Results
Five findings shape the diagnosis.
Source-reported facts, reconstructed quantities and screening results are labelled separately so a modeled sensitivity is not presented as an observation.
RQ1 · Historical balance
Electricity demand has risen materially.
Official broad consumption
Approximately +30% over four financial years.
External electricity remains structurally important.
Kerala's system balance depends on substantial electricity arriving through the wider Indian grid. That interconnection can be economically valuable; the engineering concern is dependable availability at stressed hours.
RQ2 · Chronology & weather
The difficult hour cannot be inferred from annual energy alone.
Weather adds useful—but modest—information beyond clock time.
A sparse static shape already reconstructs reported intraday extrema well. Adding ERA5 weather improves the held-out error further.
Daily energy is known and conserved on held-out days. These are intraday-shape errors, not day-ahead forecast errors.
Held-out Jan–Mar extrema RMSE
ERA5-sensitive MAE: 124.9 MW · correlation: 0.968.
FY2024–25 proxy maximum.
5,904 MW ≠ 5,797 MW
5,904 MW is the generation-RA planning/calibration reference used by the proxy. 5,797 MW is the separately reported recorded FY2024–25 peak in the newer transmission-RA study.
RQ3 · Hydropower timing
The same hydro energy can have very different adequacy value.
Full-transfer reference case
Same hydro energy; different timing. Modeled unserved energy falls 96.87% in this one case.
Benefit of same-day hydro flexibility
Percentage reduction relative to frozen-hydro counterpart at each transfer ceiling.
| Transfer ceiling | 1-day window | 3-day | 15-day | 30-day |
|---|---|---|---|---|
| 4,455 MW | 11.751 GWh | 7.422 GWh | 0.160 GWh | 0.160 GWh |
| 3,564 MW | 1,187.854 GWh | 1,167.909 GWh | 1,060.346 GWh | 965.093 GWh |
| 2,673 MW | 5,571.388 GWh | 5,542.035 GWh | 5,468.566 GWh | 5,454.820 GWh |
RQ4 · Interstate transfer
Transfer availability is one of the strongest adequacy levers in the screen.
Current ATC reference; TTC 4,575 MW.
80% of current ATC reference.
60% of current ATC reference.
Reference-demand modeled unserved-energy fraction
Scenario outcomes, not forecasts. Directional sensitivity is the result: deep transfer constraints dominate many otherwise useful local flexibility measures.
RQ5 · Internal deliverability
The public network screen identifies a persistent Shoranur-area signal.
Strongest corridor cluster
Robust 110-kV and 220-kV screening groups meet around Shornur/Shoranur, including directions toward Koppam, Perinthalmanna, Melattur, Areacode and Elankur.
Location survives the QA correction.
The historical model included a 2023 100 MVA transformer element. Current CEA reports 200+100 MVA existing transformation and still identifies Shoranur as constrained with reinforcement planned.
The geographic signal is stronger than the exact modeled loading magnitude. This is not operator congestion validation or an AC/N-1 result.
Synthesis
No single statewide annual-energy number explains the stress.
When the difficult hour arrives.
How flexible in-state energy can respond.
How much dependable supply can enter.
Whether the network can move power to where it is needed.