Portugal’s October 2026 import exposure: reconcile rounded REN statistics#

Primary data provenance (Portuguese transmission-system operator REN):

  • 2 October 2026, first nine months: demand 40.5 TWh (rounded), renewable electricity 64%, natural gas 14%, net imports 22% of consumption.

  • 3 November 2026, first ten months: renewables 62%, natural gas 14%, net imports 24% of consumption; October alone renewables 47%, all non-renewables 13%, net imports 40%. October hydro, wind and solar PV productivity indices were 0.58, 0.61 and 1.05, respectively (historical reference 1).

These are actual published rounded percentages of electricity consumption, not percentages of domestic generation, hourly records or measured plant performance. The 13% solar contribution belongs to the first ten months, not to October alone. Gas in cumulative reports must not automatically be equated to all non-renewable generation in October.

Decision problem and method#

Before a PV developer, wind investor or BESS owner contracts firm electricity or sizes flexibility, they need to distinguish genuine weather exposure from rounding/reporting effects. Using preceding demand P, subsequent month demand M, preceding percentage p and new cumulative percentage c, the implied month percentage is

monthly_pct = c + (P / M) * (c - p)

energy_methods.mix_rounding.infer_incremental_month_share propagates integer reporting precision (plus/minus 0.5 percentage points) and REN’s January-September 40.5 TWh demand rounded to one decimal (plus/minus 0.05 TWh) by evaluating all parameter corners. Output is an envelope from rounding, not a statistical confidence interval. Marginal extremes across different generation categories need not be jointly achievable.

Explicit synthetic scenario: October demand is taken as 4.50 TWh for this reproducible sensitivity calculation. REN’s cited press release does not state this monthly demand total; it is NOT a measured input and no resulting conditional October TWh volume is reported as fact.

For this scenario the central derived October shares are 44% renewables and 42% net imports. Rounding envelopes are approximately 34-54% and 32-52%, respectively. The independently reported October shares 47% and 40% lie within these conservative intervals: rounded cumulative figures do not contradict REN’s October press figures. The reported October aggregate non-renewable value 13% is not used to infer monthly gas output, since those are different reporting labels.

The runnable Sphinx-Gallery example plots the actual reported cumulative and monthly categories, rounding envelopes from the illustrative demand scenario, and REN’s reported hydro/wind/solar productivity indices. It performs no inference about the renewable/carbon content of imports.

Reproduce#

uv sync --locked --all-packages
uv run --no-sync python examples/analysis/plot_portugal_mix_rounding.py
uv run --no-sync pytest methods/tests/test_mix_rounding.py

It is listed under Analysis -> Electricity mix and imports. No keys, paid APIs, live downloads or unlicensed copied datasets are required. Sources are directly attributed. The input percentages are frozen extracts of the public dated primary statements, not a private data connector.

Limitations and next step#

No time-matched hourly generation, imports, market prices, curtailment, system losses, weather model, storage state-of-charge, degradation, arbitrage optimization or emissions are derived. A productivity index cannot be used to causally attribute import exposure to a particular renewable technology. The input consumption assumption is NOT verified. Do not infer a reliable BESS MW/MWh rating or wind/PV PPA revenue from these statistics alone.

A later study should retrieve official REN Data Hub/ENTSO-E hourly demand, generation by source, interchange and day-ahead prices (after checking licensing, revisions and time zones), reconcile the energy balance and only then run a physical/economic flexibility optimization with uncertainty.