Blue Autonomy — Insights

Added Steering Resistance: The Autopilot Tax

2026-07-06T00:00:00Z

Proceeding Phase 2 of our vessel resistance series (Added Weather Resistance), today we look at the hidden piece of the puzzle: Added Steering Resistance (R_S).

Heavy weather actually penalizes a ship twice. First, through the direct wind and wave drag we covered in our previous posts. Second, by constantly pushing the vessel off its intended course.

Wind and waves induce massive yawing moments (N_W) on the hull. To counteract these environmental forces and maintain a precise heading, the autopilot must continuously apply rudder deflections (δ).

But a ship's rudder is essentially an underwater wing. Every time it deflects to generate the transverse force needed to steer, it concurrently generates severe induced longitudinal drag.

In simple terms: the rudder acts as a hydrodynamic brake.

Crucially, this added steering resistance (R_S) is not linear. It is proportional to the square of the rudder angle (δ²). This means a 10-degree rudder correction doesn't generate twice the drag of a 5-degree correction — it generates four times the drag.

This quadratic relationship creates what we call the "Autopilot Tax."

Standard autopilot PID controllers are frequently tuned for strict course-keeping, prioritizing a perfectly straight line on the ECDIS over hydrodynamic efficiency. In adverse weather, an over-tuned autopilot will make continuous, aggressive, high-angle rudder corrections to fight every single wave.

This constant oscillation means the ship is effectively sailing with the handbrake pulled. Empirical data shows that an over-reactive autopilot can add 3% to 10% in unnecessary fuel consumption, depending on the sea state.

Optimizing efficiency in rough seas requires tuning the autopilot to accept a wider deadband — allowing the ship to yaw naturally within safe limits rather than aggressively fighting the ocean.

If you want to master fuel optimization, you must ensure your steering system isn't secretly working against your engine.

Added Wave Resistance, Part 2: Radiation

2026-06-29T00:00:00Z

Continuing Phase 2 of our vessel resistance series: Added Wave Resistance (Radiation).

In our last post, we looked at wave diffraction — the invisible blunt force of short waves hitting a hull. It proved that wave drag isn't just about ships battling massive storms.

Wave Radiation, on the other hand, is exactly the heavy-seas scenario you are picturing.

When incident waves get longer and approach the vessel's overall length (λ ≈ L), the hydrodynamic physics change entirely. The ship no longer acts as a rigid wall pushing through choppy water. Instead, it begins to experience severe first-order motions — specifically heave and pitch.

As thousands of tons of steel oscillate vertically, the hull violently displaces the surrounding fluid. In scientific terms, the ship essentially becomes a giant, dynamic wave-maker. As the bow plunges into a trough and rises over a crest, it radiates its own entirely new wave field outward into the ocean.

Generating those new waves requires massive amounts of kinetic energy. This represents a severe energy dissipation problem driven by hydrodynamic damping. The energy carried away by these radiated waves does not come from nowhere; it is stolen directly from the vessel's forward momentum and the propeller's thrust.

The severity of this penalty is strictly governed by encounter frequency. This is precisely why head seas are so punishing for fuel efficiency.

Meeting long waves head-on maximizes the encounter frequency, leading to violent vertical accelerations. The harder and faster the ship pitches, the more energy it radiates away. Conversely, in following seas, the encounter frequency drops, the ship rides the swell, and the radiation penalty is drastically reduced.

The operational impact is massive. In a Beaufort 6 sea state, wave radiation itself can easily account for over 20% of a vessel's total resistance. Because this penalty scales non-linearly with significant wave height (H_s), even a slight increase in swell height causes a severe spike in required engine load.

Every time a ship pitches heavily in a seaway, it is literally radiating your fuel away. Understanding this non-linear curve is the true foundation of effective weather routing.

Added Wave Resistance, Part 1: Diffraction

2026-06-22T00:00:00Z

Continuing Phase 2 of our vessel resistance series (Added Weather Resistance), today we dive into the first half of Added Wave Resistance: Diffraction.

When discussing wave drag, the immediate assumption is that a pitching and heaving ship requires more power. While true, that is only half the story.

Before a ship even begins to pitch or heave, it loses massive amounts of energy to Wave Diffraction.

This phenomenon dominates in short, steep sea states where the wavelength is relatively small compared to the ship's length (λ/L is low). In these conditions, the vessel acts as a rigid obstruction. The incoming waves physically strike the hull — particularly the bow — and scatter.

This scattering of wave energy creates a localized dynamic pressure shift. A distinct "pressure wall" forms directly ahead of the bow.

This pressure differential acts as a blunt force, actively opposing the vessel's forward momentum. The energy required to shatter and scatter those incoming waves is stolen directly from the propeller's thrust.

The operational reality here is critical: even in sea states where the ship feels perfectly stable, with minimal heave or pitch, wave diffraction is quietly acting as a massive hydrodynamic brake.

This blunt force resistance means the engine must burn significantly more fuel just to maintain the same speed through the water. Furthermore, added resistance in waves scales non-linearly with significant wave height (H_s) — meaning a seemingly small increase in wave height can cause an exponential spike in fuel consumption due to this diffraction penalty.

To truly optimize a voyage, we cannot just look at severe storms that cause heavy ship motions. We must account for the continuous, invisible penalty of short-wave diffraction hitting the bow.

Added Wind Resistance: The Square Law

2026-06-15T00:00:00Z

Continuing Phase 2 of our vessel resistance series (Added Weather Resistance), today we dive into the physics of Added Wind Resistance.

Wind drag is governed by one brutal hydrodynamic reality: the square law.

Aerodynamic resistance does not scale linearly; it is directly proportional to the square of the relative wind speed. If you double the wind speed, you quadruple the drag.

The standard formula makes this relationship clear:

R_AA = ½ × ρA × A_XV × C_Dwind × V_WR²

But the wind reported by the weather forecast is not the wind the ship actually fights. The critical operational metric is apparent wind (V_WR) — the vector sum of the true environmental wind and the vessel's own forward speed.

This dynamic creates a massive asymmetry in ship performance. Because drag is squared, an apparent headwind penalizes fuel consumption exponentially more than an equivalent apparent tailwind helps it. A 10-knot tailwind only reduces the relative speed and can never mathematically offset the penalty of a 10-knot headwind.

To accurately calculate this penalty, we need the aerodynamic drag coefficient (C_Dwind).

While the most precise C_Dwind values come from direct wind tunnel testing of a specific hull form, the reality is that very few commercial vessels have this data available.

When wind tunnel data is missing, the industry relies on rigorous empirical regression models. The Fujiwara method is an excellent and widely accepted approach for estimating these wind force coefficients based on specific geometric parameters of the ship's superstructure and hull profile.

Accurately estimating this coefficient is critical for separating wind drag from hull performance. Both ISO 15016 and ISO 19030 emphasize precise wind correction. If you cannot mathematically isolate the added wind resistance, your operational data will misdiagnose weather drag as hull fouling.

Understanding the specific aerodynamics of your superstructure is the only way to evaluate true engine and hull performance.

How aggressively does a Beaufort 6 headwind shift the daily fuel consumption curve on your specific vessels?

From Calm Water to Real Weather: Added Resistance

2026-05-30T00:00:00Z

We have officially wrapped up Phase 1 of our vessel resistance series, where we broke down the physics of Calm Water Resistance (friction, wave-making, form drag etc.).

But as any operator knows, ships don’t sail in towing tanks.

Today, we are kicking off Phase 2: Added Weather Resistance.

Calm water gives us a theoretical baseline, but the open ocean dictates your actual fuel bill. When the weather turns, dynamic environmental forces can destroy even the most optimized theoretical fuel curves.

So, where is that extra fuel actually going?

Swipe through today's carousel as we break down the three main culprits of real-world weather drag:

💨 Added Wind Resistance (The Square Law) 🌊 Added Wave Resistance (Diffraction & Radiation) 🧭 Steering Drag (The Autopilot Tax)

Understanding your calm-water baseline is only step one. Accurately measuring how the environment fights against that baseline is how you actually optimize voyage performance.

Swipe through the breakdown below. Which of these weather factors creates the biggest efficiency gap for your fleet? 👇

Hydrodynamic Drag from Appendages and Steering

2026-05-22T00:00:00Z

Understanding Hydrodynamic Drag from Appendages and Steering

Continuing our calm water resistance series, this post examines two often-underestimated sources of hydrodynamic drag.

Beyond common friction, waves etc. drag can arise from essential underwater appendages like bilge keels, thruster tunnels, rudders, and propeller shafts. These components, while critical, add to total resistance by increasing wetted surface and disrupting flow. Computational Fluid Dynamics (CFD) is essential for analyzing and optimizing their designs.

Furthermore, single-screw vessels also face continuous drag from the propeller's rotational wash, necessitating constant, subtle rudder deflections to maintain a straight course. These minor corrections generate continuous hydrodynamic drag, increasing propulsion power and contributing to elevated fuel consumption over time.

The cumulative impact of these steering adjustments on fuel efficiency can be significant. Optimizing rudder design and integrating advanced steering control systems are critical for mitigating this overlooked drag and enhancing propulsive efficiency. Rigorous model testing and advanced CFD analyses are vital for refining designs.

A comprehensive understanding of these subtle drag sources is crucial for maximizing calm-water performance and achieving superior operational efficiency. Proactive strategies—including optimized appendage geometries and precise steering system calibration—are essential for long-term fuel savings and reduced emissions. This ongoing focus on hydrodynamic optimization remains central to maritime efficiency.

This post concludes our calm water resistance series. Future series will explore how added weather resistance influence our vessels and increases the fuel bill.

Air Resistance: The Headwind You Can't Escape

2026-05-14T00:00:00Z

Continuing our series on calm-water resistance, today we look at a force that exists even on a perfectly windless day: Air Resistance.

It’s a common misconception that ships only face wind drag in bad weather. The reality is that simply moving forward creates an inescapable headwind. If a vessel is steaming at 15 knots on a dead-calm sea with zero true wind, it is continuously sailing into a 15-knot "apparent wind."

Think of putting your hand out the window of a moving car on a still day—the pressure you feel is apparent wind.

For a ship, this effect is massive. The frontal area of the vessel—the superstructure, the accommodation block, or towering container stacks—acts like a sail in reverse. Every square meter of that surface has to push through the air, creating aerodynamic drag.

To be clear: this isn't about battling storms or weather routing. (We will cover added weather resistance in Phase 2 of this series).

This is baseline aerodynamic drag. It is a constant, inherent component of a ship's operational profile in calm water. A container ship with a high stack will naturally have a higher baseline air resistance than a low-profile bulk carrier.

How much does the superstructure profile differ across your fleet, and have you noticed its impact on baseline fuel consumption?

Viscous Pressure (Form) Resistance

2026-05-05T00:00:00Z

Continuing our series on the breakdown of calm-water resistance, today we examine the hidden drag at the stern: Viscous Pressure (Form) Resistance.

While frictional resistance affects the entire wetted surface, form drag is highly localized. It is driven by how fluid pressure changes as water travels the length of the hull.

As a vessel moves forward, it pushes through the water, creating an area of high pressure at the bow. Ideally, if the water could flow perfectly around the hull and close perfectly behind the stern, the pressure at the aft would recover completely, pushing the ship forward and cancelling out the bow resistance.

However, because water is viscous, this perfect recovery is impossible.

As water travels along the hull, it forms a boundary layer due to friction. At the bow, this boundary layer is extremely thin, so the high pressure acts directly against the hull. But as the flow continues toward the stern, the boundary layer grows thicker.

This thickened boundary layer at the aft acts like a cushion, preventing the water pressure from fully recovering against the hull. Consequently, the pressure at the stern is always lower than the pressure at the bow.

This permanent pressure difference acts like a vacuum, continuously pulling the vessel backward.

The problem worsens dramatically if the hull geometry curves inward too abruptly at the stern. When the angle is too steep, the water can no longer follow the shape of the hull. The flow separates entirely, generating chaotic, turbulent eddies and a massive low-pressure wake directly behind the ship. This flow separation turns a moderate pressure difference into a severe suction force, wasting significant propulsive energy.

Full-block vessels, like bulk carriers and tankers, are particularly susceptible to this form drag compared to slender, high-speed designs. While naval architects use advanced CFD to design aft geometries that minimize separation, operational factors like draft and trim can also influence where this separation occurs.

At Blue Autonomy, we help you understand complex hydrodynamic realities (as the above), enabling physics-backed decisions that minimize hidden drag and maximize fuel efficiency across the fleet.

Wave-Making Resistance

2026-04-14T00:00:00Z

Wave-making resistance

Continuing our series on calm-water resistance, today we examine one of the most important hydrodynamic mechanisms affecting vessel efficiency: wave-making resistance.

As a ship advances, its hull disturbs the surrounding pressure field and generates a system of gravity waves, primarily at the bow and stern.

Those waves are not just a visual by-product of motion; they are a direct sink of energy, because part of the vessel’s propulsive power is continuously transferred into the water to create and sustain the wave system. The bow region acts as a strong positive pressure zone that pushes water upward and outward, while the aft region behaves more like a suction field, generating its own wave pattern with the same wavelength but different phase characteristics.

The resulting transverse and divergent wave systems interact with one another as they travel along the hull.

When crests from the bow and stern systems reinforce each other, wave height increases and the resistance penalty becomes larger; when crest and trough partially cancel, the resistance is reduced.

This is why wave-making resistance does not rise as a perfectly smooth curve with speed, but instead shows the well-known humps and hollows associated with interference effects.

The key parameter governing this behaviour is the Froude number, which links vessel speed to gravity and waterline length.

At lower speeds, frictional resistance usually dominates, but as speed increases, wave-making resistance becomes increasingly important.

As the generated wavelength grows and becomes comparable to the vessel’s waterline length, the ship encounters a much steeper hydrodynamic penalty. In practical terms, the vessel begins to spend a disproportionate amount of power creating and interacting with its own wave system rather than simply advancing forward efficiently.

This is exactly why small increases in speed can produce disproportionately large increases in required propulsion power and fuel consumption.

Draft and trim also matter, since they modify the pressure distribution around the hull and therefore affect the wave pattern being created.

Understanding wave-making resistance is therefore not only a design issue, but an operational one. The better we understand where the vessel sits on its resistance curve, the better we can define speed profiles that avoid unnecessary fuel penalties.

Why Fleet Performance Culture Cannot Be Purchased

2026-04-01T00:00:00Z

A presentation given at the People Tech Maritime conference at the Eugenides Foundation in Athens, on why fleet performance is an organisational capability rather than a product: the most advanced monitoring system delivers nothing if the people responsible never act on it.

What the talk covers

The trap: a company buys a state-of-the-art monitoring system, and six months later the dashboards "collect digital dust." The gap is rarely technical — it's ownership and motivation.
Performance is a triangle of three interdependent pillars: a trustworthy information backbone, expert interpretation, and genuine organisational commitment. Remove one and the structure collapses.
Garbage in, garbage out: reliable data needs proper architecture, calibrated sensors, rigorous verification, and embedded domain expertise — on some audited fleets nearly half the operational data was compromised in ways no dashboard would flag.
ISO 19030 vs AI/ML: the transparent, reproducible standard versus raw machine-learning power — but neither runs without someone who understands the physics. A model with excellent error metrics once predicted less fuel sailing into a headwind; the strongest approach is hybrid — algorithms find the patterns, experts decide which are real.
Change is human and slow: sustained, expert-led engagement over quarters and years. Collaboration produces adoption, imposition produces resistance — shown through weather-routing and generator-threshold examples where crews embraced the tools once they helped build them.

After the presentation, the topic was discussed further in a panel — you can watch the panel discussion on YouTube.

Frictional Resistance: The Dominant Component of Drag

2026-03-30T00:00:00Z

In continuation to our series for vessels' calm water resistance let's continue with the second part of our series for the Frictional Resistance component:

Foundation of Drag: Frictional Resistance's Dominance

Frictional resistance (RF) is a primary component of a vessel's total calm-water drag. It typically accounts for a significant portion, ranging from 75-85% in new, slow-speed ships and up to 50% in high-speed vessels. The ITTC-57 model-ship correlation line (CF = 0.075 / (log Re - 2)^2) is an established method for estimating this critical resistance component. This line shows its dependency on the Reynolds number (Re), wetted surface area (Sw), water density (ρ), and ship speed (U). Frictional resistance originates from viscous shear stresses. These stresses develop within the boundary layer—a thin region of water adjacent to the hull where velocity gradients occur. For full-scale ships, the flow within this boundary layer commonly transitions from laminar to turbulent at Reynolds numbers exceeding 1x10^6, influencing overall resistance.

The Impact of Hull Roughness & Biofouling

The condition of the hull surface, often quantified by Average Hull Roughness (AHR), directly influences frictional drag. An AHR of 65 µm generally indicates very good operational efficiency. Values significantly exceeding 200 µm, however, suggest a suboptimal hull condition. Industry research shows increased hull roughness can measurably increase required propulsion power, leading to higher fuel consumption throughout a vessel's operational life. Moreover, severe biofouling substantially increases power requirements. In some cases, it can double power needs compared to a clean, smooth hull. This effect typically worsens between dry dockings due to the accumulation of corrosion and marine growth.

Correlations: Hull Condition, Fuel, and Emissions

The power required to overcome total (and hence frictional) resistance is approximately proportional to the cube of the vessel's speed. This relationship implies that even modest increases in speed can lead to disproportionately higher power demands; for example, doubling a ship's speed can necessitate up to eight times the power. Consequently, increased frictional drag directly leads to higher Shaft Horsepower (SHP) requirements and a corresponding rise in fuel consumption. Therefore, poor hull condition contributes directly to increased operational costs and Greenhouse Gas (GHG) emissions. This impacts vessels' ability to comply with maritime regulations.

Optimizing hull performance is a critical strategic imperative for improving sustainability and economic viability in maritime operations.

How is your organization addressing hull efficiency to meet regulatory demands and reduce operational costs?

Frictional Resistance: Why Fouling Is a Performance Issue

2026-03-26T00:00:00Z

Our founder is going deeper into one of the most underestimated drivers of vessel inefficiency: frictional resistance. When hull condition deteriorates, drag increases quietly, fuel use rises, and the commercial impact compounds faster than many teams expect. This is exactly why fouling should be treated as a performance issue, not just a maintenance one. Check it below.

The Five Components of Calm-Water Resistance

2026-03-22T00:00:00Z

Why are your vessels losing efficiency? 🚢 Even in calm waters, hidden forces are constantly fighting your ship's progress. Understanding these resistances is the first step toward unlocking peak operational performance and significant fuel savings.

In this carousel, we break down the 5 fundamental components of Total Resistance: 🔹 Frictional Resistance: The primary drain, accounting for up to 70-90% of drag. 🔹 Wave-Making Resistance: The "tax" on higher speeds. 🔹 Form Drag: Why hull design is your best ally. 🔹 Air Resistance: The unseen wind affecting your superstructure. 🔹 Appendages & Steering: How even minor adjustments impact the bottom line.

Optimizing fleet efficiency starts with deep hydrodynamic insights. Is your fleet ready to dive deeper? Slide through to discover the science behind the drag! ➡️ Learn more at: blueautonomy.gr

Data Quality Is the Ultimate Form of Compliance

2026-03-16T00:00:00Z

"Is your vessel’s performance data truly reliable, or is it a liability? Our CEO shares his take on why data quality is the ultimate form of compliance in today’s maritime landscape. How is your team tackling the shift from manual to high-frequency data? Let’s discuss in the comments.

Measuring Hull & Propeller Performance with ISO 19030

2026-03-13T00:00:00Z

We live in the golden age of data and still poor hull and propeller performance is one of the most underestimated cost and efficiency drains in shipping. For many operators, the real problem is not just the degradation itself. It is the lack of a consistent, data-driven way to measure it properly, compare it over time, and link it to maintenance and operational decisions. This is exactly where ISO 19030 becomes important. Released to standardize the measurement of changes in hull and propeller performance, the ISO 19030 framework gives ship operators a structured methodology to move away from rough assumptions and subjective observations. Instead of relying on fragmented signals, it enables a more disciplined approach to quantifying degradation, evaluating maintenance outcomes, and improving vessel performance management. In practice, the standard is often discussed through two routes: ISO 19030-2, is the default method with stricter requirements and higher expected accuracy. ISO 19030-3, offers alternative methods with broader applicability, but potentially lower accuracy. That distinction matters. If the goal is to make sound technical and commercial decisions, measurement quality is not a detail. It is the foundation. At the core of ISO 19030 are several key performance indicators that turn raw vessel data into something operationally useful: Dry-docking performance, In-service performance, Maintenance trigger and Maintenance effect These are not abstract metrics. They directly affect maintenance timing, coating evaluation, fuel spend, emissions exposure, and the ability to make better decisions around vessel efficiency. They also support broader energy-efficiency and compliance strategies in an industry increasingly shaped by CII, FuelEU, and EU ETS. But applying ISO 19030 properly is not easy. The standard depends heavily on data quality. Primary inputs include speed through water and delivered power, and the default method requires high-frequency measurement. Secondary parameters such as wind, waves, water depth, temperature, draught, trim, and rudder activity are also essential because performance cannot be assessed meaningfully without proper filtering and normalization. This is where many implementations become weak. Common failure points include poor sensor quality, insufficient acquisition rates, weak data handling, and incomplete normalization of environmental and operational effects. In those cases, the framework may be present on paper, but the output is not reliable enough to support confident decisions. ISO 19030 is therefore more than a technical standard. It is a test of data discipline. When implemented poorly, it risks creating false confidence from noisy data. The real question question is whether your current monitoring approach is precise enough to capture what ISO 19030 is actually trying to reveal.

Why Raw Data Isn't Enough: Engineered Data Pipelines

2026-03-10T00:00:00Z

High-quality data is the only foundation for reliable insights. Our founder breaks down why raw data collection isn't enough anymore, and how engineered data pipelines are the key to unlocking true efficiency. 👇

Hull Fouling: The Hidden Tax on Fleet Operations

2026-03-07T00:00:00Z

Hull fouling isn't just an efficiency killer—it’s a massive hidden tax on fleet operations.

It quietly erodes profitability and undermines compliance efforts. When marine organisms accumulate on a vessel’s underwater hull, frictional resistance spikes. This directly translates to higher fuel consumption and reduced operational speed.

It’s a critical challenge for everyone involved: technical managers striving for optimal performance, ship owners facing escalating costs, and charterers needing contractual speed and consumption adherence.

The problem compounds quickly. It starts with microscopic slime, which can measurable increase resistance. If unchecked, it develops into macrofouling like grass, seaweed, barnacles, and tube worms. In severe cases, dense barnacles can elevate resistance by over 50-100%.

To maintain speed, your main engine has to work significantly harder.

Consider a typical bulk carrier operating at 12 knots. A moderate level of hull fouling can cause a 15-35% increase in fuel consumption just to maintain that speed. If that vessel consumes 30 MT of fuel per day, you are burning an additional 4.5 - 10.5 MT daily.

Over a single voyage, that's a major financial loss. Over a year, it’s a massive, unbudgeted expenditure.

Beyond the immediate financial hit, hull fouling profoundly impacts regulatory compliance.

Increased fuel consumption means escalated CO2 emissions. For vessels under the IMO's Carbon Intensity Indicator (CII) framework, this can degrade an operational rating from an A or B down to a D or E. Repeated poor ratings trigger mandatory corrective action plans and hurt marketability.

Similarly, higher emissions mean higher costs under the EU ETS regulation. Operational inefficiency is now a direct double financial penalty.

Traditional noon reporting - although very usable - lacks the precision to isolate these impacts, leaving you with an incomplete picture of your true performance losses.

So, how do you accurately quantify this degradation? It's difficult.

This is where advanced digital solutions change the game.

By continuously monitoring vessel performance with high-frequency data, we can establish a dynamic baseline - digital twin - and detect the subtle changes that indicate fouling development. Technical managers can quantify exact speed loss and added consumption, allowing for data-driven decisions on hull cleaning interventions.

Proactive management minimizes this "hidden tax," preserves your CII ratings, and optimizes charter party performance.

Understanding and actively managing hull fouling is no longer just good practice—it is a fundamental requirement for staying competitive and compliant.

At Blue Autonomy, we help you transform your fleet's approach to hull performance through data science and advanced analytics.

Learn how we do it: https://blueautonomy.gr/

Why Continuous Vessel Monitoring Is a Commercial Necessity

2026-03-01T00:00:00Z

Our founder explains why continuous vessel monitoring is becoming a commercial necessity, not a nice‑to‑have. Follow Blue Autonomy for more insights on vessel performance and decarbonization.

Interpretable Data-Driven Ship Dynamics Model: Enhancing Physics-Based Motion Prediction with Parameter Optimization

2025-09-01T00:00:00Z

Abstract

The deployment of autonomous navigation systems on ships necessitates accurate motion prediction models tailored to individual vessels. Traditional physics-based models, while grounded in hydrodynamic principles, often fail to account for ship-specific behaviors in real-world conditions. Conversely, purely data-driven models offer specificity but lack interpretability and robustness in edge cases.

This study proposes a data-driven physics-based model that integrates physics-based equations with data-driven parameter optimization, leveraging the strengths of both approaches to ensure interpretability and adaptability. The model incorporates physics-based components such as 3-DoF dynamics, rudder, and propeller forces, while parameters such as resistance curve and rudder coefficients are optimized using synthetic data. By embedding domain knowledge into the parameter optimization process, the fitted model maintains physical consistency.

Validation of the approach is realized with two container ships by comparing, both qualitatively and quantitatively, predictions against ground-truth trajectories. The results demonstrate significant improvements in predictive accuracy and reliability of the data-driven physics-based models over baseline physics-based models tuned with traditional marine engineering practices. The fitted models capture ship-specific behaviors in diverse conditions, with their predictions being 51.6% (ship A) and 57.8% (ship B) more accurate, and 72.36% (ship A) and 89.67% (ship B) more consistent.

Why it matters

Accurate, vessel-specific motion prediction is a prerequisite for autonomous navigation — and for any decision system that needs to know what a ship will do next. This work shows that you do not have to choose between the interpretability of physics and the specificity of data: embedding parameter optimization inside a physics-based structure delivers both.

Comparative Study of Ship Motion Prediction Models: Data-Driven Physics-Based vs Pure Machine Learning

2025-06-26T00:00:00Z

Abstract

Accurate ship motion prediction is essential for safety and efficiency in maritime navigation. Data-driven techniques offer potential accuracy improvements to ship model prediction, yet their utility and reliability are being explored. This study compares two data-driven approaches for vessel trajectory prediction: a data-driven optimization of physics-based model parameters and a purely machine learning approach. Both models are trained and evaluated based on a dataset comprising of 90 trajectories generated via a vessel simulator. The physics-based model is based on hydrodynamic principles while key parameters are fitted, using constrained nonlinear least squares, to tailor its prediction accuracy to the dataset. For the machine learning approach, a feed-forward neural network learns motion patterns directly from data without prior domain knowledge. A 5-fold cross-validation is employed for both models ensuring robust evaluation. The models' prediction accuracy is quantified using: (i) Euclidean Distance, (ii) Euclidean Distance with Heading Penalization, and (iii) Custom Vessel Distance Measure. The results across the dataset show that the data-driven physics-based model achieves higher prediction accuracy (40 per cent lower summarized absolute error) and consistency (35 per cent lower summarized error spread). Individual prediction scenarios are examined to highlight the trade-offs between accuracy, reliability and adaptability of the two approaches.

Index terms: autonomous shipping, machine learning, parameter optimization, data-driven models.

Why it matters

For autonomous and assisted navigation, a motion model has to be both accurate and trustworthy in conditions it has never seen. This study shows that grounding a model in hydrodynamics and then fitting its open parameters to data beats a pure black-box network — not only on average error, but on consistency. That consistency is what matters when the prediction feeds a collision-avoidance or control decision, where an occasional large miss is far more costly than a slightly higher average error.