model.py: reference implementation for "A Computational Framework for Perovskite Indoor Photovoltaic Energy Harvesting"
Year: 2026
Type: software
Abstract: Reference implementation accompanying "A Computational Framework for Perovskite Indoor Photovoltaic Energy Harvesting". A single Python script that regenerates every tabulated value in the manuscript from the same closed-form single-diode expressions. Two constants set every indoor result in the paper, and this script derives both from published standard spectra by direct quadrature rather than adopting tabulated values. The first is the luminous efficacy of radiation of the source, 317 lm per optical watt, computed from the CIE LED-B3 4000 K illuminant weighted by the CIE 1924 photopic luminosity function; this is not the wall-plug efficacy of a luminaire, which is roughly three times smaller and does not belong in a photometric-to-radiometric conversion. The second is the spectral utilisation factor M, computed against the ASTM G173 AM 1.5G spectrum with a step-function external quantum efficiency to the band edge, giving 1.84 for the perovskite and 1.04 for the crystalline silicon reference. M exceeds unity for the perovskite because roughly 45 per cent of the solar spectrum lies beyond its 760 nm band edge, where the absorber is transparent, while the LED emits almost nothing there. Running "python3 model.py" prints Tables 1, 3, 4, 5, 6 and 7 of the manuscript together with the extracted parameters and the intermediate quantities they depend on. Nothing is fitted; the only numerical solver in the pipeline is a one-dimensional root-find inverting Green's fill-factor expression for the normalised series resistance. Dependencies are pinned in requirements.txt. An interactive browser-based estimator, PeroCast, accompanies the same paper and is deposited separately at doi:10.5281/zenodo.22101976. The two are independent implementations of the same model in different languages, so agreement between them guards against transcription error as well as arithmetic error. PeroCast carries the two constants above as fixed values, whereas this script derives them. Limitations, inherited from the paper. No harvester hardware was constructed and no indoor measurement was performed. Indoor figures are extrapolated roughly three hundredfold below the lowest measured irradiance; over the threefold extrapolation the data do permit, the model over-predicts fill factor by up to 5.8 percentage points. The shunt resistance is assumed rather than measured. The spectral utilisation factor is modelled rather than measured, with an uncertainty of order plus or minus 5 per cent. The crystalline silicon reference is defined by nominal rather than measured parameters.
Institutions: ITMO University, Harbin Engineering University
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204198240