Synthetic hourly irradiance

<< Click to Display Table of Contents >>

Navigation:  Physical models used > Synthetic data generation >

Synthetic hourly irradiance

Previous pageReturn to chapter overviewNext page


Global Horizontal

Version 5:
For the irradiance, the generation of GHI hourly values from monthly averages is performed by using stochastic models, due to the Collares-Pereira team in the 1980's. This model generates a sequence of days, and then a sequence of hours in the day, using Markov transition matrices. These matrices have been established in order to produce an hourly sequence, with distributions and statistical properties analogous to real hourly meteo data measured on more than 30 sites all around the world.

Version 6 :
From version 6, the generation of GHI hourly values from monthly averages is performed by the Meteonorm 7.2 algorithm (included in PVsyst (7.1 up to PVsyst 6.73)). The algorithm is an improvement of the original method, it adapts the Markov matrices to take into account local conditions such as the turbidity, it also uses modern knowledge about clear sky models. The generation of hourly value from daily values is based on the Collares-Pereira algorithmv (this model consists of two parts: the first part calculates an average daily profile, the second part simulates the intermittent hourly variations). From PVSyst 6.60, the randomizer may be initialised through the first random seed parameter in the synthetic hourly data generation section of the software. 10 different first random numbers of the generation algorithm of hourly radiation can be chosen. By changing this number, different time series of all meteorological parameters are generated due to a different initialization of the stochastic process. The monthly means remain the same. When performing simulations of grid-connected systems, this may produce variations of 0.5 to 1% in the yearly result.


Diffuse Horizontal

Version 5:

When the beam horizontal or the normal beam is available in the original data, they are used to calculate the diffuse part. Otherwise, the program uses the Erbs correlation. At the end of each month, the diffuse values are renormalized in order to match the specified monthly diffuse (scaling of the Kt value): this was not the case in the version 4, and in middle Europe the Erbs correlation had a tendancy to over-estimate the diffuse. Now in PVSyst when the monthly diffuse is specified, with usually lower values compared to the Erbs one, this increases the beam component and therefore the transposition results. This renormalization procedure can be skipped but it is not recommended to do so.

Version 6:
For Synthetic generation, the Meteonorm 7.2 algorithm uses the DirInt model to separate global into diffuse and beam. The DirInt global to direct model is based on a quasi-physical model, the DISC model, which has the form of a clear sky irradiance based on a Linke turbidity factor TL equal to 2.2, attenuated by a function of the clearness index Kt. This beam component is then corrected by a function of the modified clearness index K’t as defined in Perez et al. (Perez 1990), the solar zenith angle, and a stability index that accounts for the dynamics of the time series. The corresponding coefficients are obtained from a four-dimensional lookup table consisting of a 6 6 5 7 matrix. The DirInt model was developed with a hypothesis of an average Linke turbidity factor of TL = 3. For an evaluation of the DirInt model, please download Ineichen 2008.
At the end of each month, the diffuse values are renormalized in order to match the monthly diffuse, if specified. This is done through a correction which is not Kt dependant: each hourly value is corrected by the ratio between the specified monthly value and the as-generated-hourly-cumulated monthly diffuse.
Remember that with hourly measured data, PVSyst uses the Erbs correlation to generate diffuse from global, as in version 5.