- Email: [email protected]
PV array reconfiguration method under partial shading conditions
Electrical Power and Energy Systems 63 (2014) 713–721
Contents lists available at ScienceDirect
Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes
PV array reconfiguration method under partial shading conditions Koray Sß ener Parlak ⇑ Department of Electronics and Automation, Fırat University, 23119 Elazıg˘, Turkey
a r t i c l e
i n f o
Article history: Received 7 August 2013 Received in revised form 29 May 2014 Accepted 20 June 2014
Keywords: Photovoltaic arrays PV array reconfiguration Partial shaded conditions Maximum power point
a b s t r a c t This paper proposes a new reconfiguration method for photovoltaic arrays under partial shading conditions. With the application the method, the array can produce higher power by reconfigured connections of array. In the proposed method, each row of an array is formed by connecting the panels with close short circuit current levels as possible. A novel algorithm entitled configuration scanning algorithm is used to determine the all possible connection structures. This algorithm utilizes only short circuit current values measured at particular parts of the array. The proposed method has been tested in Matlab–Simulink environment under partial shading conditions. Results of the simulation show that the efficiency of the array is considerably increased with the presented reconfiguration method. Ó 2014 Elsevier Ltd. All rights reserved.
Introduction The use of renewable energy sources has become very important as a result of increasing energy need and the environmental effects of fossil fuels. Photovoltaic (PV) modules are one of the most important renewable energy sources and their use is gradually increasing in regions that have a long sunshine duration and in rural areas. In order to be able to gain a higher solar power, PV arrays are formed by connecting the PV modules in series and parallel. However, the shadows of objects such as clouds, birds, buildings and trees on PV arrays reduce the sunlight that the array receive, inevitably resulting in a lower system efficiency. PV modules need to be operated at the maximum power point (MPP) at varying heat, insolation and load levels for an efficient use. One way of increasing the PV array efficiency is to use maximum power point tracker (MPPT). MPPT is basically a dc–dc converter which utilizes an algorithm to determine the MPP of the current–voltage (I–V) or power–voltage (P–V) characteristic in varying environmental conditions. In order to improve conventional MPPT method, the step size of MPPT algorithm is taken variable instead of fixed [1]. However conventional MPPT methods can fail under partial shading conditions because of occurring more than one local maximas. To find global MPP among local maximas, soft computing technique is developed [2]. Another developed MPPT method is based on obtaining electrical characteristics of PV array. This method works effectively in partial shading conditions [3]. ⇑ Tel.: +90 542 5856185; fax: +90 424 2188947. E-mail address: [email protected] http://dx.doi.org/10.1016/j.ijepes.2014.06.042 0142-0615/Ó 2014 Elsevier Ltd. All rights reserved.
There are reported researches results showing that system efficiency of PV modules exposed to partial shading can be increased by applying some PV module reconfiguration algorithms [4]. In some of these studies, reconfiguration of array has been made by changing connections of all modules on the array [5,6]. In these methods which is called as electrical array reconfiguration module catalog information are used along with instant current and voltage information for each module. By using these data and calculating the current radiant intensity of the modules, the array is reconfigured to form new submodules with the modules that have similar insolation levels. The disadvantage of this method is the high number of possible connections since reconfiguration is applied to all the modules. While a higher efficiency increase is provided with the reconfiguration of all modules in the PV array, quite many connection configurations are possible. In order to compensate this disadvantage, the array can be divided into two parts as the fixed part that consists of statically connected modules and adaptive modules part that can be attached to the fixed part with different configurations [7,8]. In this method, the currents of PV modules are calculated by using the measured voltage of fixed and adaptive parts, temperature of the array, and the catalog data. These currents are used in a bubble sort algorithm and reconfiguration of the array is applied. In the presented another reconfiguration method, connections of all PV modules are reconfigured according to irradiance equalization strategy [9]. This method uses some optimizations to reduce possible connections number. However this number still high since all modules are reconfigured. The method had similar reconfiguration structure to [9] uses PV module equations and number of shaded modules to some calculations [10].
714
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
In the presented method, the shaded modules are determined by measuring current and voltage of all modules in strings. All of these modules are connected to another bus by flexible switching matrix circuit and composed a new string. This string voltage is brought to nominal array voltage by dc–dc converter [11]. Another method tries to bring similar shaded modules to same string by high number switches [12]. This method is similar to [11] but uses more than one power bus and converter to get equal voltage for each string. In another method, all PV modules are reconfigured [13]. It is expected that a lot of switches are required to each module connect to another one. Additionally the algorithm consists of extra calculations to estimate module irradiations. An optimal reconfiguration logic that utilizes mathematical formulas is proposed in [14]. Reconfiguration of the array is performed by using optimization techniques with the voltage values measured at each line and current values measured in all modules in the array. It is obvious that quite many sensors will be needed in this method. In a reconfiguration method which uses ‘‘Rough Set Theory’’ based on ANN under the partial shading conditions, each PV module current in the array is compared to a reference value and proper switching signals are obtained in accordance with a certain rule. However, this method is complex and requires several measurements [15]. Load voltage and temperature parameters of the array are used as inputs to the algorithm which aims to reconfigurating an array with minimum number of switching [16]. The system simulated with four panels gives positive results at constant load but it has some disadvantages. It is difficult to obtain the parameters used in the algorithm and the performance drops when the conditions vary. In fuzzy logic controlled reconfiguration methods, selection of the most suitable connection configuration is aimed by using system and panel parameters [17,18]. Parallel connected modules that are in shade to the reconfigured PV array are proposed in [19,20] to take advantage of whatever power is available in these shaded pieces. In this way, degradation caused by shading effects is minimized. However, increased number of parallel modules in the array will raise the output current, casing in higher losses in conducting wires. In this paper, a reconfiguration method is proposed to increase the efficiency under partial shading conditions. In the proposed method, PV system is divided into two parts, namely adaptive and fixed parts. A new developed configuration scanning algorithm that scans the array and decides how the adaptive parts can be attached to the fixed part to maximize the efficiency has been developed. The main contributions of the proposed reconfiguration method as follows: The method require only short circuit currents where sensed particular nodes of array. Therefore there is no need parameter of PV panel and any additional calculations especially to estimate irradiances. The developed configuration scanning algorithm considers all possible connection configurations with defined terms. Thus the algorithm determines the proper configuration that has highest power under present shading conditions. The method needs minimum number of sensors comparing with other reconfiguration methods. Validity of the method has been tested in a 3 4 PV array which is controlled by MPPT under partial shading conditions. In the next section of the paper creation of the PV arrays and the shading effect on these arrays are explained. Then, proposed reconfiguration method and developed configuration scanning algorithm are explained. Finally, simulation results that show the effect of the proposed algorithm are given.
Connection of PV modules and PV array form Connection in parallel PV modules are connected in parallel between each other to form submodules. While the output voltage is equal in the submodule shown in Fig. 1, the output current is the total sum of module currents.
V out ¼ V 1 ¼ V 2 ¼ ¼ V n Iout ¼ I1 þ I2 þ þ In ¼
ð1Þ
n X Ii
ð2Þ
i¼1
The output power of submodule:
Pout ¼ Iout P out ¼
n X Pi
ð3Þ
i¼1
In this connection structure, as the bathos in any module will not affect the operation characteristics of other modules, all modules can transfer the power they generate to the output. In this way, the highest efficiency can be gained in any environmental conditions. The output voltage is equal to the module voltage in this connection structure and therefore it is low. Hence, the conversion ratio of converter which is going to be used to increase the voltage needs to be high. This affects the efficiency of converter in a negative way. Also, as the output current needs to be high, line losses are significant. Connection in series In the strings which consist of series connection of PV modules, a higher output voltage is obtained. In a string shown in Fig. 2, the currents of modules are equal to each other, and output voltage is the sum total of mudul voltages.
Iout ¼ I1 ¼ I2 ¼ ¼ In
ð4Þ
V out ¼ V 1 þ V 2 þ þ V n ¼
n X Vi
ð5Þ
i¼1
The output power of string:
Pout ¼ Iout P out
ð6Þ
Iout
Vout
PV 1
PV 2
PV 3
PV m
Fig. 1. Parallel connected PV modules.
PV1
PV2
PV3
PVm
Vout Fig. 2. Series connected PV modules.
Iout
715
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
4
Current (A)
3 2 1 0
5
10
15
20
25
10
15
20
25
1000 W/m2 600 W/m2 200 W/m2
60 40 20 0
Array formation
0
5
Voltage (V)
The effect of shading on PV systems The insolation on a PV module affects the module characteristic significantly. This state can be seen from the output current equation of module’s Iph-photo generated current value based on radiation (7).
G ðIscn þ K i ðT T n ÞÞ Gn
Fig. 4. I–V and P–V curves of the PV module at different irradiance levels with constant temperature value.
Current (A)
4 3 2
without bypass with bypass
1 0
0
10
20
30
40
50
0
10
20
30
40
50
100
Power (W)
In order to get a higher current and voltage from PV system, PV array is formed by connecting PV modules in series and parallel. In Series-Parallel (S&P) PV array shown in Fig. 3a, series connected PV modules form the strings (PV11, PV12, . . . , PV1n) and these strings generate PV array by being connected in parallel. Total-cross-tied (TCT) type array shown in Fig. 3b consists of submodules which are formed by parallel connected modules and these submodules (PV11, PV21, PVn1,) are connected series. Under uniform irradiance, the powers that S&P and TCT arrays generate are the same. However, under partial shading conditions which array structure generates more power depends on irradiances of string or submodules that are in the shade and the load of the array.
Iph ¼
0
80
Power (W)
As the current of series modules will be the same, any module which is under the shading effect will affect the performance of the others. As the photo-generated current (Iph) of a module which is under the effect of shade will decrease, the string current is limited with this module current. Also, according to Kircchoff’s voltage law, as a module under the effect of shading will act like a load by carrying a negative voltage, the extra current that the other modules produce flows here and causes heat accumulation. This heat accumulation is called ‘‘Hot Spot’’ effect. In the PV modules working under the effect of hot spot for a long time there can be physical damages. In order to prevent these problems parallel bypass diodes are attached to series connected PV modules. These diodes allow the flow of the current which is the difference between the string current and parallel module current. Thus hot spot effect is ironed out and the shaded modules do not affect the string [21,22].
50
0
Voltage (V)
ð7Þ
As it can be seen in Eq. (7), Iph value of PV module is directly proportional to G (W/m2) which is irradiant intensity. Here, while Iscn shows short circuit current of the PV module, Gn and Tn are rated radiant intensity (1000 W/m2) and temperature values
Fig. 5. The effect of bypass diodes to PV string.
(25°), respectively [23]. Ki is temperature coefficient of short circuit current. I–V and P–V curves for the PV module at different insolation levels at constant temperature are shown in Fig. 4. Simulated
Iout
Iout
PV11
PV12
PV13
PV1n
PV2n
PV21
PV22
PV23
PV2n
PVmn
PVm1
PVm2
PVm3
PVmn
PV11
PV12
PV13
PV1n
PV21
PV22
PV23
PVm1
PVm2
PVm3
Vout
Vout
(a)
(b) Fig. 3. S&P-(a) and TCT-(b) types array connections.
716
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
PV module parameters are Iscn = 3.74 A, Voc = 23 V, Pmax = 65 W. As it can be seen from the figure, Isc and power values of module are affected by insolation that comes to module. The effect of partial shading on series connected PV modules is shown in Fig. 5. Here, I–V and P–V curves of series connected two modules are shown for 1000 and 600 W/m2 radiant intensity levels, respectively. In order to see the effect of bypass diodes, these curves have been drawn both when the diodes exist and when they do not exist. As it is seen, bypass diodes prevent current limiting at particular operating points and increase the available power.
F1 S11
S12
S1m
F2 S21
S22
S2m
Fm
Proposed reconfiguration method
Sm1
Sm2
Smm
In this section, architecture and operation principles of the proposed reconfiguration method are explained. Architecture of the system
A1 Connection structure of the system for proposed reconfiguration method is shown Fig. 6. As can be seen from the figure, PV system consists of three parts. The first one is the fixed part which has more PV panels and its connection structure does not change during operation. This part which is formed in TCT order and shown in Fig. 3b consists of m n pieces of PV panels [‘‘m’’ serially connected submodules, each of which includes ‘‘n’’ parallel connected panels (F11, F12, . . . , Fmn)]. The second part is the adaptive part which consists of m pieces of PV panels. (A1, A2, . . . , Am). As it is seen, adaptive panel number is equal to the number of submodules (row) of fixed part. Under uniform illumination conditions, each adaptive panel is connected to a row of fixed part and the array is in the structure of m (n + 1). Third part is switching matrix circuit which provides the connection of the adaptive panels to the fixed part. S11, S12, . . . , Smm switches which are in circuit shown in Fig. 7 are designed in a way to be able to provide the connection of any adaptive panel
Adapve Part
Fixed Part
F2
Am
Isc(F1), Isc(F2), ..….. Isc(Fm)
Am
Fig. 7. Switching matrix circuit.
to any row of the fixed part. In this circuit 2 m m (due to double poles of panels) electronic switching elements are used. The reason for choosing the TCT type connection structure in this work is easy connection the adaptive panels to the fixed part by means of the switching circuit. In the S&P type connection structure adaptive panels have to be connected to the fixed part in a series way, in which case more switching elements are necessary and a more complex connection structure occurs. Additionally in the TCT structure, it can be related between short circuit current of rows and MPP because of there is not current limitation among the rows. Therefore, irradiance levels of rows can be estimated by sensed short circuit currents of rows. Reconfiguration process The proposed reconfiguration method is based on the principle of establishing the rows of array by using the panels whose insolation levels are as close to each other as possible. This would cause similar current levels at these modules, ensuring that series connection would not suffer from the current limiting problem. As the insolation affects the short circuit current of the module (Isc) directly (7), if the adaptive panels are connected to the fixed part through switches in such a way that the short circuit currents of each row is very similar, the output power of the PV array would increase. According to proposed method, the possible number of connections of the adaptive panel to the fixed part is mm. In the developed configuration scanning algorithm, the following equations are used for each possible connection structure:
Isc ðRowðiÞ Þ ¼ Isc ðF i Þ þ
Fm Switching Signals
Isc(A1), Isc(A2), ..….. Isc(Am)
Load
A2
F1 Switching Matrix Circuit
A1
A2
X
Isc ðAÞ ði ¼ 1; 2; . . . mÞ
ð8Þ
max I ¼ max½IðscÞ1 ; IðscÞ2 ; . . . IðscÞm
ð9Þ
min I ¼ min½IðscÞ1 ; IðscÞ2 ; . . . IðscÞm
ð10Þ
CVI ¼ max I min I
ð11Þ
Here Isc(Fi) is the short circuit current of each row of the fixed part.
RIsc(A) is the sum of the short circuit currents of the adaptive panels Controller Fig. 6. The architecture of proposed system model.
which can be found in every row of the fixed part for each connection configuration. max I and min I are maximum and minimum values of string calculated by Eq. (8). In the proposed method, the highest power is obtained with the connection configuration which
717
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
Isc(F1)
SA1
Isc(A1) A1
F11
F1n
F12
SF1
SL
SA2
Isc(F2) Load
Isc(A2) A2
F21
F22
F2n
SAm
SF2 Isc(Fm)
Isc(Am) Am
SFm Fm1
Fig. 8. Obtaining short current of adaptive panels.
yields the smallest ‘‘current variation index’’ which is shortly represented as the ‘‘CVI’’ (11). The scanning algorithm finds all the possible adaptive panel connections and determines the CVI value for each connection by using (8)–(11). The configuration with the smallest CVI value is the configuration being searched.
Obtainment the short circuit currents In Figs. 8 and 9, circuit diagram which is necessary to obtain the instantaneous short circuit currents (Isc) for each row of the fixed part and adaptive panels is shown. In order to measure these currents, first adaptive panels – fixed part and the fixed part – load connections are opened. Then adaptive panels and rows of fixed part are shorted and short circuit currents are measured. These currents are used in the Eqs. (8)–(11) of the configuration scanning algorithm as the input data. After the measurement of instant short circuit currents is finished, the connection of the fixed part with the adaptive panels and load is reinstated. In the proposed reconfiguration method, only the short circuit currents of the adaptive panels and rows of fixed part need to be known. Therefore, the number of current sensor which is necessary for applying this method is m + m. When it is compared especially to other proposed methods in the literature, this number is quite low. Also, there is no need for temperature and radiance sensors which require relatively high cost. The required sensor numbers of the proposed method and the existed methods of which are clearly given are compared in the Table 1. For sensible comparison, TCT structured PV array reconfiguration methods are only considered. In the method, the load is powerless only during sensing of short circuit currents and the sensing time takes quite short. Hence power degradation of load can be neglected.
Fm2
Fmn
Fig. 9. Obtaining short current of rows at fixed part.
Configuration scanning algorithm In this section, main points of the operation logic of the configuration scanning algorithm are explained. The purpose of this algorithm is to find out how many different possible connections exist between m pieces of adaptive panels (A = A1, A2, . . . Am) and the fixed part (F = F1, F2, . . . Fm) and to determine in which configuration the highest power can be obtained by using (8)–(11) equations. In the code that has been prepared for this, first in which configurations A (Adaptive) units can be connected to F units is determined. Permutation rules are utilized in this process. A units can be connected to any F unit in 2m different ways. The number of differ m 2 ent connections is P . For instance if m = 3, three units of A m can be connected to 3 units of F in 336 different ways. These placements are briefly shown in Table 2. Table 2 shows that adaptive panel placements to fixed part. For example in the sequence number 1, any adaptive panel place to F1, A3 place to F2 and A2 place to F3. In the sequence number 334, A1, A2 and A3 place to F1, A1 and A2 place to F2 and A2 and A3 place to F3. Among these possible placements, impossible placements are to be in the PV array. For instance, there are possibilities in which a specific A unit is connected more than one F unit or does not have a connection with any F unit. In the second phase of the code, arrays that can be proper among all possible placements are determined. Presence of A’s is shown as ‘‘1’’ and absence of them as ‘‘0’’ in the code. This representation simplifies implementation of Table 2 in the scanning algorithm as well as selection of the
Table 1 Comparison of sensor numbers in the reconfiguration methods. Method
Voltage sensors
Current sensors
Temperature sensor
Sum of sensors
EAR [6] Model Based [7] Optimal Reconfiguration [14] Proposed Method
m*n 2*m m –
m*n – m*n 2*m
– 1 – –
2*m*n 2*m + 1 m(1 + n) 2*m
718
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
Table 2 The connection of A units to each F units in configuration scanning algorithm for m = 3. Sequence number
F1
F2
F3
1 2 3 334 335 336
[–––] [–––] [–––] [A1 A2 A3] [A1 A2 A3] [A1 A2 A3]
[ – – A3 ] [ – – A3 ] [ – – A3 ] [A1 A2 – ] [A1 A2 – ] [A1 A2 – ]
[ – A2 – ] [ – A2 A3 ] [A1 – – ] [ – A2 A3 ] [A1 – – ] [A1 – A3]
Table 3 An example for suitability determination. F1
F2
F3
[ – A2 – ] [010] [ – A2 – ] [010]
[ – A2 A3 ] [011] [ A1 A3 ] [101]
[ A1 A2 – ] [110] [–––] [000]
010 + 011 + 110 = 131 1*3*1 = 3 010 + 101 + 000 = 111 1*1*1 = 1
U
suitable configurations. It also represents the signals which will be sent to switching matrix at the end. Suitability is determined by adding each binary code in F and then by multiplying order of them for each sequence number. If the result of this multiplication is different from ‘‘1’’ this is not an appropriate configuration. In order to understand suitable and unsuitable configurations an example is given in Table 3. The number of possible connections is mm. In this study, these connections are 27 because of m is taken as 3. After suitability determination process, the algorithm considers only these connections and so the process time of algorithm is shorten. In the third part of the code, Eqs. (8)–(11) are applied to suitable connections and the configuration with the highest power under partial shading conditions is determined. The pseudo code of the developed configuration scanning algorithm is as follows (m = n): Loop i = 1 to SequenceNumber Do If sequence(i) = suitable then For j = 1 to n Loop k = 1 to m do SumA SumA + A(k) endLoop IscRow(i) SumA + F(j) endFor min IscRow(i) min[IscRow(i)] max IscRow(i) max[IscRow(i)] CVI(i) (max IscRow(i)-min IscRow(i)) endIf endLoop BestSequence sequence[min(CVI)]
The most important feature of the developed configuration scanning algorithm is its having an adaptive structure. In other words, it is not sensitive to reconfigured panel number in the array. Therefore, whatever the adaptive panel and row of fixed part number is, there is no need for any changes. The developed algorithm requires only instantaneous short circuit current information as data entry is just enough. Proposed reconfiguration method has two important advantages. One of them is that it does not need any panel parameter or information about physical structure of the panels, or any measurement other than short circuit currents. In the previous work reported in the literature that use ‘‘bubble sort’’ or ‘‘model-based’’ methods employ voltage and temperature data as well as panel
parameters. The other advantage is that PV panels which have different models and power ratings than each other can be used in the array easily. Constructing a reasonable complexity system In the proposed approach, complexity of the control algorithm is correlated with number of rows (number of ‘‘m’’) in PV array. It is clear that we can get more power by increasing the number of adaptive panels. On the other hand it causes the system complexity scale up due to significantly increase the number of combination and electronic switching elements. Although state of art control cards are quite fast we have to scale the system complexity. In this study, m is taken as 3 due to more understandable of proposed method. The number of electronic switches employed in switching matrix circuit is correlated with m also. It is known that increasing m causes switch numbers to multiple. Cost advantages of semiconductor technology motivates one to use electronic components such as Mosfet switches. As a matter of this fact investment of the proposed system becomes relatively feasible as considering efficiency improvement of the PV system. In the proposed method only ‘‘2 m’’ switches remain in conduction at any time. Therefore switches losses are quite low with the use of proper switching elements especially have low conduction resistance. Simulations In the simulations, previously developed PV panel model has been used [24]. This model has been designed with 5-parameter PV modüle equivalent circuit. In the developed model, temperature and insolation values have been entered as input data by using PV modüle equations. The designed model of the proposed method in Matlab–Simulink environment is shown in Fig. 10. In this model, there are PV panels for the fixed part and adaptive panels, switching matrix circuit and also a dc–dc boost convertor circuit which works with MPPT algorithm. The code for the 5-parameter PV panel model mentioned above is embedded into PV panel blocks. The developed reconfiguration algorithm runs in the software (MATLAB Function). Circuit current information of the rows of the fixed and adaptive panels, control signals that are sent to switching elements in order to get this information, and signals which are sent to switching matrix are used in this block. The reconfiguration method presented in this work has been tested on a PV array. The array consists of 3 adaptive panels and a fixed part which has 3 3 TCT array, shown in Fig. 6. In the simulation, different insolation levels have been applied to PV panels in the array and a partial shading condition has been created. The insolations related to panels which have been used in simulation are shown in Table. 4. In Table 5, short circuit currents of the rows of fixed part and each adaptive panel, measured according to the method mentioned in Part 4.3, are given. These values are used in configuration scanning algorithm in order to be used in Eqs. (8)–(11) as data entry. Presented method has been simulated by creating partial shading conditions whose circuit structure in the system is shown in Fig. 6. By using the developed configuration scanning algorithm, 27 (mm) different suitable connection configurations have been determined for m = 3 and current variation index has been calculated for each configuration. In the Table 6, data obtained from the code as a result of this algorithm are given. In this table, for each possible configuration, adaptive panels (A1, A2, A3) that can be connected to the fixed part (F1, F2, F3) have been shown as binary code as explained in configuration scanning algorithm. Also, in
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
719
Fig. 10. The simulation model of the proposed method in Matlab–Simulink environment.
Table 4 Insolations of PV panels in the array. Row
Adaptive panels (W/m2)
Fixed panels (W/m2)
1 2 3
950 850 750
1000 600 400
900 500 200
800 400 300
Table 5 Short circuit current of adaptive panels and rows of fixed part. Row
Adaptive panel Isc (A)
Rows of fixed part Isc (A)
1 2 3
3.605 3.225 2.846
10.24 5.691 3.415
the event of connections, short circuit current and current variation index (CVI) values are given for each row of array. At the beginning of the simulation A1–A2–A3 is connected to F1– F2–F3. This situation represents the 8th line in Table 6. As from the table, CVI coefficient has the smallest value in 16th line and array has the highest power value with stated connection configuration in this line. For a few connection structures which are randomly selected from Table 6, MPP values and its % changes with respect to the initial condition for the PV array are shown in Table 7.
Table 7 shows that efficiency of the array for the 16th line configuration has increased 37.1% with respect to the beginning situation. Here, A2 and A3 are connected to F2 and A1 is connected to F3. When the results in table are analyzed, it is seen that the maximum power value in the array is related to CVI coefficient. The array can produce higher power with smaller this coefficient. I–V and P–V curves of array are given in Fig. 11 for before and reconfiguration process. As seen from the figure, the reconfiguration process carried out with configuration scanning algorithm minimizes the effect of bypass diodes since differences between currents of the rows is minimized with the method and higher power can be yield from the array. The increasing power of array after the reconfiguration is clearly seen form Fig. 11. Fig. 12 gives the results of the simulation where the developed reconfiguration method has been used with an MPPT algorithm [3]. A boost type dc–dc convertor is used between the PV panels and a fixed load. Beginning and end times of reconfiguration process are shown in the figure. Here, a signal which shows the beginning time of reconfiguration at 1 ms of simulation is produced. This signal is sent to switching elements, which are shown in Figs. 8 and 9, in order for the necessary short circuit currents are sensed. As it is mentioned in the previous part, the configuration scanning algorithm determines the best panel connection type and sends the signals, which will provide this connection structure, to switching matrix circuit. As seen from Fig. 12, the maximum
720
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
Table 6 Results of configuration scanning algorithm. Sequence number
F1 A1A2A3
F2 A1A2A3
F3 A1A2A3
Isc(Row(1))
Isc(Row(2))
Isc(Row(3))
CVI
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
110 101 100 011 010 001 110 100 010 000 101 100 001 000 100 000 011 010 001 000 010 000 001 000 111 000 000
001 010 011 100 101 110 000 010 100 110 000 001 100 101 000 100 000 001 010 011 000 010 000 001 000 111 000
000 000 000 000 000 000 001 001 001 001 010 010 010 010 011 011 100 100 100 100 101 101 110 110 000 000 111
17.0700 16.6910 13.8450 16.3110 13.4650 13.0860 17.0700 13.8450 13.4650 10.2400 16.6910 13.8450 13.0860 10.2400 13.8450 10.2400 16.3110 13.4650 13.0860 10.2400 13.4650 10.2400 13.0860 10.2400 19.9160 10.2400 10.2400
8.5370 8.9160 11.7620 9.2960 12.1420 12.5210 5.6910 8.9160 9.2960 12.5210 5.6910 8.5370 9.2960 12.1420 5.6910 9.2960 5.6910 8.5370 8.9160 11.7620 5.6910 8.9160 5.6910 8.5370 5.6910 15.3670 5.6910
3.4150 3.4150 3.4150 3.4150 3.4150 3.4150 6.2610 6.2610 6.2610 6.2610 6.6400 6.6400 6.6400 6.6400 9.4860 9.4860 7.0200 7.0200 7.0200 7.0200 9.8660 9.8660 10.2450 10.2450 3.4150 3.4150 13.0910
13.6550 13.2760 10.4300 12.8960 10.0500 9.6710 11.3790 7.5840 7.2040 6.2600 11.0000 7.2050 6.4460 5.5020 8.1540 0.9440 10.6200 6.4450 6.0660 4.7420 7.7740 1.3240 7.3950 1.7080 16.5010 11.9520 7.4000
The bold values represent initial state condition. The bolditalic values represent the best connection configuration. Table 7 The MPP values belonging to some configuration and% changes with respect to initial condition. Sequence number
8
2
10
16
20
22
24
27
MPP (W) % MPP Variation
350 Initial state
321 8.3
364 4
480 37.1
388 10.8
470 34.2
455 30
365 4.2
Current (A)
15 10 5 0
Before Reconfigration After Reconfiguration 0
10
20
30
40
50
60
70
Reconfiguration Signal
The bold values represent initial state condition. The bolditalic values represent the best connection configuration.
2
2
1.5 1
1 0
0.5
1
1.005
1.01 -3
x 10
0
0
1
2
3
4
5
6 -3
x 10
400 200 0
0
10
20
30
40
50
60
70
Voltage (V)
Max Array Power (W)
Power (W)
600 600 400 200 0
0
1
2
3
4
5
Time (sec)
6 -3
x 10
Fig. 11. I–V and P–V curves of array. Fig. 12. Reconfiguration signal and MPP curve of the array.
power attained from the output of PV array increases after reconfiguration process. The advantage of the developed reconfiguration method can be clearly seen from the maximum array power curve. The short circuit currents of adaptive panels and rows of fixed part are sensed in step of simulation. For this reason no more time is required to get Isc values. But this time should be considered as the proposed method is experimentally implemented.
Conclusion In this paper, a new reconfiguration method has been proposed in order to increase the efficiency of a 3 4 PV array under partial shading conditions. In this method, PV array is reconfigured by adaptive panels connect to the fixed part in such a way that the short circuit currents of each row of array is as close to each other
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
as possible. This makes it possible to extract the maximum available power from the array in any environmental condition. The method utilizes the configuration scanning algorithm that looks into all possible reconfiguration schemes and decides which one can yield the maximum power. This algorithm needs only short circuit currents of adaptive panels and rows of fixed part. The proposed method has been tested by simulation on a PV array consisting of PV panels with randomly selected insolation levels to create the partial shading conditions. For the given insolation values and array structure maximum available power value has been increased by 37.1% by using the proposed algorithm. One other advantage of the method is that it can be applied to arrays consisting of panels with different power levels and characteristics without any software or hardware changes. Acknowledgment This work has been supported by The Scientific and Technical _ Research Council of Turkey (TÜBITAK-112E214). References [1] Mei Q, Shan M, Liu L, Guerrero JM. A novel improved variable step-size incremental-resistance MPPT method for PV systems. IEEE Trans Ind Electron 2011;58(6):2427–34. [2] Salam Z, Ahmed J, Merugu BS. The application of soft computing methods for MPPT of PV system: a technological and status review. Appl Energy 2013;107:135–48. [3] Parlak KS, Can H. A new MPPT method for PV array system under partially shaded conditions. In: 3rd IEEE international symposium on power electronics for distributed generation systems (PEDG); 2012. p. 437–42. [4] Damiano LM, Vincenzo LV, Eleonora RS, Vincenzo DD, Pietro R. Reconfigurable electrical interconnection strategies for photovoltaic arrays: a review. Renew Sustain Energy Rev 2014;33:412–26. [5] Velasco G, Guinjoan F, Pique R. Grid-connected PV systems energy extraction improvement by means of an electric array reconfiguration (EAR) strategy: operating principle and experimental results. In: IEEE power electronics specialists conference (PESC); 2008; p. 1983–89. [6] Velasco G, Guinjoan F, Pique R, Roman M, Conesa A. Electrical PV array reconfiguration strategy for energy extraction improvement in grid-connected PV systems. IEEE Trans Ind Electron 2009;56(11):4319–31. [7] Nguyen D, Lehman B. An adaptive solar photovoltaic array using model-based reconfiguration algorithm. IEEE Trans Ind Electron 2008;55(7):2644–54.
721
[8] Nguyen D, Lehman B. A reconfigurable solar photovoltaic array under shadow conditions. In: Applied power electronics conference and exposition (APEC); 2008; p. 980–87. [9] Jonathan PS, Peter RW, Darren B. Improved optimization strategy for irradiance equalization in dynamic photovoltaic arrays. IEEE Trans Power Electr 2013;28(6):2946–56. [10] Srinivasa PR, Saravana GI, Chilakapati NC. Maximum power from PV arrays using a fixed configuration under different shading conditions. IEEE J Photovoltaics 2014;4(2):679–86. [11] Alahmad M, Chaaban MA, Lau SK, Shi F, Neal J. An adaptive utility interactive photovoltaic system based on a flexible switch matrix to optimize performance in real-time. Sol Energy 2012;86(3):951–63. [12] Jonathan PS, Peter RW, Darren B. The optimized-string dynamic photovoltaic array. IEEE Trans Power Electr 2014;29(4):1768–76. [13] Juan DBR, Carlos ARP, Andres JSM. Reconfiguration analysis of photovoltaic arrays based on parameters estimation. Simul Model Pract TH 2013;35:50–68. [14] El-Dein MZS, Kazerani M, Salama MMA. Optimal photovoltaic array reconfiguration to reduce partial shading losses. IEEE Trans Sustain Energy 2013;4(1):145–53. [15] Santos P, Vicente EM, Ribeiro ER. Reconfiguration methodology of shaded photovoltaic panels to maximize the produced energy. In: IEEE power electronics conf (COBEP); 2011. p. 700–07. [16] Tria LAR, Escoto MT, Odulio CMF. Photovoltaic array reconfiguration for maximum power transfer. In: IEEE region 10 conf (TENCON); 2009. p. 1–6. [17] Liu Y, Pang Z, Cheng Z. Research on an adaptive solar photovoltaic array using shading degree model-based reconfiguration algorithm. In: IEEE control and decision conf (CCDC); 2010. p. 2356–63. [18] Cheng Z, Pang Z, Liu Y, Xue P. An adaptive solar photovoltaic array reconfiguration method based on fuzzy control. In: Proc of the 8th world congress on intelligent control and automation; 2010. p. 6–10. [19] Candela R, Dio VD, Sanseverino ER, Romano P. Reconfiguration techniques of partial shaded PV systems for the maximization of electrical energy production. In: IEEE international conf on clean electrical power (ICCEP); 2007. p. 716–20. [20] Gao L, Dougal RA, Liu S, Lotova LA. Parallel-connected solar PV system to address partial and rapidly fluctuating shadow conditions. IEEE Trans Ind Electron 2009;56(5):1548–56. [21] Patel H, Agarwal V. Matlab-based modeling to study the effects of partial shading on PV array characteristics. IEEE Trans Energy Convers 2008;23(1):302–10. [22] Silvestre S, Boronat A, Chouder A. Study of bypass diodes configuration on PV modules. Appl Energy 2009;86(9):1632–40. [23] Chenni R, Makhlouf M, Kerbache T, Chenni AB. A detailed modeling method for photovoltaic cells. Energy 2007;32(9):1724–30. _ D, Parlak KS. A new numerical solution approach for the real-time [24] Can H, Içkilli modeling of photovoltaic panels. In: IEEE power and energy engineering conf (APPEEC); 2012. p. 1–4.
Contents lists available at ScienceDirect
Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes
PV array reconfiguration method under partial shading conditions Koray Sß ener Parlak ⇑ Department of Electronics and Automation, Fırat University, 23119 Elazıg˘, Turkey
a r t i c l e
i n f o
Article history: Received 7 August 2013 Received in revised form 29 May 2014 Accepted 20 June 2014
Keywords: Photovoltaic arrays PV array reconfiguration Partial shaded conditions Maximum power point
a b s t r a c t This paper proposes a new reconfiguration method for photovoltaic arrays under partial shading conditions. With the application the method, the array can produce higher power by reconfigured connections of array. In the proposed method, each row of an array is formed by connecting the panels with close short circuit current levels as possible. A novel algorithm entitled configuration scanning algorithm is used to determine the all possible connection structures. This algorithm utilizes only short circuit current values measured at particular parts of the array. The proposed method has been tested in Matlab–Simulink environment under partial shading conditions. Results of the simulation show that the efficiency of the array is considerably increased with the presented reconfiguration method. Ó 2014 Elsevier Ltd. All rights reserved.
Introduction The use of renewable energy sources has become very important as a result of increasing energy need and the environmental effects of fossil fuels. Photovoltaic (PV) modules are one of the most important renewable energy sources and their use is gradually increasing in regions that have a long sunshine duration and in rural areas. In order to be able to gain a higher solar power, PV arrays are formed by connecting the PV modules in series and parallel. However, the shadows of objects such as clouds, birds, buildings and trees on PV arrays reduce the sunlight that the array receive, inevitably resulting in a lower system efficiency. PV modules need to be operated at the maximum power point (MPP) at varying heat, insolation and load levels for an efficient use. One way of increasing the PV array efficiency is to use maximum power point tracker (MPPT). MPPT is basically a dc–dc converter which utilizes an algorithm to determine the MPP of the current–voltage (I–V) or power–voltage (P–V) characteristic in varying environmental conditions. In order to improve conventional MPPT method, the step size of MPPT algorithm is taken variable instead of fixed [1]. However conventional MPPT methods can fail under partial shading conditions because of occurring more than one local maximas. To find global MPP among local maximas, soft computing technique is developed [2]. Another developed MPPT method is based on obtaining electrical characteristics of PV array. This method works effectively in partial shading conditions [3]. ⇑ Tel.: +90 542 5856185; fax: +90 424 2188947. E-mail address: [email protected] http://dx.doi.org/10.1016/j.ijepes.2014.06.042 0142-0615/Ó 2014 Elsevier Ltd. All rights reserved.
There are reported researches results showing that system efficiency of PV modules exposed to partial shading can be increased by applying some PV module reconfiguration algorithms [4]. In some of these studies, reconfiguration of array has been made by changing connections of all modules on the array [5,6]. In these methods which is called as electrical array reconfiguration module catalog information are used along with instant current and voltage information for each module. By using these data and calculating the current radiant intensity of the modules, the array is reconfigured to form new submodules with the modules that have similar insolation levels. The disadvantage of this method is the high number of possible connections since reconfiguration is applied to all the modules. While a higher efficiency increase is provided with the reconfiguration of all modules in the PV array, quite many connection configurations are possible. In order to compensate this disadvantage, the array can be divided into two parts as the fixed part that consists of statically connected modules and adaptive modules part that can be attached to the fixed part with different configurations [7,8]. In this method, the currents of PV modules are calculated by using the measured voltage of fixed and adaptive parts, temperature of the array, and the catalog data. These currents are used in a bubble sort algorithm and reconfiguration of the array is applied. In the presented another reconfiguration method, connections of all PV modules are reconfigured according to irradiance equalization strategy [9]. This method uses some optimizations to reduce possible connections number. However this number still high since all modules are reconfigured. The method had similar reconfiguration structure to [9] uses PV module equations and number of shaded modules to some calculations [10].
714
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
In the presented method, the shaded modules are determined by measuring current and voltage of all modules in strings. All of these modules are connected to another bus by flexible switching matrix circuit and composed a new string. This string voltage is brought to nominal array voltage by dc–dc converter [11]. Another method tries to bring similar shaded modules to same string by high number switches [12]. This method is similar to [11] but uses more than one power bus and converter to get equal voltage for each string. In another method, all PV modules are reconfigured [13]. It is expected that a lot of switches are required to each module connect to another one. Additionally the algorithm consists of extra calculations to estimate module irradiations. An optimal reconfiguration logic that utilizes mathematical formulas is proposed in [14]. Reconfiguration of the array is performed by using optimization techniques with the voltage values measured at each line and current values measured in all modules in the array. It is obvious that quite many sensors will be needed in this method. In a reconfiguration method which uses ‘‘Rough Set Theory’’ based on ANN under the partial shading conditions, each PV module current in the array is compared to a reference value and proper switching signals are obtained in accordance with a certain rule. However, this method is complex and requires several measurements [15]. Load voltage and temperature parameters of the array are used as inputs to the algorithm which aims to reconfigurating an array with minimum number of switching [16]. The system simulated with four panels gives positive results at constant load but it has some disadvantages. It is difficult to obtain the parameters used in the algorithm and the performance drops when the conditions vary. In fuzzy logic controlled reconfiguration methods, selection of the most suitable connection configuration is aimed by using system and panel parameters [17,18]. Parallel connected modules that are in shade to the reconfigured PV array are proposed in [19,20] to take advantage of whatever power is available in these shaded pieces. In this way, degradation caused by shading effects is minimized. However, increased number of parallel modules in the array will raise the output current, casing in higher losses in conducting wires. In this paper, a reconfiguration method is proposed to increase the efficiency under partial shading conditions. In the proposed method, PV system is divided into two parts, namely adaptive and fixed parts. A new developed configuration scanning algorithm that scans the array and decides how the adaptive parts can be attached to the fixed part to maximize the efficiency has been developed. The main contributions of the proposed reconfiguration method as follows: The method require only short circuit currents where sensed particular nodes of array. Therefore there is no need parameter of PV panel and any additional calculations especially to estimate irradiances. The developed configuration scanning algorithm considers all possible connection configurations with defined terms. Thus the algorithm determines the proper configuration that has highest power under present shading conditions. The method needs minimum number of sensors comparing with other reconfiguration methods. Validity of the method has been tested in a 3 4 PV array which is controlled by MPPT under partial shading conditions. In the next section of the paper creation of the PV arrays and the shading effect on these arrays are explained. Then, proposed reconfiguration method and developed configuration scanning algorithm are explained. Finally, simulation results that show the effect of the proposed algorithm are given.
Connection of PV modules and PV array form Connection in parallel PV modules are connected in parallel between each other to form submodules. While the output voltage is equal in the submodule shown in Fig. 1, the output current is the total sum of module currents.
V out ¼ V 1 ¼ V 2 ¼ ¼ V n Iout ¼ I1 þ I2 þ þ In ¼
ð1Þ
n X Ii
ð2Þ
i¼1
The output power of submodule:
Pout ¼ Iout P out ¼
n X Pi
ð3Þ
i¼1
In this connection structure, as the bathos in any module will not affect the operation characteristics of other modules, all modules can transfer the power they generate to the output. In this way, the highest efficiency can be gained in any environmental conditions. The output voltage is equal to the module voltage in this connection structure and therefore it is low. Hence, the conversion ratio of converter which is going to be used to increase the voltage needs to be high. This affects the efficiency of converter in a negative way. Also, as the output current needs to be high, line losses are significant. Connection in series In the strings which consist of series connection of PV modules, a higher output voltage is obtained. In a string shown in Fig. 2, the currents of modules are equal to each other, and output voltage is the sum total of mudul voltages.
Iout ¼ I1 ¼ I2 ¼ ¼ In
ð4Þ
V out ¼ V 1 þ V 2 þ þ V n ¼
n X Vi
ð5Þ
i¼1
The output power of string:
Pout ¼ Iout P out
ð6Þ
Iout
Vout
PV 1
PV 2
PV 3
PV m
Fig. 1. Parallel connected PV modules.
PV1
PV2
PV3
PVm
Vout Fig. 2. Series connected PV modules.
Iout
715
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
4
Current (A)
3 2 1 0
5
10
15
20
25
10
15
20
25
1000 W/m2 600 W/m2 200 W/m2
60 40 20 0
Array formation
0
5
Voltage (V)
The effect of shading on PV systems The insolation on a PV module affects the module characteristic significantly. This state can be seen from the output current equation of module’s Iph-photo generated current value based on radiation (7).
G ðIscn þ K i ðT T n ÞÞ Gn
Fig. 4. I–V and P–V curves of the PV module at different irradiance levels with constant temperature value.
Current (A)
4 3 2
without bypass with bypass
1 0
0
10
20
30
40
50
0
10
20
30
40
50
100
Power (W)
In order to get a higher current and voltage from PV system, PV array is formed by connecting PV modules in series and parallel. In Series-Parallel (S&P) PV array shown in Fig. 3a, series connected PV modules form the strings (PV11, PV12, . . . , PV1n) and these strings generate PV array by being connected in parallel. Total-cross-tied (TCT) type array shown in Fig. 3b consists of submodules which are formed by parallel connected modules and these submodules (PV11, PV21, PVn1,) are connected series. Under uniform irradiance, the powers that S&P and TCT arrays generate are the same. However, under partial shading conditions which array structure generates more power depends on irradiances of string or submodules that are in the shade and the load of the array.
Iph ¼
0
80
Power (W)
As the current of series modules will be the same, any module which is under the shading effect will affect the performance of the others. As the photo-generated current (Iph) of a module which is under the effect of shade will decrease, the string current is limited with this module current. Also, according to Kircchoff’s voltage law, as a module under the effect of shading will act like a load by carrying a negative voltage, the extra current that the other modules produce flows here and causes heat accumulation. This heat accumulation is called ‘‘Hot Spot’’ effect. In the PV modules working under the effect of hot spot for a long time there can be physical damages. In order to prevent these problems parallel bypass diodes are attached to series connected PV modules. These diodes allow the flow of the current which is the difference between the string current and parallel module current. Thus hot spot effect is ironed out and the shaded modules do not affect the string [21,22].
50
0
Voltage (V)
ð7Þ
As it can be seen in Eq. (7), Iph value of PV module is directly proportional to G (W/m2) which is irradiant intensity. Here, while Iscn shows short circuit current of the PV module, Gn and Tn are rated radiant intensity (1000 W/m2) and temperature values
Fig. 5. The effect of bypass diodes to PV string.
(25°), respectively [23]. Ki is temperature coefficient of short circuit current. I–V and P–V curves for the PV module at different insolation levels at constant temperature are shown in Fig. 4. Simulated
Iout
Iout
PV11
PV12
PV13
PV1n
PV2n
PV21
PV22
PV23
PV2n
PVmn
PVm1
PVm2
PVm3
PVmn
PV11
PV12
PV13
PV1n
PV21
PV22
PV23
PVm1
PVm2
PVm3
Vout
Vout
(a)
(b) Fig. 3. S&P-(a) and TCT-(b) types array connections.
716
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
PV module parameters are Iscn = 3.74 A, Voc = 23 V, Pmax = 65 W. As it can be seen from the figure, Isc and power values of module are affected by insolation that comes to module. The effect of partial shading on series connected PV modules is shown in Fig. 5. Here, I–V and P–V curves of series connected two modules are shown for 1000 and 600 W/m2 radiant intensity levels, respectively. In order to see the effect of bypass diodes, these curves have been drawn both when the diodes exist and when they do not exist. As it is seen, bypass diodes prevent current limiting at particular operating points and increase the available power.
F1 S11
S12
S1m
F2 S21
S22
S2m
Fm
Proposed reconfiguration method
Sm1
Sm2
Smm
In this section, architecture and operation principles of the proposed reconfiguration method are explained. Architecture of the system
A1 Connection structure of the system for proposed reconfiguration method is shown Fig. 6. As can be seen from the figure, PV system consists of three parts. The first one is the fixed part which has more PV panels and its connection structure does not change during operation. This part which is formed in TCT order and shown in Fig. 3b consists of m n pieces of PV panels [‘‘m’’ serially connected submodules, each of which includes ‘‘n’’ parallel connected panels (F11, F12, . . . , Fmn)]. The second part is the adaptive part which consists of m pieces of PV panels. (A1, A2, . . . , Am). As it is seen, adaptive panel number is equal to the number of submodules (row) of fixed part. Under uniform illumination conditions, each adaptive panel is connected to a row of fixed part and the array is in the structure of m (n + 1). Third part is switching matrix circuit which provides the connection of the adaptive panels to the fixed part. S11, S12, . . . , Smm switches which are in circuit shown in Fig. 7 are designed in a way to be able to provide the connection of any adaptive panel
Adapve Part
Fixed Part
F2
Am
Isc(F1), Isc(F2), ..….. Isc(Fm)
Am
Fig. 7. Switching matrix circuit.
to any row of the fixed part. In this circuit 2 m m (due to double poles of panels) electronic switching elements are used. The reason for choosing the TCT type connection structure in this work is easy connection the adaptive panels to the fixed part by means of the switching circuit. In the S&P type connection structure adaptive panels have to be connected to the fixed part in a series way, in which case more switching elements are necessary and a more complex connection structure occurs. Additionally in the TCT structure, it can be related between short circuit current of rows and MPP because of there is not current limitation among the rows. Therefore, irradiance levels of rows can be estimated by sensed short circuit currents of rows. Reconfiguration process The proposed reconfiguration method is based on the principle of establishing the rows of array by using the panels whose insolation levels are as close to each other as possible. This would cause similar current levels at these modules, ensuring that series connection would not suffer from the current limiting problem. As the insolation affects the short circuit current of the module (Isc) directly (7), if the adaptive panels are connected to the fixed part through switches in such a way that the short circuit currents of each row is very similar, the output power of the PV array would increase. According to proposed method, the possible number of connections of the adaptive panel to the fixed part is mm. In the developed configuration scanning algorithm, the following equations are used for each possible connection structure:
Isc ðRowðiÞ Þ ¼ Isc ðF i Þ þ
Fm Switching Signals
Isc(A1), Isc(A2), ..….. Isc(Am)
Load
A2
F1 Switching Matrix Circuit
A1
A2
X
Isc ðAÞ ði ¼ 1; 2; . . . mÞ
ð8Þ
max I ¼ max½IðscÞ1 ; IðscÞ2 ; . . . IðscÞm
ð9Þ
min I ¼ min½IðscÞ1 ; IðscÞ2 ; . . . IðscÞm
ð10Þ
CVI ¼ max I min I
ð11Þ
Here Isc(Fi) is the short circuit current of each row of the fixed part.
RIsc(A) is the sum of the short circuit currents of the adaptive panels Controller Fig. 6. The architecture of proposed system model.
which can be found in every row of the fixed part for each connection configuration. max I and min I are maximum and minimum values of string calculated by Eq. (8). In the proposed method, the highest power is obtained with the connection configuration which
717
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
Isc(F1)
SA1
Isc(A1) A1
F11
F1n
F12
SF1
SL
SA2
Isc(F2) Load
Isc(A2) A2
F21
F22
F2n
SAm
SF2 Isc(Fm)
Isc(Am) Am
SFm Fm1
Fig. 8. Obtaining short current of adaptive panels.
yields the smallest ‘‘current variation index’’ which is shortly represented as the ‘‘CVI’’ (11). The scanning algorithm finds all the possible adaptive panel connections and determines the CVI value for each connection by using (8)–(11). The configuration with the smallest CVI value is the configuration being searched.
Obtainment the short circuit currents In Figs. 8 and 9, circuit diagram which is necessary to obtain the instantaneous short circuit currents (Isc) for each row of the fixed part and adaptive panels is shown. In order to measure these currents, first adaptive panels – fixed part and the fixed part – load connections are opened. Then adaptive panels and rows of fixed part are shorted and short circuit currents are measured. These currents are used in the Eqs. (8)–(11) of the configuration scanning algorithm as the input data. After the measurement of instant short circuit currents is finished, the connection of the fixed part with the adaptive panels and load is reinstated. In the proposed reconfiguration method, only the short circuit currents of the adaptive panels and rows of fixed part need to be known. Therefore, the number of current sensor which is necessary for applying this method is m + m. When it is compared especially to other proposed methods in the literature, this number is quite low. Also, there is no need for temperature and radiance sensors which require relatively high cost. The required sensor numbers of the proposed method and the existed methods of which are clearly given are compared in the Table 1. For sensible comparison, TCT structured PV array reconfiguration methods are only considered. In the method, the load is powerless only during sensing of short circuit currents and the sensing time takes quite short. Hence power degradation of load can be neglected.
Fm2
Fmn
Fig. 9. Obtaining short current of rows at fixed part.
Configuration scanning algorithm In this section, main points of the operation logic of the configuration scanning algorithm are explained. The purpose of this algorithm is to find out how many different possible connections exist between m pieces of adaptive panels (A = A1, A2, . . . Am) and the fixed part (F = F1, F2, . . . Fm) and to determine in which configuration the highest power can be obtained by using (8)–(11) equations. In the code that has been prepared for this, first in which configurations A (Adaptive) units can be connected to F units is determined. Permutation rules are utilized in this process. A units can be connected to any F unit in 2m different ways. The number of differ m 2 ent connections is P . For instance if m = 3, three units of A m can be connected to 3 units of F in 336 different ways. These placements are briefly shown in Table 2. Table 2 shows that adaptive panel placements to fixed part. For example in the sequence number 1, any adaptive panel place to F1, A3 place to F2 and A2 place to F3. In the sequence number 334, A1, A2 and A3 place to F1, A1 and A2 place to F2 and A2 and A3 place to F3. Among these possible placements, impossible placements are to be in the PV array. For instance, there are possibilities in which a specific A unit is connected more than one F unit or does not have a connection with any F unit. In the second phase of the code, arrays that can be proper among all possible placements are determined. Presence of A’s is shown as ‘‘1’’ and absence of them as ‘‘0’’ in the code. This representation simplifies implementation of Table 2 in the scanning algorithm as well as selection of the
Table 1 Comparison of sensor numbers in the reconfiguration methods. Method
Voltage sensors
Current sensors
Temperature sensor
Sum of sensors
EAR [6] Model Based [7] Optimal Reconfiguration [14] Proposed Method
m*n 2*m m –
m*n – m*n 2*m
– 1 – –
2*m*n 2*m + 1 m(1 + n) 2*m
718
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
Table 2 The connection of A units to each F units in configuration scanning algorithm for m = 3. Sequence number
F1
F2
F3
1 2 3 334 335 336
[–––] [–––] [–––] [A1 A2 A3] [A1 A2 A3] [A1 A2 A3]
[ – – A3 ] [ – – A3 ] [ – – A3 ] [A1 A2 – ] [A1 A2 – ] [A1 A2 – ]
[ – A2 – ] [ – A2 A3 ] [A1 – – ] [ – A2 A3 ] [A1 – – ] [A1 – A3]
Table 3 An example for suitability determination. F1
F2
F3
[ – A2 – ] [010] [ – A2 – ] [010]
[ – A2 A3 ] [011] [ A1 A3 ] [101]
[ A1 A2 – ] [110] [–––] [000]
010 + 011 + 110 = 131 1*3*1 = 3 010 + 101 + 000 = 111 1*1*1 = 1
U
suitable configurations. It also represents the signals which will be sent to switching matrix at the end. Suitability is determined by adding each binary code in F and then by multiplying order of them for each sequence number. If the result of this multiplication is different from ‘‘1’’ this is not an appropriate configuration. In order to understand suitable and unsuitable configurations an example is given in Table 3. The number of possible connections is mm. In this study, these connections are 27 because of m is taken as 3. After suitability determination process, the algorithm considers only these connections and so the process time of algorithm is shorten. In the third part of the code, Eqs. (8)–(11) are applied to suitable connections and the configuration with the highest power under partial shading conditions is determined. The pseudo code of the developed configuration scanning algorithm is as follows (m = n): Loop i = 1 to SequenceNumber Do If sequence(i) = suitable then For j = 1 to n Loop k = 1 to m do SumA SumA + A(k) endLoop IscRow(i) SumA + F(j) endFor min IscRow(i) min[IscRow(i)] max IscRow(i) max[IscRow(i)] CVI(i) (max IscRow(i)-min IscRow(i)) endIf endLoop BestSequence sequence[min(CVI)]
The most important feature of the developed configuration scanning algorithm is its having an adaptive structure. In other words, it is not sensitive to reconfigured panel number in the array. Therefore, whatever the adaptive panel and row of fixed part number is, there is no need for any changes. The developed algorithm requires only instantaneous short circuit current information as data entry is just enough. Proposed reconfiguration method has two important advantages. One of them is that it does not need any panel parameter or information about physical structure of the panels, or any measurement other than short circuit currents. In the previous work reported in the literature that use ‘‘bubble sort’’ or ‘‘model-based’’ methods employ voltage and temperature data as well as panel
parameters. The other advantage is that PV panels which have different models and power ratings than each other can be used in the array easily. Constructing a reasonable complexity system In the proposed approach, complexity of the control algorithm is correlated with number of rows (number of ‘‘m’’) in PV array. It is clear that we can get more power by increasing the number of adaptive panels. On the other hand it causes the system complexity scale up due to significantly increase the number of combination and electronic switching elements. Although state of art control cards are quite fast we have to scale the system complexity. In this study, m is taken as 3 due to more understandable of proposed method. The number of electronic switches employed in switching matrix circuit is correlated with m also. It is known that increasing m causes switch numbers to multiple. Cost advantages of semiconductor technology motivates one to use electronic components such as Mosfet switches. As a matter of this fact investment of the proposed system becomes relatively feasible as considering efficiency improvement of the PV system. In the proposed method only ‘‘2 m’’ switches remain in conduction at any time. Therefore switches losses are quite low with the use of proper switching elements especially have low conduction resistance. Simulations In the simulations, previously developed PV panel model has been used [24]. This model has been designed with 5-parameter PV modüle equivalent circuit. In the developed model, temperature and insolation values have been entered as input data by using PV modüle equations. The designed model of the proposed method in Matlab–Simulink environment is shown in Fig. 10. In this model, there are PV panels for the fixed part and adaptive panels, switching matrix circuit and also a dc–dc boost convertor circuit which works with MPPT algorithm. The code for the 5-parameter PV panel model mentioned above is embedded into PV panel blocks. The developed reconfiguration algorithm runs in the software (MATLAB Function). Circuit current information of the rows of the fixed and adaptive panels, control signals that are sent to switching elements in order to get this information, and signals which are sent to switching matrix are used in this block. The reconfiguration method presented in this work has been tested on a PV array. The array consists of 3 adaptive panels and a fixed part which has 3 3 TCT array, shown in Fig. 6. In the simulation, different insolation levels have been applied to PV panels in the array and a partial shading condition has been created. The insolations related to panels which have been used in simulation are shown in Table. 4. In Table 5, short circuit currents of the rows of fixed part and each adaptive panel, measured according to the method mentioned in Part 4.3, are given. These values are used in configuration scanning algorithm in order to be used in Eqs. (8)–(11) as data entry. Presented method has been simulated by creating partial shading conditions whose circuit structure in the system is shown in Fig. 6. By using the developed configuration scanning algorithm, 27 (mm) different suitable connection configurations have been determined for m = 3 and current variation index has been calculated for each configuration. In the Table 6, data obtained from the code as a result of this algorithm are given. In this table, for each possible configuration, adaptive panels (A1, A2, A3) that can be connected to the fixed part (F1, F2, F3) have been shown as binary code as explained in configuration scanning algorithm. Also, in
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
719
Fig. 10. The simulation model of the proposed method in Matlab–Simulink environment.
Table 4 Insolations of PV panels in the array. Row
Adaptive panels (W/m2)
Fixed panels (W/m2)
1 2 3
950 850 750
1000 600 400
900 500 200
800 400 300
Table 5 Short circuit current of adaptive panels and rows of fixed part. Row
Adaptive panel Isc (A)
Rows of fixed part Isc (A)
1 2 3
3.605 3.225 2.846
10.24 5.691 3.415
the event of connections, short circuit current and current variation index (CVI) values are given for each row of array. At the beginning of the simulation A1–A2–A3 is connected to F1– F2–F3. This situation represents the 8th line in Table 6. As from the table, CVI coefficient has the smallest value in 16th line and array has the highest power value with stated connection configuration in this line. For a few connection structures which are randomly selected from Table 6, MPP values and its % changes with respect to the initial condition for the PV array are shown in Table 7.
Table 7 shows that efficiency of the array for the 16th line configuration has increased 37.1% with respect to the beginning situation. Here, A2 and A3 are connected to F2 and A1 is connected to F3. When the results in table are analyzed, it is seen that the maximum power value in the array is related to CVI coefficient. The array can produce higher power with smaller this coefficient. I–V and P–V curves of array are given in Fig. 11 for before and reconfiguration process. As seen from the figure, the reconfiguration process carried out with configuration scanning algorithm minimizes the effect of bypass diodes since differences between currents of the rows is minimized with the method and higher power can be yield from the array. The increasing power of array after the reconfiguration is clearly seen form Fig. 11. Fig. 12 gives the results of the simulation where the developed reconfiguration method has been used with an MPPT algorithm [3]. A boost type dc–dc convertor is used between the PV panels and a fixed load. Beginning and end times of reconfiguration process are shown in the figure. Here, a signal which shows the beginning time of reconfiguration at 1 ms of simulation is produced. This signal is sent to switching elements, which are shown in Figs. 8 and 9, in order for the necessary short circuit currents are sensed. As it is mentioned in the previous part, the configuration scanning algorithm determines the best panel connection type and sends the signals, which will provide this connection structure, to switching matrix circuit. As seen from Fig. 12, the maximum
720
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
Table 6 Results of configuration scanning algorithm. Sequence number
F1 A1A2A3
F2 A1A2A3
F3 A1A2A3
Isc(Row(1))
Isc(Row(2))
Isc(Row(3))
CVI
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
110 101 100 011 010 001 110 100 010 000 101 100 001 000 100 000 011 010 001 000 010 000 001 000 111 000 000
001 010 011 100 101 110 000 010 100 110 000 001 100 101 000 100 000 001 010 011 000 010 000 001 000 111 000
000 000 000 000 000 000 001 001 001 001 010 010 010 010 011 011 100 100 100 100 101 101 110 110 000 000 111
17.0700 16.6910 13.8450 16.3110 13.4650 13.0860 17.0700 13.8450 13.4650 10.2400 16.6910 13.8450 13.0860 10.2400 13.8450 10.2400 16.3110 13.4650 13.0860 10.2400 13.4650 10.2400 13.0860 10.2400 19.9160 10.2400 10.2400
8.5370 8.9160 11.7620 9.2960 12.1420 12.5210 5.6910 8.9160 9.2960 12.5210 5.6910 8.5370 9.2960 12.1420 5.6910 9.2960 5.6910 8.5370 8.9160 11.7620 5.6910 8.9160 5.6910 8.5370 5.6910 15.3670 5.6910
3.4150 3.4150 3.4150 3.4150 3.4150 3.4150 6.2610 6.2610 6.2610 6.2610 6.6400 6.6400 6.6400 6.6400 9.4860 9.4860 7.0200 7.0200 7.0200 7.0200 9.8660 9.8660 10.2450 10.2450 3.4150 3.4150 13.0910
13.6550 13.2760 10.4300 12.8960 10.0500 9.6710 11.3790 7.5840 7.2040 6.2600 11.0000 7.2050 6.4460 5.5020 8.1540 0.9440 10.6200 6.4450 6.0660 4.7420 7.7740 1.3240 7.3950 1.7080 16.5010 11.9520 7.4000
The bold values represent initial state condition. The bolditalic values represent the best connection configuration. Table 7 The MPP values belonging to some configuration and% changes with respect to initial condition. Sequence number
8
2
10
16
20
22
24
27
MPP (W) % MPP Variation
350 Initial state
321 8.3
364 4
480 37.1
388 10.8
470 34.2
455 30
365 4.2
Current (A)
15 10 5 0
Before Reconfigration After Reconfiguration 0
10
20
30
40
50
60
70
Reconfiguration Signal
The bold values represent initial state condition. The bolditalic values represent the best connection configuration.
2
2
1.5 1
1 0
0.5
1
1.005
1.01 -3
x 10
0
0
1
2
3
4
5
6 -3
x 10
400 200 0
0
10
20
30
40
50
60
70
Voltage (V)
Max Array Power (W)
Power (W)
600 600 400 200 0
0
1
2
3
4
5
Time (sec)
6 -3
x 10
Fig. 11. I–V and P–V curves of array. Fig. 12. Reconfiguration signal and MPP curve of the array.
power attained from the output of PV array increases after reconfiguration process. The advantage of the developed reconfiguration method can be clearly seen from the maximum array power curve. The short circuit currents of adaptive panels and rows of fixed part are sensed in step of simulation. For this reason no more time is required to get Isc values. But this time should be considered as the proposed method is experimentally implemented.
Conclusion In this paper, a new reconfiguration method has been proposed in order to increase the efficiency of a 3 4 PV array under partial shading conditions. In this method, PV array is reconfigured by adaptive panels connect to the fixed part in such a way that the short circuit currents of each row of array is as close to each other
K.Sßener Parlak / Electrical Power and Energy Systems 63 (2014) 713–721
as possible. This makes it possible to extract the maximum available power from the array in any environmental condition. The method utilizes the configuration scanning algorithm that looks into all possible reconfiguration schemes and decides which one can yield the maximum power. This algorithm needs only short circuit currents of adaptive panels and rows of fixed part. The proposed method has been tested by simulation on a PV array consisting of PV panels with randomly selected insolation levels to create the partial shading conditions. For the given insolation values and array structure maximum available power value has been increased by 37.1% by using the proposed algorithm. One other advantage of the method is that it can be applied to arrays consisting of panels with different power levels and characteristics without any software or hardware changes. Acknowledgment This work has been supported by The Scientific and Technical _ Research Council of Turkey (TÜBITAK-112E214). References [1] Mei Q, Shan M, Liu L, Guerrero JM. A novel improved variable step-size incremental-resistance MPPT method for PV systems. IEEE Trans Ind Electron 2011;58(6):2427–34. [2] Salam Z, Ahmed J, Merugu BS. The application of soft computing methods for MPPT of PV system: a technological and status review. Appl Energy 2013;107:135–48. [3] Parlak KS, Can H. A new MPPT method for PV array system under partially shaded conditions. In: 3rd IEEE international symposium on power electronics for distributed generation systems (PEDG); 2012. p. 437–42. [4] Damiano LM, Vincenzo LV, Eleonora RS, Vincenzo DD, Pietro R. Reconfigurable electrical interconnection strategies for photovoltaic arrays: a review. Renew Sustain Energy Rev 2014;33:412–26. [5] Velasco G, Guinjoan F, Pique R. Grid-connected PV systems energy extraction improvement by means of an electric array reconfiguration (EAR) strategy: operating principle and experimental results. In: IEEE power electronics specialists conference (PESC); 2008; p. 1983–89. [6] Velasco G, Guinjoan F, Pique R, Roman M, Conesa A. Electrical PV array reconfiguration strategy for energy extraction improvement in grid-connected PV systems. IEEE Trans Ind Electron 2009;56(11):4319–31. [7] Nguyen D, Lehman B. An adaptive solar photovoltaic array using model-based reconfiguration algorithm. IEEE Trans Ind Electron 2008;55(7):2644–54.
721
[8] Nguyen D, Lehman B. A reconfigurable solar photovoltaic array under shadow conditions. In: Applied power electronics conference and exposition (APEC); 2008; p. 980–87. [9] Jonathan PS, Peter RW, Darren B. Improved optimization strategy for irradiance equalization in dynamic photovoltaic arrays. IEEE Trans Power Electr 2013;28(6):2946–56. [10] Srinivasa PR, Saravana GI, Chilakapati NC. Maximum power from PV arrays using a fixed configuration under different shading conditions. IEEE J Photovoltaics 2014;4(2):679–86. [11] Alahmad M, Chaaban MA, Lau SK, Shi F, Neal J. An adaptive utility interactive photovoltaic system based on a flexible switch matrix to optimize performance in real-time. Sol Energy 2012;86(3):951–63. [12] Jonathan PS, Peter RW, Darren B. The optimized-string dynamic photovoltaic array. IEEE Trans Power Electr 2014;29(4):1768–76. [13] Juan DBR, Carlos ARP, Andres JSM. Reconfiguration analysis of photovoltaic arrays based on parameters estimation. Simul Model Pract TH 2013;35:50–68. [14] El-Dein MZS, Kazerani M, Salama MMA. Optimal photovoltaic array reconfiguration to reduce partial shading losses. IEEE Trans Sustain Energy 2013;4(1):145–53. [15] Santos P, Vicente EM, Ribeiro ER. Reconfiguration methodology of shaded photovoltaic panels to maximize the produced energy. In: IEEE power electronics conf (COBEP); 2011. p. 700–07. [16] Tria LAR, Escoto MT, Odulio CMF. Photovoltaic array reconfiguration for maximum power transfer. In: IEEE region 10 conf (TENCON); 2009. p. 1–6. [17] Liu Y, Pang Z, Cheng Z. Research on an adaptive solar photovoltaic array using shading degree model-based reconfiguration algorithm. In: IEEE control and decision conf (CCDC); 2010. p. 2356–63. [18] Cheng Z, Pang Z, Liu Y, Xue P. An adaptive solar photovoltaic array reconfiguration method based on fuzzy control. In: Proc of the 8th world congress on intelligent control and automation; 2010. p. 6–10. [19] Candela R, Dio VD, Sanseverino ER, Romano P. Reconfiguration techniques of partial shaded PV systems for the maximization of electrical energy production. In: IEEE international conf on clean electrical power (ICCEP); 2007. p. 716–20. [20] Gao L, Dougal RA, Liu S, Lotova LA. Parallel-connected solar PV system to address partial and rapidly fluctuating shadow conditions. IEEE Trans Ind Electron 2009;56(5):1548–56. [21] Patel H, Agarwal V. Matlab-based modeling to study the effects of partial shading on PV array characteristics. IEEE Trans Energy Convers 2008;23(1):302–10. [22] Silvestre S, Boronat A, Chouder A. Study of bypass diodes configuration on PV modules. Appl Energy 2009;86(9):1632–40. [23] Chenni R, Makhlouf M, Kerbache T, Chenni AB. A detailed modeling method for photovoltaic cells. Energy 2007;32(9):1724–30. _ D, Parlak KS. A new numerical solution approach for the real-time [24] Can H, Içkilli modeling of photovoltaic panels. In: IEEE power and energy engineering conf (APPEEC); 2012. p. 1–4.