Hourly PV Power and Dust Accumulation Dataset — Laghouat, Algeria, February–August 2022

Description

Hourly dataset from a six-month dual-panel photovoltaic soiling experiment at LACoSERE Laboratory, Amar Telidji University of Laghouat, Algeria (33.8°N, 2.9°E, 767 m a.s.l.), spanning 17 February to 17 August 2022. Two identical SolarWorld SW-85 polycrystalline silicon modules were monitored continuously: one maintained clean (daily manual cleaning), one left exposed to natural dust accumulation. Dust surface density was quantified gravimetrically on 48 occasions using a KERN ABS 220-4N analytical balance (0.1 mg resolution) and surrogate glass collection plates; the sparse measurements were interpolated to hourly resolution using PCHIP interpolation. The dataset contains 4,368 hourly records and 14 columns: local timestamp, measured and PCHIP-reconstructed dust surface density, ambient temperature, cell temperature (NOCT model), in-plane irradiance (G_POA), relative humidity, wind speed, cloud cover, precipitation, clean-panel power, dusty-panel power, and two normalised power columns. Meteorological variables are from Open-Meteo ERA5 reanalysis. Dust density peaked at 22.54 g m⁻² in late June. Mean daytime power loss was 13.4%; maximum instantaneous loss was 29.7%.

Steps to reproduce

Full experimental and processing details are documented in the accompanying file Laghouat_2022_Dataset_TechnicalNote.docx. A summary follows. Electrical measurements: Both PV panels were connected to identical fixed resistive load banks. Voltage (LEM LV25-P transducers) and current (LEM LA25-NP transducers) were sampled at 1 Hz by a National Instruments USB-6009 DAQ device and averaged to hourly values using custom LabVIEW 2018 software. Dust measurements: Two 10×10 cm borosilicate glass plates were positioned co-planar with the PV modules. Every 2–6 days, plates were weighed three times on a KERN ABS 220-4N balance (0.1 mg readability) after 1-hour equilibration at 23±1°C, 40±5% RH. Dust surface density = accumulated mass / 0.01 m². PCHIP interpolation: The 48 measured dust density values were interpolated to hourly resolution using the PCHIP method (MATLAB pchip function) to produce dust_density_reconstructed_PCHIP_gm2. Meteorological data: Retrieved from Open-Meteo API (https://open-meteo.com) for grid cell 33.8°N, 2.9°E at hourly resolution; ERA5 reanalysis, observationally constrained. Irradiance: G_POA computed from solar geometry (Ineichen clear-sky model, Linke turbidity TL = 3.0–3.5), modulated by cloud cover (Kasten–Czeplak model), and transposed to the 36°/143° panel plane (Perez decomposition). The MATLAB source code used to train and evaluate all AI models is available at the same repository (Laghouat_PV_train.m, Laghouat_PV_figures.m).

Institutions

• University of Laghouat

  Laghouat, Laghouat

Categories

Licence