What percentage of Russian seaborne crude oil export loading capacity will be offline at end of June 2026 due to Ukrainian strikes (vs the H2 2025 baseline)?

TL;DR

I forecast 24% resolved offline capacity for 1–28 June 2026, because western-port loadings had mostly recovered by mid-May (Reuters/LSEG via Baird Maritime, 18 May 2026) while the fallback rule would score a normal-looking 3.6–3.9 mb/d Russian seaborne crude month as roughly 22–28% offline against the 5.0 mb/d denominator (Bloomberg summary, 27 May 2026; S&P Global, 5 May 2026).

Context

The March shock was real. Reuters, republished on 25 March, said Ukrainian attacks had suspended loadings at Primorsk and Ust-Luga, left Novorossiysk behind schedule, and put about 2.0 mb/d, or 40%, of Russian oil export capacity offline (Reuters via The Moscow Times, 25 March 2026). KSE’s Kpler-based assessment, using data accessed 7 April, showed Ust-Luga crude falling from 725 kb/d in the week of 16–22 March to 101–155 kb/d in the next two weeks, while Primorsk crude stayed at 780–848 kb/d (KSE Institute, 9 April 2026).

By late May, the system was not stuck at the March trough. Ust-Luga resumed exports after a nearly two-week suspension, with Reuters/LSEG describing the Baltic port as capable of about 700 kb/d of crude loading (Reuters via Baird Maritime, 8 April 2026). The forecast problem is therefore a strike-versus-repair problem, plus a measurement problem: a direct Reuters/Bloomberg capacity stocktake would likely resolve lower than the fallback formula, while the fallback formula uses actual June seaborne crude exports divided by 5.0 mb/d.

Evidence

The historical backbone is fast damage, then fast partial repair. Carnegie estimated that Russian port shipments of oil and products fell from 5.2 mb/d during 1 January–24 March to 3.5 mb/d during 25 March–11 April, but also said Ust-Luga had resumed loading by 7 April and sent about half of pre-strike volumes over the following two weeks (Carnegie, 20 April 2026). Reuters’ 30 April stocktake said April loadings of Urals, Siberian Light and KEBCO through Primorsk, Ust-Luga and Novorossiysk were about 2.2 mb/d, broadly in line with revised March, despite the Ust-Luga suspension and a four-day Novorossiysk/Sheskharis halt (Reuters via Business Recorder, 30 April 2026).

The full comparable IEA/Kpler/Argus public series also argues against a persistent 40% June baseline. Units are mb/d of Russian seaborne crude exports. The 2022–2025 figures are annual averages, the 2026 figures are monthly averages, and the vintage is the IEA May Oil Market Report published 13 May 2026 (IEA May 2026 Oil Market Report PDF).

The freshest flow evidence points to moderate impairment, not a renewed collapse. Reuters/LSEG estimated that Primorsk, Ust-Luga and Novorossiysk handled 2.35–2.40 mb/d during 1–15 May, up from about 2.2 mb/d in April; the same 15-day window put Primorsk just above 1.1 mb/d, Ust-Luga near 0.6 mb/d, and Novorossiysk near 0.7 mb/d (Reuters/LSEG via Baird Maritime, 18 May 2026). Bloomberg tanker-tracking data put total Russian seaborne crude shipments at 3.66 mb/d in the four weeks to 24 May, little changed from 3.65 mb/d in the four weeks to 17 May and about 100 kb/d above 2025 levels (Bloomberg summary, 27 May 2026). S&P Global’s separate Commodities at Sea series put April seaborne crude exports higher, at 3.949 mb/d, a six-month high, which is a useful cross-check but not the resolver’s preferred source (S&P Global, 5 May 2026).

The upside risk is still real. Ukraine hit Primorsk again on 3 May; Reuters reported a fire, Ukrainian claims of significant terminal damage, and Russian claims that the fire was quickly extinguished with no spill (Reuters via Investing.com, 3 May 2026). Ukraine hit Sheskharis/Novorossiysk on 23 May, with AP reporting a fire at an oil terminal and two injuries (AP, 23 May 2026). But satellite-image reporting two days later said all three Sheskharis oil-transfer berths were operating with tankers moored (Mezha citing Reuters/Bloomberg, 26 May 2026). On 31 May, hours before the forecast window, Reuters reported Ukrainian strikes on the Lazarevo pumping station serving the Surgut-Gorky-Polotsk crude line, the Saratov refinery, and a Rostov fuel depot; the Lazarevo damage is a material weak signal, but it was not yet a quantified outage at Primorsk, Ust-Luga, Novorossiysk, or Kozmino (Reuters via MarketScreener, 31 May 2026).

For the denominator, I use 5.0 mb/d. That is the fixed fallback in the question, and it is close to KSE’s 2025 port-exposure sum for Primorsk, Ust-Luga, Novorossiysk, Nakhodka/Kozmino and Murmansk, which was 4.68 mb/d of seaborne oil exports rather than a clean crude-only nameplate capacity series (KSE Institute, 9 April 2026). My submitted distribution is a six-state mixture: repaired physical capacity if a direct stocktake appears, normal-flow fallback, direct moderate impairment, fallback/moderate impairment, a renewed March-like wave, and a severe multi-terminal tail. I weighted those states 18%, 36%, 14%, 18%, 11%, and 3%, respectively, with centers of 7.5%, 24.5%, 17%, 30%, 38%, and 55%. It has mean 24%, median 24%, P10 8%, and P90 37%. The bucket mass is about 14% below 10%, 40% from 10% to 25%, 38% from 25% to 40%, and 8% above 40%.

What's non-obvious

The word “capacity” points lower than the likely resolution value. If Reuters or Bloomberg publishes a clean end-June engineering stocktake, I would expect a high-single-digit to high-teen result unless there is a fresh coordinated June hit. But if no such article appears, the fallback is $R=100\times\max(0,1-E/5.0),$ where $E$ is June average Russian seaborne crude exports in mb/d. At $E=3.66$, the Bloomberg four-week average to 24 May, $R=26.8\%$; at $E=3.76$, the IEA April value, $R=24.8\%$; at $E=3.949$, the S&P April value, $R=21.0\%$ (Bloomberg summary, 27 May 2026; IEA May 2026 Oil Market Report PDF; S&P Global, 5 May 2026).

The other trap is treating refinery damage as the same thing as export-loading damage. The IEA said Russian crude exports rose in April partly because attacks on refineries cut domestic use and pushed more crude toward export channels (IEA, 13 May 2026). That can make seaborne crude exports look strong while the broader oil system is under stress. It is also why product-export and refinery-throughput headlines are weak evidence for this question unless they identify a crude-loading terminal, a tanker loading operation, or a dedicated feed pipeline.

Limitations

Public data still measure flows better than engineering capacity. Reuters, Bloomberg, IEA, S&P Global, KSE, and CREA use different cargo timing, commodity definitions, and revision practices; CREA’s Russia Fossil Tracker uses Kpler for seaborne flows and is likely to be the fallback source, but that does not make CREA’s export-volume number a direct measure of strike-disabled capacity (CREA methodology). The 31 May Lazarevo strike could matter if it constrains feedstock to western export routes, but I found no quantified throughput loss before the forecast window began (Reuters via MarketScreener, 31 May 2026). The largest source of forecast error is a single successful June strike on Primorsk, Ust-Luga, Novorossiysk/Sheskharis, Kozmino, or a dedicated feed line; the largest measurement error is whether resolvers get a direct Reuters/Bloomberg capacity stocktake or have to use the fallback export formula.