What question did this study set out to answer?

This research aims to validate the inZORi FrozenElite strategy using real ENTSO-E data across three European countries.

March 1, 2026Open Access

inZORi-PFΔ Phase 5: ENTSO-E Real Load Validation — Romania, Germany, France (2024) on IEEE 118-bus Power Flow

Key Points

This research aims to validate the inZORi FrozenElite strategy using real ENTSO-E data across three European countries.
Tested FrozenElite on live ENTSO-E data in Romania, Germany, and France.
Conducted 48 computational jobs across 12 cores with 4 variants and 4 seeds.
Evaluated convergence performance against the NR standard.
FrozenElite achieved 75–92% convergence versus 4.5–15% for NR standard, indicating a 6× to 17× improvement.
In Romania, FrozenElite achieved 75.3% convergence (advantage of ×16.7 over NR).
In Germany, FrozenElite reached 82.5% convergence (advantage of ×7.2 over NR).
In France, the method achieved 91.7% convergence (advantage of ×6.1 over NR).
Performance varied by grid structure, better in nuclear-stable France than in renewables-heavy Germany.

Abstract

Phase 5 of the inZORi Power Flow Delta (PFΔ) research series is the first validation of the inZORi FrozenElite strategy on **live ENTSO-E Transparency Platform data** (2024). Three structurally distinct European grids are tested: Romania (thermal/hydro-dominant), Germany (renewables-heavy), and France (nuclear-dominant). 4 variants × 4 seeds × 3 countries = 48 jobs on 12 cores. FrozenElite achieves 75–92% S3 convergence vs 4.5–15% for NR standard — a **6× to 17× improvement** on real European consumption data. **Key Results (4 seeds × 3 countries):** - Romania: inZORi FrozenElite 75.3% vs NR 4.5% (×16.7 advantage) - Germany: FrozenElite 82.5% vs NR 11.4% (×7.2 advantage) - France: FrozenElite 91.7% vs NR 15.1% (×6.1 advantage) - Grid structure explains performance: nuclear-stable FR > renewables DE > thermal RO - Adaptive selector ≈ best static genome (G01) on real data — G01 generalizes well - Real ENTSO-E profiles less extreme than synthetic — validates under realistic conditions **Limitations:** Results use IEEE 118-bus academic topology, not actual TSO network topologies. ENTSO-E data is hourly national aggregate, not per-node SCADA. **Related publications:** - Phase 1: https://doi.org/10.5281/zenodo.18716837 - Phase 2: https://doi.org/10.5281/zenodo.18717007 - Phase 3: https://doi.org/10.5281/zenodo.18735120 - Phase 4: https://doi.org/10.5281/zenodo.18735099

inZORi-PFΔ Phase 5: ENTSO-E Real Load Validation — Romania, Germany, France (2024) on IEEE 118-bus Power Flow

Key Points

Abstract

Cite This Study