Optimization and Energy Management of Hybrid Photovoltaic-Diesel-Battery System to Pump and Desalinate Water at Isolated Regions

This research work aims to provide detailed feasibility, a techno-economic evaluation, and energy management of stand-alone hybrid photovoltaic-diesel-battery (PV/DG/B) system. The proposed system can be applied to supply a specific load that is far away from the utility grid (UG) connection, and it is located in Minya city, Egypt, as a real case study. The daily required desalinated water is 250 m3. The total brackish water demands are 350–500 m3 and 250–300 m3 of water in summer and winter seasons, respectively. Two different sizes of reverse osmosis (RO) units; RO-250 and RO-500, two energy control dispatch strategies; load following (LF) and cycle charging (CC); two sizes of DG; 5 kW and 10 kW are considered in the case study. The cost of energy, renewable fraction, environmental impact, and breakeven grid extension distance are the main criteria that have been considered to determine the optimal size of PV/DG/B to supply the load demand. HOMER® software is used to perform the simulation and optimization. For this case study, the minimum cost of energy and the minimum total present cost are 0.074 $/kWh and 207676 $, respectively. This is achieved by using a RO-500 unit and a LF dispatch control strategy. The related sizes to the best option of PV/DG/B are 120 kW PV array, 10 kW DG, 64 batteries, and 50 kW converter. A comparison with grid extension and installing stand-alone diesel generation is also carried out. The results of comparison have confirmed that the grid connection is better than all considered options using the RO-250 unit. However, for the RO-500 unit, all options of hybrid PV/DG/B are more economically feasible compared with grid connection, and the best cost-effective option is the one including LF strategy with 10 kW DG. Stand-alone diesel generator produces 119110 kg/year and 117677 kg/year of CO2 respectively for RO-250 and RO-500.


I. INTRODUCTION
Egypt's Vision 2030 depends on using renewable energy sources to minimize, or eliminate, the CO 2 emissions to reduce the effect of Global Warming [1]. Egypt is one of the largest producers of Oil in Africa outside of the OPEC (Organization of the Petroleum Exporting Countries) and the third large producer of Natural Gas. Also, the Suez Canal plays The associate editor coordinating the review of this manuscript and approving it for publication was Sudhakar Babu Thanikanti . the main role in the international energy market [1]. Egypt is one of the populous countries in the Middle East and faces energy demand due to rapid population growth and overcome our growing needs. This makes a great challenge to supply energy. Using renewable energy sources can help Egypt to achieve its demand for energy and create a lot of jobs with the achievement of objectives of sustainable development [2]. Global Warming is one of the biggest challenges that is facing humanity on Earth [3]. Harmful effects of using fossil fuels to generate electricity should be reduced, if not eliminated. One of the important keys to solve the problems of Global Warming is to generate energy from renewable sources and improve its efficiencies [4]. Egypt is very rich with sources of renewable energy like hydropower, wind, solar PV, concentrating solar power (CPS), and biomass energy sources. The Egyptian government started programs of using renewable energy sources since the 1970s. It depends on its resources and also on co-operation with other countries, including Germany, France, Italy, Spain, Denmark, Japan, and the United States [5]. The Egyptian government focuses nowadays on using wind energy and solar photovoltaic (PV) applications, which include water pumping, cold stores, and desalination plants [5]. Figure 1 shows the generated renewable energy sources in Egypt up to 2035 in GW. The main planned PV projects in Egypt up to 2023, with its capacity and name of companies, are listed in Table 1 [5]. Water desalination is the process of converting high saltwater to freshwater by removing salt particles. This water can be drinkable for humans or used for irrigation. Deferent desalination processes are used in industrial and commercial applications. With improvements in technology techniques, desalination processes are becoming cost-competitive and more efficient rather than other methods of producing freshwater to overcome our growing needs [6]- [8]. However, the total cost of water desalination is still high by using conventional methods of energy source. So, the new trend now of Egypt's governments is using renewable energy systems, which will decrease the total cost to reasonable values, which will decrease the cost of energy compared with grid extension and diesel generation systems. In addition to the cost of treatment of environmental effects of using fossil fuel is so high for the long term; CO 2 emissions which an effect on Global Warming [9]- [12].
Using renewable energy systems for supplying desalination plants is increased. There are about 130 desalination plants around the world opened in the last few years [13]. Table 2 lists some of them with focusing on the name of desalination plant, its location, kind of technology if it is multi-effects distillation (MED) or reverse osmosis (RO), the capacity of the plant, and kind of renewable energy used. The energy management strategies (EMS) are the process of selecting, presenting, and programming a central-controller to manage the flow of energy according to an optimal-strategy. The central-controller can be a microcontroller, microprocessor, PLC, or any other type of suitable controller [14].
Several optimization programs and mathematical techniques are used to plan and design energy management strategies [15], which include stand-alone or Utility-connection systems, as concluded in Figure 2. HOMER R software FIGURE 2. Energy management strategies (EMS) approaches commonly used. VOLUME 8, 2020  is an optimizing program that can optimize several techniques [16], [17], which are used throughout many studies to investigate the optimal design of the proposed renewable energy system based on LF [18]- [20] or CC strategies [21]- [23]. Table 3 summarizes the literature review of renewable energy systems based on hybrid configuration and simulation tools of control strategies.
In this paper, the authors' contribution is to evaluate a proposed stand-alone PV/DG/B system that supplies a real load in Al-Minya city, Egypt. The used simulation tool is HOMER R software to get the optimal size and best energy management strategy for this case study. A comparison of the proposed system with grid extension and also with installing stand-alone diesel generation has been carried out. Using PV/DG/B can significantly minimize the amount of CO 2 emissions generated in the case of a stand-alone diesel system and helps to treat Global Warming. Also, this research work is aimed to help policymakers in Egypt, location of case study, to develop and integrate the effective policy for energy-water nexus and energy-water-food security by achieving strict and fast rules of renewable energy systems for freshwater production through desalination plants and agriculture purposes, respectively. Using renewable energy in desalination water will help the economy to grow and supply sustainable water sources.

II. LOCATION AND LOAD DATA
The case study represents a flat 70 acres (283280 m 2 ) located in Minya city, Egypt, at the latitude of 28 • N and longitude of 30 • E. The site under study is positioned 12 km far away from UG connection. Minia city is characterized by a good level of solar radiation. The average daily horizontal solar radiation is around 5.97 kWh/m 2 . The mean daily solar radiation level and clearance index during the year are shown in Figure 3 [44]. The highest daily irradiance level of 8.056 kWh/m 2 is collected in June. Whereas the least daily irradiance level of 3.555 kWh/m 2 is received in December. Figure 4 shows the solar atlas of Egypt [45], which is a sun-belt and high solar radiation country. The duration of sunshine is about 9 to 11 hours per day all year except a few cloudy days.  In the location of the case study, there is a 150 m depth well with a 40 m static level of water, which produces 120 m 3 per hour of brackish water with a salinity of 2500 mg/l. It is planned to cultivate the land with some crops which can use the raw brackish water, whereas a large part of the land will be cultivated with other crops such as Wheat, which needs water with salinity less than 800 mg/l. The daily required desalinated water is 250 m 3 . The total brackish water demands are 350-500 m 3 and 250-300 m 3 of water in summer and winter seasons, respectively.
The electric power necessary to pump the required water can be estimated by the following relation [46], [47]; where P pump denotes the power of the pump (kW); H denotes the water head of the pump (m), and η denotes the pump. Based on the relation (1), the estimated daily electrical demand power to pump the water from well is around 110 kWh with a maximum of 15 kW. The seasonal load profile of the AC pump is illustrated in Figure 5-a. To desalinate the brackish water, it is planned to use reverse osmosis (RO) desalination system. There are different sizes of RO units; 50 m 3 , 100 m 3 , 150 m 3 , 250 m 3 , 500 m 3 , and 1000 m 3 . The power consumption for each unit are 4.1 kW, 7.7 kW, 10.5 kW, 15 kW, 29.5 kW and 52 kW respectively for RO-50, RO-100, RO-150, RO-250, RO-500 and RO-1000 [48]. As the required desalinated water is 250 m 3 , the three sizes, RO-50, RO-100, and RO-150, are not applicable to the study. The two sizes, RO-250 and RO-500, are compared in this research work to investigate and identify the economical option to desalinate the required quantity. RO-250 will operate 24 hours every day to get 250 m 3 , whereas RO-500 needs only 12 hours to desalinate the same amount. It is planned to operate RO-500 from 6:00 AM to 6:00 PM. The seasonal load profiles of RO-500 and RO-250 are displayed in Figure 5-b and Figure 5-c, respectively. RO unit components' schematic diagram is shown in Figure 6. Figure 7 shows a schematic diagram of the proposed renewable energy system, which consists of solar PV cells, diesel generator, converter, and batteries. The input technoeconomic parameters for all components in the proposed renewable energy system are listed in Table 4 [25], [49], [50], which is used to find the optimal sizes for the proposed system using HOMER R software [51], [52].

A. MODELLING OF SOLAR PV CELLS
The output power (P PV ) from solar PV cells at any time (t) is affected by several factors such as solar radiation (R), PV array area (A Array ), the efficiency of the converter (η c ), the efficiency of the PV cells (η PV ) and environmental factors VOLUME 8, 2020  such as ambient temperature, wind velocity [25], [53].
The output of the PV cells is affected by the ambient temperature. So, the temperature of PV cell (T c ) depends on effective transmittance absorbance of the solar PV array (α), the coefficient of heat transfer (T H ) and the efficiency of the PV cells (η PV ) which are expressed as follows: The net present cost (NPC PV ) of the solar PV cells is calculated by using the capital cost (C PV ), operation, and maintenance (C OM .PV ) costs per year as expresses by the following equation:  where, (r) is the interest, and (β) is the escalation rates; these economic aspects have been considered in the optimization process.

B. MODELLING OF THE BATTERY SYSTEM
The authors have used lead-acid batteries in this study to store the excess energy generated from the solar PV cells and DG. The battery power (P B ) can be calculated by the following equation [25], [53]: where Q B−i is the initial battery charge; V B and I B are voltage and current rating of the battery, respectively. The battery state of charge (B SoC ) can be expressed as: where Q B−max is the maximum charge of the battery. The capacity of a battery (C Wh ) can be calculated from the following formula [25]: where P L : load demand energy, kWh/day; Ad: BS autonomy per day; D d : discharge depth; η BS and η C are the efficiency of battery and converter, respectively.

C. MODELING OF DIESEL GENERATOR
The electrical power output (P DG ) from the diesel generator is AC power and depend on the fuel consumption from the DG. The fuel curve assumed a straight line in HOMER modeling for simplicity. Rate of fuel consumption (F) is calculated for electricity production [50]: where; A 1 is coefficient of the fuel curve, A 2 is the slope of fuel curve, which its values are obtained from the manufacturer's datasheet which equal to 0.246 L/kWh and 0.08145 L/kWh, respectively, C DG is the diesel generator capacity, and P DG is the electrical power output from DG. This equation can be applied when the DG is running, while the DG is at rest, the fuel consumption rate (F) is zero. The replacement cost of DG is calculated to depend on the number of operating hours.
The net present cost of diesel generator (NPC DG ) depends on the capital cost of the diesel generator (C DG ), fuel cost (F), operating and maintenance cost (C OM −DG ) and the cost of the replacement (C R−DG ) as expressed by the following equation:

IV. ENERGY MANAGEMENT STRATEGIES
To determine the energy flows, the controller compares the generated power value (P PV ) of Solar PV cells with the load demand (P L ). If the renewable energy from solar PV cells (P PV ) greater than load demand (P L ), the excess power goes to charge the batteries if its state of charge level is not reaching its maximum value (B Soc_max ). If the load demand (P L ) is lower than the renewable power (P PV ) and state of charge of batteries not reach its minimum value (B Soc_min ), batteries  can discharge its power (P B ) supplies to the load. If the state of charge of batteries reaches its minimum value (B Soc_min ), here the DG must start-up, and its power (P DG ) supplies the load. Figure 8 shows the Power Management Strategies (PMS), which include two control dispatch strategies; load following (LF) and cycle charging (CC). These dispatch strategies are used to control DG operation and battery. With the LF strategy, a DG generates only sufficient energy to meet the required load and does not charge the battery bank. The battery bank is charged only by surplus power by PV arrays Figure 8-a. With the CC strategy, the DG works at its maximum rating whenever it is switched on to supply load and charge the battery bank by the surplus energy, Figure 8-b.

V. OPTIMIZATION PROBLEM FORMULATION
The optimum size of PV/DG/B has been determined based on the minimum total net present cost (NPC) and the minimum cost of energy (COE). The NPC can be estimated based on the following relation; VOLUME 8, 2020  where C ann,tot denotes total cost per year, i denotes yearly real interest rate, N is project lifetime years), and CRF denotes the capital recovery factor. CRF is calculated as follows; The total cost C ann,tot includes initial cost, operation, maintenance, and replacement. The salvage value can be estimated by the following relation: where: C rep is the replacement cost of the components, R rem is the remaining life, and R comp is the project life span. The COE can be calculated as follows:

VI. RESULTS AND DISCUSSION
This section presents the detailed feasibility and technoeconomic evaluation of the PV/DG/B system to supply the    0.138 $/kWh-0.109 $/kWh respectively for RO-500 and RO-250. This demonstrates that using RO-500 is costeffective compared with RO-250. The minimum cost of energy and the minimum total present cost are 0.074 $/kWh and 207676 $, respectively. This is achieved with RO-500 using load following dispatch control strategy. The related sizes to the optimal configuration are 120 kW PV array, 10 kW DG, 64 batteries, and 50 kW converter. The optimal size and related costs with varying sizes of RO, control strategy, and size of DG are shown in Table 5. Using the LF strategy decreases the cost of energy with RO-500 by 22.2% and 31.48%, respectively, for DG size of 5kW and 10 kW compared with the CC control strategy. Table 6 and Table 7 display the detailed related costs of different components for the PV/DG/B system for RO-250 and RO-500, respectively, with varying the control strategy and the size of DG. Whereas the total NPC for different system  components is shown in Figure 9 and Figure 10, respectively, for RO-250 and RO-500. For RO-250 with 5 kW DG and LF strategy, the total NPC is 454729$. The replacement cost is 213795$ (47%) that represents the main part of the total NPC flowed by the initial cost (13.62%). The key reason for the high replacement cost is that the batteries need to be changed VOLUME 8, 2020 many times throughout the project lifetime. The replacement cost of batteries is 192044$, which represents 89.75% of the total replacement cost. Whereas with 10 kW, approximately, the capital, replacement, and fuel costs are very near. They are 122840$, 128939$, and 124827$, respectively, for capital, replacement, and fuel costs. The fuel cost records the maximum value of 167737$ with 10 kW of DG and CC strategy. It increased by 92.65% compared with 5 kW of DG and CC strategy. Considering Figure 10-b and Figure 10-d, it can be seen that the cost of fuel is mainly influenced by the dispatch control strategy. It is equal to 123932 $ and 31344$, respectively, for CC and LF strategies with the same size of DG (10 kW). This means that the cost of fuel reduced in the case of LF by 74.71% compared with the CC strategy.
The discounted cash flows related to PV/DG/B system with varying size of RO, control strategy, and size of DG is displayed in Figure 11. As explained in Figure 11-a, the lowest initial cost (104204$) is achieved with RO-250, CC strategy, and 10 kW size of DG. However, due to the high fuel cost (167737$) during the lifetime of the project, the total NPC has reached to 449181 $. The minimum total NPC of 403356 $ is achieved by using the load following strategy and 10 kW size of DG. Figure 11-b illustrates the variation of NPC using the RO-500 unit. It is clear from this figure that the maximum and minimum total NPC are 394873 $ and 270676 $ respectively for CC and LF control strategies with the same size of DG (10 kW).
Under the condition of using the best size of the PV/DG/B system (RO-500, LF strategy, and 10 kW DG), the total yearly generated energy is 253322 kWh. 95% (240518 kWh) of the total produced energy is delivered by the PV system, and the reminder part (12804 kWh) is powered by DG. With this configuration, the total yearly consumption energy is 166591 kWh. The AC load pump consumed around 24% (40080 kWh) of the total consumed energy, while the other portion, 76% (126511 kWh), is used to feed the RO-500 unit. The surplus energy is almost 67671 kWh (26.7%). This surplus can be used for lighting and other not considered loads, whereas the annual unmet load and capacity shortage are 2024 kWh (1.2%) and 3218 kWh (1.9%), respectively. From Table 8, the annual excess energy is very sensitive to the size of RO. The minimum annual excess energy achieved with RO-250 unit, LF strategy, and 5 kW DG. Table 9 and Table 10 illustrate the detailed performance of different components of PV/DG/B systems with varying size of RO, control strategy and size of DG.
For RO-250, the rated capacities of the PV array are 90 kW, 75 kW, 95 kW, and 70 kW respectively for LF&5kW-DG,  LF&10kW-DG, CC&5kW-DG, and CC&10kW-DG. Subsequently, the mean PV output energy is 494 kWh, 412 kWh, 522 kWh, and 384 kWh, respectively, LF&5kW-DG, LF&10kW-DG, CC&5kW-DG, and CC&10kW-DG. The mean daily produced PV power for each month with varying control strategy and the size of DG is illustrated in Figure 12 and Figure 13, respectively, for RO-250 and RO-500.
For RO-500, the nominal capacities of battery are 276 kWh, 138 kWh, 138 kWh, and 69.1 kWh respectively for LF&5kW-DG, LF&10kW-DG, CC&5kW-DG, and CC&10kW-DG. Whereas the values of the expected life of the battery are 6.56 years, 5.38 years, 3.5 years, and 2.86 years respectively, LF&5kW-DG, LF&10kW-DG, CC&5kW-DG, and CC&10kW-DG. This is confirmed that the expected life of the battery is mainly influenced by the type of control strategy. The LF strategy increases the battery lifetime compared with the CC. The monthly statistics of battery SOC are presented in Figure 14 and Figure 15, respectively, for RO-250 and RO-500.

VII. COMPARISON WITH GRID EXTENSION AND STAND-ALONE DIESEL GENERATOR
To prove the viability of the PV/FC/B system, a comparison with grid extension along with a stand-alone diesel generator has been made. The initial cost of grid connection and yearly maintenance costs are $10,000/km and $200/year/km, respectively. The Energy consumption tariff, based on the Egyptian Electricity Company, is $0.06/kWh [54]. Figure 16 illustrates a comparison among the NPC of PV/DG/B with different conditions and NPC of the grid. Figure 17 displays a comparison of the breakeven grid extension distance with varying the RO size, control strategy, and size of DG. The red line represents the distance between the location of the VOLUME 8, 2020 case study and the nearest utility grid point (12 km). From Figure 17, it can be concluded that the grid connection is better than all considered cases using the RO-250 unit. For the RO-500 unit, all options of hybrid PV/DG/B are more economically feasible compared with grid connection, and the best cost-effective option is the one including load following control strategy with 10 kW DG. The breakeven distance is 3.31 km.
The initial and replacement cost of the diesel generator is assumed to be $230/kW. Whereas, the O&M cost is considered to be $0.1/h based on an operation lifetime of 15,000 h. The diesel price in Egypt is $0.428/ [55]. However, this value can be increased in the located far regions because of the high transport fee. The optimization results confirmed that the best size of the DG is 25 kW and 45 kW, respectively, for RO-250 and RO-500. The COE and NPC for DG only with RO-250 are $0.164/kWh and $604298, respectively. Whereas with RO-500, the COE and NPC are $0.171/kWh and $630856, respectively.
From the environmental impact, the stand-alone diesel generator produces 119110 kg/year and 117677 of CO 2 respectively for RO-250 and RO-500. Such quantity can be significantly reduced thanks to PV/DG/B system. Also, the other pollutants were reduced compared to the DG system. Table 11 shows the number of different pollutant emissions by different options of PV/DG/B system compared to DG. Consequently, along with the PV/DG/B system can reduce the CO 2 emission, which affects Global Warming.

VIII. CONCLUSION
Optimization, feasibility, economic evaluation, and energy management of hybrid photovoltaic-diesel-battery (PV/DG/B) system to pump and desalinate water at isolated regions have been done in this paper. The case study represents a flat 70 acres (283280 m 2 ) located in Minya city (Egypt). The main findings can be summarized as follows; • From the economic point of view, the minimum cost of energy and the minimum total present cost are 0.074 $/kWh and 207676 $, respectively. This is achieved with RO-500 using load following (LF) dispatch control strategy.
• The related sizes to the best option of PV/DG/B are 120 kW PV array, 10 kW DG, 64 batteries, and 50 kW converter.
• The load following strategy decreased the cost of energy with RO-500 by 22.2% and 31.48% respectively for DG size of 5kW and 10 kW compared with the cycle charging (CC) control strategy.
• The fuel cost records the maximum value of 167737$ using RO-250 with 10 kW of DG and CC strategy. It increased by 92.65% compared with 5 kW of DG and CC strategy.
• The cost of fuel is mainly influenced by the dispatch control strategy. It is equal 123932 $ and 31344 $ respectively for RO-500 with CC and LF strategies with the same size of DG (10 kW). This means that the cost of fuel reduced in the case of LF by 74.71% compared with the CC strategy.
• The grid connection is better than all considered options when using the RO-250 unit.
• For the RO-500 unit, all options of hybrid PV/DG/B are more economically feasible compared with grid connection, and the best cost-effective option is the one including LF strategy with 10 kW DG. The breakeven distance is 3.31 km.
• From the environmental impact, the stand-alone diesel generator produces 119110 kg/year and 117677 kg/year VOLUME 8, 2020 of CO 2 respectively for RO-250 and RO-500. This can be significantly reduced thanks to PV/DG/B system.
• Along with the PV/DG/B system being more economical, the CO 2 emission, which affects Global Warming, is reduced.