, as the higher the number of cycles obtained, the lower
the reliability performance of the battery. A full cycle is represented by pairs of up and down half-cycles. The number
of cycles obtained have an adverse impact on the degradation of the battery incremental lifetime and subsequently, a
reduction in the actual capacity of the battery. A detailed procedure for the computational analysis of the RCA is
presented in [46] [48]. By optimally lowering the SOC cycle depths, which directly depend on the amount of power
supplied to or needed from the battery, it is possible to reduce the battery's cyclic capacity loss and the associated cost
of battery deterioration.
Figure 13: Estimation of charge-discharge cycle extraction of the BSS using state of charge
profile.
5.4 Power contribution of multi-energy generators to daily load profile

Figure 14 illustrates the actual power flow from individual components making up the DES to the energy demand
profile. Apparently, the renewable power sources (WT and MHPS) supply the energy demand between 0:00 – 05:00
due to the unavailability of solar radiation. Considerably, the power produced by the PV and wind generators is more
than sufficient to power the total load demand during the day as a result of available wind speed and solar irradiance.
Thus, these renewable energy generators contribute a significant proportion of their overall power generations to the
load demand while the excess power is conserved into the BSS using the MINLP optimization method and thus,
maintaining an efficient energy balance scenario in the DES network. During these operating hours, the MHPS
contributes an infinitesimal energy to the load while the DG does not participate in energy contribution to the load
requirement until sunset when PV power generation depleted to zero. The MINLP optimization technique clearly
utilized a multistage power sharing operation mechanism for the DES network model between its power sources with
each energy sources contributing or allocating a faction of its overall electrical power produced to the load profile
subject to availability of resources and applicable system constraints. In overall operation scenario, the BSS acts as a
storage sink conserving excess power generated and aids to improve the techno-economic performance feasibility of
the DES by complementing the energy dispatched to the load demand in the event of inadequate power generations.
In Figure 14, the sink action of the BSS is demonstrated by the charging mechanism indicated by the upward
movement of the power while the downward power movement represents the discharging operation. The results of
each energy component are further aggregated to provide a clear performance overview of the impact of the DES
network configuration under the current study and depicted in Figure 15. As depicted in this figure, the PV system
contributes the largest proportion of 45% while other energy sources such as the DG, WT, MHPS and the BSS
contribute 23%, 17%, 4% and 11% to the total energy consumption during the dispatch operation horizon. The least
energy proportion of the MHPS results from its minimal participation and rated capacity in energy generation
process. Summarily, the REF amounts to 77% of the power supplied to energy demand. Ultimately, there would be
unavoidable changes in operational cost and emission released as the number of households increases in near future
as this would result in higher energy consumption/demand.
Actual Power Contribution to Load Profile

P PV
Active Power Supplied to Load Demand [ kW ]
300 P PW
P MHP
250
P DG
200 P BSS
150
100
50
-50
-100
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Time [ h ]
Figure 14: Actual energy contribution to load demand profile by individual power
sources.
Contributing Factor of Subsystems

PV
BSS: 11%
WIND
MHP
DIESEL
BATTERY
PV: 45% DG: 23%
MHP: 4%
PW: 17%
Figure 15: Net power provided by each energy source to the load profile under MINLP.
5.5 Economic operational cost analysis of the proposed DES under two scenarios
An overall daily operational cost analysis is illustrated in Table 2. In evaluating the daily operational cost of the
system, two scenarios are considered which include (i) the use of exclusive DG system without the application of
other complementary energy sources and (ii) the utilization of the proposed DES model which include DG in
powering the load demand under a load following operational strategy. In this study, the daily operational cost
considered include diesel fuel consumed by the DG and GHG emission costs incurred in burning the fossil fuel. The
GHGe cost is a polluter-pays tax designed to reduce greenhouse gas emissions economically and sustainably by
holding companies and consumers accountable for the negative external costs associated with their production and
consumption [50][51][52][53][54][55][56].
As shown in Table 2, the total operating cost incurred in powering the total load demand by the single DG system is
$1780.26 while the total emission cost produced accounted for 5340.80 kg/kWh. Under the proposed DES model, the
DG system operates in restricted mode to satisfy excess load demand and thus, its power supply results in $509.63
operational cost. Overall, the DES considerably accounts for 77% of the total power supplied, providing
approximately $1308.17 savings in energy cost. It is worthy to note that the available RERs in the studied area, the
operational load following strategy and the efficient MINLP scheme contributed to the significant reduction in the
overall operational costs. With the MINLP energy optimization technique, the overall operational and GHG emission
costs amount to 71.53%, which shows a significant reduction in overall daily power generation costs. The MINLP
provides optimal solution that leads to reliable power supply, fast convergence speed and considerable energy cost
savings. The results presented in the table obviously shows that the DES gives best outcomes using MINLP
optimization technique compared to the DG exclusive system.
Table 2: Daily Operational and Emission Cost Analysis of the DES model under study.
Modes Fuel con- Diesel con- GHG Emis- GHG Emis- Total cost REF (%) Conver-
sumption (L) sumption sion sion Cost ($) (Fuel cost + GHG gence
cost ($) (kg/kWh) Emission Cost) Speed
(Seconds)
DG Exclusive 1156.10 1780.26 5340.80 48.60 1828.86 0 6.09
System
Proposed 330.93 509.63 1215.90 11.06 520.69 71.53 9.97
DES
Savings in 825.17 1270.63 4124.90 37.54 1308.17 71.53 3.88
incurred cost
By comparing the daily operational and emissions costs of the proposed DES model and
the exclusive DG system, we found that both the operation and emission cost essentially
depend on a number of salient factors, which are itemized as follows:
1. Optimal capacity of the multi-energy generators making up the DES
network.
2. Availablity of RERs and subsequentluy the RESs output power.
3. Proficiency of the optimization methodology adopted.
In general, the DG contributes substantially a huge portion of GHG emission with
significant lower energy generation.
5.7 Environmental evaluation of greenhouse gas emission
An analytical environmental assessment of GHG emission for the DG exclusive system and proposed DES network
configuration under the MINLP energy optimization technique is presented in Figure 16. The GHG emissions released
by the DES model is evaluated by multiplying a specific emission factor per total energy consumption with the
electrical load supplied by the DG and contrasted to the scenario of exclusive application of DG in powering the daily
load profile. The emission factor considered in this paper is taken as 0.62 kg/kWh [48] and adopted for the evaluation.
In this study, the proposed DES under the MINLP method produces siginificant
low GHG emissions due to its minimum efficiency, limited operational time frame and
minimum output power supplied during operation. Contrarilly, the total emission
released by the exclusive DG system is considerably high as a result of its inherent
maximum operation efficiency and total power supply to the load requirements.
Specifically, the DG system independently powering electrical load profile released
approximately 5340.80 kg/kWh GHG emissions, consisting of 70.4% Carbondioxide
(CO2e) and 29.6% Sulphurdioxide (SO2E). Conversely, the proposed DES emit an overall
1215.90 kg/kWh GHG emission comprising 70.4% Carbondioxide (CO 2e) and 29.6%
Sulphurdioxide (SO2E). Summarily, the DG exclusive system propagates about 4124.9 kg
GHGe (77.23%) GHG emission more than the proposed DES. The high level of GHGe
released poses great environmental consequence to the climatic atmospehric condition
which can endanger both human and animal lifespan.
A Comparison of Daily Greenhouse Gas Emission Propagation with and without RESs
Overall DG-GHGe CO2 e SO 2 e
6000
5340.80
C0 2e -70.4%
5000
GHG Emission Level [ kg/kWh ]
4000
3000
2000
S0 2e -29.6% 1215.90
1000 C0 2e -70.4%
S0 2e -29.6%
0
GHGe of Exclusve DG System without RES GHGe of the Proposed DES with RESs
Figure 16: Comparison of greenhouse gas emission for the exclusive DG system and DES
configuration.
6. Conclusion
The current paper primarily focuses on the optimal design and energy management
technique of an off-grid DES configuration comprising of multi-energy generators for
reliable supply to dynamic daily energy demand of a typical remote community. The
paper formulates multi-objective cost function as a convex MINLP aimed at minimizing
fuel consumption and emission costs through the maximization of RES power
generators in the DES while reliably satisfying load demand requirement as well as
meeting a set of operational constraints and system parameters. The MINLP technique
efficiently regulates optimal active power flows schedule to minimize operational cost
and achieve fast convergence speed. To prevent premature degradation in charge

capacity of the BSS during usage, RCA was employed to accurately determine the
number of charge/discharge cycles and monitor its health or performance status.
Performance of the MINLP energy management technique was tested under a typical
community daily load demand profile. The simulation results of the current study
shows that the exploited MINLP technique is efficient at minimizing both operational
and emission costs with 71.53% and 22.76% reductions respectively for the residential
community load scenario in contrast to the exclusive DG system. Future research is
expected to focus on overall incorporation of uncertainties synonymous with RERs into
DES formulations.
Author Contributions
AZ Obaro: conceptualization, investigation, methodology, validation, software, visualization,

writing original draft, data curation, formal analysis, writing review and editing, resources; JL
Munda: Supervision, writing review and editing, project administration, data curation, resources,
funding acquisition; AA Yusuff: Supervision, writing review and editing, data curation, project
administration, resources. All authors have read and agreed to the published version of the
manuscript.
Funding
The authors acknowledge the funding provided from Tshwane University of Technology, South Africa.
Data Availability Statement

The article includes the original contributions made and presented in the study. The
corresponding author should be contacted for further inquiries.
Acknowledgments
Not applicable.
Conflicts of Interest
The authors affirm that there is no potential dispute regarding the authorship and findings of this paper.
Nomenclature and Interpretation
Symbol Interpretation
AS Swept area of the rotor blade

BSS Battery storage system
β Blade pitch angle
CP Power coefficient of the wind turbine
DES Distributed energy system
DG GHG DG GHG emission
DG Diesel generator
EF SO / EFCO /EF NO
2 2 2
Emission factor of carbon, sulphur and nitrate oxides
E Nom Nominal energy capacity
FC DG Total diesel fuel consumption
GHG Greenhouse gas
Ηw Total water head
MINLP Mixed integer nonlinear programming
NOCT Normal operating cell temperature
NPV Number of panels
ηC /η D charge and discharge efficiencies of BSS
η conv Conversion efficiency (per unit) of the turbine
ni /mi / pi Cost coefficients of DG (specified by the manufacturer)
λ Tip speed ratio

Tref, Reference temperature
TC Temperature of the photovoltaic panel
Ta Ambient temperature
RERs Renewable energy resources
RESs Renewable energy systems
R Solar radiation
Rref Reference solar radiation
RF Renewable Factor
SoC BSS ( τ ) Initial (predefined) state of charge of BSS
SoC BSS ( n+1 ) Current state of charge of BSS
Uw wind speed (m/s) of the WTG
γ Air density
γr water flow rate of MHPS
P R , pv Rated power of the PV module
P pv ( τ ) Generated power by the photovoltaic system
PW ( τ ) Wind turbine power
PWTG ( τ ) Mechanical output power of wind turbine or generator
PWT ,Rated ( τ ) Rated electrical power of wind turbine

av Average power of load profile

PL
av Average power of the wind turbine

PWT
P DG Real power supplied by the DG
ρw Water density
Pdis ( τ ) discharge power during discharging activity
Pinj ( τ ) Injected/conserved power during charging activty
Q DG ( τ ) Reactive powers supplied by the DG
τ Discrete time horizon
αg Acceleration due to gravity
σ rated power factor (phase angle) of diesel generator
ψ blade Rotational speed of rotor blade
Rblade Radius of rotor blade
Uw real-time wind speed
Ur Rated speed of wind generator
U cin cut-in speed of wind generator
U co cut-in speed of wind turbine

WG wind generator
WASA Wind Atlas for South Africa
ESKOM Electricity Supply Commission, South Africa
HOMER Hybrid Optimization of Multiple Energy Resources

Appendix A
Table A1: Typical residential summer load profiles (kWh) in Pretoria, South Africa.
Weekdays Load Profiles Weekends Load Profiles

Aggregate
Time inter- Average STd_profile Normalized
House House House House House households
val load ( μ) (σ ) Load profile
1 2 3 1 2 House Load profile
3
0:00 – 1:00 0,54 0,57 1,03 0,54 0,57 1,01 0,71 0,26 42,37 20,16
1:00 – 2:00 0,54 0,57 1,01 0,54 0,57 1,01 0,71 0,26 42,37 61,37
2:00 – 3:00 0,55 0,67 1,09 0,55 0,59 1,03 0,72 0,27 43,44 38,37
3:00 – 4:00 0,56 0,67 1,11 0,55 0,58 1,09 0,74 0,30 44,40 54,34
4:00 – 5:00 0,59 0,83 1,14 0,55 0,65 1,11 0,77 0,30 46,20 33,09
5:00 – 6:00 0,86 1,20 2,29 0,67 0,76 2,08 1,17 0,79 70,20 78,88
6:00 – 7:00 1,16 1,57 3,14 1,20 1,57 3,14 1,97 1,03 118,22 153,45
7:00 – 8:00 1,40 1,85 4,56 2,40 2,41 4,56 3,12 1,24 187,42 283,56
8:00 – 9:00 2,03 1,89 3,71 1,93 1,89 3,51 2,44 0,92 146,66 214,21
9:00 – 10:00 1,82 1,66 2,72 1,22 1,25 2,32 1,60 0,63 95,80 77,73
10:00 –
1,30 1,21 2,83 1,16 1,02 2,13 1,44 0,60 86,20 83,26
11:00
11:00 –
1,30 0,99 2,81 1,10 0,86 2,1 1,35 0,66 81,20 135,25
12:00
12:00 –
1,26 0,92 2,84 1,21 0,81 2,18 1,40 0,70 84,00 63,68
13:00
13:00 –
1,26 0,86 2,79 1,56 1,35 2,81 1,91 0,79 114,40 80,63
14:00
14:00 –
1,18 0,88 2,99 1,73 1,72 3,8 2,42 1,20 145,00 114,74
15:00
15:00 –
1,19 1,19 2,84 1,19 1,19 2,84 1,74 0,95 104,40 84,61
16:00
16:00 –
2,00 1,40 3,1 2,00 1,86 3,1 2,32 0,68 139,16 168,03
17:00
17:00 –
2,15 3,53 4,69 2,15 2,21 4,69 3,02 1,45 180,96 141,78
18:00
18:00 –
2,53 4,19 5,68 2,90 4,19 5,68 4,26 1,39 255,44 303,97
19:00
19:00 –
3,45 4,77 6,68 3,45 4,77 6,68 4,97 1,62 298,05 306,06
20:00
20:00 –
2,75 3,83 5,22 2,75 3,83 5,22 3,93 1,24 235,90 349,07
21:00
21:00 –
2,71 2,97 4,42 2,53 2,97 4,42 3,31 0,99 198,30 193,17
22:00
22:00 –
1,53 2,18 3,37 1,43 2,18 3,37 2,33 0,98 139,71 120,80
23:00
23:00 –
1,37 1,10 2,32 1,07 1,10 1,50 0,71 89,84 50,55
24:00 2,32
References
[1] A. Z. Obaro, “Energy Dispatch of Decentralized Hybrid Power System,” Int. J. Renew. Energy Res., vol. 8, no. 4, pp. 2131–
2145, 2018.
[2] W. Zheng, J. Zhu, and Q. Luo, “Distributed dispatch of integrated electricity-heat systems with variable mass flow,” IEEE
Trans. Smart Grid, 2022.
[3] J. Cao, W. Du, H. Wang, and M. McCulloch, “Optimal sizing and control strategies for hybrid storage system as limited by
grid frequency deviations,” IEEE Trans. Power Syst., vol. 33, no. 5, pp. 5486–5495, 2018.
[4] Y. Pavankumar, R. Kollu, and S. Debnath, “Multi-objective optimization of photovoltaic/wind/biomass/battery-based grid-
integrated hybrid renewable energy system,” IET Renew. Power Gener., 2021.
[5] X. Gou, Q. Chen, Y. Sun, H. Ma, and B.-J. Li, “Holistic analysis and optimization of distributed energy system considering
different transport characteristics of multi-energy and component efficiency variation,” Energy, vol. 228, p. 120586, 2021.
[6] Z. Liu et al., “Two-phase collaborative optimization and operation strategy for a new distributed energy system that com-
bines multi-energy storage for a nearly zero energy community,” Energy Convers. Manag., vol. 230, p. 113800, 2021.
[7] O. Boqtob, H. El Moussaoui, H. El Markhi, and T. Lamhamdi, “Energy Scheduling of Isolated Microgrid with Battery
Degradation Cost using Hybrid Particle Swarm Optimization with Sine Cosine Acceleration Coefficients,” Int. J. Renew. En-
ergy Res., vol. 10, no. 2, pp. 704–715, 2020.
[8] Y. Guo, H. Gao, Q. Wu, J. Østergaard, D. Yu, and M. Shahidehpour, “Distributed coordinated active and reactive power
control of wind farms based on model predictive control,” Int. J. Electr. Power Energy Syst., vol. 104, pp. 78–88, 2019.
[9] A. Parisio, E. Rikos, and L. Glielmo, “A model predictive control approach to microgrid operation optimization,” IEEE
Trans. Control Syst. Technol., vol. 22, no. 5, pp. 1813–1827, 2014.
[10] M. Longo, M. Roscia, G. C. Lazaroiu, and M. Pagano, “Analysis of sustainable and competitive energy system,” in 2014 In-
ternational Conference on Renewable Energy Research and Application (ICRERA), 2014, pp. 80–86.
[11] W. Shi, X. Xie, C.-C. Chu, and R. Gadh, “Distributed optimal energy management in microgrids,” IEEE Trans. Smart Grid,
vol. 6, no. 3, pp. 1137–1146, 2014.
[12] F. S. Gazijahani, S. N. Ravadanegh, and J. Salehi, “Stochastic multi-objective model for optimal energy exchange optimiza-
tion of networked microgrids with presence of renewable generation under risk-based strategies,” ISA Trans., vol. 73, pp.
100–111, 2018.
[13] W. Lin and E. Bitar, “Decentralized stochastic control of distributed energy resources,” IEEE Trans. Power Syst., vol. 33, no.
1, pp. 888–900, 2018.
[14] Z. Bari and M. Ben Yakhlef, “A MAS based energy-coordination for decentralized control of the hybrid electrical system,”
in 2017 Intelligent Systems and Computer Vision (ISCV), 2017, pp. 1–6.
[15] H. Moradi, M. Esfahanian, A. Abtahi, and A. Zilouchian, “Optimization and energy management of a standalone hybrid
microgrid in the presence of battery storage system,” Energy, vol. 147, pp. 226–238, 2018.
[16] D. H. Tungadio, R. C. Bansal, and M. W. Siti, “Optimal control of active power of two micro-grids interconnected with two
AC tie-lines,” Electr. Power Components Syst., vol. 45, no. 19, pp. 2188–2199, 2017.
[17] A. Patibandla, R. Kollu, S. R. Rayapudi, and R. R. Manyala, “A multi-objective approach for the optimal design of a stand -
alone hybrid renewable energy system,” Int. J. Energy Res.
[18] S. M. Abd-Elazim and E. S. Ali, “Load frequency controller design of a two-area system composing of PV grid and thermal
generator via firefly algorithm,” Neural Comput. Appl., vol. 30, no. 2, pp. 607–616, 2018.
[19] S. Sichilalu, F. Wamalwa, and E. T. Akinlabi, “Optimal control of wind-hydrokinetic pumpback hydropower plant con-
strained with ecological water flows,” Renew. energy, vol. 138, pp. 54–69, 2019.
[20] M. Alipour, K. Zare, and M. Abapour, “MINLP probabilistic scheduling model for demand response programs integrated
energy hubs,” IEEE Trans. Ind. Informatics, vol. 14, no. 1, pp. 79–88, 2018.
[21] M. K. Dosoglu, U. Guvenc, S. Duman, Y. Sonmez, and H. T. Kahraman, “Symbiotic organisms search optimization algo-
rithm for economic/emission dispatch problem in power systems,” Neural Comput. Appl., vol. 29, no. 3, pp. 721–737, 2018.
[22] H. Pulluri, R. Naresh, and V. Sharma, “A solution network based on stud krill herd algorithm for optimal power flow
problems,” Soft Comput., vol. 22, no. 1, pp. 159–176, 2018.
[23] F. Li et al., “Multi-Objective Energy Optimal Scheduling of Multiple Pulsed Loads in Isolated Power Systems,” Sustainabil-
ity, vol. 14, no. 23, p. 16021, 2022.
[24] M. Bilal, M. Shahzad, M. Arif, B. Ullah, S. B. Hisham, and S. S. A. Ali, “Annual Cost and Loss Minimization in a Radial Dis -
tribution Network by Capacitor Allocation Using PSO,” Appl. Sci., vol. 11, no. 24, p. 11840, 2021.
[25] M. Shahzad, W. Akram, M. Arif, U. Khan, and B. Ullah, “Optimal siting and sizing of distributed generators by strawberry
plant propagation algorithm,” Energies, vol. 14, no. 6, p. 1744, 2021.
[26] M. Shahzad, A. Qadir, N. Ullah, Z. Mahmood, N. M. Saad, and S. S. A. Ali, “Optimization of On-Grid Hybrid Renewable
Energy System: A Case Study on Azad Jammu and Kashmir,” Sustainability, vol. 14, no. 10, p. 5757, 2022.
[27] O. D. T. Odou, R. Bhandari, and R. Adamou, “Hybrid off-grid renewable power system for sustainable rural electrification
in Benin,” Renew. energy, vol. 145, pp. 1266–1279, 2020.
[28] A. Kumar, M. Zaman, N. Goel, N. Goel, and R. Church, “In search of an optimization tool for renewable energy resources:
Homer vs. in-house model,” in 2013 IEEE Electrical Power \& Energy Conference, 2013, pp. 1–7.
[29] Databank World Development Indicator, “No Title,” Washington DC, United States.
[30] G. Zhang, B. Wu, A. Maleki, and W. Zhang, “Simulated annealing-chaotic search algorithm based optimization of reverse
osmosis hybrid desalination system driven by wind and solar energies,” Sol. Energy, vol. 173, pp. 964–975, 2018.
[31] M. S. Ismail, M. Moghavvemi, and T. M. I. Mahlia, “Techno-economic analysis of an optimized photovoltaic and diesel
generator hybrid power system for remote houses in a tropical climate,” Energy Convers. Manag., vol. 69, pp. 163–173, 2013.
[32] B. Boumaaraf, K. Touafek, M. S. Ait-cheikh, and M. E. A. Slimani, “Comparison of electrical and thermal performance eval-
uation of a classical PV generator and a water glazed hybrid photovoltaic--thermal collector,” Math. Comput. Simul., vol.
167, pp. 176–193, 2020.
[33] S. Dhundhara and Y. P. Verma, “Application of micro pump hydro energy storage for reliable operation of microgrid sys-
tem,” IET Renew. Power Gener., vol. 14, no. 8, pp. 1368–1378, 2020.
[34] P. A. Michael and C. P. Jawahar, “Design of 15 kW Micro Hydro Power Plant for Rural Electrification at Valara,” Energy
Procedia, vol. 117, pp. 163–171, 2017.
[35] I. Khan, Z. Li, Y. Xu, and W. Gu, “Distributed control algorithm for optimal reactive power control in power grids,” Int. J.
Electr. Power \& Energy Syst., vol. 83, pp. 505–513, 2016.
[36] A. Z. Obaro, J. L. Munda, and M. W. Siti, “Optimal Energy Management of an Autonomous Hybrid Energy System,” in
2018 IEEE 7th International Conference on Power and Energy (PECon), 2019, pp. 316–321.
[37] E. M. A. Mokheimer, A. Al-Sharafi, M. A. Habib, and I. Alzaharnah, “A new study for hybrid PV/wind off-grid power gen-
eration systems with the comparison of results from homer,” Int. J. Green Energy, vol. 12, no. 5, pp. 526–542, 2015.
[38] M. A. Hannan, M. G. M. Abdolrasol, M. Faisal, P. J. Ker, R. A. Begum, and A. Hussain, “Binary particle swarm optimiza -
tion for scheduling MG integrated virtual power plant toward energy saving,” IEEE Access, vol. 7, pp. 107937–107951, 2019.
[39] A. Q. Jakhrani, A. R. H. Rigit, A.-K. Othman, S. R. Samo, and S. A. Kamboh, “Estimation of carbon footprints from diesel
generator emissions,” in 2012 International Conference on Green and Ubiquitous Technology, 2012, pp. 78–81.
[40] C. Allison, “Climate Resilience Assessment for the 1 200 MW Thabametsi Coal-Fired Power Station in Lephalale, Limpopo
Province, South Africa,” 2017.
[41] S. Rippon and B. Design, “University of Cape Town Carbon Footprint Report 2013,” Cape T. Univ. Cape T., 2014.
[42] R. Spalding-Fecher, “What is the carbon emission factor for the South African electricity grid?,” J. Energy South. Africa, vol.
22, no. 4, pp. 8–14, 2011.
[43] M. A. Hossain, H. R. Pota, S. Squartini, F. Zaman, and K. M. Muttaqi, “Energy management of community microgrids con-
sidering degradation cost of battery,” J. Energy Storage, vol. 22, pp. 257–269, 2019.
[44] A. Hoke, A. Brissette, K. Smith, A. Pratt, and D. Maksimovic, “Accounting for lithium-ion battery degradation in electric
vehicle charging optimization,” IEEE J. Emerg. Sel. Top. Power Electron., vol. 2, no. 3, pp. 691–700, 2014.
[45] P. Belotti, C. Kirches, S. Leyffer, J. Linderoth, J. Luedtke, and A. Mahajan, “Mixed-integer nonlinear optimization,” Acta
Numer., vol. 22, pp. 1–131, 2013.
[46] T. Dragičević, H. Pandžić, D. Škrlec, I. Kuzle, J. M. Guerrero, and D. S. Kirschen, “Capacity optimization of renewable en -
ergy sources and battery storage in an autonomous telecommunication facility,” IEEE Trans. Sustain. Energy, vol. 5, no. 4,
pp. 1367–1378, 2014.
[47] S. Loew, A. Anand, and A. Szabo, “Economic model predictive control of Li-ion battery cyclic aging via online rainflow-
analysis,” Energy Storage, vol. 3, no. 3, p. e228, 2021.
[48] A. Anand, S. Loew, and C. L. Bottasso, “Economic control of hybrid energy systems composed of wind turbine and bat-
tery,” in 2021 European Control Conference (ECC), 2021, pp. 2565–2572.
[49] R. Fallahifar and M. Kalantar, “Optimal planning of lithium ion battery energy storage for microgrid applications: Consid-
ering capacity degradation,” J. Energy Storage, vol. 57, p. 106103, 2023.
[50] A. Zahar, “The Polluter Pays Principle and its Ascendancy in Climate Change Law,” Natl. Taipei Univ. law Rev., vol. 114, pp.
129–180, 2020.
[51] K. Mintz-Woo and J. Leroux, “What do climate change winners owe, and to whom?,” Econ. \& Philos., vol. 37, no. 3, pp.
462–483, 2021.
[52] M. R. Khan, “Polluter-Pays-principle: the cardinal instrument for addressing climate change,” Laws, vol. 4, no. 3, pp. 638–
653, 2015.
[53] J. F. Pinto-Bazurco, “How to Enforce the Polluter-Pays Principle,” 2022.
[54] M. Pill, “Towards a funding mechanism for loss and damage from climate change impacts,” Clim. Risk Manag., vol. 35, p.
100391, 2022.
[55] P. A. Barresi, “The polluter pays principle as an instrument of municipal and global environmental governance in climate
change mitigation law: Lessons from China, India, and the United States,” Clim. Law, vol. 10, no. 1, pp. 50–93, 2020.
[56] (SARS) South African Revenue Service, “A report on environmental-levy-products,” 2021. [Online]. Available:
https://www.sars.gov.za/customs-and-excise/excise/environmental-levy-products/carbon-tax/. [Accessed: 07-Jul-2022].

Energies

Uploaded by

Copyright:

Available Formats

Energies

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Energies

Uploaded by

Copyright:

Available Formats

Type of the Paper (Article)

Modelling and Energy Management of an Off-grid Distributed

Energies 2022, 15, x. https://doi.org/10.3390/xxxxx www.mdpi.com/journal/energies

2. Overview of CO2 Emission

Per capita (Annual) Co2 Emission in South Africa

Year (Two Decades)

Figure 1: Per capita CO2 emission in South Africa.

Co2 emission from fossil fuel combustion in South

Figure 2: Fossil fuel consumption CO2 emission in South Africa.

2.1 System Modelling and Description

Figure 3: A network structure of the proposed distributed energy system.

2.2 Mathematical modelling of the proposed DES components

2.2.1 Photovoltaic system

2.2.2 Wind Generator Model

A typical daily wind speed in Pretoria

Figure 5. A typical 24h daily wind speed in Pretoria.

1.2 Chart Title

2.2.3 Micro-hydro power system Model

2.2.4 Diesel Generator Model

A diesel generator (DG) is a synchronous electro-mechanical machine which can

2.2.4.1 Evaluation of Fossil Fuel Emissions

EC GHG ( τ )=PDG ( τ ) × ( EF SO + EFCO ) ∈ ( kg ) ( 11)

2.2.5 Electrical Power Converter

2.2.6 Battery Storage System (BSS) Model

2.2.6.1 Total Battery Cost

3 Mixed Integer Non-Linear Programming Optimization Energy Model

3.1 MINLP Optimization Algorithm

where the nonlinear objective cost function is denoted by J(x).

3.2 Objective Problem Formulation

∑ PWT ( τ ) +¿ ∑ P Pv ( τ ) +∑ P MHPS ( τ )=∑ P L ( τ )+∑ P Loss ( τ ) +∑ P BSS ( τ ) ; ( 24 ) ¿

(iii) Design variable constraint, Node power constraint

P M ( τ ) ≤ P L ( τ ) + PBSS ( τ ) ; ∀ M ∈ {WT , PV , MHPS } ( 26 )

(iv) Absolute power generation constraint

(v) Binary Switch state control variable

0 ≤ PBSS ( τ ) ≤ P BSS ; ∀ k ∈ { charging∧discharging modes } ( 31 )

(vi) Charge and discharge power flow constraint

PBSS ,∈¿ (t ) × P BSS ,out

3.3. Renewable Energy Factor

3.4 Estimation of BSS Degradation Using Rainflow Algorithm

3.5. Modelling of Electrical Load demand

sample space as follows:

Table 1: Simulation technical specifications.

Parameters Symbol Value Unit

Simulation duration Nτ 24 hours

Initial state of charge SoC BSS ( 0 ) 40 %

Capital cost (excluding installation cost) Cs 4316.13 kWh

Power factor pf 0.85 Dimensionless

Gravitational acceleration αg 9.81 m/s2

4.1 Description of Load Profile Scenario

400 Aggregate Load Demand [ P L ]

Figure 9. An aggregate daily energy demand profiles of a typical remote community.

5. Simulation results and discussion

Figure 10: Power production of renewable based multi-energy sources.

5.3 Analysis of the State of Energy of the Battery Storage System

5.3.1 Degradation assessment of State of Energy using Rainflow Algorithm

5.4 Power contribution of multi-energy generators to daily load profile

Actual Power Contribution to Load Profile

Contributing Factor of Subsystems

PV: 45% DG: 23%

Overall DG-GHGe CO2 e SO 2 e