Forecasting Solar Power Generation in India: A Path of Unnecessary Revisions

To achieve 40% of total installed capacity (power generation) from renewable energy by 2030 as per India’s NDC goal, India has a target of 175 gigawatt (GW) of installed capacity from renewable energy by 2022, of which 100GW is to come from solar, 60GW from wind. Taking India as an evolving place to put up solar businesses with international and domestic investors, the 100 GW target in Solar can not be considered as unrealistic or over-ambitious.

The major issue in large scale penetration of solar power is the variability and intermittency of power generation which creates the grid instability and the system relies on a proper energy mix of the conventional energy sources. To optimize the effect of unscheduled fluctuations, ‘forecasting and scheduling’ of the solar power generation is an essential requirement for grid management. Hence CERC, FOR and other state regulations propose the mandatory requirement of forecasting the solar power generation. If the error in scheduling the power is more than a certain limit, then there exists a penalty due to deviation, but for solar power forecasting 8 Intraday revisions (one revision in every 1 hour 30 minutes) are allowed. This revision is effective from 4^th time-block (1 time block = 15 minutes) onwards. Here the revisions are not mandatory, but revision is useful to tune the forecast accuracy.

Daily solar power generation shows a periodical trend due to which forecasting is comparatively easy using the proper pattern search algorithms which detects the possibilities of unscheduled fluctuations considering the weather models. Revision in forecasting is necessary, but the recent developments of proper pattern recognition techniques using DNN (Deep Neural Network) in forecasting methodology reduce the number of intraday revision as well as optimize the penalty due to deviation for the power generators. If the similar optimized forecasting solution is possible in lesser number of revisions, providing unnecessary intraday revision in every one or two hours can increase the indirect cost of grid stability.

To prevent the unnecessary intraday revision in the solar power forecast, the cost per revision in scheduling is already imposed at the SLDC level for some variable power generators depending on their power purchase agreements and at present this charge is same as the cost of revision of scheduling other energy sources. Here only R0 is free and other revisions are allowed but chargeable. This issue creates an extra financial burden for the power generators.

Though the present draft regulations in forecasting and scheduling of solar power generation propose the allowed revisions to be free, the unnecessary utilization of intraday revision strategy increases the complexity of demand-supply stability of the grid due to which the cost per revision is already imposed in some cases. Hence a proper forecasting in solar is an essential requirement which not only shows good forecast accuracy and reduces the penalty due to deviation, but it must optimize the penalty in the minimum number of intraday revisions for power generators and must have an acceptable level of accuracy to maintain the grid stability.

      Author

Abhik Kumar Das

Del2infinity Energy Consulting