You are on page 1of 6

2nd International Conference on Renewable Energy Research and Applications

Madrid, Spain, 20-23 October 2013

Optimal sizing of a stand-alone photovoltaic system


Ahmed bouabdallah, Salvy Bourguet, Jean-Christophe Olivier, Mohamed Machmoum
Institut de Recherche en Energie Electrique de Nantes Atlantiques - IREENA Saint-Nazaire, France Email: ahmed.bouabdallah@univ-nantes.fr

AbstractThe objective of this study is to optimize the sizing of a stand-alone photovoltaic (PV) system. The optimization of the PV energy generated by horizontal panels (or inclined) xed (or movable) for a given site is realized. The estimation of solar radiation on inclined panels based on the measurements taken on a horizontal plane is presented. Then, an optimization of the inclination angle (tilt angle) is proposed to maximize the annual PV energy production. The energy efciency of solar tracking system is analyzed and evaluated. Finally, the optimal sizing for the stand-alone photovoltaic system (SAPS) is presented based on an energy management between the PV panels, the batteries, and the load. Index TermsPV energy, tilt angle, solar radiation, tracking system, sizing system, batteries, optimization.

I. I NTRODUCTION For remote locations where the connection with the electrical grid cannot be easily realized but solar energy is sufciently available, PV technology, which is completely clean, is regarded as a better solution to produce electrical energy. For recent years, PV technology has been much improved to achieve higher energy efciency with well developed panels. However, this technology employed in a SAPS has a signicant disadvantage which is the high cost of its kWh compared to the classical electric grid [1]. One efcient solution for this problem is to size optimally the SAPS. In our study, the weather conditions of Saint-Martin en Haut (France) is considered to size the SAPS shown in Fig. 1. The solar energy is an intermittent natural source which determines that the electrical energy generated from PV panels is discontinuous, thus, in a SAPS, the use of storage device such as batteries becomes necessary. During the day when there is sufcient solar, the excess energy produced by the PV panels compared to the request from the load is stored by the batteries. For the periods when there is no sun such as cloudy or at night, the batteries supply the load with the stored energy. A DC-DC converter associated to the PV panels is controlled by a maximum power point tracker (MPPT) to maximize the power produced by the panels all the time. The batteries are connected to the continuous DC bus directly or through a bidirectional DC-DC converter. An inverter assures the interface between the load and the DC bus. Its role is to convert the DC power to any required AC power of the load. In this paper, a technical-economic optimization based on LPSP (Loss of Power Supply Probability) method is used [2] [5]. The objective is to nd the bi-objective Pareto front in the plane LPSP - Cost of SAPS, which means to nd all the

Fig. 1. Stand-alone photovoltaic system (SAPS ).

optimal numbers of PV panels and batteries for sizing the SAPS. The paper is organized as follows. In section II the solar radiation for inclined panels and the energetic model of PV panels are presented. In section III, the calculation of electric power generated by the PV panels is presented for three cases: horizontal panels, xed-inclined panels and mobile panels. Section IV is dedicated to present the optimal sizing algorithm with the energetic models of the different parts of the SAPS. In section V, simulation results are given with discussion of the three cases. Section VI is the conclusion. II. ESTIMATION OF SOLAR RADIATION ON INCLINED PANELS The determination of solar radiation requires knowledge of angle of incidence i and zenith angle z (see Fig. 2), which depend on the geographical location of the site, assuming that tilt angle and azimuth angle are known. Usually, available data on solar radiation are derived from measurements on horizontal planes Gh , such as: Gh = Gdirh + Gdif h (1)

The components Gdirh and Gdif h represent the direct radiation and diffused radiation received by a horizontal plane. Using the model of the diffused radiation (Avg-5) established in [6], the components Gdif h and Gdirh are determined h based on the measurements Gh and the coefcient kt (kt = G G0 such as G0 is the extraterrestrial radiation). Once these components are determined the total radiation Gtot received by a tilted PV panel can be deduced. Gtot is

ICRERA 2013

978-1-4799-1464-7/13/$31.00 2013 IEEE

2nd International Conference on Renewable Energy Research and Applications

Madrid, Spain, 20-23 October 2013

the sum of three components: direct radiation Gdir , diffused radiation Gdif and reected radiation Gr [6][9]. Gtot = Gdir + Gdif + Gr Where: Gdir = Gdirh Rb (3) (2)

III. SOLAR RADIATION AND PV POWER CALCULATION A Matlab program has been established based on the equations (1) to (9) with data of the measurements on the site of Saint-Martin en Haut (France). The longitude equals to 4.55 east, latitude equals to 45.66 north and elevation equals to 730 meters. These measurements are recorded hourly on horizontal plane for the whole year of 2011. The power generated by the PV chain is presented in Fig. 3. The characteristics considered of the PV chain are manuf = 13%, ch = 95%, t = 0.0045, N OCT = 47 C and GN OCT = 800 W/m2 . To better assess the energy efciency of tracking systems of two-axis (denoted TS-2), a daily energetic comparison with the case of xed and inclined panels (denoted FI) was made, considering the optimal values of tilt angle and azimuth angle ( = 35 and = 3 east), where. Emonth 1 = j
T

Gdif = Gdif h (Ai Rb + (1 Ai )Rdif )[1 + f sin3 (/2)] (4) Gr = Gh Rb (5) Where Rb =
1+cos cosi h , Ai = Gdir cosz , Rdif = 2 G0 , 1cos , and is the ratio of solar energy 2

f =

Gdirh Gh ,

Rr = reected by the ground to the incident solar energy (called albedo ( 0.2)). The total radiation then takes the following form: Gtot =(Gdirh + Gdif h Ai )Rb + Gdif h (1 Ai )Rdif [1 + f sin3 (/2)] + Gh Rr (6)

P (t)t
t=1

(10)

The power generated by a PV panel depends on the PV chain efciency pv , the surface of the panel S (or eld), and the solar radiation Gtot calculated by equation (2). This power can be expressed as follows [10] [11]: Ppv = pv SGtot (7)

The efciency pv depends on the temperature of the PV cells Tc [11] [12]: pv = manuf ch [1 t (Tc TN OCT )] Tc = Ta + [TN OCT (273 + 20)] Gtot GN OCT (8) (9)

Emonth : is daily energy for each month [W h/m2 /day ], j = 28, 30 or 31 [day], t = 1 [hour], T = j (day) 24 (hours) [hour], P(t) is the hourly average power [W]. The tracking system with two-axis (denoted as TS-2) has a considerable energy gain Emonth compared to the case of the tilted xed panels (denoted as FI), and a more important energy gain in comparison with the case of the horizontal panels (denoted as H) as show in Fig. 4 [6]. IV. O PTIMAL SIZING Several techniques such as presented in [12][14] are used for optimal sizing of SAPS. In this section, an energy management between the solar panels and batteries is presented; using a technical-economic optimization based on LPSP (Loss of Power Supply Probability) method. The proposed algorithm uses the models of each component PV, battery, load, and inverter, associated with an economic analysis part for the calculation of the cost of the SAPS.

Where manuf is the panel efciency indicated by the manufacturer, ch is the efciency of PV conversion chain (MPPT with DC-DC converter), t is a temperature coefcient, TN OCT is the cell temperature under the NOCT conditions (NOCT: Normal Operating Cell Temperature) and Ta is the ambient temperature.

200
* + , % &

X: 12 Y: 351 Z: 158.4

"


Power [ W / m2 ]

"

150 100 50 0 400

'

&

200

Days

0 0

10 Hours

15

20

25

Fig. 2. Solar angles shown with a PV panel [6].

Fig. 3. PV power on an inclined panel with = 35 and = 3 on the site: Saint Martin en Haut - France [6].

ICRERA 2013

2nd International Conference on Renewable Energy Research and Applications

Madrid, Spain, 20-23 October 2013

300 250
Power [ W ]

200 150 100 50 0 0 1000 2000 3000 4000 5000 Hours 6000 7000 8000

Fig. 4. Daily energy(Wh/m2 /day) produced by the PV chain.

Fig. 5. Data load extrapolation [W].

A. PV panel model The PV model developed in part II with the same characteristics mentioned in part III is used. Note that the data used in the algorithm sizing is introduced hour by hour for 365 days which leads to 8760 hours for a whole year. B. Battery model A lead battery model (CIEMAT model [15] [16]) is used. This model expresses the battery voltage Vbat (charge/discharge) depending on its state of charge SOC, its current Ibat , its capacity Cbat and the temperature variation T . The capacity of the battery Cbat is normalized regarding C10 . During the charge: Vbat = nb [2 + 0, 16.SOC ] + nb . |Ibat | . C10 6 1+
Ibat .C10 Cbat 0,86

100 80
Efficiency [%]

60 40 20 0 0

0.2

0.4 P /P
out

0.6

0.8

nominalinv

Fig. 6. Inverter efciency [%].

0, 48 1 SOC 1,2

D. Inverter model The inverter is modeled by its efciency inv [2] (see Fig. + 0, 036 . 6) which is expressed as the ratio between the load output power POU T and the input power PIN : POU T PIN PIN is the sum of the output power and the power losses Ploss (power consumed by the switches: IGBT or bipolar transistors) such that: inv = PIN = POU T + Ploss Ploss is the sum of a constant power which depends on the characteristics of the inverter and power that depends on the T , such as: square of the ratio POU Pn Ploss = (p0 + k.p2 ) Pn With: 1 10 1 1 out p0 = 99 ( 10 100 9)2 , k = 100 p0 1, p = P Pn Pn : rated power of the inverter; 10 and 100 : inverter efciency at 10% and 100% of Pn respectively. E. Economic model The total annualized cost Ctot an for an investment project is dened as the sum of capital annualized cost Ccap an , the

(1 0, 025T ) During the discharge: Vbat = nb [1.965 + 0, 12.SOC ] + nb . |Ibat | . C10 4 1+


Ibat .C10 Cbat 1,3

0, 27 + 0, 02 . SOC 1,5

(1 0, 007.T ) nb : is the number of cells in series; The SOC can be calculated from the previous state of charge at any time by integrating the current Ibat : SOCt = SOCt1 (+/) CQ such that: Q = I t. bat (+/): (Charge/discharge) of the battery. C. Load model The load prole taken in our study is a prole of a typical home where the heating is non-electric [17]. A database of real measurements (in France) averaged over the month is used. An extrapolation has been made to obtain the 8760 data (i.e. the same load for all the month).

ICRERA 2013

2nd International Conference on Renewable Energy Research and Applications

Madrid, Spain, 20-23 October 2013

operation and maintenance annualized cost CO&M an and the replacement annualized cost Crep an for devices which have a lifetime less than a lifetime of project: Ctot
an

TABLE III INVERTER PARAMETERS Pn (W) Cost (e/W) Replacement cost (e/W) Maintenance cost for the rst year (e/W) Lifetime (year) 1.2 Ploadmax 200 + 0.35 Pn 200 + 0.35 Pn 34.5 (3% of capital cost) 10

(Ccap

an )comp

+ (Crep
an )comp

(CO&M

an )comp

comp = PV, Batteries, Inverter. The cost is annualized for 25 years (lifetime of SAPS), it takes into account the initial capital cost, the cost of maintenance, and the cost of replacement [2] [4]. The unit cost and parameters of PV panels, batteries and inverters are available in Tables 1, 2 and 3. F. LPSP The Probability to loss a power supply (LPSP) is the ratio of the accumulation of LPS (Loss Power Supply) in all the year on the total electrical annual load: LP SP = With:
Q

<

>

>

<

>

>

8760 t=1

LP SP (t)

[ _

Eloadyear
Q

LP S = Eload (Epv + Estock )


Estock = (SOC (t) SOCmin ) Ebat


c d e f g e d h i u D G

And: Eloadyear =
8760 t=1
v w x

Pload t [kW h] 1000

Where Pload (t) is the hourly average load power [W] and t = 1 hour.
Fig. 7. Algorithm to optimize the sizing of the SAPS. TABLE I PV PANEL PARAMETERS Peak power (W) Efciency Capital cost (e/m2 ) Installation cost (e/m2 ) Maintenance cost for the rst year (e/m2 ) Lifetime (year) 130 12.35 275 100 (40% of capital cost) 2.75 (1% of capital cost) 25

G. Energy management The algorithm compares the energy Epv produced and the energy Eload demanded for each hour (see Fig.7), and decides if a battery must be charged or discharged according to the value of the difference E such that: E = Epv Eload The tests performed by the algorithm are: If E > 0 then charge the battery If E < 0 then discharge the battery If E = 0 then calculate the auto-discharge of the battery V. RESULTS AND DISCUSSION A Matlab program has been developed for sizing the SAPS (see Fig. 1) according to the algorithm of energy management presented in Fig. 7. There are two important factors (objectives) to be considered for this study: the annualized cost of SAPS and the quality

TABLE II LEAD ACID BATTERIES PARAMETERS Voltage (V/cell) Capacity (Ah) SOCmax SOCmin SOCinitial Capital cost (e/kWh) Installation cost (e/kWh) Maintenance cost for the rst year (e/kWh) Lifetime (year) 2 50 0.9 0.2 0.9 1150 1150 34.5 (3% of capital cost) 5

ICRERA 2013

2nd International Conference on Renewable Energy Research and Applications

Madrid, Spain, 20-23 October 2013

LPSP [ % ]

of service (characterized by the LPSP rate). In this study three cases of optimal sizing are presented: Case of horizontal panels (noted C-H). Case of xed and inclined panels (noted C-FI). Case of mobile panels (noted C-TS-2). A. Case of horizontal panels The measurements taken out in the horizontal plane in the site of Saint-Martin en Haut in France are used directly to calculate the power developed by the panels using the equations 7 to 9 (see paragraph III). A bi-objective Pareto front in the plane LPSP - Cost of SAPS is presented in gure 8 (starts points). Blue points represent all possibilities of sizing the SAPS, and Pareto front represents the optimal sizing of the SAPS. That means for each choice of LPSP, the minimum corresponding cost is situated in the Pareto front. B. Case of xed and inclined panels With the same energy management used in the case A, the sizing of the SAPS has been established and the bi-objective Pareto front in the plane LPSP - Cost of SAPS has been determined and presented in gure 9 (starts point). In this case the PV panels has been inclined and oriented considering the optimal values of tilt angle and azimuth angle ( = 35 and = 3 east). C. Case of mobile panels In this case the panels are movables around two axes (see paragraph III). The bi-objective Pareto front in the plane LPSP-Cost of SAPS has been determined and presented in gure 10 (starts point). This case presents a little difference regarding the two cases A and B (C-H and C-FI) in the Pareto front: for the same LPSP the optimal cost is little higher in C than in the two cases A and B. Noted that for the three cases the two factors LPSP and Cost of SAPS are inversely proportional on the Pareto front, and two points of this front can be noted: The rst where the satisfaction of the load is maximum that corresponds to LPSP = 0. The second where the satisfaction of the load is minimum. The table 4 summarizes the optimal sizing for the three cases taking into account the two points previously cited. For the C-H and C-FI the results are almost the similar for the values of LPSP, Cost of SAPS, PV numbers and batteries numbers. In fact the Pareto front is quasi the same for both cases. In the third case C-TS-2 for the same numbers of PV and batteries which correspond to 2831e the LPSP decreases to 44.27 %, on the other hand for LPSP equal to 0 the cost of the SAPS increases to 19290e (almost 8% more expensive than C-FI). These results mean that for the site of Saint-Martin en Haut in France the horizontal panels can largely satisfy the load regarding the case of xed-inclined panels and the case of mobile panels.

50 40 30 20 10 0 0

General points Pareto front

0.5

1 Cost [ euros ]

1.5 x 10

2
4

Fig. 8. Pareto front for two factors (red points): cost of SAPS and the desired LPSP (horizontal case).

50 40

General point Pareto front

LPSP [ % ]

30 20 10 0 0

0.5

1 Cost [ euros ]

1.5 x 10

2
4

Fig. 9. Pareto front for two factors (red points): cost of SAPS and the desired LPSP (xed and inclined).

50 40

Genaral points Pareto front

LPSP [ % ]

30 20 10 0 0

0.5

1 Cost [ euros ]

1.5 x 10

2
4

Fig. 10. Pareto front for two factors (red points): cost of SAPS and the desired LPSP (mobile panels).

However, the difference between the three cases is situated in the energy produced (see Fig. 4), and thereafter in the surplus energy produced SEP, where for each hour: SEP = Epv Eload And P SEP =
8760 t=1

SEP (t)

Eloadyear

Where PSEP represent probability of surplus energy produced. This PSEP is calculated for the Pareto front in the plane LPSPCost for the three cases and is shown in Fig.11.

ICRERA 2013

2nd International Conference on Renewable Energy Research and Applications

Madrid, Spain, 20-23 October 2013

TABLE IV OPTIMAL SIZING FOR THE THREE CASES C-H, C-FI AND C-TS-2 Case C-H C-FI C-TS-2 LPSP% 0 48.5 0 47.21 0 44.27 Cost of SAPS (e) 17400 2832 17880 2832 19290 2831 PV m2 70 5 75 5 90 5 Batteries (Ah) 2500 500 2500 500 2500 500

3000 2500
PSEP [ % ]

CH CFI CTS2

2000 1500 1000 500 0 0 0.5 1 Cost [ euros ] 1.5 x 10 2


4

models used for battery, load and inverter with economic analysis is presented with details, for the case of xed and movable panels. Thereafter, the Pareto front gives the optimal sizing of SAPS, and each point on this front, i.e. for a given LPSP and Cost, corresponds to an optimum combination PV/Batteries. The interest of the method is particularly in the provision of this Pareto front which is a decision support for a designer of such a system. Although this study shows that for the site of Saint-Martin en Haut in France where is heavily cloudy and for the SAPS, the horizontal or inclined panels are largely sufcient to satisfy the load regarding the tracking system, the developed methodology remains valid and interesting for other sites, where the load prole is different and the climatic conditions are most pleasant. R EFERENCES
[1] K. Branker, M. Pathak, and J. Pearce, A review of solar photovoltaic levelized cost of electricity, Renewable and Sustainable Energy Reviews, vol. 15, no. 9, pp. 44704482, Dec. 2011. [2] S. Diaf, G. Notton, M. Belhamel, M. Haddadi, and a. Louche, Design and techno-economical optimization for hybrid PV/wind system under various meteorological conditions, Applied Energy, vol. 85, no. 10, pp. 968987, Oct. 2008. [3] Z. Wissem, K. Gueorgui, and K. H edi, Modeling and technicaleconomic optimization of an autonomous photovoltaic system, Energy, vol. 37, no. 1, pp. 263272, Jan. 2012. [4] H. Yang, W. Zhou, L. Lu, and Z. Fang, Optimal sizing method for stand-alone hybrid solarwind system with LPSP technology by using genetic algorithm, Solar Energy, no. 4, pp. 354367, Apr. 2008. [5] W. Shen, Optimally sizing of solar array and battery in a standalone photovoltaic system in Malaysia, Renewable Energy, no. 1, pp. 348 352, Jan. 2009. [6] A. Bouabdallah, S. Bourguet, J. C. Olivier, and M. Machmoum, Photovoltaic energy for the xed and tracking system based on the modeling of solar radiation, IECON 2013 - 39th Annual Conference on IEEE Industrial Electronics Society, in press. [7] L. Wong and W. Chow, Solar radiation model, Applied Energy, vol. 69, no. 3, pp. 191224, Jul. 2001. [8] J. A. Dufe and W. A. Beckman, Solar Engineering of Thermal Processes. John Wiley & Sons, 2006. [9] P. Loutzenhiser, H. Manz, C. Felsmann, P. Strachan, T. Frank, and G. Maxwell, Empirical validation of models to compute solar irradiance on inclined surfaces for building energy simulation, Solar Energy, vol. 81, no. 2, pp. 254267, Feb. 2007. [10] M. Kolhe, Techno-Economic Optimum Sizing of a Stand-Alone Solar Photovoltaic System, IEEE Transactions on Energy Conversion, vol. 24, no. 2, pp. 511519, Jun. 2009. [11] H. Zhou, W. Sun, D. Liu, J. Zhao, and N. Yang, The Research of Daily Total Solar-Radiation and Prediction Method of Photovoltaic Generation Based on Wavelet-Neural Network, 2011 Asia-Pacic Power and Energy Engineering Conference, pp. 15, Mar. 2011. [12] R. Luna-Rubio, M. Trejo-Perea, D. Vargas-V azquez, and G. R osMoreno, Optimal sizing of renewable hybrids energy systems: A review of methodologies, Solar Energy, Dec. 2011. [13] S. B. Silva, M. A. G. Oliveira, S. Member, and M. M. Severino, Sizing and Optimization Photovoltaic , Fuel Cell and Battery Hybrid System, America, vol. 9, no. 1, pp. 8388, Jun. 2011. [14] I. Biswas, V. Dash, and P. Bajpai, Sizing optimization of PV-FCBattery system with hybrid PSO-EO algorithm, 2012 Annual IEEE India Conference (INDICON), pp. 869874, Dec. 2012. [15] J. F. Manwell and J. G. McGowan, Lead acid battery storage model for hybrid energy systems, Solar Energy, vol. 50, no. 5, pp. 399405, May 1993. [16] N. Achaibou, M. Haddadi, and a. Malek, Modeling of Lead Acid Batteries in PV Systems, Energy Procedia, vol. 18, pp. 538544, Jan. 2012. [17] REMODECE Homepage. [Online]. Available: http://remodece.isr.uc.pt/

Fig. 11. The surplus energy correspond to the Pareto front for the three cases C-H, C-FI and C-TS-2.

The table 5 takes the same values of LPSP in the table 4 with the corresponding values of PSEP. For the three cases CH, C-FI and C-TS-2 when the load is not satised (almost half satised) the difference between their PSEP is not much. And when the load is completely satised (LPSP=0) the difference between the three cases is remarkable: regarding the surplus of energy, the calculated systems give respectively 1100% (C-H), 1400% (C-FI) and 2400% (C-TS-2) of the load. The PSEP rate means a lost in the system studied investissment, where is no necessary to install a tracking system. The one way to recover the surplus energy is to think about installing secondary electric loads (heating, water pump) which will help to offset the overall investissment. VI. C ONCLUSION The PV power is rst based on the measurements and solar radiation model. Results can be used as a source of data for the design of the optimization algorithm. In this paper a sizing optimization method with different

TABLE V THE SURPLUS ENERGY FOR THE THREE CASES C-H, C-FI AND C-TS-2 Case C-H C-FI C-TS-2 LPSP% 0 48.5 0 47.21 0 44.27 PSEP 1176 33.63 1461 45.44 2319 72.65

ICRERA 2013

You might also like