Dataset on effects of learning curve models on clean energy technology cost developments

Published: 20-01-2021| Version 1 | DOI: 10.17632/9spxxny27f.1
Contributors:
,
Lu Aye,

Description

This dataset includes input data to estimate learning-by-doing (LbD) and learning-by-researching (LbR) rates for onshore wind and solar PV in the United States. Using different learning curve approaches the simulated technology cost developments are also presented. Coefficient of determination (R square) and Root Mean Square Error (RMSE) were applied for quantification of the agreement between simulated and observed technology costs.

Files

Steps to reproduce

1. Select energy technologies to be investigated. Onshore wind and solar PV in the U.S. were selected for this study. 2. Collect cumulative installed capacity (domestic and international), annual public RD&D expenses (domestic and international), and technology capital cost (for selected energy technologies). Data collected for conducting this study are presented in six Excel files as described in <LearningCurveDataset-description.pdf> (see ‘Input data collected’ in the ‘Description of data’ section). 3. Organise data based on analysis objectives. The aim of this study is to analyse the effects of selecting learning curve approaches to describe clean energy technology cost changes over time. 4. Select periods and learning curve approaches to be investigated. For this study four periods were selected for onshore wind, and three periods for solar PV. Also, six alternative model structure definitions and specifications were selected for investigation (description of each model formulation and periods may be found within the study). 5. Input data for simulation. Data collected are entered for each technology and period to be analysed (e.g. solar PV from 1992 to 2017). Use the analysis files (e.g. <SolarPV_analysis_1992-2017.xlsx>) to input relevant data (e.g. Sheet 1FLC-A, Column A-F). 6. Estimate technological experience (e.g. Sheet 1FLC-A, Column H-K). 7. Estimate technological-based knowledge stock using the knowledge stock calculator file <KS_calculator.xlsx>, based on parameters selected (i.e. model formulations), for each energy technology. Estimated knowledge stock is inputted within the analysis files (e.g. Sheet 1FLC-A, Column M-S). 8. Perform a (log-log) linear regression analysis to estimate LbD and LbR elasticities. A log-log linear regression analysis is conducted using the ‘Data Analysis’ Excel tool (e.g. Sheet 1FLC-A, Column U-X). 9. Estimate LbD and LbR rates based on the regression analysis (e.g. Sheet 1FLC-A, Column AA). 10. Estimate technology costs based on LbD and LbR rates based on equations presented within the study (e.g. Sheet 1FLC-A, Column AB-AC). 11. Compare observed and predicted technology costs. Difference between predicted values (e.g. Sheet 1FLC-A, Column AC) and observed values (e.g. sheet 1FLC-A, column AD) are calculated (e.g. Sheet 1FLC-A, Column AE). 12. Calculate Root Mean Square Error (RMSE) (e.g. Sheet 1FLC-A, Column AF). 13. Repeat steps 5 to 12 for each technology and period to be investigated. For this study, seven Excel files are presented for each period and energy technology to be investigated, as described in <LearningCurveDataset-description.pdf> (see ‘Simulated data’ in the ‘Description of data’ section) 14. Create a summary file with results conducted. <Wind_and_SolarPV_analyses_summary.xlsx> file contains a summary of observed and simulated technology costs, learning rates, R square, and RMSE for each one of the periods selected and each one of the model formulations proposed within the study.