Renewable Energy Focus  Volume 37  June 2021 www.renewableenergyfocus.

com

Optimization of an off-grid domestic

RESEARCH PAPER

Hybrid Energy System in suburban Paris

using iHOGA software
Heman Shamachurn

Department of Electrical and Electronic Engineering, Faculty of Engineering, University of Mauritius, Reduit 80837, Mauritius

At the COP21, France committed to reduce its greenhouse gas emissions by 50% by 2050. To attain the set
goals, it will increase the share of renewable energy in its final energy consumption. Moreover, a 50%
reduction in the energy consumption is envisaged. The National Low-Carbon Strategy aims to cut down
the emissions of the construction sector by 54% through several actions, including the acceleration of
energy renovation work. Understanding how a typical house in France could contribute to the set
objectives has not been studied so far. Therefore, an optimized Hybrid Energy System (HES) is proposed
for a house in Paris. Wind energy, solar-photovoltaic energy, battery and diesel-generator were
considered for a house which had an average measured electricity consumption of 25.81 kWh/day. A
cost of energy of 0.28 s/kWh was achieved. The multi-objective optimizations provided, at only 1.4–6%
higher Net Present Cost (NPC), 57–73.8% reduced CO2 emission and 56–68% reduced unmet load as
compared to the single-objective optimization of minimizing only the NPC. The diesel consumption-
related CO2 emission of the optimization scenarios was 2.66–10.14 g per kWh of electricity generated by
the proposed HES, which is significantly less than the average of 73 g per kWh of electricity produced in
France for the year 2016. Though currently not economically attractive compared to the grid, such an
HES, if optimized for several dwellings, will lead to decarbonising development, enabling France to
contribute significantly towards its COP21 commitments.

Introduction emissions by 2030, and a 75% reduction by 2050, as compared

The World’s energy demand is continuously increasing. However, to 1990 levels.
the fossil fuel reserves which currently supply the major propor- The variability of one RE resource leads to mismatch between
tion of energy requirements around the globe are rapidly decreas- electricity demand and generation, thereby lowering the supply
ing. Many of the environmental and climatic issues experienced reliability. This issue can be counteracted to some extent by
by the Earth are associated with the excessive use of conventional combinations of different RE sources and storage media in order
energy sources. Increasing attention is being given to measures to create a Hybrid Energy System (HES). Research and practical
such as the energy efficiency improvement and the adoption of experiences have shown that HESs are superior to a single tech-
renewable energy (RE) sources. France for instance, through its nology as regards to cost, efficiency and predictability. The size of
COP21 commitments [1], plans to achieve a renewable share of HESs typically vary from a few kWs for domestic applications to a
32% in its final energy mix by 2030 and a 50% reduction in its few MWs for an entire geographical region. Standalone systems are
energy consumption by 2050. These actions are expected to make mostly used where the grid is not available or not reliable, and
the country achieve a 40% reduction in its greenhouse gas where grid extension will be far more expensive than setting up a
HES. Grid-connected HESs benefit from feed in tariff schemes and
Shamachurn, H. ([email protected])
grid reliability. Cost being a major driver of such systems, an

1755-0084/ã 2021 Elsevier Ltd. All rights reserved. https://doi.org/10.1016/j.ref.2021.02.004

36
Renewable Energy Focus  Volume 37  June 2021 RESEARCH PAPER

Nomenclature emission can be reduced by about 29.65% and 16 tons per year
HES Hybrid Energy System respectively, as compared to conventional power plants. An off
COE Cost of energy grid Wind/Battery/Diesel HES was optimized using HOMER for 2
NPC Net Present Cost
RE Renewable energy
hotels in Malaysia [11]. The average daily load was 85 kWh/day
O&M Operation and maintenance and the peak load was 8.7 kW. A COE of 0.199 $/kWh was
LF Load-following achieved, and this was 0.051 $/kWh less than the COE for a
RF Renewable fraction conventional power plant. Moreover, the annual CO2 emission
WS Wind-speed by the proposed HES was 16 tons less as compared to conventional
SR Solar radiation
DG Diesel generator
power plants. A detailed review on recent methods to optimize the
FI Fuel inflation size of standalone HESs revealed that nowadays multi-objective

RESEARCH PAPER
AEC Average daily energy consumption optimisation is more important than the single objective of mini-
mizing the HES cost only [12]. A triple objective optimization was
carried out for a stand-alone Solar/Wind/Disesel/Battery HES to
optimal sizing strategy is a crucial element to be considered in minimize the NPC, CO2 emission and unmet load and the best
order to achieve the highest reliability at the lowest cost. COE achieved was 0.55 s/kWh [13] for a daily load of about
Optimization results of HESs for different types of loads located 6.8 kWh. However, some random load profiles were considered
in several geographical locations are available in literature. Some for the analysis and this questions the applicability of the results.
of the recent studies are reported in this section. A techno-eco- Little study has been carried out for off-grid household loads.
nomic feasibility study was carried out on a grid-connected HES for Most existing works in literature deal with huge buildings and
a huge office building in France [2]. The peak load of the building, entire villages. It should be noted that a typical single house tends
as obtained from bills, was found to vary between 350 kW and to consume a relatively smaller amount of electricity, and the idea
500 kW, and the electricity demand was estimated to be approxi- of having a HES might seem farfetched when considering only the
mately 7.4 MW h/day. The optimization results of different con- system’s cost. No study has been found on the potential of a single
figurations including PV/Hydrogen/Grid, PV/Hydro/Hydrogen/
Grid, concluded that a PV/Hydrogen/Grid system was the most
cost-effective, with the cost of energy (COE) being 0.073 s/kWh.
The COE was 10% lower compared to the utility grid and the
proposed system could reduce the CO2 emission of the building by
about 90%. An off-grid PV/Wind/Diesel Generator/Battery HES
was designed and optimized for a village consuming 213 kWh/day
in Bangladesh [3]. A COE of 0.161 $/kWh was achieved and it was
found that no CO2 would be emitted by the system. Techno-
economic feasibility analysis of off-grid HESs employing PV mod-
ules, wind turbines and diesel generators was performed for 3
villages consuming 180 kWh/day, 231 kWh/day and 379 kWh/
day in Colombia, resulting in COE of, 0.444 $/kWh, 0.448 $/
kWh and 0.473 $/kWh respectively [4]. A standalone PV/Wind/
Diesel Generator/Battery HES was optimized for a remote small
community in Malaysia [5]. The average load demand was
33 kWh/day, the peak load was 3.9 kW and the COE obtained
was 1.877 $/kWh. The implementation of an HES was assessed for
an off-grid community in Canada through 7 scenarios with vary- FIGURE 1
ing percentages of renewables in the range of 0–100% [6]. The Hourly average load.
range of the COE obtained was 0.36–1.48 $/kWh for a 4.4 MW h/
day primary electricity demand and 772 kW peak load. A stand-
alone HES for a large resort centre in South China Sea, Malaysia
with an estimated peak load of 1185 kW and average daily load of
13,045 kWh was optimized, leading to a COE of 0.279 $/kWh [7]. A
Solar/Biogas HES provided the lowest COE of 0.476 $/kWh for 124
off-grid households located in a rural village of India [8]. The
average daily demand was about 160 kWh. A PV/Wind/Battery
system was designed for a location in India having an estimated
average daily electricity consumption of 782 kWh and a peak load
of 90 kW [9]. The mono-objective optimisation resulted in a COE
of 0.4 $/kWh. A PV/Wind/Battery/Diesel HES was optimized using
HOMER for an off-grid community in Malaysia, resulting in a COE
of 0.431 $/kWh [10]. The average load demand was 80 kWh /day FIGURE 2
and the peak load was 8.1 kW. It was found that the NPC and CO2 Monthly average load.

37
 RESEARCH PAPER Renewable Energy Focus  Volume 37  June 2021
RESEARCH PAPER

FIGURE 3
Solar and Wind resource of house location.

typical house to mitigate CO2 emissions. Moreover, many works month in Figure 1 show that the patterns are similar for all
have used estimated loads rather than actual measured ones, as the the months. The monthly average load in Figure 2 shows that
latter is rarely available. Furthermore, very little reliable informa- the month of July had the lowest consumption (summer), while
tion is available about HES multi-objective optimization studies in the month of January had the highest consumption (winter).
the European climate, particularly in France. It should be noted Weather data for the site was obtained from NASA Atmospheric
that in trying to reduce CO2 emissions, cost should not be the only Science Data Center [15]. The daily solar radiation on a horizontal
factor in the optimization process. Compared to other HES opti- surface and the wind speed at the site are shown in Figure 3. The
mization software, iHOGA PRO + has unique functionalities such maximum and average daily irradiation on the plane of the PV
as pre-sizing, PV slope optimisation, probability analysis, battery
ageing modelling and multi-objective optimisation, but very few
reliable studies have employed it so far. In addition, the author is
not aware of any research employing iHOGA, for a detailed sensi-
tivity analysis. The aim of this work was therefore to contribute in
addressing the above knowledge gap by employing the latest
iHOGA PRO + version to investigate several HES optimization
scenarios for a typical house in Paris. The remainder of this paper
is organized as follows: section 2 provides the detailed methodol-
ogy adopted in this work, section 3 provides the most important
results alongside the discussions, and finally section 4 concludes
on the findings in this work.

Methodology
In this work, a house located in the community of Clamart, in the
southwestern suburbs of Paris was considered. The measured
electricity consumption information for a whole year was acquired
and processed. iHOGA was subsequently used to simulate several
HES optimization scenarios for the dwelling and the results were
analysed. Sensitivity analyses were eventually carried out to inves-
tigate the effect of varying parameters on the optimal system. Sub-
section 2.1 provides the essential details about the inputs for
designing the optimized HES, sub-section 2.2 describes the iHOGA
settings and components used in this work and sub-section 2.3
describes the optimization scenarios studied and the sensitivity
analyses.

Inputs
The electricity consumption dataset contained 2,075,259 mea-
surements at 1 min resolution gathered between December 2006
and November 2010, with approximately 1.25% missing rows [14].
Only the data only for the year 2009 was used as it contained the
least missing information. The average electricity consumption
was 25.81 kWh/day, the average yearly load was 1.08 kW and the FIGURE 4
peak hourly load was 2.6 kW. The hourly load profiles for each HES configuration.

38
 Renewable Energy Focus  Volume 37  June 2021 RESEARCH PAPER

TABLE 1 TABLE 4
PV module details. Generator details.
PV module Nominal Power Cost (s) O&M Cost Diesel Generator power rating (kVA) Cost (s) O&M Cost (s/yr)
(Wp) (s/yr) 4 1200 0.18
aSi12-Schott: ASI 100 100 110 1.1 3 1050 0.17
SiP12-Atersa: A135P 135 192 1.92 1.9 800 0.14
SiP24-Atersa: A180P 180 300 3 5.5 1300 0.22
SiP24-Atersa: A280P 280 350 3.5 7 1600 0.24
SiP24-Schott: Mono190 190 238 2.38

RESEARCH PAPER
system was 4.35 kWh/m2 and 3.16 kWh/m2 respectively. The max- TABLE 5
imum and average 10 m height wind speed at the site was 4.40 m/s Inverter details.
and 3.65 m/s respectively. Inverter Power (VA) Cost(s) Battery charger
Given that the site benefits from favourable solar and wind
VICTRON: 48/5000/70 5000 2915 Yes
resources, wind turbine technology and solar PV were selected VICTRON: 48/3000/35 3000 2190 Yes
as RE generators. Battery technology was considered for energy Generic: 3000 3000 1600 No
storage and diesel generator was selected to provide backup power
when needed. The HES configuration is shown in Figure 4.

iHOGA settings and components cost for the batteries was 50 s/yr and the equivalent CO2 emissions
iHOGA was initially used to pre-size the different components, during manufacturing was 55 kg/kWh. All the batteries had a
following which constraints were inputted prior to performing the global efficiency of 85%. The minimum state of charge for all
optimization. Based on the recommended pre-sizing information, battery banks were 20%. All of the inverters had a lifetime of 10
suitable batteries, PV modules, wind turbines, inverters and diesel years. The system was allowed to search for its own compatible
generators were selected from iHOGA’s database. The essential charge controllers and/or battery charger, if required. The
details of the considered components are provided in Tables 1–5. expected lifespan of all the generators was 10,000 h. The minimum
The lifespan of the PV modules was 25 years and the CO2 power to be supplied was 30% of a generator’s rated power. The
emissions during manufacturing was 800 kg/kWp. A loss factor diesel price was 1.3 s/L. The fuel inflation rate was 5%. The CO2
of 1.2 was considered and the PV inverter battery charge controller emission was 3.5 kg/L of diesel consumed. The equivalent CO2
was assumed to include Maximum Power Point Tracking (MPPT) emission during manufacture was 215 kg / kVA rated power and
capability. The effect of temperature on the output power was the generator was assumed to be available at all times.
taken into account. The PV modules were inclined at 65 , facing
the South (azimuth = 0 ) as recommended by iHOGA, with a Optimization
ground reflectance of 20% and a derating factor of 70%. For the The single-objective optimization minimized the NPC. The multi-
wind resource, the shape factor of the Weibull distribution was 2 objective optimizations were as follows: (1) minimizing both NPC
and the correlation factor was 0.85. For the load, the daily, hourly and CO2 emission, (2) minimizing both NPC and unmet load, and
and minute’s variability were all taken to be 30%.The fixed O&M (3) minimizing NPC, CO2 emission and unmet load. To guarantee

TABLE 2
Wind turbine details.
Wind turbine Cost (s) Replacement O&M Lifespan CO2 emission (kg)
Cost (s) Cost (s/yr) (yr)
Hummer: HWP-10 14,000 9500 280 20 4000
Bornay: 6000 12,056 10,000 224 15 3500
Bornay: 3000 7555 6000 151 15 1800
Hummer: HWP-20 22,000 17,000 440 20 8000

TABLE 3
Battery details.
Battery Nominal Capacity (Ah) Cost (s) Float life O&M Lifespan Minimum State
at 20 C (yr) of Charge (%)
Cost (s/yr) (yr)
OPZS-FIAMM:LM 3500 3500 1457 15 280 14.57 20
OPZS-FIAMM:LM 1200 1260 474 15 224 4.73 20
OPZS-FIAMM:LM 2500 2800 1080 15 151 10.8 20
OPzV-Hoppecke:3500 3030 1368 18 440 0 10

39
 RESEARCH PAPER Renewable Energy Focus  Volume 37  June 2021

the optimal solution, all the possible cases would have to be TABLE 6
evaluated, but this would take significant time. Therefore, genetic Sensitivity parameters.
algorithm was employed to optimize the components (main algo- Parameter Values
rithm) and all the combinations were evaluated to find optimum
Average wind speed (m/s) 2.92, 3.65, 4.38
operational strategy (secondary algorithm). For the genetic algo- Average solar radiation (kWh/m2) 2.53, 3.16, 3.79
rithm, 15 generations with a population size of 23 were employed Average electricity consumption (kWh/day) 22.02, 27.52, 33.03
whereby the crossover and mutation rates were 90% and 1% Fuel price inflation (%) 3.5, 5.0, 7.5
respectively. The following constraints were employed: 1% maxi-
mum unmet load, 0% minimum renewable energy fraction to
know if running the diesel generator only is cheaper than having
RESEARCH PAPER

part of the energy from renewables, and a maximum COE of 3 s/

kWh. Two diesel generator dispatch types were considered. Load
following (LF) is the type whereby the generator does not charge
the batteries. It rather produces only the required amount of power
that the renewable sources or the batteries fail to do. For the charge
cycling (CC) type, the generator starts and operates at full power
and any excess energy is stored in the batteries. The nominal
interest and inflation rates were both 2% [16,17], and the project
lifetime was considered to be 25 years. The residual cost of the
components after 25 years were considered. The installation cost
and variable initial cost were taken to be 0 s. The interest on loan
was 0%. For all the optimization scenarios, the CO2 emission per
kWh of electricity provided by the HES was obtained as the ratio of FIGURE 5
the CO2 emission by the diesel generator to the total energy System cost for single objective to minimize NPC.
supplied (including wind energy, solar energy, diesel generator
energy and energy provided by the batteries). Only the CO2
emission by the diesel generator was considered as the additional
CO2 emissions are associated with the manufacturing of the
different components for the HES. The CO2 emission per kWh
of electricity supplied by the grid does not take into consideration
the CO2 emission during the manufacture of the different grid
components, and thus, the comparison with the Diesel-generated
CO2 in the HES would be fair.
Given that the optimizations were carried out for a specific
weather, economic and load conditions, it was imperative to
analyse the effects of variations in these conditions on the opti-
mum HES configuration. Therefore, sensitivity analyses were car-
ried out for only the single-objective optimization scenario. The
parameters considered and their corresponding values are pro-
vided in Table 6. A total of 81 sensitivity analysis cases were
FIGURE 6
considered. The results of the analysis provided by iHOGA were
Total annual energy (kWh) — single objective to minimize NPC.
further processed to generate contour plots which can be more
easily studied and visually interpreted.

Results and discussion provided in Figure 5 which shows that the batteries and the wind
Single-objective optimization: minimizing the NPC turbine are the most expensive components, followed by the PV
The total number of cases evaluated was 26,978. The COE was system. The additional components have a relatively smaller NPC.
0.28 s/kWh and the total CO2 emission was 893 kg/yr, out of Figure 6 shows the total annual energy associated with the major
which 236 kg was produced by the diesel generator. The results components of the designed system. A significant amount of
concluded that a Wind/PV/Battery/Diesel system will have the excess energy is produced, mainly because of demand-supply
lowest NPC of 69,098.5 s, excluding diesel fuel cost. The initial mismatch, but the unmet load and the diesel generator energy
required investment shall be 37,250 s and the RE technologies production are relatively low.
shall provide 98% of the required annual energy. The system will Figure 7 shows the power demanded by the house and the power
comprise sixty-six ASI100 PV modules, twenty-four OPZS-FIAMM: generated by the different HES components. Figure 8 shows the
LM 1200 batteries, one DC Hummer: HWP-10 wind turbine, one battery charging and discharging powers. It can be deduced that
Diesel generator of 1.9 kVA capacity, one Generic: 3000 inverter the PV system tends to generate more energy in the autumn and
along with the corresponding compatible battery charger and spring periods, and this appears to be contrary to the load demand
charge controller. The breakdown of the total system NPC is pattern. The generated wind power, however, is much more

40
 Renewable Energy Focus  Volume 37  June 2021 RESEARCH PAPER

RESEARCH PAPER
FIGURE 7
Power generated (W) and consumed (W) over a whole year.

variable than the solar power. The diesel generator operates for Multi-objective optimization: minimizing CO2 emissions and NPC
only a few hours during the winter periods, to help in meeting the The total number of cases evaluated was 51,824 and the optimi-
highest power demands. The load is not fully met for only a very zation process resulted in 20 non-dominated solutions. The result
few hours during the winter period. The excess energy is mostly of only one solution has been presented here. The corresponding
due to the wind turbine. From Figure 8, it can be deduced that the NPC was 73257.4 s and the COE was s 0.29/kWh. The CO2
battery discharge profile resembles the load demand profile. Fig- emission was 801 kg/yr, out of which 69 kg was produced by the
ures 9 and 10 show the hourly power flow for a typical winter and diesel generator. This represents a total decrease of about 10% in
summer days respectively. The graphs clearly demonstrate that the CO2 emission as compared to the single-objective optimization.
HES tries to meet the load as much as possible. The diesel generator The LF strategy was favoured. An initial investment of 41,843 s
did not operate for any of the days. In fact, the battery operated in will be required and 99.4% of the required annual energy will be
order to compensate for the mismatch between the load and the supplied by RE technologies. The selected components were simi-
supply. The mass of CO2 emitted by the diesel generator per kWh lar to the ones obtained by the single objective optimization,
of electricity generated by the HES was 10.14 g. except for the inverter, which was replaced by one VICTRON

41
 RESEARCH PAPER Renewable Energy Focus  Volume 37  June 2021
RESEARCH PAPER

FIGURE 8
Battery charging power (W) and discharging power (W) over a whole year.

48/5000/70 inverter comprising a battery charger. The correspond- investment of 40,016 s will be required. RE technologies will
ing compatible charge controller was automatically selected. supply 99.1% of the required annual energy. The selected compo-
Moreover, the number of PV modules was increased from sixty- nents were similar to the configuration of the single objective
six to ninety-three. The breakdown of the total system NPC, is optimization, except that nintety ASI100 PV modules were
provided in Figure 11. It can be deduced that there have been slight selected. The breakdown of the total system NPC is provided in
changes as compared to Figure 5. Particularly, the share of the cost Figure 13. It can be deduced that there have been slight changes
taken by the inverter has increased by 4%, that taken by the diesel compared to Figure 5. Particularly, the share of the cost taken by
fuel cost has reduced by 3%, that taken by the PV system has the PV system has increased by 4%. The annual energy associated
increased by 4% and that taken by the battery bank has decreased with each system component is shown in Figure 14. The results
by 3%. have been compared to Figure 6 as shown in Table 10. The mass of
The annual energy associated with each system component is CO2 emitted by the diesel generator per kWh of electricity gener-
shown in Figure 12. The results have been compared to Figure 6 as ated by the HES was 4.25 g.
shown in Table 10. The mass of CO2 emitted by the diesel genera-
tor per kWh of electricity generated by the HES was 2.66 g. Multi-objective optimization: minimizing CO2 emissions, NPC and
unmet load
Multi-objective optimization: minimizing unmet load and NPC The total number of cases evaluated was 26,978 and the optimi-
The total number of cases evaluated was 26,978 and the optimi- zation process resulted in 50 non-dominated solutions. The result
zation process resulted in 12 non-dominated solutions. The result of only one solution has been presented here. The corresponding
of only one solution has been presented here. The corresponding NPC was 71275.4 s and the COE was s 0.28/kWh. The CO2
NPC was 70031.8 s and the COE was s 0.28/kWh. The CO2 emission was 828 kg/yr, out of which 111 kg was produced by
emission was 835 kg/yr, out of which 109 kg was produced by the diesel generator, The LF strategy was favoured and an initial
the diesel generator. The LF strategy was favoured and an initial investment of 41,064 s will be required. RE technologies will

FIGURE 9 FIGURE 10
Power flow for day in winter. Power flow for a day in summer.

42
Renewable Energy Focus  Volume 37  June 2021 RESEARCH PAPER

RESEARCH PAPER
FIGURE 11
FIGURE 14
System cost for multi objective to minimize NPC and CO2 emission.
Total annual energy (kWh) — multiple objective to minimize NPC unmet
load.

FIGURE 12
Total annual energy (kWh) — multiple objective to minimize NPC and CO2 FIGURE 15
emission. System cost for multi objective to minimize NPC, unmet load and CO2
emission.

FIGURE 13
FIGURE 16
System cost for multi objective to minimize NPC and unmet load.
Total annual energy (kWh) — multiple objective to minimize NPC, unmet
load and CO2 emission.

supply 99.1% of the required annual energy. The selected compo-

nents were similar to the configuration resulting from the single The results have been compared to Figure 6 as shown in Table 10.
objective optimization, except that forty-six SiP24-Schott: Mono The mass of CO2 emitted by the diesel generator per kWh of
190 PV modules were chosen for the PV system. The breakdown of electricity generated by the HES was 4.36 g.
the total system NPC is provided in Figure 15. It can be deduced
that there have been slight changes compared to Figure 5. Partic- Sensitivity analysis
ularly, the share of the cost taken by the PV system has increased The sensitivity analysis results are provided in Figures 17–20, and
by 6%. The annual energy associated with each system component details of the parameters kept constant are provided in
is shown in Figure 16 Table 7.

43
 RESEARCH PAPER Renewable Energy Focus  Volume 37  June 2021
RESEARCH PAPER

FIGURE 17
Sensitivity plots 1.

A discussion about the sensitivity analysis results is provided in parameters and system configurations. Considering the single-
Tables 8 and 9. It was noted that the battery capacity was constant objective optimization scenario as reference, the differences in
when both the average wind speed and average solar radiation diesel-generated CO2 emission per kWh of produced electricity,
were varied at a fuel inflation of 5% and an average energy unmet load, NPC, excess energy, RE percentage and initial invest-
consumption of 27.52 kWh/day. This particular result was thus ment for the multi-objective optimization scenarios are shown in
not shown. Table 10. The greatest differences between the single-objective and
The results for all the optimization scenarios yielded a minimum multi-objective optimizations occur for the CO2 emission, unmet
of 0.28 s/kWh for the COE, with slight variations in the load and excess energy, while the differences in RE percentage are

44
Renewable Energy Focus  Volume 37  June 2021 RESEARCH PAPER

RESEARCH PAPER
FIGURE 18
Sensitivity plots 2.

the smallest. The result of the multi-objective optimisation to and 10.14 g per kWh of electricity generated by the system. This is
minimize the NPC and unmet load is very close to that of the significantly less compared to the average of 73 g per kWh of
triple objective optimization when considering the parameters in electricity produced in France for the year 2016 [18]. Therefore,
Table 10. The main conclusion is that for a slightly higher NPC, the the proposed optimal HES has the ability to reduce the annual CO2
unmet load and CO2 emission can significantly be reduced at the emission of the considered house by approximately 86.1–96.4%.
expense of higher excess energy. The diesel consumption-related Furthermore, the RE percentage increased only slightly when
CO2 emissions of the optimization scenarios varied between 2.66 g multi objective optimizations are carried out. Therefore, it has

45
 RESEARCH PAPER Renewable Energy Focus  Volume 37  June 2021
RESEARCH PAPER

FIGURE 19
Sensitivity plots 3.

been shown that multi-objective optimizations are more beneficial for a typical domestic load having a power rating of 3 kVA [19],
than single-objective ones. For the house considered in this work it the proposed HES, having a COE of 0.28 s/kWh, has the
is recommended that the HES be optimized to minimize both the potential to be used for typical domestic applications, only in
NPC and the CO2 emission, as the resulting configuration leads to an attempt to become off-grid and to help France achieve its
a much lower CO2 emission and unmet load for the same COE as COP21 commitments rapidly by catering for the environment.
compared to other optimization scenarios. The main disadvantage found about iHOGA PRO + is that no
Considering the fact that the cost of electricity as from 1st information is provided about the excess energy available from
August 2016 under the ‘Tarif de Base’ in France is 0.1564 s/kWh each of the considered energy technology. This would have

46
Renewable Energy Focus  Volume 37  June 2021 RESEARCH PAPER

RESEARCH PAPER
FIGURE 20
Sensitivity plots 4.

TABLE 7 helped in providing a rough estimation of the investment

Constant parameters for sensitivity analysis. recovery through a feed in tariff scheme for the different RE
Figure Constant parameters technologies. A comparison between the results obtained in this
work and those reported in literature about recently optimized
Figures 17(a)–(f ) and Fuel inflation = 5%
18 (a)–(d) Average energy consumption = 27.52 kWh/day HESs is provided in Table 11. The COEs have all been converted
Figure 18(e)–(f ) Solar radiation = 3.16 kWh/m2 to (s/kWh) using a rate of 1 USD = 0.88 Euro for easier compari-
Average energy consumption = 27.52 kWh/day son. It can be observed that HESs for the same country, for
Figure 19 (a)–(b) Wind speed = 3.65 m/s example India or Malaysia, tend to have a decreasing COE with
Average energy consumption = 27.52 kWh/day
increasing load energy consumption. Moreover, HESs devel-
Figure 19 (c)–(d) Wind speed = 3.65 m/s
Solar radiation = 3.16 kWh/m2 oped for much higher loads in different parts of the world
Figure 19 (e)–(f ) Solar radiation = 3.16 kWh/m2 (Canada, Colombia and India) have a higher COE than obtained
Fuel inflation = 5% in this work, as shown in Table 11. Furthermore, compared to a
Figure 20 (a)–(b) Wind speed =3.65 m/s HES for approximately the same load size (3.9 kW peak load,
Fuel inflation = 5%
33 kWh/day average daily energy consumption) in Malaysia [5],

TABLE 8
Sensitivity results discussion 1.
Figure Discussion
Figure 17(a) For average wind speeds (WS) below about 4 m/s, the CO2 emission tends to increase for decreasing WS and/or average solar radiation (SR).
Figure 17(b) The unmet load tends to be minimum at WS close to 3.65 m/s and SR of 3.79 kWh/m2. For low WS, the unmet load is slightly higher at a SR of
3.16 kWh/m2 and tends to increase further when the SR increases or decreases. For WS at about 4.38 m/s, the unmet load decreases as SR
increases.
Figure 17(c) The renewable fraction (RF) tends to be high at high SR and high WS. If only SR is considered, the unmet load is the highest when SR is about
3.16 kWh/m2.
Figure 17(d) The COE tends to be lowest when WS is high and depends more on WS than on SR.
Figure 17(e) The PV system size decreases as the WS increases. For WS of about 3.65–4.38 m/s, there is a tendency for the PV system size to increases with
increasing SR. However, at WS of about 2.92 m/s, the PV system size actually decreases at higher SR.
Figure 17(f ) For WS of about 3.65 m/s and below, and for SR of about 3.16 kWh/m2 and lower, the wind turbine size remains unchanged. For higher SR and WS,
the turbine size tends to decrease with increasing SR and/or increasing WS.
Figure 18(a) The size of the diesel generator (DG) tends to be the highest at WS of about 3.65 m/s and decreases as the wind speed increases or decreases. At
WS of about 3.65 m/s, the DG size increases with increasing SR.
Figure 18(b) The excess energy is relatively higher when the WS is 3.65 m/s and the SR is 3.79 kWh/m2, and, when the WS is 4.38 m/s and the SR is 3.16 kWh/m2.
The excess energy tends to be lower for the other considered instances.
Figure 18(c) The generated wind energy is the highest at about WS of 4.38 m/s and lowest at WS of 2.92 m/s, At WS of 3.65 m/s and above, the wind energy
generated tends to decrease with increasing SR and remains constant for WS lower than 3.65 m/s.
Figure 18(d) The energy produced by the DG tends to be smallest when the SR is about 3.16 kWh/m2, irrespective of the WS. Additionally, the SR varies from
3.16 kWh/m2 to 3.79 kWh/m2 and the WS varies from 3.65 m/s to 4.38 m/s, the energy tends to remaining relatively small. A relatively significant
amount of energy is generated by the DG when the WS is 3.65 m/s and the SR is 2.53 kWh/m2
Figure 18(e) The RF tends to be a maximum when the fuel inflation (FI) is about 5%, irrespective of the WS, but decreases when the FI decreases or increases
from 5%. For FI higher or lower than 5%, there is a tendency for the RF to increase with increasing WS.
Figure 18(f ) The COE tends to be independent of the FI, but decreases with increasing WS. For WS at about 2.92 m/s, the COE only slightly decreases with
decreasing FI

47
 RESEARCH PAPER Renewable Energy Focus  Volume 37  June 2021

TABLE 9
Sensitivity results discussion 2.
Figure Discussion
Figure 19(a) For FI of about 5% to 3.5%, the RF increases with increasing SR. For FI of about 6.5%, the RF decreases when the SR increases from
2.53 kWh/m2 to a minimum when the SR is 3.16 kWh/m2 and then increases as the SR increases to 3.79 kWh/m2. At a given SR, the RF
generally decreases with increasing FI.
Figure 19(b) For SR of about 2.53 kWh/m2 and 3.79 kWh/m2, the COE remains constant with variation in FI from 3.5% to 6.5%. For FI of about 5% and
lower, the COE decreases slightly with increasing SR, irrespective of the FI. However, for FI above 5%, the COE decreases with increasing
SR and with decreasing FI.
Figure 19(c) The COE variation is more compared to Figure 19(b). Here the COE decreases with increasing average daily energy consumption (AEC)
and tends to be less affected with variation in FI, except at a low AEC of about 22.02 kWh/day, whereby the COE decreases with
RESEARCH PAPER

decreasing FI.
Figure 19(d) The RF tends to be highest at FI of about 6.5% and AEC of about 22.02 kWh/day. The RF is quite high at an AEC of about 27.52 kWh/day
and FI of 5%. The RF tends to be relatively lower at low FI and low AEC, high FI and high AEC, and, high AEC and low FI.
Figure 19(e) The RF is highest when the AEC is 27.52 kWh/day, irrespective of the WS. For WS of 2.92 m/s, there is a greater variation in RF with
variation in AEC, as compared to higher WS. For AEC greater and lower than 27.52 kWh/day, a decrease in RF with decreasing WS can be
observed.
Figure 19(f ) The COE tends to be highest at low WS and low AEC, and lowest at high WS and high AEC. The COE decreases with increasing WS and
decreases with increasing AEC.
Figure 20(a) The RF is highest when the AEC is 27.52 kWh/day and the SR varies in the rage of 3.16 kWh/m2 to 3.79 kWh/m2.
Figure 20(b) The COE tends to be highest at low AEC of about 22.02 kWh/day and lowest at a high AEC of about 33.03 kWh/day. There is a general
trend for the COE to decrease with increasing SR.

TABLE 10
Comparison of different multi –objective optimization scenarios against single-objective optimization scenario.
Parameter Multi-objective Multi-objective Multi-objective optimisation
optimisation to minimize optimisation to minimize to minimize NPC, CO2 emission
NPC and CO2 emission NPC and unmet load and unmet load
NPC (%) 6.0% higher 1.4% higher 3.2% higher
Diesel-generated CO2 per kWh 73.8% lower 58.1% lower 57.0% lower
of electricity produced (%)
Unmet load (%) 68.0% lower 56.0% lower 56.0% lower
Excess Energy (%) 38.2% higher 26.7% higher 24.0% higher
Renewable energy (%) 1.4% higher 1.1% higher 1.1% higher
Initial investment (%) 7.4% higher 10.2% higher 10.2% higher

TABLE 11
Comparison of results to those reported in recent literature. Conclusion
Reference Load energy Location COE (s/kWh) Single and multi-objective optimizations for a HES for a typical
consumption house in France to minimize the NPC; NPC and CO2 emission;
(kWh/day) NPC and unmet load; NPC, CO2 emission and unmet load resulted
[2] 7400 France 0.073 in a COE of 0.28 s/kWh for all the optimization scenarios. Com-
[3] 231 Bangladesh 0.18 pared to the single objective of minimizing the NPC, the multi-
[6] 4400 Canada 0.32 – 1.30 objective optimizations resulted in significantly reduced CO2
[4] 180 Colombia 0.39
emission and unmet load at only a slightly higher NPC. For the
231 0.39
379 0.41 considered house, it is proposed that the HES is designed such that
[7] 13,050 Malaysia 0.24 both the NPC and CO2 emissions are minimized, as this multi-
[5] 33 Malaysia 1.64 objective optimization scenario produces more beneficial results.
[10] 80 Malaysia 0.38 The proposed HES might not be currently economically suitable
[11] 85 Malaysia 0.18
for domestic application in France, given that grid electricity is
[9] 782 India 0.35
[8] 160 India 0.42 available at a lower cost of 0.1564 s/kWh and that there are
[13] 6.8 India 0.55 currently no attractive feed in tariff schemes. However, the pro-
This work 25.81 0.28 posed HES resulted in better COE compared to other recently
optimized HES reported in literature. Moreover, considering the
fact that the prices of RE technologies are gradually decreasing,
the HES for the load in this work (2.6 kW peak load, 25.81 kWh/ and that the prices of conventional energy sources are likely to
day average daily energy consumption) has an approximately continue increasing, such an HES will definitely become econom-
83% lower COE. It can therefore be deduced that the proposed ically attractive in the near future. More importantly this work has
HES is not an unviable project. shown that the optimized HES can significantly reduce the CO2

48
Renewable Energy Focus  Volume 37  June 2021 RESEARCH PAPER

emission of a typical house located in the suburban Paris by about [6] M.M. Rahman, M.M.-U.-H. Khan, M.A. Ullah, Energy 97 (2016) 151–160.
[7] M. Hossain, S. Mekhilef, L. Olatimiwa, Sustain. Cities Soc. 28 (2017) 358–366.
86.1–96.4%. Therefore, if considered for several dwellings across a
[8] R. Sharma, S. Goel, Environ. Dev. Sustain. 18 (6) (2016) 1601–1614.
large geographical area, the development of such HESs will have a [9] J.B. Fulzele, M.B. Daigave, Design and Optimization of Hybrid PV-Wind
tremendous potential to contribute in helping France to achieve Renewable Energy System, Andhra Pradesh, India, 2016, s.n..
its COP21 commitments of significant CO2 emissions reduction at [10] S. Shezan, A. Al-Mamoon, H. Ping, Environ. Prog. Sustain. Energy 37 (4) (2017)
1424–1432.
a higher pace. Sensitivity analyses revealed non-trivial changes in [11] S. Shezan, et al. Clean Technol. Environ. Policy 17 (8) (2015) 2417–2430.
the system’s cost and performance parameters, so that simulation [12] M.D. Al-falahi, S. Jayasinghe, H. Enshaei, Energy Convers. Manage. 143 (2017)
parameters have to be carefully selected. 252–274.
[13] R.J. Rathish, K. Mahadevan, Multi-Objective Optimization of Stand-alone
Renewable Energy Hybrid System, Springer, Singapore, 2018.
[14] M. Lichman, UCI Machine Learning Repository, Available at: https://archive.ics.

RESEARCH PAPER
Acknowledgements uci.edu/ml/datasets/Individual+household+electric+power+consumption.
The author gratefully acknowledge Dr Rodolfo Dufo-López for [Accessed 7 May 2017]., 2013.
[15] J.M. Kusterer, n.d. NASA Surface meteorology and Solar Energy: RETScreen Data.
providing a free licence of iHOGA PRO+. The author is also
Available at: https://eosweb.larc.nasa.gov/cgi-bin/sse/retscreen.cgi?email=rdufo
thankful to Mosche Lichman from the School of Information and %40unizar.es&step=1&lat=48.80&lon=2.26&submit=Submit. [Accessed 7 May
Computer Science, University of California, for providing access to 2017].
the electricity consumption dataset of the house in a repository. [16] tradingeconomics.com, n.d. France Interest Rate | 1998-2017 | Data | Chart |
Calendar | Forecast. Available at: http://www.tradingeconomics.com/france/
interest-rate. [Accessed 23 May 2017].
References [17] statista, n.d. France: inflation rate 2010–2020 Statistic. Available at: https://www.
[1] Gouvernement.fr, COP21: France’s National Commitments, Available at: https:// statista.com/statistics/270353/inflation-rate-in-france/. [Accessed 9 May 2017].
www.gouvernement.fr/en/cop21-france-s-national-commitments. [Accessed 14 [18] R. France, chiffres clés en | RTE France, Available at: http://www.rte-france.com/
November 2018], 2015. en/eco2mix/chiffres-cles-en#chcleco2. [Accessed 03 July 2017], 2017.
[2] M. Islam, Sustain. Cities Soc. (2018). [19] French-Property.com, n.d. Tariffs of Electricity for French Houses, EDF Prices.
[3] H.S. Das, A. Dey, T.C. Wei, A.H.M. Yatim, Int. J. Renew. Energy Res. 6 (2) (2016) Available at: https://www.french-property.com/guides/france/utilities/
402–412. electricity/tariff/. [Accessed 15 June 2017].
[4] A.H. Mamaghani, et al. Renew. Energy 97 (2016) 293–305.
[5] S.A. Shezan, et al. J. Clean. Prod. 125 (2016) 121–132.

49