Control and Optimal Sizing of PV-WIND Powered Rural Zone in Egypt

Transcription

1 Control and Optimal Sizing of PV-WIND Powered Rural Zone in Egypt Hanaa M. Farghally, Faten H. Fahmy, and Mohamed A. H.EL-Sayed Electronics Research Institute, National Research Center Building, Cairo, Egypt Center of Energy Studies, University of Trinidad of Tobago hanaa abdallah Abstract-Due to intermittent natural energy resources and energy resources seasonal unbalance, a PV-wind hybrid electrical power supply system was developed for many remote locations where a conventional grid connection is inconvenient or expensive. This paper presents the sizing of a complete PV-Wind hybrid system for supplying electricity to emergency hospital, school and home buildings according to their energy requirements. The computer program (HOMER Pro.) solves the optimization problem to minimize the objective function considering the different constraints and provides the optimum wind, solar and battery ratings. Also, a neural network controller is developed for achieving the coordination between system components as well as control the energy flows. Keywords-hybrid system, sizing, photovoltaic system, wind system, battery bank. Optimization problem, neural controller. I. INTRODUCTION Renewable energy resources such as solar and wind are abundant energy sources. The supply of renewable energy resources is inexhaustible practically, and using renewable energy poses very little environment risk. In rural remote areas, electricity could still be available if stand-alone renewable electricity power sources are used.one of the most promising applications of renewable energy technology is the installation of hybrid energy systems, where the cost of grid extension is prohibitive and the price of fuel increases drastically with the remoteness of the location. It has been demonstrated that hybrid energy systems can significantly reduce the total life cycle cost of stand-alone power supplies in many situations, while at the same time providing a more reliable supply of electricity through the combination of energy sources []. This work presents the sizing of an optimal PV-Wind hybrid system. Also, in this paper a neural network control strategy for achieving the coordination between system components as well as control the energy flows is proposed. Fig. shows schematic diagram for PV- Wind hybrid system in rural zones II. SITE CHARACTERRISTICS Figure (): Hybrid energy system components. III. RESOURCE AND LOAD DATA The Red-Sea Coast is known as a rich area of wind energy and solar energy. The available wind and solar resources greatly influence both the configuration and the cost of a hybrid system. Monthly average wind speed and solar insolation data for the selected remote area are shown in Figs &3 []. The electrical load data gives detailed information about the equipments to be powered: their number, nominal power and the number of hours of operation in a typical day. The load profiles can be deduced from the load data. Fig. 4 shows the load profiles of the three buildings for the winter and summer seasons. These data are used in system sizing, and the generation and load are assumed to keep constant in each hour interval. Figure (): Monthly average solar insolation The chosen zone is a remote area in Eastern desert was selected as the site under consideration The area is located on the Red Sea approximately km north of the city of Safaga Latitude 7 7', north and Longitude ', east [] Reference Number: W-6 88

2 cost is used for the wind turbine, but the annual cost for the PV system has been neglected. A.6 kwh Torjan L6P battery was selected for this study. The complete parameters of the battery are shown in table 3. Figure (3): Wind speed monthly average. IV. COMPONENTS The performance data of the wind turbine and PV modules used in this evaluation are provided in Tables & A kw Bergey BWC Excel-S wind generator and V, 75 W PV module were selected for the study. The rated power of the wind turbine and PV panel, and the number of turbines used are free inputs to the program and can be changed Eletrical load (W) Electrical load (W).5 x Home Hospital School Total load x School Hospital Home Total load 5 5 Figure (4): Load profile of a typical complete day s consumption in summer and winter seasons. The capital costs listed in the above tables include all installation and balance of plant costs. The annual maintenance costs for the wind turbine(s) and PV panels In the current study an annual maintenance cost of 3% of capital Table (): PV module parameters [3] Maximum power (W) 75 Efficiency (%) Area (m).67 Annual maintenance cost ($) Total capital cost ($) 384 Table (): Wind turbine parameters [4] Rated power output, watt at 9 mph (3 m/s) Cut-in speed 7 mph (3. m/s) Start- up wind speed 7.5 mph (3.4m/s) Furling wind speed 35mph (5.6m/s) Maximum design wind speed mph (53.6 m/s) Type 3 Blade Upwind Rotor Diameter 3 ft (7 m) Air density (kg/m3).5 Total capital cost ($) 7 Annual maintenance cost ($) 8 Table (3): Battery parameters [5] Voltage (V) Capacity (kwhr).6 Roundtrip efficiency (%) 85 Minimum charge (%) 3 Capital cost ($) Lifetime (yrs) 5 V. POWER OUTPUT V. Wind Turbine Average hourly wind speed data is evaluated and converted to wind turbine power. If the speed is between the cut-in and the rated speed of the wind turbine, then the power output is defined as [6] PWind ( t ) = 3 A v( t ) C p Eff ad () where is the air density (kg/m3), A is the swept area of the rotor (m ), v is the wind speed (m/s), C p is the efficiency of the wind turbine, and Effad is the efficiency of the AC/DC converter (assumed to be 9% in this study). If wind speed is between the rated wind speed and the furling speed of the wind turbine, the power output will be equal to the rated power of the turbine. Finally, if the wind speed is less than the cut-in speed or greater than the furling speed, there will be no output power from the turbine. The power calculated for each hour is multiplied by one hour to determine the energy produced (kwh) for the particular hour Reference Number: W-6 89

3 V. PV Arrays Insolation data is converted into power output from the PV array using the following equation [6] P ) PV ( t) = Ins( t AEff pv () Where Ins(t) is the insolation data at time t (kw/m), A is the area of a single PV panel (m), and Eff pv is the overall efficiency of the PV panels and the DC/DC converter. Eq.() assumes that the temperature effects (on PV cells) are ignored. VI. BASIC ECONOMICAL CONSIDERATIONS HOMER (hybrid optimization model for electric renewables), is used to identify the cost-minimizing combination of power generating technologies that would suit the load profiles. The cost minimization model is as follows: [7] Minimize cos t( C ) n = ( CIi + CRi + COMi ) (3) i= E g N The objective function to be minimized is the cost producing one kilowatt-hour of electricity [7] Subject to ( load power output) maxunserved load( %)* Load Where C is the cost that we want to minimize, i the index is made to account for wind, solar, and batteries (for our case n = 3) C I is the initial capital cost, C R is the replacement cost per power generating machine, and C OM is the operation and maintenance cost per power generation machine. Eg is the yearly energy demand and N The lifetime of the project In this analysis, the annualized capital cost is calculated by evenly spreading the initial capital costs and the fixed operating costs of a component over a projects expected lifetime. In addition, net present cost (NPC). The (NPC)o f an integrated system takes into account the initial capital investment (C I ), the present value of operation and maintenance cost (C OM ), the wind system replacement cost (C R ) and the battery replacement cost (C R3 ) The initial capital investment for the integrated system, C I is given as [8] 3 C I = CIi = S S W W + i= [( Z P ) + ( Z P ) ( Z P )] where Zs cost of solar system per kw Z w cost of wind system per kw Z B cost of battery per kwh P s total capacity of solar panels, kw P w capacity of the wind machine, kw P B total capacity of the battery bank, kwh B B (4) (5) The operation and maintenance cost OM is calculated as follows: Where, COM CN L = (6) C = mc I (7) m specified percentage of the initial capital, N L the life period of the system. Replacement costs CR and CR 3 are calculated as follows: Considering the escalated rate (e) and discount rate (d) the replacement cost of WECS, C R is given as C R n N + e = Pw Z w J = + d RW J nrw + n RW is the number of wind turbines types N J the lifetime of the wind turbine type J Similarly, battery replacement cost RP 3 is estimated from: n N + e CR3 = PB Z B J = + d RB J nrb + n RB is the number of batteries types (8) (9) C R =C R +C R3 () Then the (NPC) can be calculated from the following expression: NPC = CI + COM + CR () VII. CONTROL STRATEGY The block diagram of the PV/ Wind / Battery hybrid energy system with the associated control signals is shown in Fig.5. In this diagram, two goals should be achieved. Coordinate the interaction between system components.. Control the energy flows between the various energy sources, energy storage and the load As a basic control rule, the energy produced by renewable sources must be preferentially used to feed the loads. For every hour, if the renewable sources (PV and wind) produce more energy than is demanded, the surplus power can be used to charge the batteries. If, on the contrary, the renewable sources are unable to provide for all the energy demand, discharge, should be produced by the batteries. We have two different cases for operation control technique; - When the error is greater than zero - When the error is less than zero Reference Number: W-6 9

4 For the error in case (), the controller puts the battery in case of charge with limiter charge controller (not exceed the maximum charging current).in case the controller links the battery discharge with discharger controller. The controller can take the appropriate decision to control the battery, which depends on the load changing requirements, the solar insolation change patterns, the wind speed change and the maximum charging current of the battery. VIII. NEURAL NETWORK STRUCTURE VIII. NN Learning Strategy A basic component of a neural network is described by a set of interconnected weights, ( wij ), a node activation function, F, and a bias (qi). In this description, the output, ai of the ith element of the network is determined by mapping of the effective input, xi, through an activation function at element. The output of each basic processing element can be determined by different activation functions. A convenient choice for the activation function is the sigmoidal function given below: x ( e ) ai = F( X ) = sigm( x ) = / + () In order to establish the NN model, the interconnected weights ( wij), and biases (qi), are trained according to the existing input/output patterns. The process is intended to p minimize the error between the network outputs ( d i ), and p the actual outputs ( a i ), for the same inputs. By defining the P error measure, E, as the total quadratic error for pattern p at No output units, i.e.: E P p p ( d a ) =. 5 (3) i i The learning process is to adjust the weights and biases based P on the training pattern p to minimize the error measure E, in a gradient descent manner. It is clear that if the error measure is minimized for all patterns, the overall measure is also minimum. Due to the analytic nature of the sigmoidal activation function adopted here, the error minimization process can be backwardly traced from the output layer towards the hidden layer. As a result, using an iterative procedure, called the generalized delta rule, the adjustment of the network weights and biases for minimizing P E is obtained. VIII. NNC Model The NN controller is designed to give three outputs; these are as given the controller output current Ibattery, the PV Array switching, the wind turbine switching. It is cleared that the used PV is able to generate electricity, whenever there is solar illumination; where the electric power is proportional to the incident insolation level. The wind turbine begins to produce output when the wind speed is greater than the cut-in speed until furling speed V F is reached, at which point the machine shuts down. Off line training for the proposed NNC was applied. Data for off-line training can be obtained either by simulation or experiment. The network is trained to recognize the relationships between the input and output parameters. The BP algorithm uses the supervised training technique. In this technique, the interlayer connection weights and the processing elements thresholds are first initialized to small random values. The network is then presented with a set of training patterns, each consisting of an example of the problem to be solved (the input) and the desired solution to this problem (the output). These training patterns are presented repeatedly to the ANN model and the adjustment is performed after each iteration whenever the network s computed output is different from the desired output. This process continues until weights converge to the desired error level or the output reaches an acceptable level. For this present work, the data is obtained by simulating the proposed system. After many trials, the developed NNC, shown in Fig.6, eventually employed a 4-neuron input, an 8- neuron hidden layer, and 3-neuron output layer. The input network parameters are; the reference load and the difference between the output generation, the reference load, the wind speed and the solar insolation. For the present work, such NNC structure gave satisfactory results with small number of neurons, hence better in terms of memory and time required to implement the NN in control. The transfer function used in the hidden layer is the sigmoid, while purellin transfer function is used for the output layer. The training process has been carried out over epochs. Reference Number: W-6 9

5 Figure (5): Proposed Control system of the hybrid PW/wind/battery system Figure (6): NNC structure. IX. RESULTS In this part the results of applying the optimization program are displayed as shown in table 6. Hybrid system offers the much lower NPC and LCOE with PV and wind combined together On the other hand, power system 3 have higher Initial Capital Costs, NPC. These results are reflected in the higher LCOE. While system has higher replacement cost.it Reference Number: W-6 9

6 is cleared that The use of initial capital cost as the primary decision criteria in remote area power generation systems is misleading and inappropriate. Relying on LCOE, NPC as the key decision parameters instead of using initial capital cost will result in the better selection of remote energy technologies. As shown in fig. 5, the load profile is covered through 4 hour with the generating power from PV, wind turbine and battery charger. Also, figure 6 &7 presents the same previous profile but for power systems &3 respectively. As discussed before, the PV-wind system is designed to supply electric power for three different buildings (school, hospital and home buildings). The performance of the system is evaluated during a typical winter and summer day. Figure (6): Power generated from power system matching load through 4 hours. Figure (7): Power generated from power system 3 matching load through 4 hours. Figure (5): Power generated from power system matching load through 4 hours. Fig.8 shows the NNC outputs over 4 hours on the day the data were collected. These are as given the controller output current Ibattery, the PV Array switching, the wind turbine switching. It is cleared that the used PV is able to generate electricity, whenever there is solar illumination; where the electric power is proportional to the incident insolation level. The NNC switch PV on in the morning when sufficient solar radiation is available. While it is switched off in the evening. The wind turbine begins to produce output when the wind speed is greater than the cut-in speed until furling speed V F is reached, at which point the machine shuts down. So the switching signal of the wind turbine exists whenever the wind speed is in that range The system performance under the load profile given in Fig.4. is evaluated. The output power from the PV array in the system over the 4 hour simulation period is shown in Fig.9. The output power from the wind energy conversion unit in the hybrid energy system over the 4 hour simulation period is shown in Fig.. The plot for the battery power is shown in Fig.. Item Optimization sizing results Table (6): Optimization Sizing results. Power system PV-Wind- Battery system system Kw PV array 4 Kw wind turbine 648Kwh battery 35Kw converter Power system Wind- Battery 9Kw wind turbine 864 Kwh battery 35Kw converter Power system 3 PV- Battery system 6 Kw PV array 96 Kwh battery 35Kw converter Capital Cost ($) 36,6 354,7 55,8 O&M ($) 63,539 74,43 77,55 replacement cost ($) 6,9 89,4 54,96 Net Present cost ($) 475,64 55,865 65,5 Levelized cost of energy (COE) ($/kwh) Capacity shortage frac... Reference Number: W-6 93

7 .5 Third NNC output Second NNC output.5 Third NNC Output Second NNC output NNC outputs.5 NNC outputs NNC output (A) First NNC output NNC output (A) First NNC output Figure (8): NNC outputs PV generation (A) PV generation (A) Figure (9): PV generator power over the simulation period WIind generation (A) 3 Wind generation (A) Figure (): Wind generator power over the simulation period Reference Number: W-6 94

8 Battery power (W) x Battery power (W) x Figure (): The battery power over the simulation period X. CONCLUSION In this paper, the sizing of a complete PV-Wind hybrid system is developed. The computer program (HOMER Pro.) solves the optimization problem to minimize the objective function considering the different constraints and provides the optimum wind, solar and battery ratings. In addition, the comparison between three different suggested power systems was illustrated with details. Also, in this paper the feasibility of a neural network controller developed for achieving the coordination between system components as well as control the energy flows is introduced. XI. REFERENCES [] J. W. Twidel, and A. D. Weir, "Renewable Energy Resources", Cambridge, London, 986. [] New and Renewable Energy Authority, Minisstry of Electricity and Energy, Egyptian Solar Radiation Atlas, Cairo, Egypt, 998 [3] Sukamongkol Y., Chungpaibulpatana S., Ongsakul W.," A simulation model for predicting the performance of a solar photovoltaic system with alternating current loads", Renewable Energy, Vol.7,pp ,. [4] Education, RE components, Home power, available at [5] HOMER Pro, Ver..58, National Renewable Energy Laboratory, Golden CO, 4 April. [6] D.B. Nelson, M.H. Nehrir_, C. Wang, "Unit Sizing and Cost Analysis of Stand-alone Hybrid Wind/PV/Fuel Cell Power Generation Systems", Renewable Energy, Vol. 3,pp , 6. [7] Sami K., Carol D., "The Economics of Hybrid Power Systems for Sustainable Desert Agriculture in Egypt", Renewable Energy, Vol.3,pp. 7 8, 5. [8] Rajendra A., Natarajanb E., "Optimization of integrated photovoltaic wind power generation systems with battery storage", Energy, Vol.3,pp , 6. Reference Number: W-6 95