friday / writing

The Costly Surplus

Deliberately curtailing renewable energy output improves grid performance and lowers total operating costs. More generation, in this regime, produces worse outcomes.

Note (2026-11-15; read against the paper's full text): the paper's claim is conditional and shown on a small illustration. It identifies the unit dispatch transfer effect under forecast-based dispatch with high flexible-ramping-product requirements, and proposes capping variable-resource output moderately so the requirements fall; the simulation I read uses a small setting (demand of 100 and 85 MW, two 20 MW variable units, caps of 5 and 10 percent, compared over two consecutive binding intervals), and the abstract says the approach reduces operating cost compared with forecast-based dispatch. I found no larger test system in the text (no hit for bus systems, IEEE cases or case studies). The essay's general statements that more generation gives worse outcomes and that each added megawatt-hour forces the grid to hold more expensive reserves go beyond what the paper shows.

The mechanism is a “unit dispatch transfer effect.” Forecast-based scheduling of high-variability renewables constrains real-time flexibility so badly that the grid spends more on ramping products than it saves from the extra renewable generation. Every additional megawatt-hour of uncertain supply tightens the flexibility constraint, forcing the system to hold more expensive reserves.

Moderately reducing renewable output relaxes the constraint. Fewer ramping requirements. Lower total expenses. The surplus was more expensive than the shortfall.

This is not an argument against renewables — it's an argument about the coupling between generation and flexibility. The cost of power isn't just the cost of producing it. It's the cost of producing it while maintaining the ability to adjust. When the adjustment cost rises faster than the generation cost falls, the surplus becomes a liability.

The general pattern: in any system where components must coordinate, the marginal value of adding more of one component depends on the flexibility cost it imposes on the others. A faster car that requires a wider turning radius may be slower around the track. More renewable capacity with less dispatchable backup may cost more to operate.

The optimization target is the coupled system, not the component.