Optimal Design of a Stand-Alone Residential Hybrid Microgrid System for Enhancing Renewable Energy Deployment in Japan

: This paper aims at the optimal designing of a stand-alone microgrid (PV / wind / battery / diesel) system, which can be utilized to meet the demand load requirements of a small residential area in Kasuga City, Fukuoka. The simulation part is developed to estimate the electrical power generated by each component, taking into account the variation of the weather parameters, such as wind, solar irradiation, and ambient temperature. The optimal system design is then based on the Particle Swarm Optimization (PSO) method to find the optimal configuration of the proposed system, using the least-cost perspective approach.


Introduction
Japan's energy self-sufficiency rate was as low as 9.6% in 2017, indicating the energy security issues in this country [1]. The energy self-sufficiency rate indicates the proportion of primary energy required for daily life and economic activities that can be secured in the country [2]. Japan's electric power industry faces a wide range of challenges, including the reliance on imports of fossil fuels through the immediate nuclear power phase-out and also further focusing on reducing and decarbonizing its energy system [3,4]. Initiatives are underway to decentralize the power sector in Japan from the centralized fossil fuel-based systems to distributed ones. The evolution of the electric utility system has many drawbacks because it is vulnerable to disasters due to extreme concentration. One way to avoid this problem is to use Distributed Energy Resources (DER), which enables the decentralization of the electric power sector in Japan. The deployment of DER involves both generators and energy storage technologies. A microgrid is a combination of various interconnected DER and loads that can operate as a grid-tied (connected to the grid) or a stand-alone (disconnected from the grid) controllable system. The stand-alone microgrids are considered as the most appropriate and cost-effective ways to electrify off-grid communities. Since the stand-alone microgrids operate as the off-grid systems, matching the quantity of the supplied electricity with the load requirements is an important issue, particularly when they are used for providing reliable power in small communities or remote areas. However, the integration and hybridization of various energy sources into the microgrid system increases the complexity of the system.
Although microgrids have several advantages, including reduced maintenance costs, emissions, and increased reliability and flexibility, their initial investment costs are higher than the other conventional power systems. Therefore, finding the optimal size and configuration of a microgrid in a cost-effective way has been the main focal point of recent research activities in this field of study.
Following previous studies, this paper addresses a detailed modeling approach that is used to find the optimal configuration of a typical stand-alone microgrid system consisting of solar panels, wind turbines, battery storage and diesel generators, in order to satisfy the demand load of a residential area in Kasuga city in Japan (Figure 1). The proposed research methodology is based on a cost-effectiveness approach, which aims at finding the optimal configuration of the microgrid together with addressing the uncertainties related to the impact of variable weather conditions on the overall performance of the system and its optimal operation. The analytical framework consists of simulation and optimization models. The simulation model is based on developing a detailed power control strategy that is used to match the supplied electricity with the hourly demand load requirements in the different operating conditions. The optimization model uses the Particle Swarm Optimization (PSO) method to find the optimal configuration (optimal capacity of each DER) of the microgrid. The optimality criterion is satisfied at the minimum total cost of the system.

Wind Turbine
In this simulation model, the following equations are used to quantify the amount of power output from a wind turbine [21]: where (m/s) refers to the measured wind speed at the reference height, ( ); α is the powerlaw exponent; refers to the wind speed at the height of (m) [21]; and P r , V CIN , V rat , and V CO refer to the constant power (kW), cut-in speed (m/s), rated wind speed (m/s), and cutout speed (m/s),

Wind Turbine
In this simulation model, the following equations are used to quantify the amount of power output from a wind turbine [21]: where V re f (m/s) refers to the measured wind speed at the reference height, H re f (m); α is the power-law exponent; V refers to the wind speed at the height of H(m) [21]; and P r , V CIN , V rat and V CO refer to the constant power (kW), cut-in speed (m/s), rated wind speed (m/s), and cutout speed (m/s), respectively.

Solar Photovoltaic (PV)
The Duffie and Beckman principle model was used to calculate the global radiation incident on the PV array, using the following equation [22]: where τα is the effective transmittance-absorptance of the PV panel (%); G T is the incident solar radiation on the PV surface (kW/m 2 ); η c is the conversion efficiency of the PV array (%); U L is the overall heat transfer coefficient of the PV (kW/m 2 • C); and T c and T a are the PV cell temperature ( • C) and the ambient temperature ( • C). The equation above states that a balance exists between the solar energy absorbed by the PV array and the amount of electrical output and heat which is transferred to the surroundings. The following formula gives the cell temperature: Assuming that τα/UL is constant, this equation is substituted into the Equation (3) to calculate T c : Finally, the output of the PV array is calculated by the following equation: where in the above equations, STC and NOCT refer to the standard test and nominal operating cell temperature conditions, respectively; G pv is the rated capacity of the PV panel (kW); f pv is the PV derating factor (%); and α p is the temperature coefficient of power (%/ • C).

Diesel Generator
The diesel generator will be used as the backup system in the proposed microgrid. The rate of the fuel consumption of a diesel generator may be estimated by using the following formula [23]: where D F is the rate of fuel consumption (L/h); D R is the rated power of the diesel generator (kW); D O is the power output of the diesel generator (kW); A N and B O are the coefficients that are set at 0.2461 (L/kWh) and 0.081451 (L/kWh) [18]; and t is the time period in hours.

Power Converter
The proposed system utilizes a bi-directional converter to link the AC and DC buses to each other. The amount of the converted power by the converter is calculated by the following equation: Energies 2020, 13, 1737 4 of 18 P out (t) = P in (t)·η Conv (8) where P in and P out are the input and output power from the inverter (kW), respectively; and η conv is the converter efficiency, which is assumed to be 90%.

Battery Storage
In this study, a lead-acid battery is considered as the storage system in the proposed microgrid. The state of charge (SOC) of a lead-acid battery system should be controlled within the following range [24]: where η B is discharge and charging (round trip) efficiency (%); E CD (t) is the amount of electricity that is charged to or discharged from the battery (kWh); and P R refers to the rated capacity of the battery(kWh).

Power Control Strategy
The control strategy flowchart of electricity generation and storage in the proposed microgrid system is represented in Figure 2. If the amount of electricity generated by the renewable generators (PV and wind turbine) exceeds the load requirement, the surplus energy is sent to the battery. If the SOC of the battery reached its maximum level and there is still excess electricity, the extra electricity will be sent to a dummy load. If the amount of generated electricity is less than the load requirement, then the battery storage can be discharged to meet the demand. If the battery discharge was insufficient, a diesel generator would be added to the system as a backup.

Demand Load Calculation
The area of study was selected among the Japanese standard residential buildings located in Kasuga city, Fukuoka prefecture, Japan. The total energy consumption (electrical and thermal loads) of the selected building was estimated using EnergyPlus software developed by the National Renewable Energy Laboratory (NREL), Denver, CO, USA [25]. The 3D model of the targeted Japanese standard house was developed using Sketchup 2019, which is shown in Figure 3. This 3D model includes all walls, ceilings, floors, doors, and windows. A standard size family, including four

Demand Load Calculation
The area of study was selected among the Japanese standard residential buildings located in Kasuga city, Fukuoka prefecture, Japan. The total energy consumption (electrical and thermal loads) of the selected building was estimated using EnergyPlus software developed by the National Renewable Energy Laboratory (NREL), Denver, CO, USA [25]. The 3D model of the targeted Japanese standard house was developed using Sketchup 2019, which is shown in Figure 3. This 3D model includes all walls, ceilings, floors, doors, and windows. A standard size family, including four inhabitants, such as a father, a mother, a son, and a daughter, is supposed to live in the targeted building. An activity schedule was set for each person who should be taken into account, in order to calculate each person's internal heat gain. The occupancy schedule of each family member in this building is shown in Figure 4.   To meet this schedule, the usage plan of the electrical appliances, lightings, cooling, and heating loads are given in Figures 5-8. The internal heat generated from electrical equipment and lighting was not included in the simulation. The variable refrigerant flow (VRF) air conditioning system was considered to provide cooling and heating loads. The values of the cooling and heating Coefficient of Performance (COP) for this air conditioning system was set at 3.4 and 3.3, respectively. The Heating, Ventilation, and Air Conditioning (HVAC) operation strategy is based on the heating and cooling schedule in each room. When the HVAC is ON, cooling occurs when the temperature is higher than the cooling setpoint temperature (26 °C), and heating is performed when the temperature is lower than the heating setpoint (18 °C).
The assumed hourly electricity consumption by the electrical appliances and lighting is represented in Figures 9 and 10.   To meet this schedule, the usage plan of the electrical appliances, lightings, cooling, and heating loads are given in Figures 5-8. The internal heat generated from electrical equipment and lighting was not included in the simulation. The variable refrigerant flow (VRF) air conditioning system was considered to provide cooling and heating loads. The values of the cooling and heating Coefficient of Performance (COP) for this air conditioning system was set at 3.4 and 3.3, respectively. The Heating, Ventilation, and Air Conditioning (HVAC) operation strategy is based on the heating and cooling schedule in each room. When the HVAC is ON, cooling occurs when the temperature is higher than the cooling setpoint temperature (26 °C), and heating is performed when the temperature is lower than the heating setpoint (18 °C).
The assumed hourly electricity consumption by the electrical appliances and lighting is represented in Figures 9 and 10. To meet this schedule, the usage plan of the electrical appliances, lightings, cooling, and heating loads are given in Figures 5-8. The internal heat generated from electrical equipment and lighting was not included in the simulation. The variable refrigerant flow (VRF) air conditioning system was considered to provide cooling and heating loads. The values of the cooling and heating Coefficient of Performance (COP) for this air conditioning system was set at 3.4 and 3.3, respectively. The Heating, Ventilation, and Air Conditioning (HVAC) operation strategy is based on the heating and cooling schedule in each room. When the HVAC is ON, cooling occurs when the temperature is higher than the cooling setpoint temperature (26 • C), and heating is performed when the temperature is lower than the heating setpoint (18 • C).

Weather Data
Weather data was used in this research for the following two purposes: 1) Optimal design of the system: The optimal configuration of the system can be found based on the amount of hourly electricity generation from solar and wind power systems. To this aim, the real measured meteorological data on solar irradiation, wind speed, and ambient temperature in Fukuoka were collected from the Japan Meteorological Agency (JMA) throughout the entire period in 2019, which are shown in Figures 11-13 [26].
2) Optimal day-ahead operation: In order to be able to understand the uncertainties related to the impact of variable weather conditions on the day-ahead optimal operation and electric power dispatching of the proposed system, the GPV-MSM weather forecast provided by JMA was used in this study. The GPV-MSM weather forecasting system reproduces atmospheric phenomena using the mesoscale modeling approach, which can be applied to the selected areas in Japan and its neighboring seas, including a horizontal grid of 5 km. Hourly forecasts of seven weather-related variables are provided eight times a day (00:00, 03:00, 06:00, 09:00, 12:00, 15:00, 18: 00, and 21:00; UTC time zone). The forecast period depends on the forecast time and can be up to 39 hours or 51 hours ahead. In this research, we used the first 24 hours of forecast weather data ahead, updated at UTC 00:00 hours (JST 9:00). Temperature and wind speed are the actual meteorological data that were

Weather Data
Weather data was used in this research for the following two purposes: 1) Optimal design of the system: The optimal configuration of the system can be found based on the amount of hourly electricity generation from solar and wind power systems. To this aim, the real measured meteorological data on solar irradiation, wind speed, and ambient temperature in Fukuoka were collected from the Japan Meteorological Agency (JMA) throughout the entire period in 2019, which are shown in Figures 11-13 [26].
2) Optimal day-ahead operation: In order to be able to understand the uncertainties related to the impact of variable weather conditions on the day-ahead optimal operation and electric power dispatching of the proposed system, the GPV-MSM weather forecast provided by JMA was used in this study. The GPV-MSM weather forecasting system reproduces atmospheric phenomena using the mesoscale modeling approach, which can be applied to the selected areas in Japan and its neighboring seas, including a horizontal grid of 5 km. Hourly forecasts of seven weather-related variables are provided eight times a day (00:00, 03:00, 06:00, 09:00, 12:00, 15:00, 18: 00, and 21:00; UTC time zone). The forecast period depends on the forecast time and can be up to 39 hours or 51 hours ahead. In this research, we used the first 24 hours of forecast weather data ahead, updated at UTC 00:00 hours (JST 9:00). Temperature and wind speed are the actual meteorological data that were

Weather Data
Weather data was used in this research for the following two purposes: 1) Optimal design of the system: The optimal configuration of the system can be found based on the amount of hourly electricity generation from solar and wind power systems. To this aim, the real measured meteorological data on solar irradiation, wind speed, and ambient temperature in Fukuoka were collected from the Japan Meteorological Agency (JMA) throughout the entire period in 2019, which are shown in Figures 11-13 [26].
2) Optimal day-ahead operation: In order to be able to understand the uncertainties related to the impact of variable weather conditions on the day-ahead optimal operation and electric power dispatching of the proposed system, the GPV-MSM weather forecast provided by JMA was used in this study. The GPV-MSM weather forecasting system reproduces atmospheric phenomena using the mesoscale modeling approach, which can be applied to the selected areas in Japan and its neighboring seas, including a horizontal grid of 5 km. Hourly forecasts of seven weather-related variables are provided eight times a day (00:00, 03:00, 06:00, 09:00, 12:00, 15:00, 18: 00, and 21:00; UTC time zone). The forecast period depends on the forecast time and can be up to 39 hours or 51 hours ahead. In this research, we used the first 24 hours of forecast weather data ahead, updated at UTC 00:00 hours (JST 9:00). Temperature and wind speed are the actual meteorological data that were collected from JMA [27]. The comparison between real and forecasting meteorological data used in this study is shown in Figure 14. Energies 2020, 13, x FOR PEER REVIEW 9 of 20 collected from JMA [27]. The comparison between real and forecasting meteorological data used in this study is shown in Figure 14.    Energies 2020, 13, x FOR PEER REVIEW 9 of 20 collected from JMA [27]. The comparison between real and forecasting meteorological data used in this study is shown in Figure 14.    Energies 2020, 13, x FOR PEER REVIEW 9 of 20 collected from JMA [27]. The comparison between real and forecasting meteorological data used in this study is shown in Figure 14.

Objective Function
The optimal sizing of the power generation units in a microgrid system is essential for the efficient utilization of renewable resources. To this aim, the optimization technique was mainly founded on the basis of the minimization of the total cost of the system, subject to satisfying the technical, economic, and environmental constraints. The objective function of the optimization model can be expressed as follows: where TC is the total cost of the microgrid system over its lifetime ($); Ct, Ot, and Ft are investment expenditures, operation costs, and fuel costs in the year t, respectively; and r is the discount rate. The total lifetime of the system, n, is considered to be 20 years.

Demand-Supply Constraint
The main objective of the model is to find the optimal value of the vector of the decision variables = ( , , , ) which includes the installed capacities of the PV, wind power generator, battery and diesel generator subject to satisfying the following demand-supply equality :

Objective Function
The optimal sizing of the power generation units in a microgrid system is essential for the efficient utilization of renewable resources. To this aim, the optimization technique was mainly founded on the basis of the minimization of the total cost of the system, subject to satisfying the technical, economic, and environmental constraints. The objective function of the optimization model can be expressed as follows: where TC is the total cost of the microgrid system over its lifetime ($); C t , O t and F t are investment expenditures, operation costs, and fuel costs in the year t, respectively; and r is the discount rate. The total lifetime of the system, n, is considered to be 20 years.

Demand-Supply Constraint
The main objective of the model is to find the optimal value of the vector of the decision variables P = (P PV , P WG , P bat , P DG ) which includes the installed capacities of the PV, wind power generator, battery and diesel generator subject to satisfying the following demand-supply equality: P PV (t) + P wind (t) + P diesel (t) + P battery discharge (t) = P Demand (th) + P battery charge (t) (11) The above concept was applied to the different dispatching modes, which is visualized in Figure 15.
The above concept was applied to the different dispatching modes, which is visualized in Figure  15.

Solving Method
In this research, the Particle Swarm Optimization algorithm was used to find the least expensive combination of the decision variables. This method consists of a constant search of the best solution by moving the particles at a specific speed calculated in each iteration, which is represented as follows [28]: ( + 1) = ( ) + ( + 1) where and vid(t) represent the particle's best candidate position and the velocity of inertia, respectively; is the particle position; and u is the coefficient of inertia. The parameters c1 and c2 are positive weighting constants, described as "self-confidence" and "swarm confidence", respectively. The random values of φ1 and φ2 are between 0 and 1. ω indicates the inertia weight, which is set in the range (0.5, 1), and near 1 facilitates the global search. The first iteration ends by adjusting the speed and position of the next time step t + 1. Consistently, this process is performed until the best value of the objective function. In this paper, for all variants, fixed values considered as defaults for the PSO parameters were used as c1 = c2 = 1.5, ω = 0.8, iterations = 100, population size = 20 [9,30]. To relocate the wrong particles in the adequate solution space and evaluate the fitness function, as well as provide a valid solution to the optimization problem, the attenuation technique was used to represent the boundary condition of the proposed PSO model, as shown in Figure 16

Solving Method
In this research, the Particle Swarm Optimization algorithm was used to find the least expensive combination of the decision variables. This method consists of a constant search of the best solution by moving the particles at a specific speed calculated in each iteration, which is represented as follows [28]: where P id and v id (t) represent the particle's best candidate position and the velocity of inertia, respectively; x id is the particle position; and u is the coefficient of inertia. The parameters c 1 and c 2 are positive weighting constants, described as "self-confidence" and "swarm confidence", respectively. The random values of ϕ 1 and ϕ 2 are between 0 and 1. ω indicates the inertia weight, which is set in the range (0.5, 1), and near 1 facilitates the global search. The first iteration ends by adjusting the speed and position of the next time step t + 1. Consistently, this process is performed until the best value of the objective function. In this paper, for all variants, fixed values considered as defaults for the PSO parameters were used as c 1 = c 2 = 1.5, ω = 0.8, iterations = 100, population size = 20 [29,30]. To relocate the wrong particles in the adequate solution space and evaluate the fitness function, as well as provide a valid solution to the optimization problem, the attenuation technique was used to represent the boundary condition of the proposed PSO model, as shown in Figure 16 [30].
The above concept was applied to the different dispatching modes, which is visualized in Figure  15.

Solving Method
In this research, the Particle Swarm Optimization algorithm was used to find the least expensive combination of the decision variables. This method consists of a constant search of the best solution by moving the particles at a specific speed calculated in each iteration, which is represented as follows [28]: ( + 1) = ( ) + ( + 1) (13) where and vid(t) represent the particle's best candidate position and the velocity of inertia, respectively; is the particle position; and u is the coefficient of inertia. The parameters c1 and c2 are positive weighting constants, described as "self-confidence" and "swarm confidence", respectively. The random values of φ1 and φ2 are between 0 and 1. ω indicates the inertia weight, which is set in the range (0.5, 1), and near 1 facilitates the global search. The first iteration ends by adjusting the speed and position of the next time step t + 1. Consistently, this process is performed until the best value of the objective function. In this paper, for all variants, fixed values considered as defaults for the PSO parameters were used as c1 = c2 = 1.5, ω = 0.8, iterations = 100, population size = 20 [9,30]. To relocate the wrong particles in the adequate solution space and evaluate the fitness function, as well as provide a valid solution to the optimization problem, the attenuation technique was used to represent the boundary condition of the proposed PSO model, as shown in Figure 16 [29][30].   Figure 17 shows the interrelationship between the optimization and simulation models. The global solution of the PSO model, the best place that every individual in the flock has ever achieved, is adopted for all particles.
Energies 2020, 13, x FOR PEER REVIEW 12 of 20 Figure 17 shows the interrelationship between the optimization and simulation models. The global solution of the PSO model, the best place that every individual in the flock has ever achieved, is adopted for all particles.

Estimation of the Annual Electricity Demand
EnergyPlus software was used to perform a load simulation by computing an hourly energy balance in the building. The amount of required electricity for providing necessary cooling/heating loads to maintain the building at the desired temperature was estimated based on the given scheduled plan of cooling/heating and the difference between the outdoor and indoor temperatures. The total electricity consumption for cooling and heating purposes in the selected residential building in Kasuga city was estimated at 600 kWh/year. This estimated value was summed up with the total electricity consumption by electrical appliances, considering their usage plans and rated power to calculate the total electricity consumption in the building. Figure 18 shows the hourly electricity demand load. The total electricity consumption in this building was estimated at 2303 kWh/year. The annual electricity consumption in the selected residential building is represented in Figure 18.

Estimation of the Annual Electricity Demand
EnergyPlus software was used to perform a load simulation by computing an hourly energy balance in the building. The amount of required electricity for providing necessary cooling/heating loads to maintain the building at the desired temperature was estimated based on the given scheduled plan of cooling/heating and the difference between the outdoor and indoor temperatures. The total electricity consumption for cooling and heating purposes in the selected residential building in Kasuga city was estimated at 600 kWh/year. This estimated value was summed up with the total electricity consumption by electrical appliances, considering their usage plans and rated power to calculate the total electricity consumption in the building. Figure 18 shows the hourly electricity demand load. The total electricity consumption in this building was estimated at 2303 kWh/year. The annual electricity consumption in the selected residential building is represented in Figure 18.
According to the Japan Agency for National Resource and Energy, the average annual electricity consumption per household in Japan is approximated at 4618 kWh, and the electricity demand of the main electrical appliances accounts for 57% of the total, which is about 2632 kWh [31]. The comparison between this value and the estimated value of total electricity consumption indicates the good agreement between the results of the simulation model and the standard data. According to the Japan Agency for National Resource and Energy, the average annual electricity consumption per household in Japan is approximated at 4618 kWh, and the electricity demand of the main electrical appliances accounts for 57% of the total, which is about 2632 kWh [31]. The comparison between this value and the estimated value of total electricity consumption indicates the good agreement between the results of the simulation model and the standard data.

Optimal Design of the Proposed Microgrid
The technical specifications of the main components of the microgrid system are given in Tables  2-4. The hourly power output per unit of wind generator and the solar panel is shown in Figure 19. Table 2. Main input data used in the solar panel simulation [32].

Optimal Design of the Proposed Microgrid
The technical specifications of the main components of the microgrid system are given in Tables 2-4. The hourly power output per unit of wind generator and the solar panel is shown in Figure 19. Table 2. Main input data used in the solar panel simulation [32].  Table 3. Main input data used in the battery storage simulation [24].

Battery Type Lead-Acid
Nominal capacity (kWh)  Table 5 shows the cost analysis of the proposed system. The optimal size of each component is given in Table 6. Figure 20 demonstrates the pathways towards reaching the optimal solution by each element (particle in the PSO model) based on satisfying the minimum total cost of the system. The total cost of the proposed microgrid is estimated at USD 42,300.  Table 5 shows the cost analysis of the proposed system. The optimal size of each component is given in Table 6. Figure 20 demonstrates the pathways towards reaching the optimal solution by each element (particle in the PSO model) based on satisfying the minimum total cost of the system. The total cost of the proposed microgrid is estimated at USD 42,300.  Reference height (m) 10 Figure 19. Estimated power output per each unit of PV panel and wind turbine. Table 5 shows the cost analysis of the proposed system. The optimal size of each component is given in Table 6. Figure 20 demonstrates the pathways towards reaching the optimal solution by each element (particle in the PSO model) based on satisfying the minimum total cost of the system. The total cost of the proposed microgrid is estimated at USD 42,300.   Table 5. Cost analysis of the system.

Components Capital Cost 1 ($/kW) O&M Cost ($/KW) Fuel Cost ($/KW) Lifetime
Wind Turbine [20] 2300 2 0 20 years PV [34] 5100 10 0 20 years Diesel [23] 300 0.5 1. 3 15,000 h Battery [23] 120 10 0 4 years Converter [23] 127 1 0 20 years 1 Including both purchase and installation costs. As can be observed from Figure 21, the solar panel represents the largest share in the total cost of the system, followed by the battery storage and diesel generator. The levelized cost of electricity (LCOE) of the proposed microgrid is estimated at 0.88 $/kWh, which is much higher than the average electricity tariff in Japan (0.2 $/kWh ≈ 22 JPY/kWh). As can be observed from Figure 21, the solar panel represents the largest share in the total cost of the system, followed by the battery storage and diesel generator. The levelized cost of electricity (LCOE) of the proposed microgrid is estimated at 0.88 $/kWh, which is much higher than the average electricity tariff in Japan (0.2 $/kWh≈22 JPY/kWh). The comparison between the estimated LCOE in this study and other similar off-grid residential microgrids is given in Table 7. Table 7. Comparison between the estimated LCOE by the model and other references.

System
LCOE ($/kWh) The proposed system in this paper (PV + wind + battery + diesel) 0.88 Typical off-grid microgrid in Japan: 4kW of PV + 4kWh of battery [35] 0.55-0.72 Typical off-grid microgrid in Pacific Island: PV + diesel [36] 1-1.7 The average monthly electricity generation by the system is shown in Figure 22. The battery SOC is represented in Figure 23. The comparison between the estimated LCOE in this study and other similar off-grid residential microgrids is given in Table 7. Table 7. Comparison between the estimated LCOE by the model and other references.

System LCOE ($/kWh)
The proposed system in this paper (PV + wind + battery + diesel) 0.88 Typical off-grid microgrid in Japan: 4kW of PV + 4kWh of battery [35] 0.55-0.72 Typical off-grid microgrid in Pacific Island: PV + diesel [36] 1-1.7 The average monthly electricity generation by the system is shown in Figure 22. The battery SOC is represented in Figure 23.  The model results revealed that the power loss due to the charging and discharging efficiency of the battery is considerable, which is estimated at 719 kWh per year, which indicates that improving the round-trip efficiency of the battery is necessary for the effective utilization of the proposed microgrid system (Figure 24).  The model results revealed that the power loss due to the charging and discharging efficiency of the battery is considerable, which is estimated at 719 kWh per year, which indicates that improving the round-trip efficiency of the battery is necessary for the effective utilization of the proposed microgrid system (Figure 24). The model results revealed that the power loss due to the charging and discharging efficiency of the battery is considerable, which is estimated at 719 kWh per year, which indicates that improving the round-trip efficiency of the battery is necessary for the effective utilization of the proposed microgrid system ( Figure 24). The model results revealed that the power loss due to the charging and discharging efficiency of the battery is considerable, which is estimated at 719 kWh per year, which indicates that improving the round-trip efficiency of the battery is necessary for the effective utilization of the proposed microgrid system (Figure 24).  The microgrid's fuel consumption and energy storage requirements are examined as a function of the atmospheric conditions. Weather data is thus necessary to establish optimal operating and  The microgrid's fuel consumption and energy storage requirements are examined as a function of the atmospheric conditions. Weather data is thus necessary to establish optimal operating and dispatching plans according to the operational objective of the microgrid. Day-ahead weather forecasts are also responsible for deviations from these plans, thereby being a valuable source of uncertainty in the scheduling process. Figure 25 represents the weather satellite images taken by the satellite Himawari-8 at 12:00 on 3 July 2018 (real data) and 2019 (forecasts). On 3 July 2018, Typhoon No. 7 went north over the East China Sea and approached Kyushu, and strong winds and heavy rain took Fukuoka. Therefore, there was insufficient solar electricity generation, but since the strong wind blew in the afternoon, a sufficient amount of electricity was generated by the wind power generator. However, on the same day in 2019, Fukuoka was covered with heavy clouds, and the average wind speed was very weak at 2-3 (m/s). Therefore, there was no power generation from the wind turbine on this day. Besides, since the sky had been covered with dark clouds for a long time, there was no sufficient power output from the PV panels, which resulted in reducing the SOC of the battery.
Energies 2020, 13, x FOR PEER REVIEW 17 of 20 dispatching plans according to the operational objective of the microgrid. Day-ahead weather forecasts are also responsible for deviations from these plans, thereby being a valuable source of uncertainty in the scheduling process. Figure 25 represents the weather satellite images taken by the satellite Himawari-8 at 12:00 on 3 July 2018 (real data) and 2019 (forecasts). On 3 July 2018, Typhoon No. 7 went north over the East China Sea and approached Kyushu, and strong winds and heavy rain took Fukuoka. Therefore, there was insufficient solar electricity generation, but since the strong wind blew in the afternoon, a sufficient amount of electricity was generated by the wind power generator. However, on the same day in 2019, Fukuoka was covered with heavy clouds, and the average wind speed was very weak at 2-3 (m/s). Therefore, there was no power generation from the wind turbine on this day. Besides, since the sky had been covered with dark clouds for a long time, there was no sufficient power output from the PV panels, which resulted in reducing the SOC of the battery. As shown in Figure 26, the diesel generator was used to offset the shortage of battery discharge. The comparison between the historical (2018) and forecasting data (2019) highlights the remarkable impact of weather conditions on both fuel consumption and energy storage requirement of the proposed microgrid. Based on the forecasting data, the stationary battery storage's SOC decreases to 40%, since it cannot be charged by solar and wind in the morning. Therefore, the diesel generator should be used in order to meet the demand load during the evening. It highlights that forecasts of weather conditions at the site location would be required to know in advance the amount of power that the wind turbine or the PV will feed into the battery over the next hours and days. Day-ahead forecasting of the weather data will help in managing the battery operation through monitoring its SOC condition and lowering the usage of the diesel generator to reduce its cost and environmental impacts on the system. As shown in Figure 26, the diesel generator was used to offset the shortage of battery discharge. The comparison between the historical (2018) and forecasting data (2019) highlights the remarkable impact of weather conditions on both fuel consumption and energy storage requirement of the proposed microgrid. Based on the forecasting data, the stationary battery storage's SOC decreases to 40%, since it cannot be charged by solar and wind in the morning. Therefore, the diesel generator should be used in order to meet the demand load during the evening. It highlights that forecasts of weather conditions at the site location would be required to know in advance the amount of power that the wind turbine or the PV will feed into the battery over the next hours and days. Day-ahead forecasting of the weather data will help in managing the battery operation through monitoring its SOC condition and lowering the usage of the diesel generator to reduce its cost and environmental impacts on the system.

Conclusion
This research addressed the optimal design of a stand-alone microgrid system that can be used in order to meet the electrical load requirement in a selected Japanese standard building in Kasuga city, Fukuoka prefecture, Japan. Based on the results, the optimal size of the main components of the system was estimated for the PV as 2.65 kW; for wind power as 2.01 kW; for the battery as 14.86 kW; for the diesel generator as 3.6 kW; and for the converter as 2.8 kW. The total cost of the proposed system was estimated at USD 42,300. The LCOE of the proposed system was estimated at 0.88 $/kWh, which is much higher than the average electricity rate in Japan. The percentage of power provided by each power unit was estimated at 43.4% by the solar PV, 16.7% by the wind power, 4.9% by the diesel generator, and 35% by the battery discharge. The model results show that the operation of the proposed microgrid system is highly dependent on batteries and solar power, due to the high potential of solar energy in Kasuga city. Furthermore, the results revealed the remarkable impact of weather conditions on the optimal operation of the proposed microgrid, especially during the windy and rainy seasons. Due to the issue of the high LCOE of the proposed microgrid in this study, the investment demand of the system might be insufficient in the early stages, which makes customers lack the motivation to participate in the project. In order to promote the development of such a system, the corresponding incentive mechanism should be designed through an optimal subsidy from the Japanese government and an optimal cooperation incentive from the energy supplier. Another solution would be the introduction of the community microgrid, including some prosumers to a local grid, which allows its participants to achieve a greater outcome than they would individually. In this scheme, participants who generate excess electricity are able to share their generation with participants of their choosing. They are also able to take advantage of shared energy-storage systems in the community to improve the operational reliability and the economy. However, it demands extra efforts regarding a detailed cost analysis of the system, which can be considered as the future work.