Flexibility in generation planning: Identifying key operating constraints

This paper utilizes the recent methodology of clustered integer unit commitment (UC) - which tractably captures operational flexibility within generation expansion planning - to explore which operating constraints are the most important for generation investment decisions. Flexibility has been previously shown to alter the optimal generation mix, particularly in scenarios with high flexibility required by significant renewables (>=20%) and/or decreased flexibility of some low-carbon technologies (e.g. traditional nuclear). This work explores the potential to relax some operating constraints to speed-up computation while controlling errors. It is found that operating reserves and maintenance are the most important constraints, while hour-to-hour ramping is least important. However, relaxing integers provides the best accuracy vs. performance trade-offs, but only when the linear program (LP) relaxation includes the full problem with UC constraints. This reduces computation time ~50× with the lowest errors across all metrics: cost (-1%), CO₂ (-1%), capacity (-9%), and energy mix (-9%). Considerably high speed ups are possible using the LP relaxed formulation with selected subsets of UC constraints. For cost and CO₂, the combination of grouped reserves, ramping, maintenance, and LP provides errors around 2% with ~1500× speedup. However, for capacity and energy mixes, only LP provides reasonable errors (<;25%).