In my knowledge there is not restriction to use this three-diode model. But I recommend you to use the two-diode model, because the number of parameters to estimate are less when compared to the three-diode model. Also, some assumptions needs to be made in order to obtain all the parameters required by the this three-diode model, specially if you are using only the open-voltage, short circuit and maximum power point. I made a comparison of both models and the difference is not significant.

If you are looking for a really good quality reference of this topic, I recommend you to take a look of this research paper of Prof. Villalva.

Villalva, M.G.; Gazoli, J.R.; Filho, E.R., "Comprehensive Approach to Modeling and Simulation of Photovoltaic Arrays," Power Electronics, IEEE Transactions on , vol.24, no.5, pp.1198,1208, May 2009

doi: 10.1109/TPEL.2009.2013862

Two-diode model is just enough and  has a good physical meaning for the parameters. What is the theoretical meaning of three-diode model?

the third diode s is to consider contribution of the diode current component (Id3), due to

recombination in the defect regions, grain sites, etc. 

why we use it only to represent the pv cell only why not use it to represent  also pv modules ????

Of course you can use it in modules as well as in cell. IMO it just makes problem more complicated and does not bring much precision improvement. And, theoretically, does recombination in the defect regions or grain sites have a different factor a other than 1 or 2 in the item exp(-qV/akT)? Why? I will appreciate it if you give some references.

Dear Lu Wei 

'' I will appreciate it if you give some references.''

my problem is to find lot of refrence talk in this part 

thanks for your response

Theoretically, I think, it will work.  But I will appreciate to compare the results with a real PV.

yes tou can

Yasser A. Abdel-Hadi: I have this article at hand:

Analytic solutions of solar cell model parameters, Zhai Zaiteng et al., Vol.30 No.8 Aug.2009 Acta Energiae Solaris Sinica

It's a chinese article though, but you can see the formula and chart. The article just use one-diode model, and the fit is good enough (deviation

Lu Wei,   

now, i think the stochastic  algorithms become more popular than the analytical methods in the parameters estimation probleme and thier results achieve more accurate results 

Maybe different target: at system level, an accurate IV representation is more important , while at module and cell level, analytical methods which gives physical meaning is more important for analysis and improvement.

Dear Dalia,

All these models are an approximations to the real IV characteristics. The two diode model is physically justified by the recombination current in the space charge region and in the neural regions of the solar cell materials. If there is any leakage current at the surface of the cells or at the grain boundary it will be taken into consideration by the shunt Resistance. The series resistance is semiconductor material resistance of the neutral regions of the cell. I read the attached paper on the three diode model and their arguments for using three diode model in not convincing for me. please look at the link:https://www.researchgate.net/publication/272405523_A_Three_Diode_Model_for_Industrial_Solar_Cells_and_Estimation_of_Solar_Cell_Parameters_using_

The ideality factors can be be different from n1=1 and n2=2, if there is recombination at the interface between the n and p asides. However, strictly speaking, the recombination may be in the neutral regions and in the space charge region and on interfaces and grain boundaries.  But one has to find that n1=1, n2=2 and n3>2.

More important is that they cosidered RS is dependent on the current. Taking RS as a function of the current may lead to a better fit.

The three diode model may be more justified for multicrystalline solar cells and heterogeneous solar cells or for large area cells.

Since long time i developed a distribute solar cell model containing a didtributed network of diodes and resistors. 

please look at the link: https://www.researchgate.net/publication/3062547_A_distributed_SPICE-model_of_a_solar_cell

In conclusion, the tree diode model may be justified for jeterojunction and multicrystalline solar cells. Otherwise the single and the double diode model are sufficient. It is so that one has to make a compromise between the accuracy and complexity. 

wish you success

Article A Three Diode Model for Industrial Solar Cells and Estimatio...

Article A distributed SPICE-model of a solar cell

 great answer

i read the first paper you attach it , thanks alot prof for your benefit  response , i conclude this  note of using the three diode model when i read paper of 2007 for modeling multicrystalline solar cells, so i think why we don't apply three diode for different multicrystalline solar modules. 

thanks 

Dear Dalia,

Abdelhalim's answer is very thorough and impressive. I have used various models to simulate dye sensitized solar cells with interesting results. It depends very much what you want to use the simulation for. Don't forget in the real world, manufacturing tolerances and variations in the weather and sun position totally mask any minor variations you might get from adjusting internal cell parameters. On the other hand, if you are researching the cells themselves and maybe comparing technologies, then a simulation approach using diodes is essential.

Simulating large arrays, I found the best way to handle these is to simply measure a batch of cells, derive typical responses for the ranges of solar input and applied voltage and plug the numbers into a multidimensional lookup table. Then you can simulate large arrays with some confidence.

Best wishes with your work!