Deploying high penetration of variable renewables represents a critical pathway for decarbonizing the power sector. Hydro power (including pumped-hydro), batteries, and fast responding thermal units are essential in providing system flexibility at elevated renewable penetration. How to quantify the merit of flexibility from these sources in accommodating variable renewables, and to evaluate the operational costs considering system flexibility constraints have been central challenges for future power system planning. This paper presents an improved linear formulation of the unit commitment model adopting unit grouping techniques to expedite evaluation of the curtailment of renewables and operational costs for large-scale power systems. All decision variables in this formulation are continuous, and all chronological constraints are formulated subsequently. Tested based on actual data from a regional power system in China, the computational speed of the model is more than 20,000 times faster than the rigorous unit commitment model, with less than 1% difference in results. Hourly simulation for an entire year takes less than 3 min. The results demonstrate strong potential to apply the proposed model to long term planning related issues, such as flexibility assessment, wind curtailment analysis, and operational cost evaluation, which could set a methodological foundation for evaluating the optimal combination of wind, solar and hydro investments.
Deploying high penetration of variable renewables represents a critical pathway for deep decarbonizing the power sector. The conflict between their temporal variability and limited system flexibility has been largely ignored currently at planning stage. Here we present a novel capacity expansion model optimizing investment decisions and full-year, hourly power balances simultaneously, with considerations of storage technologies and policy constraints, such as carbon tax and renewable portfolio standards (RPS). Based on a computational efficient modeling formulation, all flexibility constrains (ramping, reserve, minimum output, minimal online/offline time) for the 8760-hour duration are incorporated. The proposed model is applied to the northwestern grid of China to examine the optimal composition and distribution of power investments with a wide range of renewable targets. Results indicate that the cost can increase moderately towards 45% of RPS, when properly designing the generation portfolio: prioritizing wind investments, distributing renewable investments more evenly and deploying more flexible mid-size coal and gas units. Reaching higher penetrations of renewables is expensive and the reductions of storage costs are critically important for an affordable low-carbon future. RPS or carbon taxes to reach a same target of emission reduction in China will result in similar overall costs but different generation mixes.
The Harvard-China Project adopted an Open Access policy in September 2017. Most journal articles published henceforth are available in the Harvard University open-access repository, DASH.