Working example data for generating syntehtic 4-second utility-scale PV output time series from hourly solar irradiance data

Published: 7 January 2021| Version 2 | DOI: 10.17632/49hv5brf92.2
Contributor:
Kanyawee Keeratimahat

Description

Generation of synthetic PV output time series at 4 seconds using hourly solar irradiance. This dataset is a working example for the methodology proposed in the article "Generation of synthetic 4-second utility-scale PV output time series from hourly solar irradiance data". The codes and dataset can be used to generate synthetic 4-second utility-scale PV output time series for a PV plant based on the hourly solar irradiance and the 4-second training data derived from the reference PV plants that are operating in the Australian National Electricity Market (NEM). Corresponding author: Kanyawee Keeratimahat (k.keeratimahat@unsw.edu.au).

Files

Steps to reproduce

PV plants notation: 277 - Nyngan, 295 - Broken Hill, 299 - Moree, 309 - Royalla. 'example - input of reference data' contains parameters such as tilt angle, azimuth of the four reference plants which are used as the input to 4.2.2. Generate reference/training data 'code - Generate reference data' contains subfolders of steps in generating statistical distribution from the reference data 4.2.1 contains code which caluclate bin number from hourly solar irradiance data. 4.2.2 contains codes to calculate 4-second clear sky PV output, then arrange the 4-second clear sky output into hourly interval and also filter the hours to be binned. 4.2.3 contains code which sorts the hourly interval according to each pair of GHI and DNI. BIN_INDEX.mat contains the bin numbers that physically exist. 4.2.4 contains code that fits q-exponential distribution to the measured distributions of the reference PV plant (training data). 4.2.5 & 4.2.6 contains the code which scale the distributions to the desired DC plant capacity. Generate synthetic time series 'code-Synthetic time series' contains: - model_4_sec.m which generate 4-second time series with input of the hourly GHI, DNI and PV output of the plant to be modelled and the scaled distributions from the previous step. - prepare_for_validation.m which calculate PV output variability from the generated 4-second time series and other statistical measures including RMSD, NMBD and KSI. Sample outputs Each folder contains some examples of output file of the code(s) within the same folder which work as the input files to the next step in the flow. The user may be required to change input directory if the input filenames are different to the sample outputs. Where a reference plant and a test plant are paired, the file name can be read as follow: bin_dist_pool_(yyyy)_(reference plant)_(test plant).mat Hourly_model_(yyyy)_(test plant)_(reference plant).mat The flow chart diagram in "Generation of synthetic 4-second utility-scale PV output time series from hourly solar irradiance data" should be used to assist the understanding of the modelling steps detailed above. Note: We do not have the permission to provide raw data of Satellite derived irradiance and AWS weather data from the Australian Bureau of Meteorology and 4-second SCADA of generation output in the NEM. However, the 4-second SCADA is publicly available on NEMWEB. The codes involve the use of funtions from PV_LIB Toolbox for Matlab which can be downloaded from https://pvpmc.sandia.gov/applications/pv_lib-toolbox/.