Notes from the third edition of the Demurrage Innovation Forum, in Mumbai, India.
Insights from Deepak Chelladurai, Commercial Shipping Consultant, and Amitvikram Chitnis, Inchcape Shipping Services
Consider two effectively identical fixtures: the same vessel class, the same lane, the same charter party, all signed within the same week. Six weeks later, one delivers cleanly while the other generates six figures of demurrage. Nothing in the contract accounts for the divergence; the legal terms are identical, and the claims team will spend months reconstructing what happened, only to deliver its analysis long after the next fixture has been signed against the same set of assumptions.
This is the dimension of the demurrage problem that the industry has been reluctant to examine directly. Demurrage is commonly understood as an event that occurs at port. In practice, it is created considerably earlier. At the point of the trade decision, when a fixture is committed against an execution picture that the market can no longer reasonably assume to be stable.
The underlying assumption no longer holds
For most of the past two decades, freight desks operated on a quiet but load-bearing assumption: that two similar routes, run with similar tonnage under similar conditions, would produce similar outcomes: not identical, but close enough that historical analogues could price the next voyage with reasonable confidence.
That assumption no longer holds, because the execution conditions between fixture and discharge have ceased to be stable. Red Sea disruptions force reroutes around the Cape that add fifteen days and reshape laycans that appeared clean only three weeks earlier; the Strait of Hormuz printing risk premiums that determine whether a parcel crosses at all; sanctions-driven rerouting and shadow-fleet inefficiencies have made voyage duration a function of far more than distance; and beneath all of it, persistent congestion in the major discharge hubs has become structural rather than episodic.
Freight volatility was once the variable that commanded a desk’s attention. It is now compounded by execution volatility, and the second is considerably harder to price. Two identical fixtures can today produce entirely different execution outcomes, and the industry has not yet fully absorbed what that shift means for where demurrage is actually created.
Congestion is frequently a symptom, not the root cause. The cause lies in the interconnected behaviour beneath it, and none of which currently reaches a charterer's screen at the moment of fixture.
Mature tooling, applied at the wrong moment
It is worth considering what a charterer actually has available today. For freight pricing, the tooling is mature: indices, benchmarks, broker reports and historical analogues give a fixture desk reasonable visibility into the state of the market. For post-voyage reporting, the tooling is equally mature: statements of facts are parsed, laytime is calculated, claims are raised and disputes are worked, and the numbers eventually arrive. It is the interval between these two well-served moments that remains unguarded.
Between the signing of a fixture and the vessel’s arrival at the discharge berth, execution risk compounds in ways that no system in the average chartering workflow is designed to surface: the congestion probability at the specific terminal nominated, the berth-allocation behaviour at that port over the preceding forty-five days, whether a given terminal is currently performing or quietly under-performing against its declared productivity, and whether the route the fixture assumed is in fact the route the vessel will run.
The raw inputs (AIS, port intelligence, terminal analytics, weather and congestion patterns) exist in volume and grow richer with each quarter. The gap, rather, is one of timing.
The industry has built systems that report what the freight market is doing before the trade and what the operational truth was after the voyage, but it has not built, at scale and with the operational depth required, an intelligence layer at the decision point itself. The central inefficiency, in other words, is not demurrage but the absence of execution visibility at the moment the commitment is made.
The thesis
Two well-served moments, one unguarded interval
- Indices & benchmarks
- Broker reports
- Historical analogues
- Congestion probability at the nominated terminal
- Berth-allocation behaviour, last 45 days
- Real terminal productivity vs. declared
- Whether the assumed route is the route run
- SOF parsing
- Laytime calculation
- Claims & disputes
The raw inputs already exist in volume: AIS, port intelligence, terminal analytics, weather. The gap is one of timing; nothing surfaces them at the moment the commitment is made.
Six categories of AI, one missing layer
Mapped honestly, the application of AI to demurrage divides into six categories.
- Voyage Understanding covers SOF parsing, event reconstruction and laytime alignment.
- Voyage Prediction covers congestion forecasting, ETA-versus-reality modelling and the assessment of geopolitical disruption.
- Contract Intelligence covers if-then simulation of clauses, pre-fixture exposure mapping and ambiguity detection.
- Commercial Optimisation covers negotiation support, counterparty modelling and dispute detection.
- Portfolio View covers fleet-exposure forecasting, fat-claim recovery and cash-flow optimisation.
- And Pre-Fixture Intelligence covers high-risk fixture alerts, port-anomaly detection and laytime-overrun probability.
Three of these six categories sit pre- or during-voyage, but one (Pre-Fixture Intelligence) sits upstream of every other, and it is precisely the layer that the industry’s current tooling barely touches. The present centre of gravity is Voyage Understanding: SOF parsing is real, commercially valuable and increasingly automated.
The landscape
Six categories of AI.
Pre-Fixture Intelligence
High-risk fixture alerts · port-anomaly detection · laytime-overrun probability.
Contract Intelligence
If-then clause simulation · pre-fixture exposure mapping · ambiguity detection.
Voyage Prediction
Congestion forecasting · ETA-vs-reality · geopolitical disruption.
Voyage Understanding
Centre of gravity todaySOF parsing · event reconstruction · laytime alignment.
Commercial Optimisation
Negotiation support · counterparty modelling · dispute detection.
Portfolio View
Fleet-exposure forecasting · fat-claim recovery · cash-flow optimisation.
What pre-fixture intelligence looks like in practice
Amitvikram Chitnis, who has run operations at Inchcape and observed ground truth from the port side across thousands of calls, describes the operational reality plainly: the vessel, the terminal, the agent, the charterer and the trader each work from a different version of the operational truth, and by the time those versions align, the commercial exposure has already materialised.
Teams react to congestion rather than anticipate it, and two terminals within the same port can perform very differently without the trader knowing in advance which they will be assigned.
Once that reality is structured rather than scattered, the questions a desk can begin to ask are neither subtle nor exotic: they are the questions that determine whether a fixture is well priced:
- Which terminals consistently create hidden delay exposure?
- Which charter party clauses generate the highest dispute frequency?
- Which ports show early indicators of congestion before vessels begin queuing?
- Which vessel classes underperform at certain berth types?
- What operational factors most strongly correlate with demurrage escalation?
- Which counterparties consistently resolve claims faster or slower?
- What is the actual cost of operational uncertainty across the network?
None of these questions is operationally unusual, yet all of them are effectively unanswerable in the average fixture workflow today. That is the gap. Chitnis’s observation from the port side is the one worth dwelling on: congestion is frequently a symptom rather than the root cause.
The cause lies in the interconnected behaviour beneath it, berth allocation, tidal restrictions, pilot constraints, labour shifts, terminal-productivity variance, documentation readiness, weather windows and inland logistics, each of which is forecastable, and none of which currently reaches a charterer’s screen at the moment of fixture.
The vessel, the terminal, the agent, the charterer and the trader each work from a different version of the operational truth — and by the time those versions align, the commercial exposure has already materialised.
The organisational implication
Once intelligence exists, the question of who consumes it becomes a structural one. Trading owns the fixture; demurrage owns the claim; and the intelligence that would have improved the fixture sits in neither team’s tooling, nor within either team’s incentive window.
This is the more uncomfortable element of the thesis: the function best positioned to read execution risk. The demurrage team, which spends its days reconstructing what actually happened at port, is also the function furthest removed from the decision that created the exposure.
Moving pre-fixture intelligence into trading workflows is therefore not a tooling problem but an organisational one; it asks the most upstream desk in the business to slow down enough to consume a signal it has not historically been accountable for. The industry will resist this, as industries reliably do, but it will happen nonetheless: because the alternative is to continue pricing fixtures against an execution picture that no longer holds.
The question for the trading desk
For a head of claims, the immediate work is unchanged: SOFs still require parsing, laytime still requires alignment, and claims still require raising and defending. Voyage Understanding remains real and consequential work.
The harder question belongs to the trading desk: which of the fixtures signed this quarter was the one the demurrage team could no longer save? If that question cannot be answered without reaching for the claims register, the leverage in the demurrage function is in the wrong place.