The following processes were taken into consideration while calculating PDISPL model used to optimize its parameters:
1. RI + hvi = R + I ; fi	8. I + I + M = I2 + M ; ka
= R + I ; (1-fi)	9. I* + I + I2 = I + I + I2 ; kg
2. I2 + hv2 = I + I ; f2	10. I + I + RI = I2 + RI ; kio
= I + I* ; (i-f2)	11. I + I + M = I2 + M ; kii
3. I* + R = I + R ; кз	12. I + I + I2 = I + I + I2 ; ki2
4. I* + RI = I + RI ; k4	13. R + I = RI ; ki3
5. I* + M = I + M ; k5	i4. R + R = R2 ; ki4
6. I* + I2 = I + I2 ; k6 7. I* + I + RI = I2 + RI ; ky	15. R +I2 = RI + I ; ki5

The constants of recombination velocity with the excited iodine atoms participating in it are substantially lower (for about the order) than those of non-excited I atoms. That is why the reaction I* + I* +M = I2 + M is not being taken into consideration. Above all, it is being assumed that the temperature of the operational laser medium is insufficient for pirolysis processes to be playing the noticeable role.

Gas flow over the constant cross-section channel was considered in 1-D approximation. Two options of operational substance flow were explored.

1. It is assumed that the direction of gas flow is perpendicular both to resonator axis and the direction of light pumping of the operational substance (cross-sectional pumping of the operational substance).

Differential equations describing changes in concentration of I* and I atoms enclose, alongside with the members considering for the above mentioned processes, those members, which describe the induced radiation and radiation absorption. Thermal conductivity, diffusion and viscosity are not being considered here. All the processes are assumed to be stationary, meaning that all the derivatives are equal to zero.

These processes were added by the differential equations describing concentration of reaction components versus distance and the equation for field density in resonator and generation power.

2. It is assumed that gas flux goes along the axis of the cylindrical cell, which coincides with resonator axis (longitudinal pumping of the operational substance). This option reflects the acting facility, on which the experiments were set.

In this case generation power was calculated with the assumption that the average amplification ratio for the operational substance is equal to the ratio of losses for the solicited specific resonator. This requirement is the expression of the condition for stationary generation mode, when the power of the induced radiation, generated by the active medium throughout the entire resonator volume is equal to the power of radiation output by the resonator in the form of outer radiation and radiation losses as it is being reflected by the mirrors.

Above all, diffusion of excited iodine atoms to the cell walls was considered alongside with the increased intensity of light pumping to the cell axis.

The system of differential equations was integrated by Gere method /2,3/ and by Runge-Kutt method /3/.

To assure the maximal energy yield of grid connected PV systems, system faults have to be identified as quickly as possible. For this task several procedures that offer an almost continuous performance check are in development or application (see e. g. [1], [2]). They are mostly based on the identification of reference values of the performance that are estimated from information on the meteorological conditions (irradiance, temperature) that may either be gained by on site measurements or be remotely sensed, i. e. derived from the images of meteorological satellites. Within the project PVSat-2 (see. [3]), a continuation of the PVSat project [1] funded by the European commission, a respective procedure based on remotely sensed hourly irradiance data is subject to further improvements. One field of improvements is the modelling of the performance of the generators.

Whereas in PVSat, main emphasis was put on the modelling of modules with cells of the crystalline Silicon (cSi) type, PVSat-2 aims to cover the whole range of different cell technologies. Thus a model, able to reflect the efficiency characteristics of both the classic crystalline silicon and the various thin film technologies is required. In the following a simple procedure for the estimation of the efficiency of PV-generators operated under MPP conditions in dependence of irradiance and module temperature n(G, T) is proposed. The application of this model is tested for Generators with cells made of crystalline silicon (cSi), amorphous silicon (a-Si) and copper-indium-diselenite (CIS).

A simple procedure for the simulation of the performance of grid connected systems

The modeling of the DC-output of the PV generator with a physical model either calls for a large effort devoted to the identification of the parameters of a 2-Diode model (if cells made of crystalline silicon are concerned) for each module type or — as was done within PVSat I — the use of a ‘standard’ parameter set, referring to the module that is most commonly applied. The errors associated with the use of this standard set for the modeling of crystalline cells had been discussed in [4].

In view of a broader application of thin film technologies, the 2-Diode model has its limitations and has to be modified (see [5]). This modification, even for the most simple thin film technologies (single active layer) increases the number of model parameters and the complexity of the parameter identification scheme.

In [6] an approach which consists of the use of a set of semi-empirical equations to characterize the dependence of the MPP current and voltage on inradiance and temperature, was presented. It has its advantages in a straightforward (no iterative fitting) procedure for the identification of the model parameters from data sheet information. This procedure is however not exactly transparent and is partly based on equations derived for ideal cell characteristics (no resistances, ideal diode factor). But despite these shortcomings it has proven its applicability for the modeling of the MPP performance of crystalline silicon cells and modules (see e. g. [3] 7]. Therefore another approach for the modeling of the DC performance of grid connected, MPP-tracked PV generators is analyzed here.

Based on the analysis of the characteristics of modules made from both crystalline and amorphous silicon and suggestions for a respective model by [8], a model for the dependence of the efficiency at MPP operation on the Irradiance G is proposed by [9].

Пмрр(G) = ai + a^G + 03 ln G (1)

a1-a3 are device specific parameters.

This equation is primarily applied to represent the efficiency at 25°C. The performance at operation temperatures other than 25°C may then be modeled by the standard approach:

WG, T) = n(G,25°C) (1+a(T-25°C)) (2)

Thus this model has in total 4 parameters, which may be determined in straightforward manner from e. g.:

— the MPP-power at STC

— 2 values for the MPP power at other irradiances for 25°C

— the MPP-power at 1000 W/m2 and a temperature ^ 25°C.

If data for the measured power output of operating systems are available, the parameters may as well be fitted for that set. Due to the structure of the model simple linear fit procedures may be applied. A similar model proposed by [11], in which instead of the term a1+a3lnG the a1Ga3 is used, calls for nonlinear fit procedures.

In the following we will present the application of this model for both, different module types and different input data sets.

Example for the application to the modeling of a c-SI generator based of data sheet information

Using data from systems out of the WEBlog supervision scheme [2] the model for the Пмрр(С, Т) performance is applied to c-Si generators.

For these generators, the necessary input data for the 3-point evaluation of the model parameters are available due to the application of the SolEM modeling tool [7] based on [3] (except one: the MPP-current, necessary for calculating the MPP power the at the intermediate radiation level (500 W/m2) had to be estimated from the values of the MPP — current at 1000 and 100 W/itP) .

Figure 1 shows an example for the estimated efficiency characteristics of a standard cSi module at 25°C together with the 3 efficiency values used for the model determination.

Fig. 1: Example for an efficiency characteristic of a cSi module at 25°C extrapolated from three sample values (single points connected by straight lines), using the efficiency model discussed here.

For the modeling of the temperature dependence (see eqn. 2) a default temperature coefficient of 0.45%/°C is assumed.

As within the WEBLOG scheme only data of he AC performance are available, a model for the simulation of the complete system performance was implemented using the INSEL simulation tool. Hereby the inverter efficiency is simulated according to the standard model [8] based on the respective manufacturers data. For the system losses it is assumed that 2% of losses have to be accounted for both, the mismatch and MPP-tracking errors. The maximum losses in the cables are assumed to be 4%. Due to these rough estimates in absence of more detailed information, the results have to be interpreted with some care.

Fig. 2 gives an example for the comparison of the performances of the new n(G, T) model and the SolEM procedure. Monthly sets (October 2002) of measured data from 2 generators that are supervised by WEBlog are used to test the models. The performance of the new model is equivalent to slightly superior to the performance of SolEM.

70

■ Solem new model

0 10 20 30 40 50 60 70

g 60

50

40 О

30

20 ГО

Q.1 10 0

P_ac_meas. [kW]

P_ac_meas. [kW]

Fig. 2 Scatter plot of simulated and measured power output of a PV generator taking part in the WEBLOG supervision scheme. The simulations are performed with SolEM (squares) and a system model applying the n(G, T) model for PV performance discussed here (diamonds).

Example for the application to the modeling of an a-SI generator based on measured data

At the Universidade Federal de Santa Catarina at Florianopolis, Brazil a PV generator made of amorphous silicon modules is in operation and monitored since 1997. Its initial nominal power output was 2 KW (see. e. g. [13]).

eta measured ■ eta model irradiance [W/m2] Fig. 3: One month of measured MPP-DC-efficiencies for the 2kW a-Si generator of the UFSC at Florianopolis, Brazil. The line indicated the efficiency model fitted to this data set.

The model for the irradiance dependence of the MPP power was applied while testing the applicability of the PVSat procedure for South America [14]. As no significant temperature dependence of the MPP-power could be identified [see [13], the operation temperature was not taken into account by this fit (i. e: a = 0).

Figure 3 shows measured MPP-DC-efficiencies as a function of the irradiance together with the efficiency model. The irradiance data are measured by a cS-sensor (Matrix). A correction is applied to transfer these data to a set that presents values equivalent to pyranometer measurements. It can be remarked, that for irradiances > 200 W/m2 the efficiency of the generator is almost constant, with a small tendency to decrease with higher irradiances. The model performs well, however the drop of the efficiency at higher irradiances is somewhat exaggerated. It should be remarked, that the DC data presented already include the losses due to the wiring, which are therefore also included in the model fitted to these data.

— PVSat measured data

time [h]

Fig. 4: Measured and PVSat derived data of the DC power output of the a-Si generator at Florianopolis for the first 15 days of September 2002.

The performance of this model was tested in the framework of the PV-Sat procedure, by applying Meteosat derived irradiance data to calculate the DC-power output of this system.

An example for this comparison is given in figure 4.

The long time performance of the model — coupled here to the PVSat prodecure — is given in table 1., showing the monthly DC gain for this System. The MBE is smaller than 7%.

	PVSat	measurements
	E_dc [kWh]	E_dc [kWh]	MBE%
July	198.80	187.60	5.97
August	180.35	171.85	4.94
September	201.24	197.17	2.06
October	205.88	198.05	3.95
November	257.50	251.24	2.49
December	252.99	236.49	6.98

Table 1: Monthly DC-gains for the a-Si generator at the UFSC.

From these data it can be concluded that the model proposed can be used to present the performance of a-Si generators, at least for a time span, at which the generator shows no dynamics in its aging process (which is supposed to be the case for the generator of the UFSC, which for the time of the measurements presented had reached an age of 5 years).

Example for the application to the modeling of a CIS generator based on measured data

For the analyses of the performance of a generator using modules with CIS technology, a data set had been made available by the University of Applied Sciences Frankfurt. These data stem from a generator with a nameplate rating of 513 W, but is supposed to have a real rating of about 600 W. The measuring system uses a reference cell for the registration of the irradiance. The data are directly used in the following. Two month of data had been analyzed (February and August 2003). For Due to the increased relative uncertainty of the irradiance sensor radiation sensor at low radiation level, the respective data are skipped.

The DC performance for these months is given in fig. 5. The differences of the performance at these 2 month may be traced back to variations of the operating temperatures (max. values: February: 42°C, October: 69°C).

lin » data lin ♦ data

0 200 400 600 800 1000 irradiance [W/m2]

0 200 400 600 800 1000 irradiance [W/m2]

700 600 „ 500 400

Ф

300 200

100 0

Fig. 5 a, b

Measured DC-performance of a generator using CIS modules. 5a (left) refers to August 2003, 5b to February 2003. The straight line in the plot is added for orientation.

The data for the month of August are used for the identification of the model parameters a1- a3 and a by a least square fit. The application of this data set to the modelling of the DC power is presented in the scatter plot in figure 6a. Applying the same parameter set to the data of February results in the comparison given in fig. 6b. The bias of the modelled data is smaller that 1% for August and 3.4 % for February. The applicability of the model is also confirmed by the fact, that the value for temperature coefficient a (-0.36 %/°C) is in close agreement to the value of -0.3%/°C that was determined for periods of constant irradiance conditions by [15].

Conclusion

The simple n(G, T) model for the MPP performance of PV generators seems to offer the potential to give a good representation of the dependencies of the system output on irradiance and temperature. Its performance as standard model in procedures for the performance check of PV systems is currently evaluated within the PVSat-2 project.

♦ data——- ident measured power DC [W]

♦ data ————— ident.

0 200 400 600 measured power DC [W]

Fig. 6 a, b Comparison of measured and simulated DC performance of the CIS generator analyzed. The straight line indicates the identity.

Acknowledgements

Thanks are due to the people of the LABSOLAR of the Federal University of Santa Catatina, Florianopolis (Brazil) and to Prof. Dr.-Ing. habil. J. Lammel of the University of Applied Sciences at Frankfurt am Main (Germany)

The simulation study has been carried out according to the voltage collapse scenario and table (1) using MODES. The simulation results are illustrated in figures (2),(3),(4),(5), where under these figures, the subfigures are:

(a) Active & reactive power of nodes N204 , N206 , N8; showing the ramp change of load.

(b) Active power of generating units; where the generating unit M4 is switched off at t=180 sec. and the response of the other units is recorded.

(c) Voltage profile of nodes N204 , N206; where the constant value lines indicate the voltage set value, to which the OLTC of the transformers try to adjust the fluctuating instantaneous values.

(d) Reactive power & voltage profile of generating units; where a stable or unstable voltage collapsed behavior of generating units can be detected.

An analysis of the simulation results reveals the following remarks:

1- Referring to case A, figure (2-a), it can be seen that after the switching off of M4, all generating units have to increase the real power injected to compensate for the loss of one of the major units in the network. Also, the balance in the reactive power has caused that one of the units monitored M5 has suffered of fluctuating reactive power and terminal voltage (figure 2-d); this unstable behavior specially beyond t =300 sec has as a result the disconnection of unit M5. Figure (2-c) shows a continuous adjustment carried out by OLTC for the voltage values at N204 & N206 , but in a narrow band. At t=300sec, a speedy deterioration in voltages have been recorded; a voltage collapse problem in the network.

2- For case B, figure (3-a) shows that, with the reduced load at N204 , the ramp change is still satisfied. For this case, all generating units have also to compensate for the power loss, but as expected the change will be less than that recorded for case A because the introduction of DG to cover half the load at N204. After the disconnection of the double line at t =300 sec, the generating unit M5 started to suffer fluctuating voltage, but less fluctuating than that in case A. A voltage collapse problem has recorded in this case too.

3- For case C, where the DG can fully supply the demand at N204 , the voltage collapse scenario is still followed. Nevertheless, all units have maintained a stable operation except M5 showed ever fluctuating voltage trend beyond t = 300 sec. Moreover, OLTC had to adapt the voltage of nodes N204 & N206 in a wider band.

4- Case D shows a better performance of the network. In this case, DG can fully supply the demand at N206 keeping the voltage collapse scenario valid. All generating units including M5 achieved a successful adaptation of their individual performance after switching off M4 and double line outage, without going in a voltage collapse problem for the network. It should be noted that, OLTC had also in this case to move in a wider band.

5- The powers generated by units N12 & N15 & N16 are almost constant, because the points of disturbance are rather far from these units. Whereas, for unit M5 being closer, its output must significantly change to compensate for the power loss. The presence of DG has affected the performance of M5 depending upon the point of injecting DG.

2. Conclusion

This paper represents a study of one of the problems associated with interconnected power systems; the voltage collapse problem. The study has been carried out taking into consideration a real simulation of power systems with different types of loads; static as well as dynamic loads, using an advanced specialized tool. Also, the scenario suggested in this work is not a simple one, but a real scenario of what can happen in real practice. Different cases were analyzed, showing that, DG is not an absolute grant for network topologies. Generally, in steady state operation for some networks, DG is a plus added value, reducing loads on generating units and maintaining an improved voltage profiles. In case of common disturbances, DG could cause a worth situation or on the contrary an improved system operation depending upon the system configuration.

Therefore, as a conclusion, the study shows that, off-line analysis of steady state and under common faults operation should be carried out to decide: how much added value is

expected in introducing DG, depending upon the topology of power systems, and at what injection points DG will be connected with recommended penetration levels.

References:

[1] R. Karki, R. Billinton, "Reliability/ Cost implications of PV and wind energy utilization in small isolated power systems”, IEEE Trans. On Energy Conversion, Vol. 16, No. 4, December 2001.

[2] Hirohi Asano, Koji Yajima and Yoichi Kaya, "Influence of Photovoltaic Power Generation on Required Capacity for Load Frequency Control", IEEE Trans on Energy Conversion, Vol. 11, No 1, March 1996, pp 188-193.

[3] Bjorn H. Bakken, Michael M. Belsnes & Jarand Roynstrand, "ENERGY DISTRIBUTION SYSTEMS WITH MULTIPLE ENERGY CARRIERS", SINTEF Energy Research, N-7465 Trondheim, Norway, 2002.

[4] G. Ernst Palomino, John Wiles, John Stevens and Frank Goodman, “Performance of a grid connected residential photovoltaic system with energy storage”, the 26th IEEE Photovoltaic Specialists conference, September 29- October 3, 1997, Anaheim, California, USA.

[5] Fadia M. A. Ghali, M. Said Abd El Moteleb, “Stability Region Estimation of Hybrid Multi­Machine Power System”, 9th IEEE International Workshop on Robot and Human Interactive Communication, IEEE RO-MAN 2000, pp. 149-154, Japan.

[6] B. H. Chowdhury, A. W. Sawab, “Evaluating the Value of Distributed Photovoltaic Generation in Radial Distribution Systems”, IEEE Trans. On Energy Conversion, Vol. 11, No. 3, September 1996, pp. 595-600.

Jih-Sheng Lai and Fang Zheng Peng, “Multi-Level Converters — a New Breed of Power Converters”, IEEE Trans. IA, Vol. 32, No. 3, May 1996, pp. 589-517.

[7] A. S. Neris, N. A. Vovos and G. B. Giannakopoulos, ”A Variable Speed Wind Energy Conversion Scheme for Connection to Weak AC Systems”, IEEE — Trans. on Energy Conversion, Vol. 14, No. 1, March 1999, pp. 122-127.

[8] Giancarlo Ferrari-Ttrecate, et al, “Modeling and Control of Co-generation Power Plants: A Hybrid System Approach”, PV in Europe Conf., October 2002, Rome.

Figure (2): Case A

Figure (3): Case B

Figure (2-a) : Active & reactive power of nodes
N204 , N206 , N8 in p. u.

Figure (2-b) : Active power of generating units,
in p. u.

Figure (2-c) : Voltage profile of nodes N204 ,
N206 in p. u.

Figure (3-a) : Active & reactive power of nodes
N204 , N206 , N8 in p. u.

Figure (3-b) : Active power of generating units,
in p. u

Figure (3-c) : Voltage profile of nodes N204 ,
N206 in p. u.

Figure (2-d) : Reactive power & voltage profile
of generating units

Figure (3-d) : Reactive power & voltage profile
of generating units

Figure (4): Case C Figure (4-a) : Active & reactive power of nodes N204 , N206 , N8 in p. u.

Figure (4-b) : Active power of generating units, in p. u.

Figure (4-d) : Reactive power & voltage profile of generating units in p. u.

Figure (5-c) : Voltage profile of nodes N204 , N206 in p. u.

Figure (5-d) : Reactive power & voltage profile of generating units in p. u.

Figure (5-b) : Active power of generating units, in p. u.

Figure (4-c) : Voltage profile of nodes N204 , N206 in p. u.

Figure (5): Case D

A comparison between different RES requires a data analysis in a standardised manner. Different terminology and different representations of the same data make it difficult, if not impossible, to compare data from different locations with potentially different applications. For this reason, a stringent, formalised manner of presenting data is required and has to include all possible combinations of designs and components. Graphs, histograms and performance figures are used to show how RES and its components are used. A description of the Standard Evaluation Report is given in /1/ and examples are available at www. benchmarking. eu. org/reports. asp.

As the standard evaluation report will be used to identify categories of similar use of components, e. g. batteries, fuel operated generators or any other component in RES, good data are required. One of the minimum requirements /2/ is data availability for a period which also reflects seasonal variations. Where these are noticeable, one year of data with little loss of data is required. In addition, at least one data set per hour and the exact position of the measurement point must be known. The value for energy provided from a PV module, for instance, depends on whether the sensor is before or after the charge controller. In order to be able to easily compare different RES systems, it is important to produce normalised performance indicators. The following definitions give an example for batteries:

• The battery capacity is given at its 10 hourly rate C10.

• The battery voltage is normalised to the cell voltage, i. e. battery voltage divided by the number of cells in the battery pack.

• The charge and discharge current is normalised to the C10 rate of the battery.

The standard evaluation report is generated by means of the software tool THESA /3/. THESA is the short form of Technical Hybrid Energy System Analysis. It is an online application for analysing logfiles containing measured time-series from hybrid energy systems. THESA also needs a parameter file describing the system, its components, and the format of the logfile. To generate such a parameter file in an easy way, a graphical user interface was developed which provides assistance to generating the parameter file for a certain system. Due to the number of different systems, the range of renewable energy sources (PV, wind, biomass, hydro) and the range of data formats and data averaging algorithms, THESA needs to and does provide a high level of flexibility. THESA can not only be used for measured data but also for data generated by a simulation software with a sufficient level of detail and high enough time resolution.

The system design is limited to one AC busbar and none or one DC busbar and a design with one DC busbar. The system with a DC busbar is called a DC system even if it includes an AC busbar, the system with a single AC busbar is called an AC system. Figure 1 shows, for example, a DC system with all possible components and the measuring points with their labels which THESA supports.

An automatically generated Standard Evaluation Report has a pdf Format and a size of approx. 6 MB. Figure 2 shows an example of one time series for a PV module and figure 3 an example of one time series for a battery. The two figures show that unexpected deviations can be interpreted directly to make possible improvements to the system and to detect a component which is at the end of its lifetime.

■* Phydro dc * Pwtg dc

•dcgen_dc ■* Pdcgen dc

location

storage

vwind

"flow

wind

jrnb

pv-generator

micro hydro

wind generator dc-generator

bio generator

)Tpv

Tbatt

Ubatt

ban

Pbatt

Ufivdra

U..-у

Udcgen

Ibatt

■batt d

busbar

Pbjtt

‘pv

Ihydro

Idcgen

Pbatt d

IS0C|

ш

Ppv

I’hyd

Pdcgen

Ibatt dc

“• Pbatt dc

acgenerator

■pv dc

‘ "• Ppv dc

Ihydro dc

Iwtg dc

lbio_dc w Pbio dc

U|o<id

latgen.

dc-load

rectifier

pfac

‘ acqen ac

acgen ac

I mv

“load

diludd

ac-load

ac busbar

О

flatload I

Figure 2: Scatter diagram of current output of a PV module as a function of solar insulation. The line showing a lower than normal current output may be the result of shading at certain times of the day or year.

Figure 1: Basic design of the DC system and all measuring points with their labels which are supported by THESA. The data files have to be assigned to one of the measuring points. In most cases, only a fraction of the data that could be handled will be available, however, this does not invalidate the result of the analysis.

The presence of power electronic components in Photovoltaic plants for DC/AC power conversion can bring forth interferences at radio frequency and can, under different ways, produce electromagnetic noises to the nearby electronic and electric appliances [1]. The problem can be more relevant for the Building Integrated Photovoltaics applications due to the shortest distances toward the domestic appliances.

During recent years the integration of PV on buildings has reached a very interesting level and has been a challenging alternative to the traditional plants. Since their electric nature and the presence of other components they should comply with electromagnetic compatibility. The problem is more specific for the Building Integrated Photovoltaics for the shortening of the distances between the electric and electronic appliances ant the PV plant, sharing the same cables.

The presence of switching components with frequencies ranging up to some tens of kHz generates the most of noises in form of conductive type, traveling along the power supply and the latching cables, within the range from 150 kHz to 30 MHz [2,3]. These interferences, moving along the cables, could head toward the direct connected part of the plant, the modules, causing them to radiate like antennas, so broadcasting disturbances toward the near electronic equipments. It is important to evaluate the radiated and conducted effects and compare them with the existing standards for electromagnetic compatibility. In addition to the national rules, the electronic equipment must comply with the EN 50081-11 and EN 55011 for the radiated emissions and with the EN 50011 IEC 1000-3-2 and IEC 55014 for the AC and DC sides as conducted emissions [4].

The paper intends to present a work that ENEA, the Italian Agency for New Technologies the Energy and the Environment, has begun in the effort to give more understanding and some answers to the EMC BIPV issue, by making measurements on a PV facade, provided of two different inverters: one with a single DC/AC Pulse Width Modulation, PWM, inverter and the other with one by one module integrated inverters. Radio frequency measurements in different points of the PV plant have been accomplished by means of a RF probe to
evaluate the presence of noise in the connection cables. The conducted emissions inside the plant and the radiated emissions transmitted from the electric background have been estimated by means of the Line Impedance Stabilisation Network, LISN, located before the supplying of the grid [5].

The LISN has been used to intercept and evaluate the disturbance at RF levels from the plant to the grid and to avoid that noises present in the grid could interfere with the PV plant [6].

Grid Connected Solar Inverter AC data Output Voltage 230 Vac Frequency 50 Hz Efficiency max 95 % Power factor 0.99 Total Harmonic Distorsion < 3% DC data Typical String length 5-9 72 cells modules Rated Power 700 — 1200 Wp Starting Power 2 W Input voltage range 100- 380 Vdc DC connection PCB Spring Terminals Main Connection PCB Spring Terminals Table 1. Grid Connected Solar Inverter

AC Module Inverter type A AC data Output Voltage 190-270 Vac Frequency 49-51 Hz Efficiency max 94% Power factor 0.99 Total Harmonic Distorsion < 3% DC data PV-Module type in series 72 cell Rated Power 100W Starting Power 0.15 W Input voltage range 24-50Vdc DC connection Cable Water Resistant Main Connection Cable Water Resistant Table 2. Module Inverter Type A

Noises in the range between 150 kHz and 30 MHz have been compared, according to the Italian standard CEI 110-6, related to the European 55011 class B (appliances to be used on buildings for domestic use and directly connected to the low voltage grid) and related to the principles of measurements for radio signals noises of Industrial, Scientific and Medical devices, the so called ISM [7]. Different measurements have been performed on a grid connected fagade with PWM inverter (see table 1) and distributed module inverters (see tables 2 and 3); these last ones are small inverters that are inserted on each module and are provided with a normal socket [8].

AC Module Inverter type B
AC data
Output Voltage	230 Vac
Frequency	49-51 Hz
Efficiency max	92%
Power factor	0.99
Total Harmonic Distorsion	< 5%
DC data
PV-Module type in series	72 cell
Rated Power	110W
Starting Power	0.85 W
Input voltage range	24-40Vdc
DC connection	Screw Terminals M4
Main Connection	PCB Terminal Block

Table 3. Module Inverter Type B

The PV fagade is located at the CR ENEA at Portici; it is sized 11.7 square meters and it is composed by 6 glass-glass modules realized by Schuco using the sggPROSOL® technology in a sandwich structure with the two external panes made of heat-strengthened glass (sggPLANIDUR®) and the solar cells bonded between them by means of an high — transparent resin [9].

Ahmed K. A., Motaz* M. S., Mohamed E. A. and Alaa S. H.

Faculty of Engineering, Alexandria University, Egypt.

* Institute of Graduate Studies and Research, Alexandria University, Egypt.

ABSTRACT

This paper presents a genetic algorithm optimization for MIS/IL solar cell parameters including doping concentration NA, metal work function фт, oxide thickness dox, mobile charge density Nm, fixed oxide charge density Nox and the external back bias applied to the inversion grid V. The optimization results are compared with theoretical optimization and shows that the genetic algorithm can be used for determining the optimum parameters of the cell.

KEY WORDS

Photovoltaic, Solar cells, MIS/IL, Optimization and Genetic Algorithm.

1. INTRODUCTION

Aluminum/thin SiOx/P-Si structure with an induced inversion layer MIS/IL has been studied [1-4]. Its distinctive features are formed by shallow n-p junctions of about 0.1 pm thickness arising under the influence of a positive fixed charge in the insulating layer. However, practical achievement of high efficiency of MIS/IL solar cells was complicated by the presence of significant loss of generated power in the high-resistance inversion layer (IL). Some authors [5-7] have reduced the IL specific resistance by an increased of the built-in charge density and this made the MIS/IL technology complex and increase the cost of the MIS/IL solar cell. Several different improved methods were proposed. Among them Ref.[8], Built-in positive charge in the insulating layer was used for improving the conductivity of inversion layer of the underlying semiconductor surface. In addition, the difference of work functions between the semiconductor and the thin metal strips of the inversion grid was used. The inversion metal grid with a low work function was formed on top of an insulating layer of tunneling thickness between the strips of the collecting grid as shown in Fig.1. The aim of the work is to optimize the silicon MIS/IL solar cell parameters. The MIS model is based on the use of internal factors such as built-in charge including fixed oxide charge, mobile charge, interface trapped charge, doping concentration, oxide thickness and metal work function. External parameters will be included namely, external back bias on the inversion grid [9].