How a combined cycle plant navigates Spain's electricity markets

At 18:00 on August 22, 2024, Algeciras 3 scheduled 700 MWh in the day-ahead market. By delivery, it was producing 780. The extra megawatt-hours came from replacement reserves (+60), real-time constraints (+50), and intraday corrections — offset slightly by tertiary balancing (-30).

This is what flexibility looks like in the I90 settlement data. A combined cycle plant doesn’t just generate — it adjusts across every market program, responding to system needs in real time. This article follows Algeciras 3 (unit code ALG3), a 750 MW plant owned by Repsol in the south of Spain, through the entire market cascade over a full year.

Where ALG3 participates

Over the past year, ALG3’s absolute energy volume splits across every market program — but not evenly.

The treemap makes the proportions immediate. PDVP — technical constraints imposed by REE — accounts for more volume than the day-ahead market itself. ALG3 doesn’t earn primarily by scheduling energy; it earns by being the plant REE calls when a constrained zone needs more power.

After PDVP, the next biggest programs are PHF1 (intraday corrections) and RTR (real-time constraints). But the waterfall only shows net values — what happens when we split upward and downward?

The butterfly confirms what the waterfall suggests: ALG3 is overwhelmingly upward. PDVP alone adds 310 GWh upward against just 14 GWh downward. Even in balancing programs (RR, BT), where most plants provide downward regulation, ALG3 is net positive. This is a plant that the system calls on to produce more, not less.

A day in the life: 24 hours of adjustments

Each row in the heatmap adds one program layer to ALG3’s position on August 22 — from the day-ahead schedule at the top to the final dispatch at the bottom.

The heatmap reads top to bottom. The first row — PDBF — is the day-ahead schedule: 0.7 GWh during night hours, dropping to 0.4 GWh at midday when solar pushes prices down. By PHF1, the position has nearly doubled overnight as intraday corrections add volume. Each subsequent row adds another layer of adjustment.

By the bottom row (BS), the final dispatch has taken shape. Night hours peak around 1.4 GWh — twice the day-ahead schedule. Afternoon hours settle around 0.5 GWh. The difference between the top and bottom rows is the flexibility premium: everything the plant earns beyond its initial auction bid.

Notice the evening ramp at 20:00-22:00 — PDBF jumps back to 0.7, then RR and RTR push the position above 1.4 GWh. This is when solar drops off and the system needs gas to fill the gap.

The PHF1 pattern

How much does ALG3 add on top of its day-ahead schedule through intraday corrections? The ratio tells the story month by month.

ALG3 is a summer plant. The empty months on either side of the lollipops tell the story — from October through May, the plant barely operates in the day-ahead market. When it does run, PHF1 corrections add substantial volume on top of PDBF.

September stands out at +137% — ALG3 more than doubled its day-ahead schedule through intraday. This isn’t a correction; it’s a strategy. Gas turbines schedule conservatively in the day-ahead auction, then add volume in intraday sessions where they have better visibility on system conditions and prices.

Is this pattern unique to ALG3, or do other combined cycle plants do the same? Compare with San Roque 2, a plant that operates year-round.

SROQ2 (Endesa) shows consistent PHF1 additions every month — modest but persistent, ranging from 3% to 20%. ALG3’s strategy is more extreme: inactive most of the year, then aggressive corrections when it does operate. Two plants, same technology, different commercial strategies — both visible in the I90 data.

What happens when you compare not just two gas plants, but five different technologies? Nuclear, hydro, solar, wind, and gas each leave a completely different fingerprint. See How Different Power Plants Trade in Spain’s Electricity Markets for the full comparison.

Monthly patterns

Zooming out to the full year, each cell shows ALG3’s net energy per program per month — excluding PDBF to reveal the adjustment layer.

The monthly heatmap confirms the seasonal concentration. Activity peaks sharply in July-September across every program. PDVP dominates — September alone reaches 130 GWh of upward redispatch, the largest single month-program combination.

The absence of data in winter months is itself the insight. When nuclear and renewables cover demand, there’s no need for gas flexibility. ALG3 steps aside. But when summer heat drives air conditioning load and solar variability creates system stress, gas becomes essential.

ALG3 among its peers

Spain operates over 40 combined cycle plants. The treemap shows where ALG3 sits in the fleet — sized by day-ahead volume, grouped by company.

ALG3 (highlighted in red) sits mid-fleet at 244 GWh PDBF — the 16th largest among 19 combined cycle plants with significant output. The treemap reveals the competitive landscape: Endesa leads with San Roque and Besós, TotalEnergies operates the two Castejón groups, and Repsol’s fleet includes Escatrón alongside ALG3.

The fleet spans six major companies and nearly 9 TWh of combined day-ahead volume. Every one of these plants follows a similar pattern — schedule in the auction, adjust through intraday and constraints, respond to TSO signals in real time. The magnitudes differ, but the business model is the same: sell flexibility.

What the cascade reveals

Tracing one combined cycle plant through the I90 settlement data reveals a business model hidden behind simple generation statistics. ALG3 earns more from PDVP constraints resolution than from its day-ahead schedule. It adds volume through intraday, responds to real-time TSO signals, and operates seasonally — active in summer when the system needs flexibility, stepping aside in winter when renewables and nuclear suffice.

This pattern isn’t unique to ALG3. Across the combined cycle fleet in Spain, flexibility is the product. The I90 data shows not just what each plant produced, but how — and that story is far more nuanced than any headline about gas generation suggests.

---

For the system-wide view of each market program, see Inside Spain’s Electricity Market Programs. For company-level portfolio analysis, see Who Generates What in Spain?.

All data queried via the datons Python library. The same analysis works for any of the 3,600+ programming units in the system — generation, demand, and interconnections.