Regarding the duck curve --
Suppose (contrary to current retail tariffs) that all power injected into the system, from whatever source, was paid the real-time price (RT LMP) by the system operator (
CAISO,
PJM,
ISO New England, etc.).
Suppose further that the costs of operating reserves (
regulation,
flexi-ramp, etc.) are assigned to market participants (generators or loads) that deviate from their hourly dispatch set-point or consumption average. Simply put, if your generation is going down while system load is going up, you (and load) get charged for operating reserves.
(In case you think that my hypotheticals are extreme, what I've described is exactly how the
Midcontinent ISO pays transmission-level generation resources, be they renewable, nuclear, or fossil.)
Finally, suppose that there are no federal or state subsidies for solar, but there is a $40/ton (CO2 equivalent) carbon tax.
Given these (radical) hypothetical changes in how solar gets paid, will the penetration of solar be as deep as CAISO projects in the duck curve?
My guess is no. When we get to an "over generation" point, the LMP goes negative -- that is, you have to
pay to inject power on the system. In all those hours of high insolation, LMPs will either be negative or fairly low, probably about $0.03/kWh (=marginal dispatch cost of a modern combined-cycle plant). So, there's not a lot of revenue to PV owners. Then, during the evening ramp, PV suppliers will be charged for part of the substantial costs of all that ramping.
So the problem is not that nuclear isn't flexible enough. It's that retail tariffs and subsidies are promoting uneconomically high levels of PV adoption, at least in California. If the real costs and value of PV were charged/credited to PV owners, there will be a natural equilibrium point of PV adoption. By masking these true prices, the retail tariff structure has the potential to create serious grid operational issues.