When does a non-ice-class tanker need an icebreaker? The shoulder-season decision

The season between the rules

Twice a year, seasonal ice seas put operators in a regulatory gap. In full winter the rules are unambiguous: traffic restrictions are in force, icebreakers convoy traffic, and a vessel below the declared ice-class and tonnage bar gets no icebreaker assistance – which in a hard winter amounts to not sailing at all. In summer the sea is open water and ordinary navigation applies. The problem is the transition on either side – weeks when the ice chart shows one to three tenths of drift ice, a scatter of floes on a mostly dark sea, and a berth window that was fixed months ago.

The regulatory frameworks were not written for that window. The Polar Code applies inside defined polar-water boundaries – and those boundaries exclude several seas that freeze every winter; the Baltic, with its mature ice regime, sits entirely outside it. National escort systems switch on when restrictions are declared and off when they lapse. Between the two states, the paperwork says open water while the radar occasionally says otherwise. The decision to sail a non-ice-class tanker or gas carrier through that picture, with or without support, lands on the operator.

Both easy answers are wrong. Mandating icebreaker support for every shoulder-season transit is the reflex of a risk register, not an operation: escort capacity is scarce exactly when everyone wants it, and holding a laden carrier for an escort that buys little on a near-clear day is real money spent on the wrong margin. Waving traffic through because the chart is “mostly water” is worse, because the consequence side of the ledger is not symmetric. Bow shell plating meeting an old-ice or ridged first-year fragment at transit speed is a penetration scenario – into the fore peak and ballast spaces first. On a laden tanker or gas carrier, the escalation path from a holed bow does not stop at a repair bill; the tail of that path reaches the cargo containment.

The two shoulders are also different problems, and a regime that treats them as one will be wrong twice a year. At freeze-up the hazard is the rate of change: new ice forms and consolidates fast, and the exposure is a transit window that closes while the vessel is inside it. That race has recent casualties: an early freeze-up in November 2021 caught roughly twenty vessels on the Northern Sea Route, and icebreaker relief took weeks. At break-up the ice is dying but the fragments are at their worst – ridged and rafted pieces released as the pack decays, old floes that keep their strength deep into the melt season, drifting through what the chart calls open water. Autumn is a race against the calendar; spring is a collision hazard wearing a reassuring concentration number.

What separates a defensible decision from a guess is not the answer. It is the method.

Three answers the world already has

The industry has solved adjacent versions of this problem three different ways, and an operator designing a shoulder-season regime should steal from all three.

The Baltic answer: graduated restrictions. The Finnish-Swedish winter navigation system combines ice class rules, declared traffic restrictions and icebreaker assistance into one machine. Restrictions are announced port by port as ice develops – a typical step requires at least Finnish-Swedish ice class II and 2,000 tonnes deadweight before a vessel is eligible for assistance, and the bar rises as winter deepens. The elegance is in the graduation: the system does not flip from “open” to “closed” but tightens in steps that track the observed ice, and in Finland the declared service level targets an average waiting time for assistance of around four hours. The lesson for an operator is that access rules can follow the ice week by week, rather than a calendar or a hunch.

The IMO answer: an index that includes your ship. POLARIS – the Polar Operational Limit Assessment Risk Indexing System, issued as guidance in MSC.1/Circ.1519 – scores an ice regime numerically. Each ice type present is assigned a risk value for the vessel’s ice class, weighted by its concentration in tenths, and summed into a Risk Index Outcome. Zero or better means the regime is acceptable for normal operation. For polar-class ships (PC1 to PC7) a negative outcome opens a band of elevated-risk operation with speed limitations – but a ship below PC7, and a ship with no ice class at all, falls in the column the tables reserve for ships not assigned an ice class, where a negative outcome puts the transit into operation subject to special consideration, and the circular’s voyage-planning advice for such regimes is to avoid them. For the vessels this article is about, the outcome is close to binary. Escorted and independent operation are assessed differently too – under escort the regime scored is the ice immediately ahead in the icebreaker’s track, and for voyage planning only the circular allows a ten-point credit to the outcome. And the tables explicitly include ships with no ice strengthening at all: the methodology was built to answer, among other things, our question.

Two honest caveats belong next to that endorsement, because both shape the gate an operator builds on top. Concentration weighting dilutes: at one or two tenths, even dangerous ice types leave a non-ice-class outcome positive – push the same old ice to three tenths and the outcome goes negative – so the index discriminates least in the very window this article is about, and a gate that leans on it must add its own trigger for old and deformed ice at low concentration. Glacial ice, meanwhile, sits outside the index altogether: the circular handles growlers and bergy bits separately, through caution and documented stand-off distances, not through the outcome. POLARIS itself insists it is decision support, not a go/no-go switch – which is why the operator’s gate, with named authority and documented inputs, is the missing layer rather than the index alone.

The Canadian answer: regime-by-regime arithmetic. Canada’s Arctic Ice Regime Shipping System, the TP 12259 standard that predates and informed POLARIS, works the same way with ice numerals: concentrations multiplied by per-ice-type multipliers for the vessel category, summed, and compared to zero. A negative ice numeral means the regime ahead is not for that ship, whatever the calendar says. AIRSS matters to this argument for one reason above all: it normalised the idea that the decision is made against the ice actually present, documented at the time of entry, not against a season or a zone alone.

The Russian answer: the per-voyage permit. The Northern Sea Route adds a fourth model – an administration issues each voyage a permit against the vessel’s ice class and the forecast conditions, specifying where the ship may sail independently and where only under escort, with vessels carrying no ice class admitted in the easier seasons. It answers the title question of this article voyage by voyage; what keeps it from transplanting is the machinery behind it – a single authority over a single route, with its own icebreaker fleet to back the permit’s promises.

None of these transplant directly to every seasonal sea. But together they establish the principle: the world’s mature ice regimes all quantify, and they all write the decision down.

Ice is a metocean parameter

The same operators who improvise shoulder-season ice decisions run textbook metocean discipline everywhere else. A North Sea lift does not proceed because the sky looks fine. It proceeds because the forecast significant wave height over the operation reference period sits below the operational limit derated by an alpha factor for forecast uncertainty – with a contingency margin added to the operation’s duration before the comparison is even made. That is the DNV-ST-N001 machinery, and offshore crews apply it as routine: limit, forecast, derating, window, documented decision.

Now list what the shoulder-season transit decision actually needs: a hazard field (ice type, concentration, floe size), observation of it (charts, satellite, aerial and shipborne reconnaissance), a forecast of its evolution (drift and deformation over the transit duration), a vessel-specific threshold (what this hull may meet at what speed), a derating for forecast uncertainty, and a written go/no-go against the lot. Point by point it is the same structure. The nearest prior art stops one step short: OCIMF’s guidance on operating large low- and no-ice-class tankers in seasonal first-year ice covers winterisation, preparation and the realities of ice passage, but it offers judgement, not a threshold engine. The missing piece is the gate itself – missing partly because the shoulder season is short, partly because “mostly open water” does not feel like an environmental limit, and partly because responsibility for the call sits ambiguously between charterer, master, terminal and operator.

The packaging is the fix. An operator can write an ice-transit gate the way it writes a weather gate: POLARIS or an equivalent index as the threshold engine, the ice service’s chart and the drift forecast as the input, a defined derating when the forecast is stale or the reconnaissance is thin – in ice terms: assume the next-more-severe ice type where identification is uncertain, add tenths for chart age, and widen the assumed regime boundary by the observed drift rate times the hours since the last pass – and named decision authority. The components are all standard practice somewhere; only the assembly is missing. There is even a precedent for crediting human capability in the derating itself – marine operations practice allows a better alpha factor when qualified environmental monitoring is on site. The ice equivalent of that provision is a person, and we come to them below.

What the bridge cannot see

Any gate built on detection has to be honest about the detection floor, because the hazard that defeats the whole scheme is the ice the bridge never saw.

The dangerous object in a shoulder-season sea is rarely the floe on the chart. It is glacial or deformed ice sitting low in the water: by WMO definition a growler shows less than a metre above the water across an extent of some twenty square metres – ice-patrol practice calls anything under five metres of length a growler – and a bergy bit carries one to five metres of sail. In any seaway those numbers translate to an intermittent radar target at best – a paint that appears on one sweep, drowns in wave clutter on the next, and never forms a track. Dark hours multiply the problem; so does snow in the air, which costs the lookout and the radar at the same time. Satellite support has its own floor: open-water SAR scenes resolve floes, not growlers, and revisit intervals leave gaps that matter when ice is drifting at a knot or more across a planned track. High-latitude operations add a further layer of degraded positioning and augmentation coverage – conditions we have mapped separately in why positioning degrades above 70°N – but the core detection problem is universal to every seasonal ice sea.

The operational consequences are not subtle. Safe speed in suspected ice is not a formality but the single strongest lever the bridge holds, because impact energy rises with the square of speed – and speed is also what berth windows quietly pressure crews to keep up. Routing is the second lever: the difference between crossing a reported ice tongue and giving it ten miles of sea room is minutes of schedule against the tail risk that defines the whole exercise. Both levers need information the bridge of a non-ice-class carrier does not natively have, which is the honest argument for the next section.

The escalation ladder

Between “sail as normal” and “hold for icebreaker” sits a ladder of intermediate responses, and a shoulder-season regime is essentially a set of rules for climbing it. In our experience supporting year-round export operations in seasonal ice, the rungs look like this.

Rung one – monitored transit. Ice chart review and satellite tasking as part of voyage planning, a defined reporting scheme, and a shore-side watch that owns the ice picture for the transit corridor. This is the floor for any shoulder-season sailing, and it is cheap.

Rung two – shore-side routing support. The ice desk – the operator’s own or a contracted service – issues corridor guidance and updates it against new imagery and drift forecasts, the marine equivalent of the weather routing every ocean passage already uses. Effective while the picture is sparse and the corridors wide.

Rung three – an ice advisor on board. At some point the picture is too dynamic for shore support alone: floes are moving between chart issues, and the judgement calls – what that return on the radar is, whether the lead ahead closes, what speed the next three hours permit – have to be made on the bridge in real time. The practical answer is to embark a specialist in ice navigation on the non-ice-class vessel for the passage.

Canada formalised the role decades ago: its Arctic shipping rules have long required tankers in shipping safety control zones – and vessels applying the ice regime system – to carry a qualified ice navigator, with the qualification defined in sea time that includes days spent manoeuvring in ice. In most seasonal seas outside such regimes the embarked specialist is an operator’s own mitigation, and it has to be prepared before the season: contracted in advance, communications with the shore ice desk agreed, and the trigger conditions for embarkation defined in the transit regime rather than improvised when the chart turns ugly.

One line in that document must be unambiguous: the advisor advises, and the master’s overriding authority stands untouched – the regime defines not who commands but what the advice triggers procedurally, such as a mandatory speed and routing review whenever the advisor calls the conditions marginal. Placed on the ladder this rung is precisely the ice counterpart of the on-site meteorologist that marine operations practice already credits: capability embarked, uncertainty reduced, and a defensible basis for operating closer to the limit than an unsupported bridge should.

Rung four – escort. Icebreaker or ice-capable support vessel ahead of the tanker. The step change in cost and scarcity is obvious; less obvious is that escort is not a solved end state – convoy operations carry their own failure modes, which we examined for Arctic escort separately. Escort belongs on the ladder, not on a pedestal.

Rung five – suspend. The transit does not sail, or the laden carrier holds at a safe waypoint. Writing this rung into the regime in advance is what makes it usable in anger; a suspension criterion invented during a schedule crisis will lose the argument against the berth window. The parallel with written operating limits that crews bypass under commercial pressure is exact, and the industry already knows how that failure works in DP operations.

The gate that moves traffic between rungs is the metocean machinery of the previous section: index thresholds on the observed regime, forecast-quality derating, and a named owner of the decision. Its output is rarely a bare yes or no – conditions are part of the vocabulary, and the mature regimes already use them: speed caps tied to the ice actually reported, daylight-only passage, corridor discipline, and tighter rules for the terminal approach, where traffic concentrates and sea room to avoid a sighted floe runs out first. Rung transitions can be written against the same inputs: shore routing engages when any ice type carrying a negative risk value for the hull is reported in the corridor, the advisor embarks when the conservatively computed outcome approaches zero or the chart outages exceed what the regime allows, and escort or suspension follows a negative outcome. The two shoulders get different tests, cashing out the asymmetry the opening described: the freeze-up gate is a window-closure check – forecast ice development over the transit duration plus contingency, the direct transplant of the operation reference period – while the break-up gate weights identification quality and triggers on old or deformed ice at any concentration. The rungs are not exotic. What is rare is finding all five written into one document before the season starts.

What remains unsolved

Honesty about the open problems is part of the method, because a gate that pretends to more certainty than the inputs carry is worse than judgement.

Drift forecasting at decision scale. Basin-scale ice models are respectable; forecasting where a specific band of drift ice will sit relative to a corridor twelve hours out is still poor, and the shoulder season – mobile ice, open water, wind-driven – is the hardest case. Until that improves, derating and reconnaissance carry the load.

The detection floor. Nothing in routine service reliably finds a growler in a seaway at transit-decision range. The practical mitigations remain indirect: assume glacial or deformed fragments wherever their sources exist, treat reported bergy water as a regime boundary rather than a data point, and let speed absorb the residual.

Liability and the split incentive. The charterer wants the schedule, the terminal wants the berth turned, the master carries the legal responsibility, and the operator carries the consequence. Behind all of them sits the insurance layer: hull cover is written with trading warranties and ice clauses, and a non-ice-class vessel taken knowingly into reported ice can put the operator on the wrong side of its own policy – a constraint that often decides the question before any risk index does, and one the regime should state rather than discover. A transit regime with a named decision authority does not dissolve these tensions, but it moves the argument from the bridge at midnight to a document agreed in daylight – which is where an operator wants it.

No common yardstick between seas. The Baltic, Canada and the polar-water regimes each quantify their own way, and the seasonal seas outside them mostly do not quantify at all. An operator running assets across several ice-margin regions ends up writing its own translation layer. Joint industry guidance for seasonal-ice tanker operations exists and proves the appetite – what no document yet does is couple that guidance to a quantified decision gate. Until one does, the in-house regime is the yardstick.

The threshold engine itself. POLARIS carries no floe-size term and no speed term, and treats ridged and decayed ice coarsely. The outcome can be the gate’s backbone, but speed caps and fragment assumptions have to be layered on top – the index does not know that impact energy rises with the square of speed; the regime has to.

The decision itself, though, is already writable. Treat drift ice the way the industry treats every other environmental hazard it has learned to respect: observe it, forecast it, threshold it against the actual hull, derate for what the forecast cannot promise, climb the ladder when the numbers say so, and write every step down. A non-ice-class tanker needs an icebreaker when the gate says it does – and the operator’s real job, in the weeks between the rules, is to make sure a gate exists.