Resource adequacy and capacity-mechanism modelling.

Loss-of-load expectation (LOLE), effective load-carrying capability (ELCC), and capacity-mechanism clearing on the same 27-zone European fleet model that powers CEGridSight. CESentinel is what you reach for when the question is can the system actually keep the lights on, and what's the capacity any new asset will be derated to — successor in spirit to CANOPI-class adequacy studies, run on demand against today's fleet rather than last year's snapshot.

Get on the early-access list CEAtlas Expert pricing

Why this exists

ELCC numbers shouldn't take six months and live in a PDF.

Effective load-carrying capability is the number that decides what a battery, wind farm, or hydrogen-fired peaker is worth in a capacity market. Today, getting one means commissioning a study from a consultancy on a 6-month cycle against last year's fleet. CESentinel turns ELCC, LOLE, and capacity-mechanism clearing into reproducible runs on the live fleet model — auditable, replayable, and updatable when the fleet changes.

What it does

Three pillars.

LOLE & LOLP

Loss-of-load expectation in hours/year and loss-of-load probability hour-by-hour, computed by Monte-Carlo on weather, demand, and forced-outage state — over the same 27-zone European topology and 6-ISO US topology used by CEGridSight.

ELCC by class & per asset

Effective load-carrying capability for any technology (wind, solar, BESS-by-duration, demand response, interconnector) — computed both as class average and per individual asset, with the marginal-vs-average distinction made explicit.

Capacity-mechanism clearing

GB Capacity Market, French CRM, Italian CapMech and PJM-style RPM clearing simulated as a sealed-bid auction over the derated fleet. Outputs clearing price, awarded capacity, and per-asset award status.

Methodology

For the quants in the room.

Plain-English first; this section for anyone vetting the maths. Skip if you're not running the study yourself.

Sampling

Sequential Monte-Carlo on hourly state.

Each replication: 35 historical weather years sampled with replacement, hourly demand and renewable capacity factor draws conditioned on temperature/wind/irradiance, forced-outage state evolved as a two-state Markov chain per unit. 2,000 replications by default; convergence checked against LOLE standard error.

ELCC

Equivalent firm-capacity reliability match.

For each candidate asset: solve for the firm-capacity injection that produces the same LOLE as the asset itself, holding the rest of the fleet fixed. Marginal ELCC computed at the current fleet point; class-average ELCC computed by stripping all assets of the class and re-solving. Both numbers reported because both are misleading if used in isolation.

Storage adequacy

Inventory-coupled, not steady-state.

BESS adequacy contribution evaluated with state-of-charge carried hour-by-hour through each Monte-Carlo replication, not derated against a steady-state availability factor. Duration matters — a 2-hour battery and a 4-hour battery produce different ELCC numbers, and CESentinel reports both honestly.

Auction

Capacity-mechanism MIP with derating curves.

Sealed-bid clearing implemented as a MIP with administratively-set demand curves, derating factors, and minimum/maximum capacity-obligation rules per market. GB CM, French CRM, Italian CapMech, and PJM RPM templates included; user-defined market rules supported via a YAML mechanism spec.

Research

Working papers.

CESentinel grew out of academic work on capacity-mechanism design and storage adequacy. The methodology papers behind the model are open-access and will be linked here as they're submitted; early-access partners get the drafts ahead of public release.

In preparation

Marginal vs average ELCC under high storage penetration.

Why the gap between marginal and average ELCC widens as storage share grows, and why capacity-mechanism design needs to clear on the right one. Working paper, target submission Q4 2026.