Valencia shore power: what the public data says.

The question was how much maritime and GIS information can be had from public sources for every ship calling at Valencia: who calls, when they arrive and leave, and how long they stay. The sources fall into three tiers.

Each PCS record is one call. The arrival and departure flags distinguish estimates from actual times, and for completed calls both are actual. A check against the cruise schedule shows the stored times are UTC, and that the recorded arrival is roughly 1.3 hours before the scheduled berth time, so the arrival is at the port or pilot station rather than all-fast at the quay.

Across the 71 months the download returned between 435 and 582 calls per month for Valencia alone, with no failed months. After removing forecast calls, estimated times, harbour craft, and stays over 240 hours, 35,570 calls and 1.05 million ship-hours remain for the model.

3,412 distinct vessel names appear, 3,392 after merging renamed ships on their IMO number. Identity was resolved by bulk Wikidata lookup, the cruise forecast PDFs, and a rate-limited pass over a free vessel search for the ships with the most berth time. Coverage is measured in ship-hours because a handful of large container ships matter far more than hundreds of one-off callers.

Two thirds of ship-hours are tied to a confirmed IMO number and 64% to a registry gross tonnage. Ferries and cruise ships are essentially complete. The gap was in container ships: 942 vessels, 41% of container visits, had no size and fell to a default bin.

Container services run sister-size ships, so the model now takes the median GT of the sized vessels on the same liner service. Of 104 services with at least three sized ships, the middle one has a size spread (75th to 25th percentile) of only 1.14. This fills 887 vessels and leaves 313 container visits unsized, under 2%. The imputed sizes range from 8,000 to 211,000 GT, median 45,000. Each such visit is tagged in its evidence reference so it can be audited or replaced by registry data later. The effect on energy was small, a 5% reduction, because the IMO at-berth curve for container ships is flat: a 1,000 TEU feeder is rated at 370 kW and a 20,000 TEU ship at 1,400 kW. Size matters less than the kW-per-size coefficients themselves.

By terminal the medians are 34 h at CSP Iberian, 27 h at APM, 33 h at MSC, around 5 h at the Trasmed and Balearia ferry berths, and 12 h at Valencia Terminal Europa. The 2025 container jump, with slightly fewer calls, looks like congestion and waiting counted inside the stay rather than longer working time, so 2025 container energy is best read as an upper bound.

Every call becomes a visit with a connection window: the recorded port arrival plus one hour to reach the berth (1.3 h for cruise ships, from the schedule comparison), minus half an hour at each end to connect and disconnect. Each visit's power is then resolved through a five-level evidence waterfall: a consultant override, a measurement of the same ship at this port, a measurement of the same ship elsewhere, a category and size model, and finally installed generator capacity times a demand factor. Today every visit resolves at level four because no overrides or measurements exist yet; the tables for them are in place and tested.

The category models carry three power levels. Mid follows the IMO at-berth benchmarks as reproduced in TNO 2025 R12590, replaced by TNO's own measured values where it has them (cruise, RoPax, RoRo, large tankers, general cargo). Low is roughly two thirds of mid. High keeps the EMSA ranges and the EALING Valencia design values, and is close to what version 0.1.0 of this page called mid. Energy, within-call profile shape and peak power are resolved independently, so a supply limit can expose unmet demand instead of hiding it.

The full table has 36 rows across nine categories and is in the project as an editable register with source, scope note and an "is assumption" flag per row. Peak kW is a separate column: measured interval maxima where TNO gives them, otherwise 1.35 to 2 times the mid average, marked as an assumption. The biggest judgement call is cruise: the mid rests on three ships of one owner, probably in mild weather, and Valencia summer air-conditioning could push real loads back toward the IMO figure.

Container ships form a flat 16 MW base around the clock. Cruise adds up to 4 MW between 08:00 and 17:00. Ferries add up to 10 MW between 18:00 and 22:00 as the Balearic night boats load, and are the clearest daily feature. The port-wide average swings between 19 MW at night and 28 MW at 20:00.

Individual days are far lumpier than the average. Cruise calls arrive in blocks of two or three ships and, with the evening ferries, push the total toward 37 MW for a few hours, then disappear. A weekday with no cruise ship sits near 20 MW all day.

Scenarios run on two separate axes. The power axis (low, mid, high) says how much a connected ship draws. The adoption axis says which ships connect: only those with confirmed OPS today (none, since no installation evidence has been entered yet), the FuelEU 2030 obligation set, a phased ramp, or every merchant visit.

The power axis moves the answer far more than the adoption axis: high is 2.9 times mid, while going from the FuelEU set to every ship adds only 10%. Calibrating kW per ship against real meters is therefore worth more than refining who connects.

Energy grew from 135 GWh in 2021 to 193 GWh in 2025, almost entirely from longer container stays rather than more calls. Cruise energy tripled between 2021 and 2022 as the season returned and has held at about 12 GWh a year since. Ferries are a steady 17 to 26 GWh.

The table shows, for a given contracted-power cap, how much energy above the cap a battery would have had to supply in 2025 under the mid, full-adoption case, assuming perfect foresight and full recharge whenever demand is below the cap.

Three conclusions follow, all conditional on the mid power assumptions.

• Shore power at Valencia is a base-load problem first. Container ships alone hold the floor near 16 MW day and night. A cap at or below the 22 MW average would be exceeded for hundreds of consecutive hours, which no battery can carry.
• Storage starts to make sense against caps between the 90th and 99th percentile, 30 to 40 MW. At 35 MW the cap is exceeded on 36 days, at most 10 hours in a row, and the worst day needs 71 MWh. At 40 MW it is 6 days, 6 hours and 28 MWh. The 45 MWh battery Steven just delivered for one ferry is the right order of magnitude for the whole port at a 35 to 40 MW cap.
• The two load shapes a battery could usefully shave are the ferry evening block and cruise days. Both are scheduled months ahead, which favours contract design and day-ahead scheduling over oversized storage.

Under the high power case none of this holds: the average is 64 MW and 60 MW is exceeded for 5,113 hours. Until Valencia meters its first connections, the contract decision should be framed as a range, with the mid case as the planning basis and the high case as the stress test.

• No berth numbers. PCS names the terminal operator, not the quay, so per-connection-point peaks cannot be derived yet.
• Arrival precedes all-fast. The recorded arrival is about 1.3 h before the scheduled berth time for cruise ships; a one-hour offset (1.3 h for cruise) is now applied, but container stays in 2025 may still include anchorage waiting.
• Literature power values, thin measured base. TNO's measured samples are a few ships per category at unnamed northern ports, read from charts. Measured container loads vary by up to 300% with reefers and season. The low-to-high spread (125 to 563 GWh) is the honest range until Valencia meters its first OPS connections.
• Adoption is a scenario, not a forecast. No vessel has installation evidence yet, so "current confirmed" is zero and the FuelEU and full-adoption cases bracket the outcome. The register has the fields to hold installation status as it is learned.
• Timezone. Stored times were shown to be UTC by comparison with the cruise schedule across the daylight-saving change, and converted to local time.
• Identity gaps. A third of ship-hours still lack a confirmed IMO. Their size now comes from the liner service rather than a default bin; 452 of the 887 imputed vessels sit on services that mix ship sizes and are flagged low confidence.

The pilot is organised as the registers and workflow in the Bridge Power development brief, so that better evidence replaces assumptions row by row without rebuilding anything. Every hourly value can be traced back to the visits and the evidence level that produced it.

A bilingual data-request letter to the Autoridad Portuaria de València has been drafted. It asks for six things, in order of value to the model:

1. Call-level data with IMO, GT, berth number and the phase times the port already records (pilot on board, all fast, last line), plus anchorage waiting and whether OPS was used.
2. The underlying tables of the "Average call times by phase" report.
3. A GIS layer of quays, berths and terminal concessions with identifiers matching the call data.
4. Electrical data: metered OPS energy per connection once the MSC quay installation runs, the port's hourly demand at the existing substation, rated power per planned connection point, and the EALINGWorks demand study datasets.
5. The structure of current and planned supply contracts, as far as shareable.
6. Access to the planned-calls web service or a bulk export of the historic search.

In return it offers the model, the hourly curves per terminal and the battery scenarios, under a data-sharing agreement. With items 1 and 4 the model moves from literature values to calibrated, berth-level curves; without them it remains a well-grounded envelope.

Two decisions are open with Steven and Gridcog: the export interval and format (hourly or 15-minute, kW or kWh, totals or components), and the pilot period (calendar 2025 or the full 2021 to 2026 history).