AIS DARKMARITIME INTELLIGENCE FUSION

CASE REF AISD-GLOBALSEAWAY-PART5-2026TRACK COVERAGE

§1 §2 §3 §4 §5 §6 §7 §8

2026-08-15

The managers converged. The routes didn't.

// SHADOW FLEET //

STATUS — ACTIVE — BACKGROUND-RATE COMPARISON RETIRED (NO SHARED PORT SET); THREE SEPARATE TRADE THEATERS INDEPENDENTLY RECONSTRUCTED INSTEAD

7 MIN READ — CAPTURED 15 AUG 2026

A short video walks through the reconstruction. Part 3: Eleven Designated. Part 4: The Shell Layer Dissolved.

Part 4 closed one thread and left another open: whether this network's port-state record differs from an unrelated tanker cohort calling the same ports. That comparison assumes a shared set of ports to compare against. It doesn't hold. A targeted AIS backfill and independent port-call reconstruction on DENVER, MIMOZA, MAGNOLIYA, and CETUS/MERIDIAN — the same four hulls Part 4 traced to a converging set of managers — finds these hulls operating in three separate trade theaters, not one shared route a background-rate comparison could be built against.

Three hulls, three oceans

DENVER (IMO 9382712) shows a short Mediterranean sequence in May 2026: Skhira, Tunisia (4 May), Rades, Tunisia (7 May), then Ambarli, Turkey (16 May) — Tunisia twice, then east to Turkey, inside a two-week window. No Black Sea or Baltic Russian port appears in this hull's own reconstructed record at all.

MIMOZA (IMO 9323314) returns to the same port three times inside three months: Samsun, Turkey, on 13 April, 7 May, and 19 June 2026. The first two stays run roughly a day and a half each. The third, starting 19 June, runs at least 55 days — the reconstructed dwell extends to the edge of the backfilled window (14 August), so whether the vessel has since departed Samsun is not established here.

MAGNOLIYA (IMO 9293997) is the outlier, and the sharpest fact in this piece: 14 of its 19 reconstructed port calls between late March and mid-August 2026 are at Slavyanka, Russia — a Pacific-coast port near Vladivostok, not anywhere near the Black Sea or Baltic theater every other piece in this series has documented. The remaining five split across Nakhodka and Vladivostok (Russia, same coast) and Longkou Gang, Zhoushan (China), and Busan (South Korea). One Slavyanka stay, beginning 8 July, runs roughly eight days. This hull is not working the same corridor as the rest of this network — it is running a Russian Far East-to-China/Korea route, under the same St Petersburg manager (Ocean Solutions LLC, per Part 4) that holds hulls Part 1 through 3 place squarely in the Black Sea and Baltic.

CETUS/MERIDIAN — excluded, not included with a caveat

CETUS/MERIDIAN's own MMSI changed when the hull renamed and reflagged to Russia (Part 4's own central finding) — Equasis dates that change to on or after 30 June 2026. The new MMSI's backfilled position history returns three port calls at Primorsk, Russia, dated 15 March, 6 June, and 8 June 2026 — all of them before the earliest date Equasis records for this MMSI actually belonging to this hull. That means this data cannot be safely attributed to CETUS/MERIDIAN; MMSIs get reassigned, and a position history predating a hull's own confirmed reflag is a documented risk in this project's own prior work, not a new caveat invented for this piece. Rather than present a fourth theater on data that might belong to a different vessel entirely, this piece excludes CETUS/MERIDIAN's port-call pattern outright. It is not stated here, hedged or otherwise.

What this doesn't establish

Three trade theaters across three hulls, under a converging set of managers, is a real, dated, independently reconstructed fact. It is not, on its own, evidence of coordination, a single fleet-wide sanctions-evasion route, or a rate comparison against anything. Each hull's own theater could reflect nothing more than which charterer or cargo owner it happened to be working for at the time — the same shape ordinary commercial tramping takes. What it does establish is narrower and more useful: the premise Part 4 carried forward from Part 2 — that a background-rate comparison could be built against "an unrelated tanker cohort calling the same ports" — doesn't fit these four hulls, because they don't share a port set to compare against in the first place.

Evidence and confidence

Claim — evidence basis	Grade
DENVER (IMO 9382712): three reconstructed port calls, all Mediterranean/North Africa -- Skhira, Tunisia (4 May 2026), Rades, Tunisia (7 May 2026), Ambarli, Turkey (16 May 2026). No Black Sea or Baltic Russian call in the reconstructed record.	A
MIMOZA (IMO 9323314): three reconstructed port calls, all at Samsun, Turkey -- 13 Apr, 7 May, 19 Jun 2026. The third dwell extends at least 55 days, to the edge of the backfilled window (14 Aug 2026).	A
MAGNOLIYA (IMO 9293997): 19 reconstructed port calls, 14 of them at Slavyanka, Russia (Pacific coast), spanning 29 Mar-12 Aug 2026. The remaining 5 split across Nakhodka and Vladivostok, Russia, and Longkou Gang, Zhoushan (China) and Busan (South Korea).	A
MAGNOLIYA's Slavyanka pattern includes several same-day, short-duration entries likely reflecting one continuous stay fragmented by signal gaps, not distinct visits -- the true visit count is likely lower than 14, though Slavyanka remains the dominant pattern.	B
CETUS/MERIDIAN's new-MMSI position history (273273370) returns three Primorsk, Russia calls (15 Mar, 6 Jun, 8 Jun 2026), all dated before Equasis's own confirmed reflag date (on or after 30 Jun 2026) -- excluded from this piece's claims as unsafe to attribute to this hull.	B
MIMOZA's and MAGNOLIYA's MMSIs were not previously on file in this project's own corpus -- both resolved directly from Equasis's own management-detail pages before this piece's backfill was run.	A
No shared port set exists across the four hulls Part 4 traced to a converging manager set -- the background-rate comparison as framed in Part 2 cannot be built against these hulls.	A

What would prove this wrong

This reconstruction covers a roughly six-month backfilled window (mid-February to mid-August 2026), not each hull's full service history. A different window could show different theaters, or overlap where this one shows none.
Port-call detection depends on proximity to a known port coordinate (World Port Index) and a minimum dwell time. A vessel anchored well outside a charted port's own coordinate — the case for at least one hull checked during this same investigation — would not register here even if functionally present at that port. Absence of a detected call is not proof of absence at sea.
MAGNOLIYA's Slavyanka pattern includes several same-day, short-duration entries that likely reflect one continuous stay fragmented by signal gaps, not several distinct visits in one day. The 14-of-19 count is not adjusted for this; the true number of distinct visits is very likely lower than 14, though still the dominant pattern in this hull's record.
CETUS/MERIDIAN's exclusion is a precaution, not a finding that its position data is wrong. It may well belong to the hull; this piece simply cannot confirm that from the dates alone, and states the gap rather than guessing.
No claim is made about why these three hulls are in three different theaters — commercial assignment, charterer instruction, and coincidence are all consistent with the pattern as stated.

Method

Sources for this piece: a targeted historical AIS backfill (Datalastic, mid-February to mid-August 2026) for each hull's own current MMSI, newly linked to its IMO number in this project's own identity table before the backfill was run; this project's own port-call detector, run independently against the backfilled positions — proximity to a World Port Index coordinate plus a minimum dwell, the same detector this project uses throughout its case files; and Equasis's own management-change and flag-change dates, used only to bound which portion of CETUS/MERIDIAN's position history is safe to attribute to the hull.

MIMOZA's and MAGNOLIYA's MMSIs were not on file in this project's own corpus before this piece — both were resolved directly from Equasis's own management-detail pages, the same primary source used throughout this series, not inferred or guessed.

Part 6

The background-rate comparison, as originally framed in Part 2, is retired by this piece rather than carried forward — there is no shared port set left to compare against. What remains open: whether MAGNOLIYA's Pacific pattern recurs across other hulls under the same St Petersburg managers, and whether MIMOZA's 55-day Samsun dwell resolves to a departure, a lay-up, or something else once a later backfill window covers it. Neither is started here.

Use

This reconstruction and video are free to use, but you must credit AIS Dark.

Want the same reconstruction run on a fleet or counterparty you are assessing? Get in touch.

As of 15 August 2026.

Want the full case file on DENVER, MIMOZA, MAGNOLIYA — three hulls, three trade theaters, one shared shell layer?

One email, sent once. No list, no follow-up spam.

DENVER, MIMOZA, MAGNOLIYA, CETUS — WHERE THE SIX-COMPANY HANDOFF ACTUALLY WENT

Part 2 left open whether six shell companies installed across DENVER, MIMOZA, MAGNOLIYA, and CETUS in early 2024 managed tonnage beyond those four hulls. All six have since vanished from every one of those hulls' own registry pages. Where the tonnage went is not new ground — two hulls now sit with the same Mumbai company that led to Part 3's own discovery, one sits fully inside the same St Petersburg building, and the fourth, already carrying its own EU designation under its own name, has since been renamed.

MARGINALIA — ANALYST NOTES

LAST RECORDED STATUS

DENVER, MIMOZA, MAGNOLIYA — three hulls, three trade theaters, one shared shell layer

Active — background-rate comparison retired (no shared port set); three separate trade theaters independently reconstructed instead

EVIDENCE GRADES

Data table

Grade	Meaning
A	PRIMARY RECORD
B	CORROBORATED
C	REPORTED, NOT LOAD-BEARING

OPERATION CLOCK

Data table

Date	Label	Grade	Signal
2026-03-29	MAGNOLIYA's Slavyanka, Russia pattern begins (first of 14 reconstructed calls at this port)	A	transmit
2026-04-13	MIMOZA's first Samsun, Turkey call (of three)	A	transmit
2026-05-04	DENVER's Mediterranean sequence begins — Skhira, Tunisia	A	transmit
2026-06-19	MIMOZA's third Samsun call begins — dwell extends at least 55 days to the edge of the backfilled window	A	hold
2026-07-08	MAGNOLIYA's longest single Slavyanka stay begins — roughly 8 days	A	hold
2026-08-12	MAGNOLIYA's most recent reconstructed Slavyanka call, within the backfilled window	A	transmit

EXCLUDED FROM THE FINDING

Data table

Date	Reason

ANALYST NOTE — BACKGROUND-RATE COMPARISON RETIRED, NOT RESOLVED

Part 2 and Part 4 carried forward a plan to compare this network's port-state record against an unrelated cohort calling the same ports. That plan assumed a shared port set. It doesn't exist -- DENVER, MIMOZA, and MAGNOLIYA operate in three separate trade theaters. This piece retires that comparison rather than completing it.

ANALYST NOTE — MAGNOLIYA'S PACIFIC PATTERN IS THE SHARPEST NEW FACT IN THIS PIECE

14 of 19 reconstructed calls at Slavyanka, Russia -- a Pacific-coast port -- under the same St Petersburg manager (Ocean Solutions LLC, per Part 4) that Parts 1-3 place squarely in the Black Sea/Baltic theater. This is a genuinely new geographic fact, not previously documented anywhere in this series.

ANALYST NOTE — CETUS/MERIDIAN EXCLUDED AS A PRECAUTION, NOT A FINDING

All three of this hull's reconstructed port calls under its new MMSI predate Equasis's own confirmed reflag date. MMSI reassignment creating false continuity is a documented risk this project has flagged before -- the data may well belong to the hull, but this piece cannot confirm that from the dates alone, and states the gap rather than guessing.

ANALYST NOTE — DETECTION METHOD HAS REAL LIMITS, STATED PLAINLY

Port-call detection requires proximity to a World Port Index coordinate plus a minimum dwell. A vessel anchored well outside a charted port's own coordinate would not register -- confirmed directly during this same investigation on a different hull. Absence of a detected call in this piece is not proof of absence at sea.