Search

Ultimate limit state (ULS) assessment examines the maximum load-carrying capacity of structures considering inelastic buckling failure. Contrary to the traditional allowable stress principle which is mainly based on experiences, the ULS assessment focuses on explicitly evaluating the structural safety margin and thus enables a consistent level of safety/risk between conventional and novel structural designs. Modern structures are usually designed as a network of plates and stiffeners (e.g., ship structures, offshore and onshore wind turbine, and land-based bridge) joined by welding which induces a residual stress field. Hence, predicting the ultimate strength reduction of stiffened panels caused by welding residual stress is a crucial problem addressed by many scholars with different approaches, among which the Nonlinear Finite Element Method (NLFEM) is the prevailing approach within the community of structural engineering. Unfortunately, the NLFEM has a high computational requirement which prevents its use in the design, appraisal, and optimisation phases of stiffened panels. To well approximate the nonlinear finite element method, a data-driven method is proposed in this paper, with a functional which is computationally expensive to build but computationally inexpensive to use allowing its application at design stage. Results obtained in different (i.e., interpolation and extrapolation) scenarios using data generated by a state-of-the-art NLFEM on a series of stiffened panels will support the proposed method.

Hull and propeller performance have a primary role in overall vessel efficiency. Vessel fouling is a common phenomenon where undesirable substances attach or grow on the ship hull. A clear understanding of the extent of the degradation of the hull will allow better management of assets and prediction of the best time for dry docking and hull maintenance work. In this paper, the authors investigate the problems of predicting the hull condition in real operations based on data measured by the on-board systems. The proposed solution uses an unsupervised Machine Learning (ML) modelling technique to eliminate the need for collecting labeled data related to the hull and propeller fouling condition. Two anomaly detection methods based on Support Vector Machines and k-nearest neighbour have been applied to predict the hull condition using the available parameters measured on-board. Data from the Research Vessel The Princess Royal has been exploited to show the effectiveness of the proposed methods and to benchmark them in a realistic maritime application.

The maintenance of the several components of a Ship Propulsion Systems is an onerous activity, which need to be efficiently programmed by a shipbuilding company in order to save time and money. The replacement policies of these components can be planned in a Condition-Based fashion, by predicting their decay state and thus proceed to substitution only when really needed. In this paper, authors propose several Data Analysis supervised and unsupervised techniques for the Condition-Based Maintenance of a vessel, characterised by a combined diesel-electric and gas propulsion plant. In particular, this analysis considers a scenario where the collection of vast amounts of labelled data containing the decay state of the components is unfeasible. In fact, the collection of labelled data requires a drydocking of the ship and the intervention of expert operators, which is usually an infrequent event. As a result, authors focus on methods which could allow only a minimal feedback from naval specialists, thus simplifying the dataset collection phase. Confidentiality constraints with the Navy require authors to use a real-data validated simulator and the dataset has been published for free use through the OpenML repository.

In this paper the authors investigate the problems of predicting the fuel consumption and of providing the best value for the trim of a vessel in real operations based on data measured by the onboard automation systems. Three different approaches for the prediction of the fuel consumption are compared: White, Black and Gray Box Models. White Box Models (WBM) are based on the knowledge of the physical underling processes. Black Box Models (BBMs) build upon statistical inference procedures based on the historical data collection. Finally, the authors propose two different Gray Box Model (GBM) which are able to exploit both mechanistic knowledge of the underlying physical principles and available measurements. Based on these predictive models of the fuel consumption a new strategy for the optimisation of the trim of a vessel is proposed. Results on real world operational data show that the BBM is able to remarkably improve a state-of-the-art WBM, while the GBM is able to encapsulate the a-priory knowledge of the WBM into the BBM so to achieve the same performance of the latter but requiring less historical data. Moreover, results show that the GBM can be used as an effective tool for optimising the trim of a vessel in real operational conditions.