CESentinel will compute loss-of-load expectation, capacity value and capacity-market clearing: the numbers that decide what a battery, wind farm or peaker is worth to system reliability. Today they usually come from a six-month consultancy study run against last year’s fleet. CESentinel will make them reproducible runs on the current one.
Loss-of-load expectation and expected energy not served come from sequential Monte Carlo over decades of historical weather years, on the same fleet model the rest of the platform uses. Demand and renewable output are conditioned on temperature, wind and irradiance, and unit outages are modelled as Markov chains. Battery state of charge is carried hour by hour through each replication, so a two-hour and a four-hour battery get different capacity values.
Capacity value is computed as ELCC by equivalent-firm-capacity matching: we solve for the firm injection that gives the same reliability as the asset, per technology class and per asset. The marginal value is taken at the current fleet point and the class average by re-solving. Both are reported, because either one alone misleads.
Capacity markets in GB, France and Italy, and PJM-style designs, are simulated as auctions over the derated fleet, with administrative demand curves and each market’s obligation rules built in. The output is the clearing price, the awarded capacity and the outcome for each asset. Other mechanisms can be defined in a declarative spec.
Early-access partners run pilot adequacy studies with our engineers.