ICCAS 2026: International Conference on Computer Applications in Shipbuilding

Reliable perception is an essential prerequisite for the safe navigation of Autonomous Surface Vessels (ASVs). Obstacle detection in maritime environments is challenging due to water reflections, illumination variations, wave-induced occlusions, and large scale variations. As per our literature review perception systems typically rely on bounding-box detectors or semantic segmentation. This paper investigates maritime obstacle perception using instance segmentation across two sensing modalities: visible daycam/electro-optical (EO) imagery and thermal/infrared (IR) imagery. Two YOLOv11m-seg models were independently trained on domain-specific datasets. The daycam model was trained on the LaRS dataset, augmented with locally collected daycam images. The Thermal /IR model was trained on the MaSSMIND dataset, introducing instance segmentation to a benchmark where previous studies have primarily focused on semantic segmentation. Both models target critical maritime objects including ships, buoys, floating debris, and potential man-overboard targets. Bounding box models often fails to detect small or out-of-distribution (OOD) obstacles due to coarse localization. Semantic segmentation improves this by providing pixel-level classification but cannot distinguish or track individual object instances. Our model uses instance segmentation to deliver precise boundaries while separating objects of the same class, enabling detection of small or partially occluded obstacles and tracking of multiple maritime obstacles. The resulting outputs generate structured spatial metadata—including object identifiers, segmentation masks, and image-plane coordinates—that can be directly utilized by situational awareness modules for navigation, tracking, and collision avoidance. Additionally, unassigned regions can be analyzed through a post-processing stage to identify potential out-of-distribution (OOD) obstacles. Experimental evaluation shows that the IR model achieves a mask mAP$_{50-95}$ of 0.677, while the EO model achieves a mask mAP$_{50-95}$ of 0.436. Real-world sea trials were conducted using a vessel equipped with an EO/IR camera system, processing a live RTSP stream on an edge laptop with an RTX 4060 GPU. Despite its compact architecture, the system consistently delivered inference speed of 30 FPS while reliably identifying multiple obstacle instances. The contributions of this work are: (1) demonstration of high-performing, edge-deployable EO and IR instance segmentation models for maritime perception; (2) incorporation of locally collected EO data to improve regional robustness; and (3) practical validation through real-time sea trials. These findings indicate that perception at the instance-level effectively enhances situational awareness for ASV operations.

Modern cruise ship design is increasingly focusing on the passenger experience as an important part of the design process, in addition to technical performance. Computer-aided ship design tools do a good job of meeting structural, hydrodynamic, and operational needs, but passenger experience is not often used as a way to judge designs in the early stages. At the same time, large amounts of unstructured passenger review data give detailed information about how well the ship works in terms of space, crowding, accessibility, and service interaction, but they are still mostly separate from the decision-making process for building ships. This paper presents an experience-centric, computer-assisted text analytics framework that converts unstructured passenger review narratives into design-relevant experiential indicators for cruise ship design. Utilizing a dataset of 773 publicly accessible passenger reviews from three cruise ship classes, the method methodically extracts experiential patterns concerning spatial legibility, congestion perception, amenity distribution, and service accessibility. The analysis does not look at satisfaction or loyalty outcomes. Instead, it looks for design-sensitive experiential stressors and enablers that are linked to ship scale and layout configurations. The results show that different types of ships have different performance profiles when it comes to experiences. This is because there are trade-offs between how intense the experiences are, how efficient the space is, and how easy it is to use. The suggested framework acts as a decision-support layer that works with current computer-aided cruise ship design tools by allowing for an evaluation of layout and capacity options based on experience during the early stages of concept development.

Maritime digitalization is frequently implemented through vertical solutions addressing specific operational domains related to total cost of ownership (TCO), such as energy management, maintenance, performance monitoring and guest services. This paper presents Navis Sapiens, a digital ecosystem applicable to both civil and defence vessels, centered on a Foundational Digital Platform (FDP) conceived as a unified, open and evolvable digital backbone for ships. The objective is to illustrate how a platform-based approach can reduce systemic integration complexity while enabling Vessel-to-Port (and Vessel-to-Naval-Base) interoperability and digital services across multiple ship domain . Rather than introducing isolated applications, the platform establishes a common onboard and ashore-based digital environment integrating heterogeneous OEM systems, automation, other onboard networks, and legacy applications through a shared data layer and standardized interfaces. Building on the Foundational Digital Platform, the proposed architecture follows a hierarchical approach composed of three complementary platforms, replicated both onboard and ashore: o Foundational Digital Platform (FDP) - providing the core infrastructure for vessel multi-domain data acquisition and integration across ship systems, supported by onboard and ashore computing infrastructure enabling incremental activation and over-the-air updates; o Application Platform - supporting the development and execution of modular digital services, including Digital Twins, performance optimization services, lifecycle management tools, Guest services and AI-based decision-support systems; o Aggregation Platform - enabling integration and controlled distribution of applications, including third-party, OEM and legacy digital services, through a BtoB marketplace. The approach is suitable for both newbuild programs and retrofit scenarios, supporting gradual digital transformation of existing fleets. A first full-scale civil implementation was delivered in Q1 2026 on a next-generation cruise vessel. In parallel, the same architectural principles are being aligned with defence program requirements, demonstrating applicability across both commercial and military contexts while accommodating differentiated operational constraints. The paper presents the platforms and explains the systemic synergies of a shipbuilder-led platform strategy, showing how the Fincantieri's role of Digital Vessel Design Authority can ensure coherence across onboard digital assets and enables structured long-term evolution of ship capabilities.

Mitigating Underwater Radiated Noise (URN) while maintaining propulsive efficiency requires resolving complex hull-wake-propeller interactions, making full-ship Large Eddy Simulation (LES) computationally prohibitive. Furthermore, identifying an optimal design requires a vast number of simulations, making the search for a high-performance, low-noise configuration both time-consuming and expensive. Interaction effects are especially notable for small, fast craft where the free surface elevation and running trim angle greatly affect the propeller inflow. This paper presents a novel, highly efficient numerical framework to evaluate propeller hydroacoustics and generate high-fidelity datasets. The methodology utilizes a two-stage approach. First, an unsteady RANS self-propulsion simulation using an overset mesh captures the appended hull's dynamic sinkage, trim, free surface, and effective wake. Second, the computed free surface and hull geometry are extracted and "frozen" to form the upper boundary of a confined domain—effectively a virtual hydroacoustic tunnel. Inside this domain, a targeted high-fidelity simulation is performed using a Rigid Body Motion (RBM) approach and an acoustic solver to compute URN. By isolating the propeller within the re-computed flow field, the macro-physics of full self-propulsion are bypassed, drastically reducing computational overhead. The methodology presented in this study allows for the rapid testing of multiple propeller geometries to train a machine learning model, ultimately enabling a fast, predictive tool for simultaneous hydroacoustic and propulsive optimization.

U.S. national political discourse has seen a renewed bipartisan push for a U.S.-flagged commercial fleet. Proponents of this capability have made it clear that these vessels should be produced domestically, as seen in the proposed SHIPS for America Act and recently released Maritime Action Plan. However, U.S. domestic commercial shipbuilding output has dwindled in recent decades, and the current policy environment is fiscally constrained. There is little political support for a perpetually subsidized industry, leading some to advocate for collaboration with allied nations that possess profitable, self-sustaining commercial shipbuilding industries. The bipartisan interest in revitalizing domestic commercial shipbuilding exists alongside varying proposals of which ships should be procured, on what schedule, and with which supportive government policies or allied partnerships. The policy push is rooted in an urgent concern for emerging maritime security threats. A strategic industrial policy plan must account for disruptive geopolitical scenarios; a tool is needed that can inform policymakers of the relationship between fleet procurement objectives, disruptive scenarios, and policy levers. This paper presents a scenario-based techno-economic model (TEM) to evaluate the cost and feasibility of U.S. commercial fleet procurement objectives under disruptive geopolitical scenarios. The TEM models national shipbuilding capacity of the U.S., Japan, and Korea as the number of ships that can be concurrently constructed, derived from historical production data and modulated by a scenario-driven capacity growth factor. Over a user-defined time horizon, a mixed-integer linear optimizer identifies the minimum-cost sequence of ship production that satisfies scheduled fleet objectives while respecting capacity, lead time, and policy constraints. Policy levers allow decision-makers to evaluate the impact of permitting allied newbuilds and open-market reflagging alongside domestic shipbuilding. Geopolitical scenarios are implemented as user-defined conflict timelines with prescribed effects on national shipbuilding capacity and vessel procurement priority. An initial case study applying the TEM to the SHIPS for America Act demonstrates the model's utility in evaluating policy options and objectives. Results show that a mixed strategy of domestic newbuilds, allied newbuilds, and reflagged vessels can substantially reduce program cost and enable more ambitious near-term objectives. The analysis further reveals that meeting proposed timelines under domestic-only constraints requires capacity growth rates not typical of current shipbuilding paradigms, motivating questions about the role of disruptive manufacturing technologies and what kind of supportive policies could yield a self-sustaining and cost-competitive domestic commercial shipbuilding industry.

Marine Electrification is taking place at a rapid pace, and guidelines set by MPA to transform domestic harbour crafts to either use net-zero fuel or convert to full electric by 2030 have driven an uptick in full-electric new builds, with future fuel-readiness. In this context, developing accurate digital representations of electric propulsion systems has become essential for de risking design decisions and validating system performance under realistic operating conditions. This study presents an integrated co-simulation for Digital twin development of an electric tug's power and propulsion system. Dynamic models of the electric drivetrain components—including battery systems, converters, motors, and control strategies—are coupled with propeller and hydrodynamic models of the electric tug. Such multi-domain integration enables the capture of transient interactions among torque demand, shaft speed, vessel acceleration and deceleration, and resulting variations in ship speed. The co-simulation enables analysis of load-RPM coupling effects and voltage bus disturbances under realistic operational scenarios. In addition, steady-state operational models are developed using Automatic Identification System (AIS) data from existing conventional tugs to evaluate mission profile suitability, charging scheduling strategies, and energy consumption characteristics. The combined dynamic and steady-state simulations provide insights into battery state-of-charge (SOC) evolution, peak power demands, and system stability throughout typical harbour operations. The proposed co-simulation approach supports early-stage design optimisation, improves power system sizing accuracy, and enhances performance verification for next-generation electric tugs.

Modern marine vessels operate complex machinery systems that require reliable monitoring, control, and fault management to ensure safe and efficient operations. The Integrated Platform Management System (IPMS) is widely adopted in ships to provide centralized monitoring and automation of critical systems such as propulsion auxiliaries, power generation, ballast operations, fuel management, and alarm systems. Conventional IPMS implementations typically rely on distributed hardware Programmable Logic Controllers (PLCs) installed across machinery spaces and control rooms. While this architecture has proven reliable over several decades, it introduces challenges related to hardware dependency, system scalability, maintenance complexity, and lifecycle cost. Recent developments in industrial computing and virtualization technologies have introduced new possibilities for automation system architectures. Virtual PLC technology enables the implementation of control logic as software applications executed within virtualized computing environments rather than on dedicated PLC hardware. This approach allows multiple PLC instances to operate on centralized industrial server platforms while interfacing with distributed input/output (I/O) modules through industrial Ethernet networks. This paper presents a Virtual PLC-based IPMS architecture designed specifically for marine applications. The proposed architecture integrates distributed I/O modules, high-reliability shipboard communication networks, supervisory control systems, and cloud-enabled data analytics platforms. The virtualization of control systems allows improved flexibility, easier scalability, centralized maintenance, and enhanced redundancy through virtual machine failover mechanisms. In addition, the integration of shipboard automation with digital technologies enables the development of advanced digital ship capabilities including predictive maintenance, operational optimization, and fleet performance monitoring. The paper discusses system architecture, implementation considerations, advantages over conventional approaches, and future opportunities for smart ship automation.VIRTUAL PLC BASED IPMS ARCHITECTURE A Virtual PLC is a software implementation of a programmable logic controller that runs on general-purpose computing platforms rather than dedicated PLC hardware. In a Virtual PLC architecture, control programs are executed within virtual machines hosted on industrial servers. These servers communicate with distributed I/O modules installed throughout the vessel. The architecture typically consists of: o Industrial computing servers o Virtualization software platform o Virtual PLC runtime environments o Distributed I/O modules o Industrial Ethernet communication network o SCADA and HMI monitoring stations The distributed I/O modules collect signals from sensors and actuators and transmit them to the virtual PLC controllers through Ethernet-based communication protocols. The virtual PLC executes control logic and sends control commands back to the field devices through the I/O network.The adoption of Virtual PLC technology enables the development of Digital Ship or Smart Ship automation systems. In a digital ship architecture, shipboard automation systems are integrated with advanced digital technologies including cloud computing, big data analytics, and artificial intelligence. The architecture consists of several layers: Field Device Layer Sensors and actuators installed on machinery systems collect operational data. Edge Control Layer Distributed I/O modules and Virtual PLC controllers process real-time control logic. Supervisory Control Layer SCADA and HMI systems provide visualization and operator interaction. Data Integration Layer Operational data is stored in centralized databases for historical analysis. Cloud Analytics Layer Advanced analytics platforms process operational data to extract insights related to equipment performance and operational efficiency.

SUMMARY Dry docking is a critical operation in ship maintenance where vessels are positioned precisely on dock blocks for inspection and repair. Traditional docking methods rely heavily on manual alignment, plumb line measurements, and diver inspections, making the process labour-intensive and prone to inaccuracies. This paper proposes a multi-sensor-based docking system integrated with laser crosshair projection to enhance docking accuracy and operational efficiency. The system employs laser measurement sensors, coordinate mapping, and real-time data monitoring to guide vessel positioning during docking operations. By comparing real-time sensor data with pre-calculated reference values derived from vessel hull geometry, the system enables precise alignment within millimetre-level tolerances. The proposed solution reduces human dependency, improves safety, minimizes docking time, and enhances operational reliability in shipyards. NOMENCLATURE X - Coordinate along dock width (m) Y - Coordinate along dock length (m) d1, d2, d3, d4 - Measured distances from laser devices to vessel reference points (m) p1, p2, p3, p4 - Reference points on vessel A1, A2, A3 - Dock block reference positions 1. INTRODUCTION Dry docking is an essential maintenance activity for ships and offshore structures, allowing inspection and repair of underwater components. The process involves guiding a vessel into a flooded dock and positioning it accurately on docking blocks before the dock is dewatered. Traditional docking procedures rely on manual alignment methods including plumb lines, visual observation, and diver verification. These methods are time-consuming and susceptible to inaccuracies caused by human error. With increasing vessel size and operational complexity, shipyards require more precise and efficient docking systems. Automation and sensor-based technologies offer promising solutions for improving the reliability and accuracy of docking operations. 2. EXISTING DRY DOCKING PROCESS The conventional dry docking process involves arranging dock blocks according to the vessel docking plan. The dock is flooded to allow the vessel to enter and floatable guides assist in initial alignment. Operators rely on plumb lines and visual inspection to estimate alignment. Divers are often required to verify the vessel's position relative to the dock blocks. Although widely practiced, the process depends heavily on manual judgement. 3. CHALLENGES OF THE EXISTING DOCKING PROCESS The traditional docking process presents several operational challenges. Manual alignment can result in misalignment between the vessel hull and dock blocks, potentially causing structural stress or damage. The process requires extensive labour including divers and manual measurement teams. Docking operations also take longer due to repeated adjustments and verifications. Inaccurate docking can introduce safety risks and operational delays. 4. MULTI-SENSOR-BASED DOCKING SYSTEM SOLUTION The proposed system integrates laser projection technology and sensor-based measurements to automate vessel positioning during docking. The system operates using predefined coordinate mapping and real-time monitoring of vessel position relative to dock block locations. 4.1 DETAILED WORKING PHILOSOPHY Position Transfer and Reference Coordinates A coordinate system is established within the dry dock environment and a 3D model of the dock floor and block positions is generated. The coordinate system is transferred to the vessel reference frame enabling consistent alignment measurements during docking Laser Crosshair Projection Laser crosshair projectors project intersecting lines onto the vessel hull, identifying precise docking reference positions and assisting operators in vessel alignment. Sensor-Based Measurement System Laser sensors installed around the dock measure distances between predefined dock locations and reference points on the vessel.

Ship pipe routing is a complex task that requires placing pipes with diverse functions within limited spaces while simultaneously satisfying design regulations, efficiency, and space utilization. Designers repeatedly revise pipe routes until all requirements are met, considering pipe interferences, routing priority among pipes, and connection constraints. In this process, expert knowledge accumulated from past project data plays a critical role. For example, designers route pipes to follow ship structures such as decks and pillars, or group and align pipes with similar purposes. However, consistently applying this expert knowledge is time-consuming and cognitively demanding, as designers must simultaneously balance many competing constraints. To reduce this manual burden, prior studies have proposed automated pipe routing methods based on rule-based cost functions or heuristics. However, preferences in real-world pipe routing are often context-dependent and implicitly applied, making them difficult to represent consistently as rules or cost functions. As a result, incorporating expert knowledge typically requires iterative manual tuning, and methods that fit specific design cases may exhibit limited generalization to other cases. To address these limitations, this study proposed a learning based automated pipe routing method that incorporates expert knowledge. The proposed method employs a Transformer-based model to learn design tendencies from past project data that embody this expert knowledge, thereby internalizing it during the pipe routing process. This method enables the generation of pipe routes that implicitly capture expert knowledge, such as following ship structures and grouping and aligning pipe bundles, without manual rule crafting or cost-function tuning. To improve generalization, the model is trained on pipe routes collected from spaces with diverse equipment arrangements, promoting robust routing behavior under varying layout constraints. Given information about the target pipe and the surrounding space, the proposed model sequentially predicted pipe waypoints and generated pipe routes that sufficiently reflect expert knowledge. Overall, the proposed method effectively incorporated expert knowledge without manual intervention, thereby significantly improving the efficiency of the pipe routing process.

This paper presents a data-driven methodology for sizing shipyard facilities based on vessel type, production rate, and manufacturing strategy. The objective is to enable empirically grounded estimation of the physical plant required to sustain a specified naval or commercial shipbuilding program, or to assess what vessel types and production rates a constrained site can support. The work addresses a gap in the published literature: no publicly available quantitative method links vessel program requirements directly to facility design parameters using coefficients derived from a large, systematically collected dataset. The empirical foundation is a dataset of 310 shipyards across 38 countries, collected via agent-based web scraping spanning seven domains: facility area, build positions, production throughput, cranes, outfitting infrastructure, workforce, and geographic access. The yards were classified into six manufacturing strategy classifications based on how each yard builds rather than what it builds, distinguishing mega-yard block construction, covered-dock precision facilities, modular integration yards, distributed build networks, series production line yards, and repair facilities. Sizing equations were calibrated from this dataset using median statistics. Key equations address build position count, dock dimensioning, site area allocation, crane sizing, outfitting berths, and workforce. A library of 34 vessel types provides dimensional and cycle-time inputs. The methodology operates in two modes: Requirements-Based Sizing computes the facility needed to achieve a target production rate, while Site Capability Assessment inverts the equations to determine achievable rates given site constraints. The six manufacturing strategy classifications produce meaningfully distinct facility profiles: site area per build position ranges from 60,000 m2 to 450,000 m2, covered construction fractions from 15% to 75%, and dock occupancy fractions from 0.25 to 0.70. The finding that a yard's physical layout follows its manufacturing strategy, not its vessel product, is validated by cross-regional analysis. The sizing equations reproduce known configurations at benchmark yards, including a Korean mega-yard sustaining 40 ships per year across 10 build positions and a Japanese yard achieving 38 bulk carriers per year from a single dock with a four-month cycle. This work contributes an original dataset of 310 shipyard facility profiles and calibrated empirical equations indexed by manufacturing strategy and vessel type. The two-way sizing capability provides a structured approach applicable to greenfield planning, brownfield evaluation, and industrial base assessments. The manufacturing strategy classification offers a multidimensional alternative to traditional single-axis build-approach taxonomies, with direct application to shipbuilding capacity planning.

Programmatic decisions in naval vessel design can be understood as options: the right, but not the obligation, to commit to a decision at a predetermined expenditure on or before the expiration date. Valuing these options requires an approach that accounts for the interdependence of design decisions, the temporal resolution of information across disciplines, and the emergent uncertainties that create and destroy option value throughout the design process. Traditional financial option pricing relies on observable market prices and well-defined stochastic processes, neither of which exists in the design domain. This paper presents a valuation framework for design options that operates directly on the structure of the design process itself, using information-theoretic measures to quantify the value of design interventions under uncertainty. To address this challenge, we model the design process as a directed network using Influence Diagrams and Network Theory. Information Entropy serves as the unit of measurement for the resolution state of the network. Monte Carlo simulation propagates uncertainty through the network across thousands of possible design futures, producing empirical distributions of entropy trajectories from which option values can be computed. Each option, representing a structural change to the design process such as adding or removing activities, modifying dependencies, or shifting schedules, is valued by comparing the entropy trajectory when the option is exercised against a baseline where it is not. This counterfactual approach naturally captures cascading effects of delayed or incomplete information and emergent disruptions arising from cross-disciplinary dependencies. The valuation framework is demonstrated through a case study based on the design of an uncrewed surface vehicle, comprising 124 design activities and 181 knowledge dependencies across a biweekly epoch structure. Preliminary results indicate that option values are highly state-dependent, varying significantly across epochs as information is resolved. Notably, options that reduce network coupling (removing dependencies) consistently produce positive value, while options that increase network complexity (adding activities) can carry negative value at the system level despite local benefits to individual disciplines. These findings reveal a measurable tension between local and global perspectives in design management. This work contributes an original, computationally tractable method for valuing programmatic design options using network-based entropy measures rather than financial market analogies. The approach extends prior work on information-based ship design by treating design interventions as contingent claims on the information resolution process, enabling quantitative comparison of design management alternatives under identical uncertain futures.

Digitalization, although present for decades, remains the most prevailing trend across all industries with Industry 4.0 and Digital Twins (DT) focusing on optimizing human-system integration while improving operational efficiency. However, the transformation is not uniform across all industries. This disparity largely depends on the specific characteristics of a given industry, including the existing digital infrastructure, knowledge, adaptability, mental readiness, among others. Recognizing these individual characteristics is crucial to identify the steps and define a roadmap towards digitalization. Shipbuilding is a deeply traditional industry where the level of incorporating novel technologies has been gradual for the vast majority of shipbuilding facilities. Despite the significant work towards Shipyard 4.0, there exists a gap between academic research and practical implementation due to the numerous operational complexities. Existing studies predominantly focus on fully autonomous and digitalized smart yards but for the vast majority of facilities, this leap is hindered by high initial investments, lack of digitalized data, unclear implementation strategies, uncertain return on investments (ROI) metrics, etc. The paper intends to propose a comprehensive solution, 'Shipyard 4.0 Readiness Rating Framework' which would assess the digitalization maturity of shipyards. The framework evaluates yards based on multiple factors and core metrics such as the existing product data management (PDM) practices and the available infrastructure. Despite the varying unique constraints present in shipyards globally, the aim is to find a broader holistic starting point for yards at similar maturity levels. To validate the proposed framework, the study aims to incorporate empirical data obtained from industrial experience at a shipyard currently navigating a digital transition. By analyzing data across various yards and correlating with the readiness ratings, the paper intends to demonstrate the practical capabilities of the framework. This aims to provide shipyards with feasible, ROI justified strategies for adopting digital tools and having a starting point such as transitioning from a PDM system to a Product Lifecycle Management (PLM) system before attempting complex DT or AI integrations. This approach aims to standardize the Shipyard 4.0 transition, offering a step-by-step blueprint for yards at different levels of digitalization to build intelligent, efficient and interconnected production workflows.

Ship design drawing check is a critical process in shipbuilding; however, it still relies heavily on manual visual check. This dependence increases the risk of overlooking defects, which may lead to significant losses. Therefore, manual check alone has inherent limitations. This study aims to develop an AI-based design drawing check support system for ship hull drawings that can detect defects, classify their types, and assist in proposing corrective actions. Shipbuilding is a made-to-order industry in which design requirements differ for each ship owner, resulting ship structural drawings that contain diverse and complex design information. To address these challenges, this study proposes an AI-based framework based on three fundamental aspects of design drawing check: "Where" defects are located, "What" types of defects they are, and "How" they can be corrected. For the Where drawing check model, a color-mapping method was developed to visualize the likelihood of defects within drawings. The drawings were first divided into meshes to extract structural features, and k-means clustering was applied to group regions with similar characteristics. Convolutional neural networks (CNNs) were then applied to each cluster to detect potential defects. This cluster-based approach improved detection performance, achieving an average accuracy of approximately 80%. For the What drawing check model, a CNN-based classifier was developed to identify defect types within sections that include meshes detected by the Where model. Forty types of defect labels, identified as important by designers, were defined. Training data were constructed using a cluster-based structure consisting of primary clusters based on section features and secondary clusters based on defect labels. The model outputs the top three most probable defect types for each section. The results show that the correct defect type is included within the top three predictions with a probability of 98%. For the How drawing check model, a retrieval-based system was developed to support defect correction. For sections identified as defective, the system searches past drawings to retrieve structurally similar sections along with comments from design reviewers, thereby providing suggestions for corrective actions. A prototype user interface integrating these models was implemented, and case studies were conducted to evaluate the effectiveness of the proposed system. In Japanese shipyards, the design drawing check process can sometimes account for approximately 50% of total design effort. The proposed AI-based system demonstrated sufficient accuracy for practical use, and its implementation is expected to significantly reduce design workload while improving design quality.

Vessel performance monitoring remains an essential yet resource-intensive aspect of performance optimisation, typically constrained to discrete sea trial events. Traditional methods are largely reliant on manual data collection, human-driven post-processing, and manual plotting - workflows that only offer fragmented assessments of vessel performance. Once delivered, operators have limited visibility into the true manoeuvring capability of their vessels, leaving performance potential unrealised. This paper presents the application of an automated, real-time vessel performance intelligence system deployed aboard a series of high-speed patrol boats, demonstrating how a new paradigm for data-driven performance evaluation can transform operational technique and drive sustained performance optimisation. The system leverages onboard sensors alongside plug-and-play instrumentation to continuously capture propulsion, navigation, and motion data - including vessel speed, position, heading, and control inputs. This data is processed in real time to perform automated, multi-variable analysis grounded in a naval architecture core and aligned with IMO manoeuvring trial procedures. Insights are generated through the use of six standard trial types: turning circle, coursekeeping, acceleration, stopping, zig-zag, and speed runs. Sea trials form the performance baseline, with continuous in-service monitoring extending these insights throughout the vessel's operational lifecycle. The system outputs high resolution, time synchronised performance data instantaneously upon completion of each manoeuvre. The reports enabled a deeper understanding of real world vessel behaviour, including a detailed analysis of angle of attack variation over time and speed profiles across turns. These insights drove the development of tailored driving techniques, unlocking tighter achievable turning circles of up to 46%. The resulting performance intelligence was subsequently leveraged to establish a more consistent and teachable vessel handling protocol, translating data-driven manoeuvring insights into institutional knowledge. This work demonstrates the integration of automated, instantaneous computation with established naval architecture principles to deliver continuous performance intelligence. The paper contributes original field data and a methodology applicable across all vessels, showing how innovative data capture and advanced analytics can inform both operational technique and structured training curricula.

The primary goal for conducting dynamic interface analysis is to expand existing operating envelopes and increase ai and sea vehicle availability thereby improving overall naval effectiveness. The design and configuration of ship platforms can both enhance or detract from these objectives. Validation results are discussed and compared in confirming the tendency of certain ship design parameters being impacting Vehicle deck limits with the actual at-sea result. The primary objective of this field of study is to determine the feasibility of applying full motion simulators and plug and play simulations in support ship design criteria. A typical simulation consists of modular plug and play components and is assembled into a High Level Architecture (HLA) run system with the components describing the ship environment coupled with the ship's response mechanisms as described by computed transfer functions. The use of 6 degree-of-freedom motion flight simulator to forecast physical deck motion and deck limits, is discussed. The robotic and piloted systems studied focused on specific simulated vehicle-ship interface responses to various environmental stimulate. At-sea validation study results are displayed and discussed comparing the simulated scenario with that recorded during the sea trial. Recorded and analysed results are presented in relation to the deck interface stability problem encountered by, for example, a helicopter on recovery to a moving deck. Using simulation methods, the vehicle model was exposed to progressively difficult deck environment conditions. A brief synopsis of several of the integrated HLA modules representing various stages in evolutions on the deck, are presented. The computational method employs sufficient performance criteria and correlates well with forecasted Quiescent windows of deck and wind motions.

This report summarises the ship-based Quiescent Period Prediction (QPP) computational airwake methodology and corresponding simulation results. Quiescent Period Prediction (QPP) is a class of technology that use measured remote air and sea surface profile measurements to predict the future air and wave forces acting on a vessel in space and time. Such technologies have prediction time horizons measured in minutes and offer predictions of the actual environment of a target vessel. The objective of deploying QPP is to safely launch and recover (L&R) air and sea vehicles on a moving vessel within reasonable safety margins regardless of the seaway. The purpose of the present study is to focus on the predicted simulation results of the airwake portion of the QPP device. Specifically, the ship-based test and evaluation (T&E) system covered by this study concerns the transient prediction of arriving winds structure at a vessel. A key objective being the simulation of Quiescent Periods (QPs) where for short periods the arriving wind is typically lower than the prevailing average values. The fundamental prediction calculation is based on the arrival time of wind air flow including certain types of gust structures. The main results and corresponding consequences are summarized: 1-the computational cost of the real-time prediction cycle was significantly lower than originally anticipated. This was owing to the efficiency of the numerical propagation algorithm developed for this project. 2-For a typical operational prediction of two minutes, it was only necessary to explore the size of the timestep interval. It was shown there was no strong sensitivity to the timestep parameter so there was little benefit to optimise it. 3-A study was undertaken to explore the benefits of Closed Loop adaptation based on a running average value of the start and stop of Quiescent Period prediction. Given the measured values of the arriving wind, the running averages of the errors in the predicted start and duration of the Quiescent Period were determined. The results show clear benefits from the real-time trend-correction process.

Ensuring the quality of welded connections in shipbuilding and offshore structures is crucial for maintaining safety and meeting the standards set by classification societies. Radiographic testing (RT) is one of the main non-destructive testing (NDT) methods used to check for weld defects. However, traditional inspection methods often rely heavily on manual analysis. This process can take a lot of time, require much effort, and be vulnerable to mistakes. Additionally, because of disjointed and manual documentation, ship surveyors face a higher risk of misinformation and inefficiencies in operation. This paper presents a decision-support system, MARINA (Maritime Regulatory Inspection and NDT Assessment), for automated radiographic weld evaluation and regulatory compliance. The system converts analogue RT films into high-resolution digital formats and integrates metadata from the ship's Shell Expansion Plan to enable spatially referenced inspection records within a digital-twin framework. A hybrid two-stage machine learning architecture is employed, combining YOLO-based defect localisation with a convolutional neural network (CNN) based classification using domain-specific features. Calibrated pixel-to-physical scaling ensures reliable defect sizing, while image pre-processing improves detection consistency. Experimental evaluation on a curated dataset of radiographic weld images achieved a defect localisation mAP@0.5 of 94.2% and a classification accuracy of 96.1%, providing reliable inputs for downstream regulatory assessment. The novelty of this research lies in the introduction of an Automated Regulatory Compliance module. This module encodes the acceptance logic of ISO 5817:2015, IACS Unified Requirements W32 and W33, and ASME Boiler and Pressure Vessel Code Sections V and VIII. This capability allows for automated, clause-referenced acceptance decisions, providing transparent justifications for surveyors. An explainability interface and a human-intervened backtracking mechanism support adaptive learning and performance refinement. Validation against expert-reviewed inspections demonstrated an automated compliance agreement of 92.8% with certified surveyor decisions. By creating a structured digital workflow and enhancing regulatory traceability, MARINA improves inspection reliability, operational efficiency, and professional learning. It contributes to safer, digitally enabled shipbuilding practices.

The development of new digital solutions and of the Artificial Intelligence (AI) algorithms are shaping new ships-related challenges and opportunities, leading to the birth of the so-called Shipping 4.0. This new paradigm consists of creating a new maritime space, where all the main subjects (ships, harbours, cranes, power sources etc.) are interconnected, realizing a cyber-space where the shared information make easier a correct and intelligent management of all the navigation phases, increasing the safety and the security of the maritime field. The proposed work aims to give its novel contribution to this challenging topic proposing, for the first time, the development of a cyber-physical system for marine power sources. It consists of a Remote Monitoring and Management System where all the interconnected marine engines send information then processed by AI algorithms located into a Cloud. Then, the AI functionalities process the received information extrapolating important information on how the engine is working; furthermore, the developed system has also prescriptive, predictive and communicative capabilities. In fact, it does not provide only information on the current state-of-the-health of the engine but also suggests working conditions more suitable in case of anomalies, predicts the upcoming failure, and, finally, communicate with the onshore operators if something wrong is detected. In this way, the engine will be always protected and covered by the maintenance team, avoiding unexpected loss of power and guaranteeing quick intervention, increasing the safety of the navigation. The novelty of the proposed work does not rely only in the development of this cyber-physical system, but also in the deep phase of study necessary for realizing a solution suitable for the present and the future of the marine power sources. In fact, the high versatility is its biggest advantage, making its adoption easily for marine Diesel engine, hydrogen engine, electrical batteries, and Fuel Cells. To obtain this big advantage, a deep study has been conducted for each asset, identifying the most meaningful sensors measurement for maintenance purposes and correlating them with the most common failures. This operation allows also to reduce the computational cost of the data to be processed, reducing the amount of measurements for the information-extraction phase. In summary, the proposed work gives its contribution in the digitalization of the maritime sector, proposing the development of a Remote Monitoring and Management systems, and reviewing the current and future marine power sources with the aim to obtain a comprehensive overview of the most common failures and of the sensors measurement able to detect them.

Damage stability assessment is a critical component of ship design and safety verification, particularly for passenger and cargo vessels where survivability, functionality, and spatial constraints must be balanced. The process of evaluating damage scenarios typically involves generating large numbers of flooding cases and iteratively assessing vessel response under prescribed conditions. In parallel, the placement and optimization of bulkheads and compartment arrangements present complex trade-offs between stability performance, usable space, operational requirements, and aesthetic considerations. These factors contribute to a lengthy and computationally intensive design process. This paper presents a computational approach for rapid damage stability assessment that uses direct calculations on three-dimensional ship geometry and focuses on the most relevant damage cases, enabling faster design evaluation. The method operates directly on triangulated ship models and simulates flooding scenarios by modelling compartmental damage and resulting buoyancy redistribution. Damage cases are prioritised based on established regulatory guidelines (e.g., SOLAS damage case definitions) combined with simplified heuristic filtering, reducing the need for exhaustive enumeration of all possible scenarios. Numerical equilibrium is determined through enforcement of displacement conservation and force-moment balance, enabling estimation of vessel response under damaged conditions. Initial results are compared qualitatively with expected stability trends from conventional design tools to verify physical consistency. The framework is designed to integrate with parametric ship design environments, allowing hull forms and compartment layouts generated in CAD-based tools to be evaluated iteratively for stability performance. This enables stability considerations to influence design decisions earlier in the development process, rather than serving solely as a downstream compliance check. By enabling comparison of alternative bulkhead configurations and compartment layouts, the approach supports more informed trade-offs between survivability, usable space, and design constraints. The primary contribution of this work is the development of an extensible computational platform for damage stability analysis that combines first-principles hydrostatics with targeted scenario evaluation. In particular, the framework enables systematic exploration of bulkhead number and positioning by allowing alternative compartment arrangements to be evaluated rapidly for their impact on stability performance. This supports informed trade-offs between survivability, usable space, structural constraints, and design considerations such as weight and material cost. By enabling stability-driven iteration of compartment layouts within the design loop, the proposed methodology contributes toward a more integrated and efficient ship design process.

Shipbuilding programs increasingly begin with third party designs delivered in multiple CAD authoring formats, which are then evolved through shipyard specific standards, systems, and constraints. This creates a core requirement for modern product lifecycle management (PLM): PLM and downstream engineering systems must be CAD agnostic, able to ingest and govern CAD content without forcing redesign, and support established shipyard workflows with minimal disruption to downstream stakeholders. This paper presents practical steps for enabling multi-CAD within a shipbuilding-specific PLM environment. It focuses on building and maintaining a stable product structure that is independent of any single authoring tool, while still preserving clear traceability to the source models. It outlines how to define and protect the right levels of granularity (part, assembly, and SWBS aligned structures) so that change control, production planning, and supply chain activities can be executed consistently across programs. The paper also describes approaches for capturing, mapping, and normalizing key attributes so they remain queryable and traceable across tools and lifecycle stages. These attributes include classification, material, weight, approvals, identifiers, and other metadata needed for downstream engineering, analysis, and reporting. Special attention is given to how attribute governance reduces rework and prevents loss of intent when information moves between design, planning, and production phases. Configuration management is addressed as a central risk area in multi-CAD environments. The paper explains methods to keep configuration rules, baselines, and revisions consistent as models transition from concept to detail design, fabrication, and construction. It also highlights common multi-CAD pitfalls, such as limited metadata, inconsistent identifiers, attribute fragmentation, and uncontrolled derivative outputs, and provides mitigation strategies that can be implemented without disrupting active programs. Real-world business drivers frame the technical approach, using SSI ShipbuildingPLM as an illustrative example, including the need to reuse proven designs across regions and yards while accommodating local standards. The outcome is a shipbuilding-focused digital thread that leverages existing information, supports shipyard specific outlines, and improves traceability from design through manufacturing and change.

Building a Digital Shipyard for next-generation naval platforms is not primarily a technology challenge - it is a challenge of structuring collaboration, governance, and scalable integration across complex system landscapes. This paper presents the strategic partnership between ASC Pty Ltd and PROSTEP AG as a practical case study in implementing systems engineering capabilities within the AUKUS-class submarine program. ASC as system owner defines the digital vision, governance framework, and target architecture, including the transition to Teamcenter as the new PLM environment and the adoption of model-based systems engineering. The integration of heterogeneous systems such as AVEVA (CAD) and IFS (ERP) introduces significant complexity that cannot be addressed by traditional point-to-point integration approaches. PROSTEP as integration enabler provides OpenPDM as the backbone of a federated integration architecture and translates strategic objectives into an operational integration landscape. The collaboration is characterized by clear role allocation, shared architectural responsibility, and a structured enablement approach that progressively transfers integration capabilities to ASC. The paper reflects on key insights gained during implementation so far. It shows that establishing a strategic integration platform early is essential to prevent fragmentation and manage long-term complexity. Equally important are clear governance structures and role definitions, which enable efficient decision-making and scalable growth of the integration landscape. A phased implementation approach, delivering early and tangible value, proves critical for reducing transformation risk and building organizational confidence. Finally, sustained capability transfer emerges as a prerequisite for ensuring long-term independence and continuous evolution. Initial results, including PLM-ERP integration and the introduction of a central data integration layer, demonstrate measurable improvements in data consistency, process efficiency, and transparency. The case study provides practical guidance for organizations aiming to operationalize systems engineering in complex naval programs.

During ship piping installation, inserting pipes are often required to connect separate pipe sections due to hull assembly deviations, equipment installation errors, and spatial constraints in the construction site. Conventional inserting pipe design mainly relies on manual measurement and experience-based calculations. This approach is time-consuming and often fails to ensure sufficient design accuracy when dealing with complex spatial configurations, which may lead to rework and assembly difficulties. Therefore, developing a method that can rapidly acquire on-site spatial information and automatically generate inserting pipe designs is of great significance for improving the efficiency of ship construction and maintenance. To address this issue, this paper proposes a parametric design method for ship inserting pipes based on a light-pen measurement system. First, the spatial pose and dimensional information of the flanges at both ends of the inserting pipe are obtained using the light-pen measurement system. The collected feature-point data are then processed to reconstruct the spatial position and orientation of the two flanges. Based on the coordinates of the flange centers and the relationship between their normal vectors, a mathematical model of the inserting pipe center path is established, enabling a parametric representation of the pipe route. Furthermore, by analyzing the spatial relationship and geometric constraints between the two flanges, different types of inserting pipes are automatically classified, and corresponding path models are constructed to determine the pipe route automatically. Next, the spatial relationships and geometric constraints between the two flanges are analyzed to automatically classify different types of inserting pipes. Corresponding mathematical models are then constructed to solve the pipe path for each type. On this basis, a parametric design program for ship inserting pipes is developed using C#. The program accurately reconstructs the flange positions from the measured pose data and automatically generates a pipe model that satisfies the design requirements, enabling automated and visualized inserting pipe design. Furthermore, the measured flange pose can be reproduced in the workshop using a numerical reproduction device, allowing the inserting pipe to be fabricated before installation. This enables the pipe to be installed successfully on site without repeated adjustments. Application results show that the proposed method improves both design efficiency and modeling accuracy, providing reliable technical support for the design and fabrication of ship inserting pipes and contributing to improved efficiency and quality in ship piping installation.

Surveyors of classification societies review shipyard drawings during the design stage to ensure safety and regulatory compliance. When revisions are required, the surveyors issue comments, and the designers modify and resubmit the drawings; this process is repeated until all requirements are satisfied. In practice, designers are expected to mark each revision with symbols such as revision clouds; however, these marks may be unintentionally omitted. Moreover, iterative revisions may lead to additional changes occurring beyond the marked areas, making the revision markings incomplete and less reliable. As a result, surveyors typically must rely on direct visual comparison of the original and revised drawings to verify the revisions, which is a significant challenge. Particularly when the size of drawings is vast, it not only consumes immense time and effort but also carries a high risk of overlooking revisions or making errors in judgment. In addition, repeated scanning or exporting during resubmissions can introduce non-revision inconsistencies that further obscure true changes. Therefore, this study proposed a method for automatically identifying revisions in ship drawings using deep learning-based change detection technology. To overcome the limitation of acquiring a large number of real revised drawings, the proposed method introduced a technique utilizing image inpainting and segmentation from the field of computer vision to generate synthetic revised drawings from original drawings. Specifically, we defined revisions as three representative types—addition, removal, and replacement—and generated synthetic revisions that closely resemble real revision patterns by reflecting these types. By generating numerous revised drawings with natural modifications applied to the original, we effectively constructed a synthetic training dataset. The change detection model, trained on the synthetic training dataset, rapidly and accurately identified revisions in actual drawings and presented them to the surveyors. Consequently, the surveyors could perform rapid, intensive reviews centered on the identified areas, significantly enhancing the efficiency of the iterative drawing review process.

In recent years, the strong demand for highly efficient operations in ocean studies and offshore development has driven a trend toward designing research and offshore support vessels as multi-purpose platforms. Deep-sea studies require next-generation vessels capable of concurrently operating multiple survey systems, such as autonomous underwater vehicles and human-occupied vehicles, within a single voyage while minimizing crew size. However, such vessels, which integrate diverse and concurrent survey operations, remain largely unprecedented. This introduces complex and tightly coupled challenges, including human resource constraints, spatial limitations, and uncertainties in operational processes. Conventional planning approaches, which rely heavily on empirical knowledge and rules, are insufficient to systematically capture these interdependencies. Moreover, approaches to predicting concurrent multi-system operations in detail, thereby proactively visualizing and quantifying operational issues, remain largely unexplored. To address these challenges, this study proposes a model-based operational evaluation framework using discrete-event simulation. The proposed framework integrates multiple layers of shipboard operations including survey systems, crew members' skillsets, individual crew tasks, workflows, and spatial constraints of onboard facilities, within a unified simulation environment. By explicitly modeling the interactions and resource couplings among these elements, the framework enables the reproduction of complex and concurrent deep-sea operations under realistic conditions. The simulation generates quantitative time-series outputs, such as system utilization and crew workload, based on predefined operational scenarios. Furthermore, an iterative evaluation process is established, in which bottlenecks identified from simulation results are used to refine operational plans and system configurations. As a case study, concurrent operations of multiple deep-sea survey systems were modeled based on actual deep-sea survey operational data. The results reveal quantitative trade-offs between crew composition and mission duration under varying operational and spatial conditions, providing insights into optimal operational configurations. The proposed framework provides a computational foundation for the systematic evaluation of next-generation deep-sea survey vessels under complex constraints. It further supports data-driven decision-making in the early-stage ship design process by linking operational feasibility with design parameters.

In shipyard block operations, accurate estimation of block position and orientation is essential for improving the efficiency and safety of assembly, transportation, and erection processes. In particular, the 3D pose of blocks can serve as a key input for various systems, including assembly monitoring, dimensional quality control, AR-based drawing visualization, production progress assessment, and crane operation support. Although a variety of sensors, such as distance sensors, total stations, 3D laser scanners, depth cameras, and monocular cameras, have been used for pose acquisition, depth-sensing approaches are often limited in real-time field applications due to accuracy degradation under varying measurement conditions and their low interpretability for workers. Monocular RGB images, by contrast, offer advantages in terms of real-time processing and practical applicability in field environments. Nevertheless, existing RGB-based block pose estimation studies still suffer from performance degradation under illumination changes and in the presence of occlusion and truncation. To address these limitations, this study proposes a monocular camera-based 6D pose estimation method for monitoring large shipyard blocks. Rather than relying on conventional feature extraction techniques such as Scale-Invariant Feature Transform (SIFT) and Line Segment Detection (LSD), the proposed method employs a keypoint-based voting network that is robust to changes in brightness, occlusion, and truncation, enabling stable feature extraction even when shape information is partially missing. In addition, to overcome the difficulty in obtaining large-scale training data prior to actual block production , an automated synthetic dataset generation pipeline is developed based on 3D CAD models and a game engine-based virtual environment. The pipeline automatically generates RGB images, masks, and pose annotations, while also incorporating occlusion and truncation scenarios into the training data to improve robustness under complex working conditions. Furthermore, domain randomization that reflects variations in brightness, shadows, backgrounds, and sensor noise is applied to enhance generalization performance, and diffusion model-based domain adaptation is introduced to mitigate the domain gap between synthetic and real-world data. The proposed method provides a shipyard-specific framework for 6D block pose estimation and demonstrates its potential for integration into block monitoring systems and digital twin-based operational support systems.

The maritime industry is under increasing pressure to transform its fleet operations in alignment with the International Maritime Organization carbon neutrality target for 2050. However, shipowners face significant uncertainty in fuel availability, infrastructure maturity, and regulatory trajectories when making long-term investment decisions. Despite these challenges, decision-support methodologies capable of addressing such compound uncertainties remain insufficiently developed. This study proposes an agent-based simulation model combining pre-defined scenario inputs with dynamically evolving market variables to support strategic fleet composition decisions for shipping companies navigating this fuel transition period. The model integrates three layers: fuel supply, market demand, and decision-making agent. The supply layer integrates port fuel infrastructure, production regions, and transportation networks, representing cost, availability, and supply stability of five alternative marine fuels: hydrogen, ammonia, methanol, biofuel, and LNG. The demand layer captures fuel demand in the maritime sector, shaped by competitor fleet dynamics across the same five fuel types, together with regulatory scenarios. The decision-making agent represents the focal shipping company and updates fleet composition ratios at each time step based on an objective function that incorporates total cost of ownership (TCO), carbon reduction potential, and supply chain stability. Simulation results based on five regulatory pathways aligned with IPCC Shared Socioeconomic Pathways (SSP) scenarios indicate that fleet transition trajectories depend not only on regulatory intensity but also on the maturation speed of supply infrastructure and the asynchronous nature of market adoption. Intermediate pathways exhibit delayed diffusion of low-carbon fuels, leading to prolonged mixed-fuel strategies. Stronger scenarios generate nonlinear transitions with abrupt fuel shifts, while weaker conditions are associated with slower transitions and continued reliance on conventional fuels. Furthermore, fleet composition trajectories vary significantly depending on the relative weighting of cost efficiency, carbon reduction, and supply stability within the agent's objective function, highlighting the importance of firm-level strategic preferences. This study proposes a simulation framework that captures supply chain heterogeneity and multi-fuel infrastructure dynamics while incorporating competitive market dynamics. The framework complements conventional static optimization approaches and data-driven statistical models by enabling systematic exploration of fleet transition pathways under data-scarce and rapidly evolving regulatory conditions. As such, this study provides actionable insights for shipowners on investment timing, fuel diversification, and transition risks under different regulatory environments, while serving as a practical decision-support tool for fleet managers and maritime policymakers, as demonstrated through case studies of five regulatory pathways.

In recent years, 'weather routing (WR)' has attracted increasing interest as a means of reducing costs and environment impacts through optimum voyage planning that takes meteorological and oceanographic conditions into account. To achieve high-quality WR service, it is essential to accurately predict the ship's speed over ground (SOG) during the voyage based on the ship's control variables and forecast meteorological and oceanographic information. However, with current technology, the accuracy of measured ship's speed through water (STW) and the accuracy of predicted ocean currents are insufficient. Consequently, this has a detrimental effect on the prediction of ship speed over ground, which is the sum of the ship's speed through water and the drift component of the ocean currents. In this study, we propose a method for accurately determining STW using drift data automatically collected from ships that regularly operate on the same route, such as ferries. We also propose a machine learning method for predicting SOG by considering collected previous voyage's drift data, the accurately determined STW and collected past operational data. To validate the effectiveness of the proposal method, a domestic ferry in Japan was applied. Proposed method can improve the accuracy of STW, leading to improved accuracy in estimating ship propulsion resistance in actual sea conditions. Consequently, this is expected to facilitate the monitoring of changes in ship performance over time and improve the accuracy of evaluation of the energy-saving effects of corrective action for energy saving such as WR. It is considered useful for ship operation management. Improving the accuracy of STW can enhance the accuracy of estimating ship propulsion resistance in actual sea conditions. This is expected to facilitate the monitoring of changes in ship performance over time and improve the accuracy of evaluating the energy-saving effects of WR using energy efficiency indicators; therefore, it is considered useful for ship operation management.

Growing demand in naval and commercial shipbuilding is increasing the need for more efficient, data-driven processes to boost regional shipbuilding capacity. This paper introduces a computational framework that fills a gap in current academic research: how to use a data-informed approach to guide future decisions on shipyard policies across regions. This framework develops a site-identification method for greenfield, brownfield, and active shipyard locations, linking them with manufacturing strategies and vessel types to shape future capacity-focused shipbuilding strategies. The framework is based on site-identification data collection methods for greenfield, brownfield, and active shipyards. Greenfield sites are identified through an automated shoreline segmentation algorithm that divides a coastline or waterway into candidate segments. Each segment is evaluated against pass-fail criteria including terrain elevation, land cover, proximity to rail access, and regional workforce availability, using globally accessible datasets like NASA SRTM, land cover databases, and OpenStreetMap. Brownfield and active shipyard locations are determined through analysis of manufacturing strategies. Data gathered via agent-based web scraping from shipyards worldwide is used to establish relationships that characterize active shipyards. These yards are classified into manufacturing strategies such as mega-yard block construction, covered-dock facilities, modular-integration yards, distributed-build networks, series-production lines, and flexible repair and conversion yards. A library of vessel types is then mapped to compatible strategies, based on vessel size, cycle time, and maximum block weight. This classification supports the process of identifying active and brownfield sites and explores the relationships between manufacturing strategies and greenfield layouts. Once shipyard sites are identified, an interactive layout design tool allows planners to arrange component parameters—including steel fabrication shops, block assembly halls, blast and paint facilities, staging areas, and ship-lift infrastructure—within a site boundary. Component sizes are calculated from regression equations calibrated with the global shipyard dataset. A site capability assessment then estimates vessel production rates for each layout, considering build locations, quay length, waterway shape, and workforce availability. The results provide a ranked list of candidate sites designed against constraints. Existing, reactivatable, and new sites are matched with a shipyard layout and production capacity estimates, offering planners a data-driven basis for developing new shipbuilding strategies at the national or regional level.

Warships are becoming more outfit-dense, networked, and precision-manufactured, yet the production strategies many nations use to build them have not kept pace with that shift. The prevailing instinct to benchmark naval shipyard performance against the high-volume yards of Korea and China conflates steel throughput with production effectiveness in ways that systematically underserve warship programs of other nations. For nations that lack the industrial scale to replicate those models, and for programs defined by outfitting complexity rather than hull volume, a different production logic is needed. This paper argues that a distributed, microfactory-based approach to block outfitting, exemplified by Irving Shipbuilding in Halifax and BAE Systems in Glasgow, offers a more practical and scalable path for nations that cannot and should not replicate the Korean or Chinese volume model. European cruise shipbuilders have been successful in solving a distributed outfitting problem similar to that of a warship. Analogous to cruise ships, warship production is dominated by outfitting complexity, design instability, and late-arriving government-furnished equipment. The correct production objective is not maximum steel throughput, but a hybrid strategy of maximum upstream outfitting completion under controlled shop conditions and segmented block construction. The paper proposes the block microfactory as the physical and managerial embodiment of that objective and develops an outfit dense production concept based on replication, resilience, and incremental expansion rather than monolithic shipyard growth. Existing analyses of U.S. defense shipbuilding delays have focused predominantly on workforce shortages, funding instability, and program management failures. Instead, this paper locates the root cause of schedule growth in the physical and organizational architecture of U.S. naval shipbuilding network, a production-systems explanation largely absent from both the policy debate and the engineering literature. By introducing the block microfactory concept and grounding it in both the Halifax and Glasgow reference cases, and European cruise production practice, the paper provides shipyard planners and international naval acquisition officials with a concrete, evidence-based alternative to the legacy campus model, bridging a gap between industrial engineering theory and U.S. defense shipbuilding practice.

Ship performance evaluation is an essential procedure for verifying compliance with shipowner requirements and enhancing operational safety and efficiency. However, conducting sea trials to evaluate the performance and verify compliance is often impractical for repeated verification. This is primarily due to substantial costs, time constraints, and environmental limitations that often hinder consistent data collection. To address these challenges, this study proposed a method for virtual modeling and performance prediction of a ship to replace sea trials. To achieve this, a virtual ship was modeled by emulating its major elements (e.g., power and propulsion systems, control system, and sensor system) in detail. Furthermore, a virtual marine environment was modeled by replicating the characteristics of a real marine environment. During this process, the virtual ship and marine environment incorporated environmental loads induced by various sea conditions, such as wind, waves, and currents. In particular, these environmental loads were accurately reflected by utilizing actual oceanic environmental data to ensure the reliability of the modeling method. This method allowed the virtual ship to replicate the complex physical responses of a ship under the sea condition. Subsequently, the performance of the virtual ship, including resistance, FOC (Fuel Oil Consumption), and maneuvering performance, was predicted in accordance with ISO 15016, the international guidelines for speed and power performance analysis of ship based on sea trial data. To validate the proposed method, ship performance was predicted for a ship operating along a specific path under various sea conditions, and the resulting variations in ship performance were analyzed. The results confirmed that the proposed method effectively predicts ship performance. Also, it is possible to significantly reduce reliance on costly sea trials by adopting this method. Simultaneously, it provides a reliable and scalable means of evaluating ship performance in a virtual marine environment.

Most traditional shipbuilding CAD systems focus mainly on modeling, without including analysis features. This means designers have to use separate commercial analysis tools after completing the model. In this process, they often need to re-model the 3D model or convert it into a neutral format before importing it into the analysis tool. This can be very time-consuming and effort-intensive for designers, and there's also a risk of errors or mistakes during the conversion or re-modeling steps. This study aims to address this issue by integrating a structural analysis solver directly into the CAD system. The focus is on analyzing structures like pipe supports, cranes, hoists, platforms, and electrical seats. The approach involves extracting the necessary geometric data and conditions from the 3D model, running the analysis using the built-in solver, and then visualizing the results on the 3D model while generating detailed reports for designers. For steel sections, elements and nodes are separated for each part, and for symbolic models, symbol information is analyzed to separate elements and nodes. Machine learning technique is used to organize the data, which is then used for analysis. The results show that designers can now perform structural analysis on their own, without needing external tools or help from research institutes. This not only saves a lot of time compared to the old methods but also helps ensure the structural stability of the models. Additionally, the system provides immediate feedback during the design process, which greatly improves design efficiency. Future work will focus on expanding the analysis capabilities to cover more types of structures and creating a more user-friendly interface to make it easier for designers to use. There are also plans to integrate other types of analyses into the CAD system, beyond just structural analysis. The goal is to bring innovation to the shipbuilding design process and make the industry more competitive.

Optimization research using deep reinforcement learning has been actively conducted in the ship design field recently, but existing studies have been limited in accurately reflecting complex flow characteristics as they mainly use simple resistance estimation models based on empirical equations. To overcome this, this study proposes a novel hull form optimization architecture that simulates the expert design process by combining a Proximal Policy Optimization algorithm and a U-Net-based high-precision surrogate model. This framework effectively controls the degree of freedom of hull forms by defining the latent space of the generative model as an action space of reinforcement learning. Toward this end, this study proposes Auto-Encoded-DeepSDF, an implicit neural representation framework for generating three-dimensional hull forms with diverse topologies. The model combines a variational autoencoder-based latent representation with a signed distance function network, enabling direct shape encoding from surface point clouds without iterative optimization. A dataset of 1,247 synthesized hull forms, including commercial ships, a trimaran, and a submarine, is used for training. The learned latent space shows strong correlations with key hull parameters, indicating physically meaningful representations. Smooth, watertight surfaces are reconstructed using an adaptive marching cubes method, and continuous shape variation is achieved via latent interpolation. The results demonstrate a topology-agnostic, data-driven approach for hull form generation and design optimization. In the environmental configuration stage, the pressure and shear stress distribution information on the hull surface are trained and then predicted by U-Net as well as simple resistance values. This can drastically reduce the computational cost of numerical analysis while providing a high-fidelity evaluation mechanism close to the conventional process in which the designer directly judges the physical distribution of the computational fluid dynamics analysis results and modifies the hull forms.

The growing demand for versatile marine platforms capable of operating across multiple domains has motivated the development of vehicles that combine surface and underwater capabilities within a single hull. This paper investigates the hydrodynamic design of an Autonomous Underwater-Surface Vehicle (AUSV), a novel dual-mode platform that integrates the functions of an Autonomous Surface Vehicle (ASV) and a Remotely Operated Vehicle (ROV) into a single 6-metre displacement hull. The study focuses on the central design challenge of AUSVs: balancing the conflicting hydrodynamic requirements of surface and submerged operation, where hull forms favourable for reducing wave-making resistance at the free surface may not be optimal for minimising viscous resistance underwater. To address this challenge, a systematic CFD-based parametric study is conducted to evaluate the influence of key hull-form parameters on hydrodynamic characteristics across both operating regimes. Numerical simulations are performed in ANSYS Fluent by solving the Reynolds-Averaged Navier-Stokes (RANS) equations with the k-ω SST turbulence model. For surface operation, a Volume of Fluid (VOF) multiphase approach is adopted to resolve the free surface, and resistance is assessed in surge. For submerged operation, a single-phase formulation is used to examine hydrodynamic resistance in surge, sway, heave, and yaw. A structured design-of-experiments framework is used to explore the parameter space, while mesh-independence studies are conducted in accordance with the ITTC Recommended Procedures and Guidelines to ensure numerical reliability. The study identifies hull-form trends and parameter combinations that provide favourable resistance trade-offs across both surface and underwater modes. The results highlight the extent to which conventional single-domain optimisation principles must be reconsidered for cross-domain platforms and provide practical guidance for the preliminary design of AUSVs. By establishing a CFD-based framework for evaluating dual-mode hydrodynamic performance, the paper contributes new insight into the design compromises and opportunities associated with marine vehicles that must operate efficiently both at and below the free surface.

Accurate numerical prediction of hull-propeller interaction and cavitation phenomena remains a critical challenge in computer aided ship design, particularly when transitioning from isolated component analyses to integrated, scale consistent propulsion simulations. The primary objective of this paper is to assess and consolidate a computational methodology capable of predicting propeller performance, cavitation patterns, and hull-propeller interaction effects across both model scale benchmark conditions and full scale self propulsion scenarios. The scope of the study combines two complementary investigations conducted within the framework of the Wageningen 2025 workshops. The first focuses on the Nawigator XXI benchmark case, where a systematic progression is applied from open water propeller simulations to nominal wake and in behind ship configurations under cavitating and non cavitating conditions. The second addresses the full scale JoRes1 tanker, using an unsteady self propulsion approach to resolve propeller loading, delivered power, and interaction with a complex stern arrangement, including rudder and appendages. The methodology relies on unsteady Reynolds Averaged Navier-Stokes simulations with sliding grid techniques for propeller motion and carefully constructed hexahedral meshes. Grid convergence studies are performed independently on hull and propeller configurations to quantify numerical uncertainty and support mesh selection. Cavitation is modelled using a transport equation approach, complemented by adaptive mesh refinement to resolve vapor structures without prior assumptions on cavity location. For the full scale case, surface roughness effects are explicitly incorporated through a modified wall function formulation, and a self propulsion controller is employed to achieve thrust-resistance equilibrium. Key findings indicate that the adopted CFD approach reproduces propeller thrust, torque, and efficiency trends with good agreement against available experimental and sea trial data. Cavitation patterns observed in both open water and in behind conditions are consistent with expected physical behaviour. For the full scale tanker, results highlight the dominant impact of surface roughness modelling on powering predictions and demonstrate stable convergence of coupled hull-propeller simulations. The novelty of this work lies in the unified analysis of model scale cavitation benchmarks and full scale self propulsion simulations using consistent numerical practices. The paper provides practical insights into mesh strategy, uncertainty quantification, and physical modelling choices, contributing to improved best practices for CFD based propulsion assessment in modern ship design.

Dassault Systèmes is the PLM world leader and a key player of the MBSE (Model Based System Engineering) implementation in the industry. This presentation introduces a Proof of Concept (PoC) developed through close collaboration between the Mechanical Engineering Department of UoA (University of Adelaide-South Australia) and Dassault Systèmes Marine & Offshore Industry Solutions team. The project aims to build a Virtual Twin Submarine Drone starting from the stakeholder requirements and specifications to system and sub-system definition. The objective is to enable early-stage simulation of the behavior of a submarine drone allowing to evaluate the performance and handling from the initial phases of the design cycle. This underwater drone simulator enhances the pre-design phase thanks to a new process that combines two traditionally separate processes: o Marine Concept Performance: Imagine Simulate critical Systems concepts to optimize safety and performances. o Virtual Rendering: Promote the concept in realistic 3D. Virtual Rendering process is typically used at the end of pre-design to present a concept to clients using the rendering (3DExcite) tool, which creates photorealistic environments for visualizing and promoting designs. This immersive tool allows interaction through devices such as controllers or virtual reality headsets, enabling stakeholders to explore and validate design concepts before prototyping. In parallel, the Review Systems Engineering process relies on a Model-Based Design approach using CATIA Magic and Dymola. Dymola uses the Modelica language to simulate Multiphysics systems and analyze their dynamic behavior through mathematical models. This enables engineers to evaluate different design strategies before physical testing. Instead of relying on predefined animations, the drone's movements in the 3D environment are driven by real simulation data, resulting in more realistic behavior. This integration enables the creation of an interactive simulator where users can control the drone and observe its performance dynamically. Such a tool offers significant advantages in early design stages. It allows engineers and pilots to test performance under various scenarios, identify improvements, and validate design choices before building prototypes. It can also be used for training purposes. Ultimately, this approach reduces the need for costly prototypes and physical testing while enabling broader design exploration. The project follows the MBSE methodology and aligns with the V-cycle development process, ensuring comprehensive system validation from concept to real-world testing and deliver in what we call a Virtual Twin environment.

This work presents the development of an integrated process for the structural optimization of ship transverse section scantlings, specifically: plate thicknesses, stiffener scantling, spacing of longitudinal secondary stiffener and webframe. The main objective is to minimize the total weight of these structures while simultaneously ensuring compliance with the structural design requirements established by classification society rules, through the coordinated optimization of geometric configurations and construction parameters. The developed process implements an automated workflow using the exploratory design software HEEDS. This workflow enables the integration of structural analysis through two independent paths: one in accordance with Bureau Veritas criteria using the MarsMili software, and another in compliance with Lloyd's Register requirements using the NSR software, followed by a post-processing stage and automatic result exploitation using Python. The entire workflow is automatically constructed by means of a graphical user interface specifically developed for this process. The SHERPA algorithm combines evolutionary algorithms, intelligent sampling techniques, and stochastic methods to adaptively explore the design space of the structural variables defined in the parametric model of the midship frame. These variables include plate thickness, spacing and number of web frames, scantlings and spacing of longitudinal stiffeners, identifying configurations that simultaneously satisfy all structural constraints. This leads to a weight reduction of 15% compared to a baseline frame with no prior optimization, and 5% compared to a frame optimized through four days of manual work by a junior engineer, which is replaced by a computational time of two days. This approach enables the capture of complex nonlinear relationships and dependencies between variables that traditionally require intensive manual iterations and enables the exploration of design solutions not initially considered. Once the optimal midship frame design has been established and validated by the engineer, a second workflow is implemented that reuses the design variables common to the remaining transverse sections of the ship (such as web frame spacing). This enables consistent global optimization, significantly accelerating the structural design process of the vessel. The automated frame design optimization process allows the achievement of high-performance structural solutions, significantly reducing design time and costs, and facilitating data-driven decision-making.

*** OBJECTIVE AND SCOPE: *** The maritime industry faces unprecedented pressure to meet IMO 2050 decarbonization targets, with CII ratings directly impacting vessel operational viability and market value. Current CII prediction methods rely on simplified empirical formulas lacking uncertainty awareness and regulatory adaptability. This paper presents a Neuro-Symbolic AI framework for CII rating prediction with Bayesian Uncertainty Quantification and automated regulatory compliance verification using operational data. +++ APPROACH AND METHODOLOGY: +++ We develop a Neuro-Symbolic architecture combining: (1) Neural Networks trained on IMO GISIS CII database (5,000+ vessels) and EU MRV emissions data for operational carbon intensity prediction; (2) Monte Carlo Dropout for Bayesian Uncertainty Quantification providing 95 percent confidence intervals on CII ratings; (3) Knowledge Graphs encoding IMO CII thresholds, EU ETS regulations, and LNG well-to-wake factors as machine-readable symbolic constraints. Validation employs real CII ratings from 2023-2024 reporting cycles. +++ KEY FINDINGS AND CONCLUSIONS: +++ Results demonstrate CII rating prediction accuracy of 91 percent within one grade (A-E) against verified IMO GISIS data, with computational time 95 percent faster than traditional compliance assessment workflows. The Bayesian Uncertainty Quantification module successfully identifies vessels with insufficient data for reliable prediction with 87 percent precision. Integration of regulatory Knowledge Graphs enables automated compliance scenario analysis for LNG fuel conversion and operational optimization strategies. +++ NOVELTY AND CONTRIBUTION: +++ This work provides three original contributions: (1) First Neuro-Symbolic framework for CII rating prediction with auditable regulatory logic; (2) Uncertainty-aware CII predictions enabling risk-informed operational decisions; (3) Knowledge Graphs integrating IMO CII, EU ETS, and LNG emissions factors for comprehensive decarbonization compliance. The framework offers practical value for ship operators seeking to optimize CII ratings and avoid regulatory penalties. +++ KEYWORDS: +++ Neuro-Symbolic AI, IMO CII, Decarbonization, Uncertainty Quantification, Knowledge Graphs, LNG Emissions

This study proposes a novel framework for transforming ship design processes from a drawing-centered paradigm to a design-information-centric paradigm, enabling more efficient and transparent engineering workflows. Traditionally, ship design has been managed at the level of drawings, where multiple design elements are bundled together. While this approach facilitates holistic understanding, it results in coarse information granularity, implicit dependencies, and delays caused by waiting for entire drawings to be completed. The objective of this research is to address these limitations by introducing the concept of design information exchange relations, which explicitly represent dependencies among individual design information units, and by integrating this concept with a design agent that supports process execution. The proposed approach consists of three main components. First, design information is decomposed into fine-grained units. Second, exchange relations among these units are constructed using a hybrid methodology combining domain knowledge templates, designer interviews, and large language model (LLM)-assisted extraction and normalization from actual drawings. Third, a prototype design agent is developed to utilize the resulting network, enabling functions such as next-step recommendation, dependency checking, and impact analysis. A demonstration using representative engine room drawings shows that meaningful design information networks can be constructed. The results indicate that the proposed framework enables earlier initiation of downstream design tasks, reduces waiting times, and enhances visibility of design dependencies. Furthermore, evaluation reveals that an appropriate level of granularity is critical for balancing usability and scalability. The novelty of this study lies in formalizing ship design as a network of design information units and demonstrating how such a network can serve as the foundation for an intelligent design agent. This work contributes a new perspective on design process modeling, offering a pathway toward more modular, traceable, and AI-assisted engineering practices in shipbuilding.

This paper presents the development and validation of a comprehensive digital twin for ships as integrated transportation systems, aimed at supporting operational optimization and maintenance decision‑making. Most merchant ships engaged in international trade are equipped with slow-speed, reversible two-stroke diesel engines directly driving large propellers. The propulsion performance of such systems is highly sensitive to environmental disturbances, including wind and waves, as well as to hull and propeller fouling and engine torque limitations. Accurately capturing the coupled influence of these factors is essential for realistic performance assessment but remains challenging for conventional steady state modeling approaches. To address this issue, a comprehensive mathematical model of the propulsion system was developed using MATLAB/Simulink. The model consists of more than 50 coupled equations that describe both translational ship motion and rotational dynamics of the propulsion train. In contrast to simplified models, the proposed digital twin explicitly incorporates time varying environmental loads and detailed main engine characteristics, including torque limit behavior and its interaction with wave induced resistance. This enables realistic simulation of propulsion performance under actual and fluctuating sea states rather than idealized operating conditions. The digital twin was validated using measured onboard data obtained under actual seas, including wave height and direction. Comparison between simulated and measured results demonstrates quantitative agreement across a wide range of operating conditions, confirming that the model accurately reproduces the ship's speed-power relationship and transient propulsion behavior. Based on the validated model, three key applications are demonstrated. First, the digital twin supports operational limit prediction by identifying discrepancies between commanded and achievable engine speeds near the torque limit, where performance degradation and fuel inefficiency are most pronounced. This capability is particularly relevant for recently built ships, which often employ lower installed power relative to displacement to improve greenhouse gas emission indices. Second, the model enables speed optimization by incorporating forecasted environmental conditions and optimizing daily speed allocation, achieving fuel consumption reductions of up to 3% while maintaining schedule reliability. Third, the digital twin supports fouling management by estimating hull and propeller fouling levels from performance deviations, enabling cost effective maintenance planning and potentially reducing fuel consumption by up to 12% over a ship operational period. Overall, this study demonstrates that a physics based digital twin can provide actionable insights into coupled propulsion dynamics and support data driven ship operation, contributing to improved fuel efficiency, emissions reduction, and economically sustainable fleet management across various ship types and trading routes.

As the shipbuilding industry shifts toward highly outfitted vessels, the volume and complexity of pipe fabrication processes have increased, making their systematic planning a critical challenge. Pipe fabrication involves diverse geometries, materials, and spatial constraints, and the determination of manufacturing workflows has relied on the tacit knowledge and heuristic judgments of craftsmen. While this has historically sustained operations, such person-dependent processes pose risks and hinder objective production management. In particular, the lack of explicitly defined BOP (Bill of Processes) data prevents the direct application of conventional production planning systems in pipe fabrication. To address this challenge, this study develops an automated system for generating BOPs based on pipe product and factory operational data. The proposed system generates candidate BOPs using M-BOM (Manufacturing Bill of Materials) tree structures and graph-based algorithms, and determines suitable BOPs through simulation-based evaluation. The methodology consists of three main steps. First, the structural transitions of intermediate products are modeled as an M-BOM tree, encoding assembly constraints such as installation sequences. Second, manufacturing tasks are extracted from part relationship data stored in a graph database and represented as a task network. Based on this representation, feasible process sequences are generated using graph algorithms, such as topological sorting, while satisfying precedence and spatial constraints. A key feature of the proposed method is the treatment of inspection timing. Conventionally, inspection timing is implicitly determined based on craftsmen's experience. In contrast, this study explicitly formulates inspection timing as a design variable defined by intermediate assembly states of pipes. In other words, inspection is not scheduled arbitrarily in time, but is systematically associated with specific joining states of intermediate products. By varying the inspection timing defined in this manner, multiple alternative BOP candidates with different process structures are generated. Third, discrete-event simulations are applied to the generated BOP candidates. By evaluating metrics such as lead time and transportation frequency, the impact of inspection timing, defined by intermediate assembly states, on operational performance is quantitatively analyzed, and suitable BOPs for the target factory are identified. A case study on a virtual pipe factory demonstrates that the proposed method can generate feasible BOPs and reveals the relationship between inspection timing and process performance. The proposed system transforms tacit, experience-based process planning into a formalized and computational framework, enabling systematic BOP generation. By integrating product structure and process definition, the method supports data-driven production planning and contributes to the standardization of pipe fabrication processes.

The design of battery-electric harbour craft is strongly influenced by hull hydrodynamics, internal arrangement, and the integration of onboard energy storage systems. Unlike conventional vessels, electric craft must accommodate large battery capacities within constrained hull volumes, making battery placement a key design driver. For a fixed total battery capacity, multiple battery placement configurations can be developed within the hull, such as different longitudinal, transverse, or vertical arrangements of battery racks. Each configuration imposes distinct requirements on hull dimensions, displacement distribution, structural layout, and internal volume allocation, which in turn affect hull form geometry and hydrodynamic resistance characteristics. Battery placement configurations therefore play a direct and active role in shaping the hull and should be explicitly considered during the early stages of the design and optimisation process. This paper presents a design-focused framework that integrates battery-room driven general arrangement development with hydrodynamic optimisation for battery-electric harbour craft during early-stage development. A representative electric harbour craft concept is developed based on a fixed total battery capacity selected to satisfy intended operational requirements, such as endurance and energy demand, while allowing flexibility in battery placement during hull form development. By maintaining a constant battery capacity, the study isolates the influence of battery configuration and associated hull geometry changes on hydrodynamic performance. Reynolds-Averaged Navier-Stokes (RANS) Computational Fluid Dynamics (CFD) simulations are conducted to evaluate calm-water resistance and the corresponding propulsive power requirements across a range of operating speeds for multiple hull configurations. To reflect realistic vessel usage rather than idealised single-speed operation, an operational speed profile is introduced, where individual operating speeds are weighted according to their probability of occurrence. This probabilistic weighting enables energy efficiency to be assessed using a weighted average metric, providing a more representative estimate of average energy demand under real operational conditions. The results demonstrate that, even with a fixed total battery capacity, variations in battery placement can lead to significant differences in hull form characteristics, resistance behaviour, and weighted energy efficiency. Once a feasible hull-arrangement configuration is identified, operational parameters such as mean operating speed can be further adjusted to enhance overall performance. The proposed framework highlights the importance of coupling general arrangement and hydrodynamic optimisation early in the design of battery-electric harbour craft, supporting informed, energy-efficient, and practical vessel design decisions.

In the shipbuilding process, welding is a critical process that accounts for the majority of total man-hours and has a direct impact on production efficiency and safety. As ship hull structures contain numerous long weld joints and are manufactured as one-off products, it is difficult for factories to introduce inspection lines equipped with expensive automated inspection systems such as 3D scanners. Consequently, visual inspection to verify the quality of welded joints has traditionally relied on visual assessment by skilled inspectors. However, this method is time-consuming, and training inspectors capable of making rapid and accurate judgments remains a challenge. If defects overlooked during the shipyard's self-inspection are subsequently discovered during supervisory inspections by classification societies or shipowner, repair work becomes necessary and leads to a decline in production efficiency. To address these challenges, this research focuses on the quality inspection of fillet welds, which are abundant in ship hull structures and are prone to becoming the initiation points of fatigue cracks, with the aim of developing a method to evaluate weld leg length based on welding process data. Welding logs (time series of current, voltage and wire feed rate) obtained from the power supply during welding, together with welding sounds recorded by an external microphone, are used as input data. Weld bead geometry was measured at fine intervals using 3D scanner and correlated with aforementioned input data via welding speed. Deep learning is then employed to correlate the input data with the leg length of fillet welds. By incorporating welding sounds into the input data, it is expected that this will provide supplementary information regarding welding phenomena that cannot be fully captured by electrical signals alone, thereby improving the reliability of the estimation of leg length. Focusing on welding operations at the laboratory level, the effect of the window size used for processing the input data on the accuracy of leg length estimation is investigated. By adopting a window size that yields relatively good estimation results, it is confirmed that relatively accurate leg length estimation is also possible for the welding by using simple automated carriage in shipyard. The analysis results showed that the model using only electrical signals achieved high estimation accuracy for fillet leg length. Furthermore, the model incorporating welding sounds showed a tendency toward improved accuracy. Conversely, while welding sound alone could estimate leg length, its accuracy was limited and susceptible to environmental noise.

In shipbuilding, the erection process of grand assembly blocks in the dock is the most critical stage, as it affects the schedules of other construction stages and requires strict adherence to delivery deadlines. To ensure smooth erection process in the dock, it is essential to properly plan the grand assembly process. This study proposes a method for optimally platen allocation planning of the grand assembly. The problem involves determining whether each block can be placed on the platen before its erection start date, under multiple constraints, such as time, size of blocks, and facility limitations. Even for experienced platen scheduling planners, solving this large-scale combinatorial optimization problem manually requires significant time and effort. Most conventional studies on scheduling focus on load leveling. They assume fixed resource capacities for each stage and aim to level the workload of individual processes. However, these approaches tend to produce locally optimal solutions and often fail to maintain consistency with facility constraints. Even if one stage is leveled, it may cause a shortage of space during platen allocation planning. This study focuses on shipyards where workers are multi-skilled and capable of handling multiple tasks. In this case, it is possible to flexibly reallocate worker resources across construction stages, thereby achieving load leveling over the entire shipbuilding stages. Based on this premise, this study does not fix the number of workers for each stage. Instead, it proposes a framework to search for feasible platen allocation plans under constraints such as usage periods, space, and facilities. A genetic algorithm (GA) is adopted as the search method, enabling discovery of high-quality solutions within practical computation time, even as the search space grows exponentially. The evaluation function includes weighted penalty terms representing violations of plate utilization limits, deviations from completion dates, and changes in plate usage periods. By adjusting these weights, the method reflects planners' priorities. A case study using real shipyard data over six months, including approximately 120 blocks of different ship types, showed that the proposed method reduced planning time from over a week to a few hours. Sensitivity analysis of the penalty weights confirmed that the solution responds as expected to changes in planning priorities. The proposed method formalizes experience-based scheduling into a precise optimization problem, enabling fast, transparent, and reproducible planning and supporting decision-making in shipbuilding production management.

In current naval operations, forces must adapt to rapidly evolving threats and technology refresh cycles, while operating platforms that were built for multi‑decade lifetimes. A trade-off is created between the need to deploy new capabilities onboard and the difficulty of integrating legacy systems, which increases the effort required for each software and interface change. To address this challenge, new European warships require a digital architecture that supports long lifecycles, heterogeneous legacy systems and continuous evolution of their capabilities, complying with strict security and safety constraints, through their complete life cycle. The Digital Ship Reference Architecture separates user, access, service and resource layers and makes cross‑cutting concerns (security, orchestration, monitoring and lifecycle management) architectural responsibilities. The architecture is operationalized through a standardized Digital Platform defined as a service catalogue, enabling applications to consume common capabilities with interfaces, while allowing staged evolution from an initial baseline to deeper IT/OT convergence and, ultimately, service‑oriented system‑of‑systems composition. To make the reference architecture actionable, Model‑Based Systems Engineering (MBSE) is implemented, using Unified Architecture Framework (UAF). Motivations and constraints are traced to operational roles and activities, then to platform service families and underlying resources, providing end‑to‑end traceability from intent to implementation. This model‑driven approach supports impact analysis, consistency management and controlled change across stakeholders and lifecycle phases, supporting reuse and reducing integration ambiguity. The contribution is a reference architecture that can be instantiated across shipboard, ashore and naval key integrator environments.

The shipbuilding sector is experiencing a paradigm shift as digital technologies increasingly replace traditional design and production practices. This paper examines how the integration of advanced digital tools, such as 3D model-based design environments, simulation platforms, digital twins, artificial intelligence, and data-centric production management systems, is transforming the way ships are conceived, engineered, and constructed. The study analyzes the transition from fragmented, document-driven workflows to fully integrated digital ecosystems that connect design, engineering, procurement, and manufacturing. Emphasis is placed on the role of virtual prototyping, real-time data exchange, and collaborative platforms in reducing design errors, improving interdisciplinary coordination, and accelerating project delivery. The paper further explores how automation, robotics, and digital fabrication technologies are reshaping production processes within modern shipyards. In addition, the research addresses the organizational and technical challenges associated with digital transformation, including interoperability between software systems, data governance, cybersecurity risks, and workforce upskilling. The findings suggest that digitalization not only enhances efficiency and quality but also enables more sustainable and adaptable shipbuilding practices in response to evolving regulatory and market demands. By synthesizing current technological trends and industry practices, this paper provides a comprehensive perspective on how digital tools are redefining both the design philosophy and production logic of contemporary shipbuilding.

An important part of planning the layout of a shipyard is simulating its production process to estimate its productivity and find possible bottlenecks. The simulation of different production processes involves an experienced planner estimating the cycle time of each station depending on the processed parts. These estimations are formulas-based and depend on part parameters such as the size, weight or weld length, but the accuracy is highly dependent on the individual and is hard to validate. This paper presents a shipyard planning approach based on high fidelity, quasi-continuous simulations to replace formula-based estimations. Our solution consists of a typical shipyard process simulation built around a modular Discrete Event Simulation (DES), where an initial estimation of the cycle time of stations is calculated by formula-based assumptions as described above. Then, the cycle time is updated by using a more accurate, part-specific, and detailed simulation of the welding process. A regression analysis between the accurate simulation and the formula-based outputs is done to improve the formulas used initially to estimate the cycle time. The refined formulas are then used to accurately estimate the cycle times of each station. Consequently, combining them allows for a detailed and precise shipyard simulation without compromising the simulation speed of the DES. To test the developed solution, a shipyard has been modeled in the DES to manufacture multiple ship blocks. A highly accurate and detailed digital model of one of the production stations, calibrated on real data, was used to simulate the welding of several panels, to calculate the cycle time, and then update the DES formulas of that station. Our solution provides two major outcomes. The first is more accurate refined formulas to calculate cycle time compared to the traditional formula-based approach. The second is the option to use the highly accurate model for feasibility studies when new parts are considered, or other performance KPIs should be evaluated. In shipyards, cycle time estimation can vary significantly between formula-based estimations and real-world behavior. To get better cycle time estimations without significant computational load increase or the availability of a digital twin, a new approach is presented. Our approach can be used for both planning new shipyards or restructuring existing ones for performance optimization. It can also be used to improve the scheduling of existing shipyards, if no digital threads or other means of feedback from the real operations can or may be established.

In this study digitalization of compliance check with the new ship intact stability in waves based on non-linear time-domain simulations is addressed. In first place, the roll damping of the post-Panamax C11 container ship is estimated using a commercial available CFD software tool, and the ship dynamic stability scenarios in waves are simulated based on non-linear time-domain methods. This method's reliability is validated by contrasting the simulation results with the experimental data of C11 container ship. Finally, the simulated roll motions of ship based on a time-domain method can be applied to check compliance with the Second-Generation Intact Stability criteria (SGISc), and this approach is critically discussed as well.

In the shipbuilding industry, particularly within ship design offices, the Digital Twin (i.e the 3D product model and associated design data) is only one part of the engineering workload. A substantial fraction of time is consumed by the surrounding context required to make correct and optimal modelling decisions. This context is shaped by project specifications, classification and statutory requirements, vendor constraints, internal practices and biases, historical references, and the continuous tension and reconciliation of what the project requires versus what the design is becoming. This work is typically manual, fragmented across file systems and tools, and heavily dependent on a limited number of senior personnel. The result is a throughput bottleneck, elevated review burden, and a growing risk of institutional knowledge loss as senior experts retire. These challenges are amplified by increasing vessel complexity and by the need to deliver higher first-pass quality under multiple regulatory regimes. This paper presents an approach for building an AI-enabled knowledge and decision-support layer over ship design projects. The approach converts unstructured project packages into structured, connected project knowledge and makes that knowledge continuously usable during design execution. It employs multiple function-specific AI workflows to ingest heterogeneous artifacts such as requirements, standards, drawings, vendor data, engineering lists, and prior project material. These workflows organize information into a project context that can be queried and reasoned over, and they encode repeatable procedures derived from observed senior expert practice, including how experienced designers scope relevant requirements, identify conflicts, interpret constraints, and decide what to escalate early. The intent is not to replace senior expertise, but to reduce low-leverage context labour so that human effort is concentrated on high-value judgment and complex design problem solving. The paper further describes methods for linking structured project knowledge to engineering artifacts, including 2D drawings and 3D model data. This enables proactive guidance and issue surfacing while designers retain full design authority and changes remain inexpensive. In addition, we introduce a framework for design decision traceability that captures rationale and constraint pathways at the moment decisions are made. This produces a navigable design decision tree that supports future modifications, improves review and decision explainability, and enables organization-specific learning. Such traceability is rarely captured in a systematic form in ship design projects, in part because of the large number of decisions made between mission definition and delivery. We outline an evaluation programme spanning academic and industrial deployments, including measured reductions in time spent assembling and reconciling project context, improvements in first-pass quality indicators, and practitioner feedback on usability and trust. The work provides a practical pathway toward AI-native ship design workspaces that integrate with existing CAD and information systems while creating a durable, explainable knowledge layer that scales with project complexity and workforce constraints.

Water-lubricated stern tube bearings have been increasingly adopted as an environmentally friendly alternative to conventional oil-lubricated systems. Despite their environmental benefits, vibration problems have been observed in a low-speed small craft equipped with a water-lubricated stern tube bearing system. To identify the source of the vibration observed during operation, a series of experiments and analyses was conducted. The investigation was carried out in several stages; an initial audio frequency analysis was first performed to detect abnormal vibration signature and identify potential sources. Next a detailed mechanical vibration measurement was done using sensors to investigate further and diagnose the vibration behaviour. Lastly, a hammer test was carried out to investigate the natural frequency of selected structural plates. Aside from the test conducted, natural torsional vibration frequencies of the propulsion system are determined based on the governing equations of free torsional vibration. The calculated torsional frequencies were compared with the measured vibration spectra, and no alignment was observed with the dominant excitation frequencies. This indicates that torsional vibration is not a contributing factor to the observed vibration behaviour. Friction-induced vibration was considered as a potential contributing factor, particularly for the high-frequency vibration observed between 600 RPM and 800 RPM. Based on literature, water-lubricated stern bearings are susceptible to unstable lubrication conditions, which may lead to stick-slip behaviour and generate vibration in the form of chatter or squeal. The observed vibration trend in this study shows a similar pattern to those reported, where vibration occurs within a specific intermediate speed range and reduces as rotational speed increases. To mitigate potential friction-related effects, several measures can be considered, including reducing the friction coefficient through material selection or surface treatment, optimising radial clearance, adjusting contact stiffness, and modifying bearing material properties such as rubber layer thickness and hardness. In addition, inspection of the stern tube system and lubrication condition is recommended to ensure stable operation and minimise the likelihood of vibration.

In recent years, Japanese shipyards have increasingly adopted 3D-CAD systems to improve design quality and productivity. Although 3D viewers are used in production sites for preliminary reviews and outfitting installation work, traditional 2D drawings are still predominantly employed for fabricating outfitting components such as piping and steel structures. Consequently, converting 3D models into 2D drawings requires additional effort. While automatic drawing-generation functions in 3D-CAD systems have reduced drafting time, issues remain, including information loss due to discrepancies between 3D models and 2D drawings, as well as difficulties for workers on the shop floor in interpreting the drawings. Currently, experienced designers supplement 2D drawings with additional information tailored to the skill level of the production site, while workers rely on their own experience to infer the design intent. This situation occasionally leads to unnecessary communication costs. Moreover, due to Japan's demographic trends, a shortage of experienced personnel is inevitable, and these challenges are expected to become even more pronounced. The goal of this study is to address these issues by introducing a 3D manufacturing drawing system that can replace conventional 2D drawings in production design and piping fabrication processes, and by evaluating its effectiveness in actual shop-floor environments. In this study, a 3D manufacturing viewer application ("3D drawing") was developed by constructing piping 3D models based on data extracted from a 3D-CAD system, enabling workers to directly reference fabrication information. The 3D drawings contain geometric shapes, bending details, welding points, and other attribute information required for fabrication, thereby functioning as a viable alternative to traditional 2D drawings. To evaluate applicability, validation tests were conducted through actual fabrication work in a shipyard environment. Workers performed fabrication tasks using the 3D drawings, and the results were compared with those of tasks executed using conventional 2D drawings. The evaluation confirmed that the 3D drawing approach reduced design-side drafting time, shortened the time required for workers to interpret drawings, and decreased fabrication errors arising from drawing misinterpretation. These findings demonstrate that 3D drawings incorporating fabrication information can contribute to streamlining the piping fabrication process. Future work includes expanding this approach to outfitting components beyond piping.

Objective and scope. Ship lifecycle data remains fragmented across design, shipbuilding, and operations. Complexity of engineering models, one of the kind production data, gaps of the data ownership and over reliance on 2D old fashioned documents are among some of the reasons behind this status quo situation. Decades of using intent- driven CAD as a de facto hub to transfer design intent into production are typical for the industry. Currently, the maritime industry undergoes a rapid transformation and is expanding its data ambition—driven by robotics, automation, and AI-enabled generation and semantic linking of data models. Big PLM platforms promise end to end lifecycle continuity. In practice, maritime data complexity, the number of collaborators, and vessel lifetimes create discontinuity: data is required for decades, often outliving multiple generations of software, making interoperability and long term accessibility dependent on robust, unified standards rather than the software platform. This paper presents a discussion about the path toward lifecycle digital continuity as goal oriented: define high - value lifecycle use cases first, then engineer the minimum necessary "bridges" between authoritative design models and operational needs. Such use and business cases have a potential to make a foundation and provide business models for the use of design data models, digital handover of data from design and building to operators and digital asset management for ship owners. The data already created at the design stages could lay the foundation for the digital twin in operations and unlock the potential of multi-dimensional digital twin. Approach and methodology. We analyze lifecycle data flows and classify use cases by (i) stakeholder value (yard/owner/operator), (ii) required fidelity and configuration granularity, and (iii) data longevity and standardization needs. Based on this classification, we propose a reference approach for connecting design models to operational applications. Key findings and conclusions. Three use case groups demonstrate tangible benefits when design and operational data are connected: (1) ship stability calculation engines and loading-condition management, (2) use of performance models for voyage optimization and fleet performance management, and (3) operational simulation for validation of ship performance models in operational conditions. Across these, design data provides direct operational value and forms a stable foundation for a Digital Twin in operation. Using the already available design data models in operational decision support provides a value for the ship operators and owners and forms digital asset management. AI is already improving the feasibility of linking unstructured operational information to complex engineering structures, particularly through semantic mapping and conversational interfaces, but it amplifies (rather than replaces) the need for governed standards and configuration control. Novelty and contribution. The paper contributes a use case driven roadmap for building a multidimensional Digital Twin as layered lifecycle datasets, enabling practical benefits early while establishing durable foundations for long-lived maritime data interoperability.

The maritime industry is undergoing a fundamental shift from 2D drawing-based classification approval to 3D model-based workflows. At the center of this transformation is the Open Class 3D Exchange (OCX) standard - an open interoperability format developed and maintained by a consortium of more than 30 industry leaders, including major classification societies, maritime software providers, shipyards, and design offices. Early pilot projects have demonstrated that OCX can drastically reduce the use of 2D drawings, enable automated rule checking through direct 3D model import, and reduce basic design time in a range of 10-30% , however, the standard has not yet reached mainstream adoption. The transition from successful pilots to industry-wide implementation raises critical questions about readiness, incentives, and barriers across the stakeholders. This paper presents a multi-stakeholder impact analysis of adopting the OCX as the global standard for 3D model-based approval in maritime classification. The study examines the major stakeholder groups such as shipyards, software providers, and classification societies , and maps both the anticipated benefits and the barriers to adoption for each. Benefits analysed include reduced time to market, improved design traceability, earlier classification engagement, elimination of redundant data conversion, and the foundation for digital twin and lifecycle data management. Barriers examined include legacy workflow dependencies, investment costs for software and training, data quality and conformance challenges, regulatory inertia, and uneven regional adoption rates. The OCX vision extends well beyond the single focus of structural hull approval to include a holistic approach for classification in the maritime industry. Recent developments point toward an expanded scope encompassing stability analysis, finite element model data exchange, regulatory compliance automation, and integration with Product Lifecycle Management (PLM) systems, positioning OCX as a standard for 3D model data exchange to support the potential digital backbone for the entire ship design-to-operation lifecycle. Realizing this vision requires coordinated adoption across a diverse ecosystem of stakeholders with differing incentives, capabilities, and risk profiles. The analysis draws on published pilot project results, consortium documentation, and industry expertise of the co-authors and the organizations behind them. The paper proposes a stakeholder readiness framework that identifies the dependencies and sequencing required for OCX to progress from its current early-adopter phase to global industry adoption. The paper concludes with recommendations for the OCX Consortium, classification societies, and policymakers to accelerate the transition to the 3D Model-Based classification while managing risks for all stakeholders.

This paper addresses the growing complexity of production planning in the shipbuilding industry, particularly in the hull block assembly process. As production scales expand, efficiently arranging large and heavy blocks within limited workspace while optimizing overall schedules has become increasingly challenging. Traditionally, such planning has relied heavily on the experience of skilled workers; however, the diversification and scale of modern shipbuilding demand more systematic and automated approaches. The problem is characterized by the need to simultaneously consider spatial layout and temporal scheduling under resource constraints, including limited workspace and workforce capacity. Previous studies have approached scheduling problems using frameworks such as the Resource-Constrained Project Scheduling Problem (RCPSP), applying methods like linear programming, tabu search, genetic algorithms, and discrete-event simulation. While these methods effectively handle precedence relationships and resource limitations, they often treat layout and scheduling separately. More recent research has attempted to integrate these aspects, using techniques such as polygon-based placement, heuristic bin-packing approaches, and rectangle approximations combined with precedence constraints. Despite these advances, existing methods remain limited in their ability to handle polygonal block shapes and to simultaneously optimize layout, scheduling, and resource allocation. To address these limitations, this paper proposes a tree search-based algorithm that integrates simulation. The proposed method enables simultaneous optimization of two-dimensional block layout and scheduling while explicitly considering resource capacity constraints. By incorporating both spatial and temporal decision-making within a unified framework, the approach aims to generate more feasible and efficient production plans. The effectiveness of the proposed method is evaluated through numerical experiments using real shipyard data. The results demonstrate that the algorithm can produce improved schedules and layouts compared to conventional approaches, achieving better resource utilization and reduced overall make span. This study contributes to the advancement of integrated production planning methods in shipbuilding and highlights the potential of combining tree search techniques with simulation for solving complex industrial optimization problems.

The rapid growth of electric propulsion in small marine craft and unmanned surface vessels has led to an increased demand for efficient and application-specific propeller designs that can be quickly adapted to changing operational requirements. Conventional propeller series, primarily developed for larger vessels and traditional manufacturing processes, provide limited flexibility for low-power, weight-sensitive electric propulsion systems. This paper presents a computer-aided digital design and validation framework that integrates parametric modelling, computational fluid dynamics (CFD), optimisation, and experimental testing to facilitate the development of additively manufactured (AM) marine propellers. A fully parametric propeller model is implemented within an integrated computational design environment using CAESES, enabling systematic variation of key geometric parameters including pitch-to-diameter ratio, expanded area ratio, blade camber, thickness distribution, skew, and rake. The parametric geometry is directly linked to automated CFD simulations using Reynolds-Averaged Navier-Stokes solvers in Siemens STAR-CCM+ to evaluate open-water performance across a target advance-coefficient range relevant to small electric craft. Initial design exploration is conducted using a structured design-of-experiments approach to map performance trends and identify influential design variables. This is followed by a dedicated optimisation stage, in which multi-objective optimisation algorithms are applied to maximise thrust and open-water efficiency while constraining torque within electric motor power limits. The optimisation process identifies a set of Pareto-optimal propeller designs, from which candidate geometries are selected for physical validation. The digital workflow enables the computational assessment of large numbers of propeller variants before physical fabrication. Selected optimised designs are manufactured using AM techniques and evaluated in a controlled experimental setup. The results show consistent performance trends between computational predictions and experimental measurements, confirming the effectiveness of the digital workflow. The analysis emphasizes the significant impact of factors such as blade area ratio, pitch distribution, and camber on propeller performance. This integrated workflow greatly reduces the number of physical prototypes needed and shortens the design-to-validation cycles when compared to traditional build-test-modify methods. The framework supports a digital-twin-oriented design philosophy, where numerical models are iteratively refined using experimental feedback to enhance predictive capability over multiple design cycles. By combining parametric modelling, automated simulation, optimisation, and targeted testing, the approach reduces reliance on fixed propeller series and enables rapid, data-driven design iteration. The methodology contributes to broader shipbuilding digitalisation initiatives, including virtual prototyping, model-based engineering, and computer-assisted decision-making, and applies to a wide range of small electric marine propulsion systems.

The rapid growth of river cruise tourism in India has led to an increased demand for luxury river cruise vessels in Indian waterways. To enhance passenger experience, interior arrangements of vessels must prioritize accessibility, scenic views, and luxury amenities, without compromising vessel safety and regulatory requirements. This balancing act makes layout generation a primary challenge in river cruise vessel design, typically requiring numerous manual iterations and often consuming a lot of time. This paper presents a Python-based parametric framework designed to automate the preliminary layout generation for inland river cruise vessels. The framework utilizes total passenger capacity and functional requirements such as dimensional constraints, cabin categories, choice of amenities, etc. as primary design drivers. At the core of the system is a Particle Swarm Optimization (PSO) algorithm, utilized to ensure that space allocations in the layout meet specific area targets, aspect ratio, and horizontal/vertical access requirements. Data from regression analysis of similar river cruise vessels is utilized in the parametric framework as an input for PSO algorithm. To ensure technical viability of the resulting arrangement, the parametric layout framework is integrated with submodules for preliminary weight estimation, stability assessment, powering, and electrical load calculations. This approach ensures regulatory compliance and technical robustness during early stages of design. As future scope, the parametric layout framework is extended to link to a parametric hull modelling tool and CFD tool for creating a multi-disciplinary optimization system that simultaneously evaluates hull form performance and spatial arrangement, thereby making this framework a useful tool for river cruise vessel designers and vessel operators/owners looking for rapid vessel concept evaluations in the emerging Indian inland cruise vessel market.

The rapid evolution of unmanned drone warfare—demonstrated in recent conflicts—has shifted priorities from platform sophistication toward production scalability and industrial resilience. Additive Manufacturing (AM) has been increasingly adopted for unmanned aerial vehicle (UAV) production to mitigate disrupted supply chains and enable rapid design iteration for fast operational adaptation. Building on the same principles, Singapore's Defence Science and Technology Agency (DSTA), in collaboration with Odense Maritime Technology (OMT), are reimagining the design approach for an unmanned surface vessel (USV) to optimise for AM production. This paper presents the design considerations for an AM‑produced USV hull, thus facilitating rapid design evolution and a more efficient production.

An increasing number of shipyards want to apply PLM principles to their daily work. They aim to better manage engineering changes, ensure data consistency, or preserve company knowledge. Additionally, shipyards seek to improve collaboration between engineering disciplines, streamline production processes, and reduce costly rework caused by outdated or conflicting information. The promise of PLM lies in creating a single source of truth that spans the entire vessel lifecycle—from initial design through construction, operation, and eventual decommissioning. Defining the initial scope when introducing PLM principles or deciding on the next extension remains challenging, though. Nearly every shipyard starts from scratch when specifying the target solution. One reason is the lack of an established best practice for the shipbuilding industry. Whereas the PLM methodologies applied in the automotive or aerospace industry have matured over decades, most shipbuilding PLM implementations are still in their infancy. Additionally, shipbuilding faces unique challenges, including one-of-a-kind vessel designs, complex supply chains, and the integration of numerous engineering disciplines working in parallel. In this paper, we introduce a blueprint to describe and categorise shipbuilding PLM, addressing that gap. The vendor-neutral method is based on a capability map and can be used as a checklist for new PLM initiatives or to compare PLM implementations across different organisations. This structured approach enables shipyards to identify their current maturity level, prioritise improvement areas, and develop a roadmap for systematic PLM adoption. To validate this blueprint, we present real-world implementation examples found today at shipyards, demonstrating practical applications and measurable benefits achieved through structured PLM adoption.

In response to increasing demands for energy efficiency and emission reduction in the maritime industry, accurate and efficient prediction of ship hydrodynamic performance has become essential. However, conventional approaches often rely on towing tank experiments or fragmented CFD analyses, where resistance, self-propulsion, trim, and energy-saving device (ESD) effects are evaluated separately. This paper presents the development of an integrated CFD-based system, Jupiter ShipStar, built on the TechnoStar Jupiter CFD platform using the open-source CFD software OpenFOAM. The system provides a fully automated end-to-end workflow covering geometry import, mesh generation, simulation execution, post-processing, and report generation, supported by an integrated database for systematic data management and result traceability. The proposed framework uniquely integrates resistance prediction, BEMT (Blade Element Momentum Theory)-based self-propulsion analysis, trim optimization, and ESD performance evaluation within a unified computational environment. The hydrodynamic simulations are conducted using a RANS-based approach with free-surface modeling, while the propulsion performance is evaluated through BEMT coupling. This tightly coupled framework enables consistent evaluation of interactions between hull, propeller, and ESDs. The workflow automation enables high-throughput simulations, significantly reducing manual effort and allowing efficient exploration of design and operational parameters, even for users with limited CFD expertise. The integrated database further supports systematic comparison of multiple design variants and operating conditions. The numerical methodology is verified through mesh convergence studies and validated against towing tank experimental data and reference benchmarks. The results demonstrate strong and consistent correlation between numerical simulations and tank test results across nine different ship types, with prediction errors generally within predictable ranges for resistance and propulsion performance. Applications of the system demonstrate its capability to support ship designers in evaluating resistance and propulsion characteristics, optimizing ESD configurations, and assisting ship operators in trim optimization and fuel efficiency improvement under various operating conditions. The developed system provides a scalable and practical solution for modern ship design and operation and establishes a foundation for future integration with data-driven approaches, digital twins, and machine learning-based optimization.

Shipbuilding relies on complex and globally distributed procurement supply chains involving multi-tier suppliers, specialised equipment manufacturers, and tightly coupled logistics processes. The increasing frequency of geopolitical and economic disruptions, such as trade restrictions, supply shortages, and regulatory constraints, highlighted the vulnerability of these procurement systems, particularly for critical components with limited sourcing alternatives and high supplier concentration. Supplier disruptions often propagate through interconnected supply networks with cascading effects, leading to production delays and cost overruns in the shipyard. Meanwhile, conventional risk assessment methods in shipbuilding remain largely static and siloed, often failing to capture dynamic interactions, interdependencies, and system-wide ripple effects across multi-tier supply structures. To address these limitations, this paper proposes a supply chain digital twin framework to map and analyse procurement insights in shipbuilding under dynamic disruption scenarios. The framework provides End-to-End (E2E) visualization on procurement processes, integrating multi-tier supplier networks, component dependencies, and logistics bottlenecks associated with vessel construction. The three-tier framework starts with the supplier layer that models network relationships and concentration structures. Next, a component layer represents bill-of-material dependencies and lastly, the process layer analyses procurement lead times, transportation dynamics, and assembly operations to provide decision support capabilities. These layers are formalised using a Petri Net structure to model concurrency, synchronization, resource constraints, and disruption propagation. Through large language model (LLM) orchestrations, an adaptive decision-support mechanism then analyses system states to recommend mitigation strategies such as supplier diversification, buffer adjustments, and production rescheduling. In a case study featuring an LNG vessel shipbuilding network, the simulation employs an empirically grounded synthetic benchmark anchored in real-world data. Under severe geopolitical disruption scenarios, such as the Red Sea rerouting, the average total project completion time increases by 7.6%. However, when an autonomous dual-sourcing mitigation strategy is dynamically implemented by the proposed cognitive digital twin, the extreme tail risk of delay propagation, measured by the 95% conditional value-at-risk, is substantially mitigated by a reduction of 55.9 days. Furthermore, this intervention yields a notable system throughput improvement by reducing the average project delay by 21 days, successfully stabilizing the joint confidence level for schedule and cost targets. These findings demonstrate the effectiveness of the proposed cross-scale framework to capture disruption ripple effects, quantitatively evaluate supplier concentration risk, and facilitate proactive, AI-driven actions for shipyards operating in uncertain and fragmented global supply environments.

Ensuring the reliability of full-electric harbour craft (eHC) is essential for maintaining operational availability, safe vessel performance, and cost-effective ship operation under varying service conditions. Failures in critical electrical subsystems can lead to propulsion loss, charging interruption, or power instability, resulting in unplanned downtime, reduced vessel availability, and increased maintenance burden. As electric harbour craft rely heavily on integrated electrical propulsion and power conversion systems, systematic identification of critical failure risks is necessary to support maintenance planning and improve operational resilience. This paper presents a ship-level Failure Modes, Effects, and Criticality Analysis (FMECA) framework for a generic eHC architecture to support reliability-oriented maintenance prioritisation and operational risk reduction. The analysis covers four major subsystems: propulsion, energy storage, charging, and power distribution. Critical failure modes are identified through a combination of system functional analysis, engineering judgment, and literature-based failure knowledge, and ranked using occurrence, severity, and detectability criteria through risk priority number (RPN) evaluation. Within the propulsion subsystem, stator winding degradation, rotor demagnetisation, and bearing faults are identified as major motor-related risks, while converter power module failures and DC-link capacitor failures exhibit the highest criticality. Similar high-risk failure modes are also observed in the energy storage and charging subsystems, particularly in DC/DC and bidirectional AC/DC converters, indicating that power conversion stages are dominant contributors to vessel operational risk. Across subsystems, converter-related components consistently dominate the criticality ranking, highlighting their strong influence on propulsion continuity, charging reliability, and overall power stability. In addition to criticality ranking, early fault indicators associated with each major failure mode are also summarised, including measurable thermal, electrical, and mechanical signals that can support future onboard condition monitoring and predictive maintenance development. The proposed framework provides practical guidance for prioritising monitoring targets, maintenance planning, and reliability-oriented decision making, contributing to improved ship operational performance and reduced lifecycle maintenance cost for electric harbour craft.

Title: Towards Net Zero: Modelling Offshore Carbon Costs Using an Integrated Decision-Support Dashboard Authors: 1. Denise Chew Ying Xuan (Corresponding Author: Denise Chew Ying Xuan and Email: 2301726@sit.singaporetech.edu.sg), Singapore Institute of Technology, SIT, Singapore 2. Arun Kr Dev (Co-Author and Email: a.k.dev@newcastle.ac.uk), Newcastle University in Singapore (NUiS), Singapore Abstract: Offshore support vessels (OSVs) are essential to offshore energy operations but exhibit complex operational profiles characterised by multiple activity modes and highly variable power demand. This variability leads to significant fluctuations in fuel consumption and associated carbon dioxide emissions. With the introduction of carbon pricing mechanisms such as the European Union Emissions Trading System (EU ETS), vessel operators are increasingly exposed to both environmental obligations and financial risks. This study develops an integrated analytical framework to evaluate the relationships among operational behaviour, emissions generation, and carbon cost exposure in offshore vessel operations. The proposed methodology adopts an operational mode-based modelling approach, in which vessel activities, including dynamic positioning (DP), transit, standby, towing, and port operations, are represented with distinct fuel consumption characteristics. Carbon dioxide emissions are estimated using standard fuel-based emission factors, and corresponding carbon costs are calculated in accordance with the principles of emissions trading. The framework is implemented within an interactive decision-support dashboard that integrates operational data modelling, emissions estimation and financial analysis into a unified visual analytics platform. Results indicate that operational mode significantly influences emissions output and operational expenditure. DP operations are identified as the dominant contributors to both fuel consumption and carbon cost exposure, driven by sustained high power demand. Scenario-based analysis demonstrates that variations in operational intensity, fuel price and carbon price can lead to substantial changes in total operating expenditure, highlighting the sensitivity of offshore operations to both market and regulatory drivers. The developed dashboard provides a practical yet analytically robust tool for supporting data-driven decision-making. By linking operational performance with emissions and cost implications, it enables vessel operators to evaluate operational strategies, assess carbon-pricing risks, and examine the economic feasibility of emissions-reduction technologies within a coherent decision-support framework.

Naval, offshore, and defense platform programs are increasingly challenged by rising system complexity, extended development lifecycles, and globally distributed engineering and supply chain environments. As engineering data is spread across numerous domain-specific tools - such as requirements management, systems engineering environments, simulation platforms, software development tools, and PLM systems - organizations often lack a consistent, cross-domain view of system dependencies and program progress. This fragmentation limits the ability to manage engineering changes efficiently and makes it difficult for program leadership to reliably track planned versus actual progress across complex platform developments. Conventional integration strategies frequently attempt to address these challenges by consolidating lifecycle data into a central platform. While this approach promises a unified backbone, it often results in high implementation effort, rigid system dependencies, and complex data migration initiatives that are difficult to sustain in heterogeneous engineering environments and multi-tier supplier ecosystems. This paper presents an alternative Digital Thread architecture based on a lightweight knowledge graph-driven integration layer that connects existing engineering systems without centralizing their data. The architecture preserves existing tools as Systems of Record (SoR) and Authoritative Sources of Truth (ASoT), while the knowledge graph captures semantic relationships between distributed lifecycle artifacts - including requirements, system elements, software components, verification results, and planning information - enabling an integrated view across engineering domains. To accelerate implementation, AI agents support the semi-automated and guided construction of the knowledge graph, enabling organizations to progressively establish and evolve the Digital Thread across complex tool landscapes and extended supply chains. The resulting architecture enables cross-domain traceability, integrated plan tracking with continuous planned-versus-actual visibility, and improved change impact analysis, providing a scalable foundation for more effective program steering, earlier risk detection, and more predictable delivery of complex naval, offshore, and defense platforms.

Title: CFD Analysis of Battery Room Ventilation for All-Electric Vessels: From Rule Compliance to Goal-Based Safety Authors: 1. Khor Zhi Liang Ivan (Corresponding Author and Email: 2301572@sit.singaporetech.edu.sg), Singapore Institute of Technology, SIT, Singapore. 2. Arun Kr Dev (Co-Author and Email; a.k.dev@newcastle.ac.uk), Newcastle University in Singapore (NUiS), Singapore Abstract: The increasing use of lithium-ion battery energy storage systems (BESS) in all-electric vessels introduces safety risks, including thermal runaway and the release of flammable off-gases. Current classification rules typically specify a minimum ventilation rate of six air changes per hour (ACH) for battery rooms. While this ensures adequate overall air exchange, it does not necessarily guarantee effective removal of hazardous gases in confined and obstructed spaces. This study evaluates the effectiveness of ACH-based ventilation design and proposes a practical framework to support goal-based safety assessments for marine battery installations. Three battery room configurations—baseline, compact, and extended—were assessed under identical ACH conditions to examine airflow distribution and gas dispersion behaviour. The analysis focuses on key safety indicators relevant to industry practice, including the presence of poorly ventilated regions, the persistence of hazardous gas concentrations, and gas clearance time following a release event. The results show that meeting the 6 ACH requirement does not ensure uniform ventilation. Low-velocity regions persisted in all configurations, occupying approximately 16-21% of the room volume, particularly near ceilings and behind battery racks, where gas accumulation is most likely. The compact configuration demonstrated improved performance, with reduced stagnant zones and faster gas clearance, while the extended configuration showed higher accumulation risk and slower dispersion. The study highlights the limitations of relying solely on ACH as a design criterion and recommends incorporating additional performance-based indicators, such as clearance time and identification of stagnant zones. These findings support the adoption of goal-based safety approaches, enabling more robust ventilation design and improved risk management for battery-powered vessels.