This work conducts a comprehensive study on harnessing the carbon mitigation potential of facade photovoltaics (FPV). We select all 102 largest cities in China for the study. We first show the high power generation potential of FPV compared with rooftop photovoltaics (RPV), and then we determine the cost-effective deployment pathway for FPV in 2030-2050.
Codes for the submitted paper "Mitigating Carbon Emissions in China’s Large Cities with Facade Photovoltaics".
Authors: Xueyuan Cui, Hao Li, Zeyang Long, Xiaohua Liu, Chris P. Nielsen, Michael B. McElroy, Dabo Guan, Jinyue Yan, Chongqing Kang, Xiaochen Liu, Yi Wang.
Version: 3.8.17
Required libraries include osgeo, fiona, shapely, pyproj, scipy, geopandas, sklearn, pyomo, etc.
Version: 3.36.3
Required plugin for solar radiation simulation: UMEP
Sources of required data and outputs of the study are publicly available in Baidu Netdisk (Code: jx8f)
1.1 UMEP_input
It includes the required data for calculating solar irradiance based on UMEP. All data for the selected 30 cities are zipped with .zip format. The data serve as the input source for the code of #Codes/Evaluation/UMEP_QGIS;
1.2 Sampled_input_output
It includes the required data for training the regression algorithm. All data are divided into .npy and .xlsx formats as the input features and labels. The data are generated by running the code of #Codes/Evaluation/Sample_label_read_.py and #Codes/Evaluation/Sample_feature_read_.py, and they are used as the input source for the code of #Codes/Evaluation/ML_training.py;
1.3 Regression_model
It includes the trained model that extrapolates the results of facade irradiance, rooftop irradiance, facade power, and facade capacity, respectively. The models are generated by running the code of #Codes/Evaluation/ML_training.py;
1.4 All_input
It includes all required inputs of the trained model for extrapolation. All feature metrics for the 102 cities are in the .npy format. The models are generated by running the code of #Codes/Evaluation/ML_input_read.py, and they serve as the input source for the code of #Codes/Evaluation/ML_output_power.py and #Codes/Evaluation/ML_output_capacity.py;
1.5 All_output
It includes all the output data that are extrapolated by the trained model. A more detailed introduction of each type of data is presented in #Introduction.txt in the folder.
2.1 Electricity_price
It includes all the required data on electricity prices in all provinces (2024).
2.2 Loads
It includes all the required data to generate load profiles of grid cells. The data are used as the input source for the code of #Codes/Optimization/Load_generate.py;
2.3 Other_parameters
It includes all other required data information for the optimization problem, including the land use, climate zones, etc.
All data in 2.1-2.3 serve as the input for the optimization problem when running the code of #Codes/Optimization/Multi_period_planning.py;
2.4 Fig_input_data
It includes all the required data to generate Figures in the main text and the supplementary of the manuscript, and the corresponding codes are in #Codes/Figures.
The results of the work are presented in our developed website: link.
The results are presented in ~2km*2km grid cells of all studied cities. They include the evaluated power generations, capacities, and carbon mitigation potential of facade and rooftop PVs, and include the deployment capacities of facade and rooftop PVs during 2030-2050, with a 5-year interval.
To reproduce the experiments of the proposed methods and generate the figures, please go to folders
cd #Codes/Evaluation
cd #Codes/Optimization
cd #Codes/Figures
respectively.
The Evaluation folder covers the procedure for solar irradiance simulation and power generation potential in the study, and the Optimization folder covers the procedure of the optimal deployment model for FPV's development pathway. The Figures folder includes the codes for generating the figures in the manuscript.
The introduction on the running order and each file's function is explained in Readme.md in the sub-folder in #Codes.