State-of-the-Art Digital Twin Applications for Shipping Sector Decarbonization

Karakostas, Bill (Ed.); Katsoulakos, Takis (Ed.) Book State-of-the-Art Digital Twin Applications for Shipping Sector Decarbonization Advances in Logistics, Operations, and Management Science Provided in Cooperation with: IGI Global, Hershey, PA, USA Suggested Citation: Karakostas, Bill (Ed.); Katsoulakos, Takis (Ed.) (2024) : State-of-the-Art Digital Twin Applications for Shipping Sector Decarbonization, Advances in Logistics, Operations, and Management Science, ISBN 978-1-6684-9849-1, IGI Global, Hershey, PA, https://doi.org/10.4018/978-1-6684-9848-4 This Version is available at: https://hdl.handle.net/10419/305341 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ State-of-the-Art Digital Twin Applications for Shipping Sector Decarbonization Bill Karakostas Inlecom, Belgium Takis Katsoulakos Inlecom, Belgium A volume in the Advances in Logistics, Operations, and Management Science (ALOMS) Book Series Published in the United States of America by IGI Global Business Science Reference (an imprint of IGI Global) 701 E. Chocolate Avenue Hershey PA, USA 17033 Tel: 717-533-8845 Fax: 717-533-8661 E-mail: [email protected] Web site: http://www.igi-global.com This book published as an Open Access Book distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/) which permits unrestricted use, distribution, and production in any medium, provided the author of the original work and original publication source are properly credited. Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or companies does not indicate a claim of ownership by IGI Global of the trademark or registered trademark. Library of Congress Cataloging-in-Publication Data British Cataloguing in Publication Data A Cataloguing in Publication record for this book is available from the British Library. All work contributed to this book is new, previously-unpublished material. The views expressed in this book are those of the authors, but not necessarily of the publisher. For electronic access to this publication, please contact: [email protected]. CIP DATA IN PROCESS State-of-the-Art Digital Twin Applications for Shipping Sector Decarbonization Bill Karakostas, Takis Katsoulakos 2024 Business Science Reference ISBN: 9781668498484 eISBN: 9781668498491 This book is published in the IGI Global book series Advances in Logistics, Operations, and Management Science (ALOMS) (ISSN: 2327-350X; eISSN: 2327-3518) Advances in Logistics, Operations, and Management Science (ALOMS) Book Series Operations research and management science continue to influence business processes, administration, and management information systems, particularly in covering the application methods for decisionmaking processes. New case studies and applications on management science, operations management, social sciences, and other behavioral sciences have been incorporated into business and organizations real-world objectives. The Advances in Logistics, Operations, and Management Science (ALOMS)BookSeries provides a collection of reference publications on the current trends, applications, theories, and practices in the management science field. Providing relevant and current research, this series and its individual publications would be useful for academics, researchers, scholars, and practitioners interested in improving decision making models and business functions. Mission John Wang Montclair State University, USA ISSN:2327-350X EISSN:2327-3518 • Organizational Behavior • Networks • Operations Management • Decision analysis and decision support • Production Management • Computing and information technologies • Information Management • Political Science • Services management • Risk Management Coverage IGI Global is currently accepting manuscripts for publication within this series. To submit a proposal for a volume in this series, please contact our Acquisition Editors at [email protected] or visit: http://www.igi-global.com/publish/. The Advances in Logistics, Operations, and Management Science (ALOMS) Book Series (ISSN 2327-350X) is published by IGI Global, 701 E. Chocolate Avenue, Hershey, PA 17033-1240, USA, www.igi-global.com. This series is composed of titles available for purchase individually; each title is edited to be contextually exclusive from any other title within the series. For pricing and ordering information please visit http://www.igi-global.com/book-series/advances-logistics-operations-management-science/37170. Postmaster: Send all address changes to above address. Copyright © 2024 IGI Global. All rights, including translation in other languages reserved by the publisher. No part of this series may be reproduced or used in any form or by any means – graphics, electronic, or mechanical, including photocopying, recording, taping, or information and retrieval systems – without written permission from the publisher, except for non commercial, educational use, including classroom teaching purposes. The views expressed in this series are those of the authors, but not necessarily of IGI Global. Titles in this Series For a list of additional titles in this series, please visit: www.igi-global.com/book-series/advances-logistics-operationsmanagement-science/37170 Convergence of Industry 4.0 and Supply Chain Sustainability Muhammad Rahies Khan (Bahria University, Karachi, Pakistan) Naveed R. Khan (UCSI University, Malaysia) and Noor Zaman Jhanjhi (Taylor’s University, Malaysia) Business Science Reference • copyright 2024 • 491pp • H/C (ISBN: 9798369313633) • US $275.00 (our price) Drivers of SME Growth and Sustainability in Emerging Markets Sumesh Dadwal (Northumbrian University, UK) Pawan Kumar (Lovely Professional University, India) Rajesh Verma (Lovely Professional University, India) and Gursimranjit Singh (Dr. B.R. Ambedkar National Institute of Technology, Jalandhar, India) Business Science Reference • copyright 2024 • 308pp • H/C (ISBN: 9798369301111) • US $250.00 (our price) Navigating the Coaching and Leadership Landscape Strategies and Insights for Success Andrew J. Wefald (Staley School of Leadership, Kansas State University, USA) Business Science Reference • copyright 2024 • 306pp • H/C (ISBN: 9798369352427) • US $295.00 (our price) Strategies for Environmentally Responsible Supply Chain and Production Management Yanamandra Ramakrishna (School of Business, Skyline University College, Sharjah, UAE) and Babita Srivastava (William Paterson University, USA) Business Science Reference • copyright 2024 • 309pp • H/C (ISBN: 9798369306697) • US $275.00 (our price) Industry Applications of Thrust Manufacturing Convergence with Real-Time Data and AI D. Satishkumar (Nehru Institute of Technology, India) and M. Sivaraja (Nehru Institute of Technology, India) Engineering Science Reference • copyright 2024 • 353pp • H/C (ISBN: 9798369342763) • US $365.00 (our price) Utilization of AI Technology in Supply Chain Management Digvijay Pandey (Department of Technical Education, Government of Uttar Pradesh, India) Binay Kumar Pandey (College of Technology, Govind Ballabh Pant University of Agriculture and Technology, Pantnagar, India) Uday Kumar Kanike (Independent Researcher, USA) A. Shaji George (Crown University International, Saudi Arabia & Chartered International Da Vinci University, Nigeria) and Prabjot Kaur (Birla Institute of Technology, Mesra, India) Business Science Reference • copyright 2024 • 336pp • H/C (ISBN: 9798369335932) • US $395.00 (our price) A Critical Examination of the Recent Evolution of B2B Sales Joel G. Cohn (Everglades University, Boca Raton, USA) Business Science Reference • copyright 2024 • 276pp • H/C (ISBN: 9798369303481) • US $255.00 (our price) 701 East Chocolate Avenue, Hershey, PA 17033, USA Tel: 717-533-8845 x100 • Fax: 717-533-8661 E-Mail: [email protected] • www.igi-global.com Table of Contents Foreword .............................................................................................................................................xiv Preface .................................................................................................................................................. xv Chapter 1 Shipping Digital Twin Landscape ........................................................................................................... 1 Takis Katsoulakos, Inlecom, Belgium Georgia Tsiochantari, Inlecom, Belgium Fearghal O’Donncha, IBM Research, Ireland Eleftherios Kaklamanis, EURONAV, Belgium Allesandro Maccari, RINA, Italy Marcin Mucharski, RMDC, Poland Chapter 2 A Digital Twin Approach for Selection and Deployment of Decarbonization Solutions for the Maritime Sector .................................................................................................................................... 26 Anargyros Spyridon Mavrakos, Inlecom, Belgium Theodosis Tsaousis, Inlecom, Belgium Nicolo Faggioni, Ficantieri, Italy Alessandro Caviglia, Ficantieri, Italy Eros Manzo, CNR, Italy Chapter 3 Shipping Digital Twin Data Management With the Use of Knowledge Graphs .................................. 53 Antonis Antonopoulos, Konnecta, Greece Antonis Mygiakis, Konnecta, Greece Chapter 4 Towards Intelligent Ship-Edge Computing Enabling Automated Configuration of Ship Models and Adaptive Self-Learning .................................................................................................................. 73 Fearghal O’Donncha, IBM Research, Ireland John D. Sheehan, IBM Research, Ireland Maroun Touma, IBM, USA Sofiane Zemouri, IBM Research, Ireland Rob H. High, IBM, USA  Chapter 5 Shipping Green Fuel Strategies and Benchmarking Supported by Digital Twins ................................ 94 Anargyros Spyridon Mavrakos, Inlecom, Belgium Maxime Woznicki, CEA, France Chapter 6 Enhanced and Holistic Voyage Planning Using Digital Twins ........................................................... 112 Dimitris Kaklis, DANAOS, Greece Antonis Antonopoulos, Konnecta, Greece Chapter 7 Digital Twins for Synchronized Port-Centric Optimization Enabling Shipping Emissions Reduction ............................................................................................................................................ 137 Efstathios Zavvos, VLTN BV, Belgium Konstantinos Zavitsas, VLTN BV, Belgium Charilaos Latinopoulos, VLTN BV, Belgium Veerle Leemen, VLTN BV, Belgium Aristides Halatsis, VLTN BV, Belgium Chapter 8 Application of Digital Twins in the Design of New Green Transport Vessels ................................... 161 Austin A. Kana, Delft University of Technology, The Netherlands Wenzhu Li, Delft University of Technology, The Netherlands Isabel van Noesel, Delft University of Technology, The Netherlands Yusong Pang, Delft University of Technology, The Netherlands Aleksandr Kondratenko, Aalto University, Finland Pentti Kujala, Estonian Maritime Academy, Estonia Spyros Hirdaris, American Bureau of Shipping, Greece & Department of Mechanical Engineering, Aalto University, Espoo, Finland Chapter 9 A Ship Digital Twin for Safe and Sustainable Ship Operations ......................................................... 192 Spyros Hirdaris, American Bureau of Shipping, Greece Mingyang Zhang, Aalto University, Finland Nikos Tsoulakos, Laskaridis Shipping Co. Ltd., Greece Pentti Kujala, Taltech, Estonia Chapter 10 Shipping Applications of Digital Twins ............................................................................................. 221 Angeliki Deligianni, Glafcos Maritime Ltd., Greece Leonidas Drikos, Glafcos Maritime Ltd., Greece George Mantalos, Starbulk, Greece Eleftherios Kaklamanis, EURONAV, Belgium  Chapter 11 Enhancing a Digital Twin With a Multizone Combustion Model for Pollutant Emissions Estimation and Engine Operational Optimization .............................................................................. 247 Theofanis Chountalas, National Technical University of Athens, Greece Compilation of References ............................................................................................................... 287 About the Contributors .................................................................................................................... 312 Index ................................................................................................................................................... 318 Detailed Table of Contents Foreword .............................................................................................................................................xiv Preface .................................................................................................................................................. xv Chapter 1 Shipping Digital Twin Landscape ........................................................................................................... 1 Takis Katsoulakos, Inlecom, Belgium Georgia Tsiochantari, Inlecom, Belgium Fearghal O’Donncha, IBM Research, Ireland Eleftherios Kaklamanis, EURONAV, Belgium Allesandro Maccari, RINA, Italy Marcin Mucharski, RMDC, Poland The evolution of ship computerization towards digital twinning (DT) has been gradual, having its roots in the 1970s, when the first automated navigation and control systems were developed. Over the course of the past decades, the increased automation of ship functions, coupled in ICT advances, paved for the development and analysis of highly realistic ship models. These models are now enhanced and supported by sensor technologies that provide real-time data from ships. This chapter explores the transformative potential of digital twin technology to create virtual replicas of ships, their systems and broader shipping processes. These digital twins empower decision-makers in various stages, including ship design and operational management, both onboard and ashore, regarding ship and fleet management as well as optimised integration in multimodal transport networks. Additionally, they facilitate optimized integration within multimodal transport networks. The chapter provides insights into current state-of-the-art (SOTA) solutions, recent advancements, and emerging approaches in the maritime industry. Furthermore, the chapter delves into the regulatory aspects associated with the adoption of digital twins in the shipping sector, shedding light on potential risks and limitations. To assist in understanding and implementing digital twins effectively, the chapter introduces a comprehensive shipping digital twining architecture and a capabilities model. These frameworks can accommodate diverse technologies, enabling different levels of ambition and customization in the realm of shipping digital twins. Preface Shipping accounts for 80% of global trade. Roughly 50,000 ships carry 90% of the world’s traded cargo every year, and most of these ships run on heavily polluting oil known as bunker fuel. According to the International Maritime Organisation (IMO), the share of shipping emissions in global anthropogenic emissions has increased from 2.76% in 2012 to 2.89% in 2018. IMO has mandated therefore, that all ships built in the future must reduce pollution from today’s averages. The World Commission on Environment and Development (WCED), defines sustainability as development that meets the needs of the present generation without compromising the ability of future generations to meet their own needs. Technology whose use is intended to mitigate or reverse the effects of human activity on the environment is known as green technology. Consequently, technological developments that reduce environmental effects due to waterborne transportation is known as green ship technology. There has been recently a new impetus for green technology application in shipping, towards halving current emission levels by 2050. An information technology with a great potential to support the acceleration of green shipping technologies is digital twins. A digital twin is a representation of a physical entity in a digital format. A ship digital twin is a digital replica of the real (physical) ship in terms of its structure (e.g. hull type, component layout, hull parameters), its equipment (e.g. engines, propeller, rudder) and its behaviour and functions (e.g. propulsion, navigation, loading), as well as its integration in a fleet management system or multimodal supply chains. It provides a unique, intelligent ship model merging technical specifications, component models, and parameters with management information on the components and processes of the ship, ultimately enabling computerised simulation and optimisation of all its functions along a matrix of performance metrics that can emphasise environmental goals. Digital twins (contrary to similar concepts such as ‘digital shadows’) use bi-directional communication links with the ICT infrastructure on the physical ship. The communication link from the ship to the digital twin is used to monitor the physical ship constantly through several data collection techniques and devices, and communication channels to transfer this information. This allows the (virtual) digital twin to constantly learn from its physical counterpart and evolve mirroring its lifecycle. A digital twin can therefore be used in place of its physical counterpart to carry out simulations, analyse data and make predictions in order to prevent unnecessary outcomes, reduce downtime, redesign and improve equipment or processes. Recently, digital twins have been receiving attention in the context of green shipping. This has been largely accelerated by the emergence of the smart ship, i.e. ships where a large number of sensors are installed to collect all possible information about the state of the vessel, various indicators, etc. At the same time, the maturity of digital twin technologies has been advancing rapidly, with the growth of digital twin underpinning technologies such as Internet of Things (IoT), Big Data Machine xv Preface Learning, Cloud and Edge computing. This has made possible the use of digital twins in areas such as new ship design. Digital twin technology has started to penetrate through the whole ship lifecycle to include activities such as the design of more energy efficient ships, the installation and retrofitting of energy efficiency subsystems, greener fuels, the optimisation of the ship’s voyage, operation, and finally the decommissioning or retire stages. While digital twining advances have been mainly concentrating to manufacturing (i.e. in the emergent Industry 4.0), research in the application of digital twins to the shipping sector lags behind, with the maritime industry registering lower levels of digital twin research and development than aerospace or automotive engineering. To reduce this gap, several national and international initiatives have been introduced, notably the European Commission’s research and development in waterborne transport (https://research-and-innovation.ec.europa.eu/research-area/t ransport/waterborne_en), aiming to balance optimal energy use with environmental impact. Within this context, the theme of this book, i.e. the application of digital twin technologies to green shipping is both novel and timely. The shipping and maritime ecosystem is a complex one. Decarbonisation of shipping with the use of digital twins is a paradigm that spans technical and business disciplines. The book therefore Is both authored by a truly interdisciplinary team and addresses a broad audience of readers, including: • Ship owners /operators. • Ship Designers & Builders (academia, consultancies, shipbuilders). • Ship building service providers (engineering services, retrofitters, maintainers). • Shipping Decarbonisation solution providers, including engine manufacturers, green fuel producers, alternative energy manufacturers. • Shipping ICT services and application providers • Classification societies. • EU and national government policy makers • Maritime Authorities such as port authorities. • Researchers in a broad range of disciplines including naval architecture and engineering and computer science. To understand the context and scope of this book, the diagram shown in the figure needs to be consulted. According to that diagram, green shipping can be classified along a number of parallel streams of innovations, products and technologies that address different perspectives. Underlying all these alternative green shipping approaches is however, Information Technology and more specifically, digital twins. The book therefore covers all different spheres of green shipping activity under the prism of a unified paradigm of digital twins. In the following section we illustrate how the different chapters of the book address the above green shipping areas and collectively construct an integrated view of green shipping through the prism of the digital twin paradigm. Chapter 1 (“Shipping Digital Twin Landscape”), provides a historical overview, as well as the current state-of-the-art solutions, recent advancements, and emerging approaches in the maritime industry. Furthermore, the chapter delves into the regulatory aspects associated with the adoption of digital twins in the shipping sector. Chapter 2 (“A Digital Twin Approach for Selection and Deployment of Decarbonisation Solutions”), main focus is on how Digital Twining can support the selection of decarbonisation technologies and xvi Preface operational strategies in designing decarbonisation solutions in a rolling time-horizon to meet regulations with the goal of achieving green shipping (zero-emission shipping) by 2050. Furthermore, the chapter illustrates the application of DTs to energy production, distribution, recovery onboard process management, and hull performance prediction utilising simulations. Chapter 3 (“Shipping Digital Twin Data Management With The Use Of Knowledge Graphs”), presents a model for shipping operations management expressed as a Knowledge Graph that provides data management functionality, and links ship data to operational models and automates model pipeline executions. The Knowledge Graph serves as a host for the digital twin of the vessel and functions as a semantic layer that integrates structural and operational ship models and facilitates model pipeline execution as well as interlinking of digital twins. Chapter 4 (“Towards Intelligent Ship Edge Computing Enabling Automated Configuration Of Ship Models And Adaptive Self-Learning”), delves into the significance of edge computing in the shipping industry, outlining its ability to enhance operational efficiencies. It explores the specific user requirements Figure 1. A classification of green shipping technologies xvii Preface for an effective edge computing solution and highlights the role of AI in enabling scalable and continuous computation. Additionally, the chapter provides a comprehensive overview of edge infrastructure, platform requirements, and considerations pertaining to data and AI at the edge. Chapter 5 (“Shipping Green Fuels Strategies And Benchmarking Supported By Digital Twins”), explains how digital twin’s simulation capabilities, can be used to model complex energy systems and alternative fuels and compute emissions, power consumption/output, etc., virtually. The Chapter provides a comparison of alternative marine fuels, in terms of storage requirements and suitability of energy converters such as combustion engines and fuel cells. Chapter 6 (“Enhanced and Holistic Voyage Planning Using Digital Twins”), presents techniques and approaches to optimize a ship’s voyage using digital twins in terms of environmental and business parameters controlling ship speed, trim, route, as well as estimated time of arrivals. It demonstrates how voyage planning is enhanced taking into account actual current vessel status reflected by the digital twin. The theoretical backbone of Voyage Planning entails a multitude of state-of-the-art processes from trajectory mining and shortest path algorithms to multi constraining optimization by including a variety of parameters to the initial problem, such as bunkering, JIT and hull condition. Chapter 7 (“Digital Twins For Synchronised Port Centric Optimisation Enabling Shipping Emissions Reduction”), discusses the context for emission regulations, port operations and hinterland capabilities, and frames a general methodology for port authorities, terminal operators, shipping companies and logistics operators to implement collaborative, emission reducing practices such as slow steaming and synchro-modality. Chapter 8 (“Application Of Digital Twins In The Design Of New Green Transport Vessels”), discusses the role of optimization in the early design phases. Two complimentary design approaches are proposed aimed at the concept design phases of green ships, and an optimization based DT design approach covering the early concept design phase, are presented. Chapter 9 (“A Ship Digital Twin For Safe And Sustainable Ship Operations”), discusses the potential of AI based digital twin models to monitor ship safety and efficiency. A paradigm shift is introduced in the form of a model that can predict ship motions and fuel consumption under real operational conditions using deep learning models. A bi-directional Long Short-Term Memory (LSTM) network with attention mechanisms is used to predict ship fuel consumption and a transformer neural network is employed to capture ship motions in realistic hydrometeorological conditions. Chapter 10 (“Shipping Applications of Digital Twins”), presents three applications of digital twin in shipping, namely the predictive maintenance of ship machinery, cargo load area cleaning and hull biofouling treatment. The Chapter discusses the business importance of each of these applications and surveys the current state of the art practices. The Chapter then illustrates novel approaches that utilise digital twins and bring improvements in terms of costs, safety and environmental impact. Lastly, Chapter 11 (“Enhancing A Digital Twin With A Multizone Combustion Model For Pollutant Emissions Estimation And Engine Operational Optimization”), presents a sophisticated multizone combustion model used to estimate engine performance and NOx emissions. This model enables realtime assessment of emissions based on basic engine operational parameters and functions as a decision support system for forecasting emissions across various fuel types and engine operational schemes. This book aims to inform researchers and practitioners alike about the State of the Art in digital twin applications for shipping sector decarbonisation, as well as all important emerging developments in the area. It contains expert views from a broad array of disciplines such as naval architecture, naval engineering, shipping operations management, ship modelling and simulation, and Computer Science, xviii Preface on how to apply digital twinning for maximum waterborne industry decarbonisation. The book addresses a timely and important issue at the intersection of a societal and business imperative (maritime industry decarbonisation) with a promising emerging technology of digital twins. Review of current developments of digital twins in the maritime industry, indicates that mostly conceptual papers are available and even fewer industry applications and case studies. This indicates an imbalance between the levels of research and development of digital twins for maritime compared to other areas such as aerospace and manufacturing. The book aims to address such imbalance with a comprehensive review of the current state of digital twins in waterborne transportation, in a green shipping context, together with a set of key industrial applications of digital twins, both existing as well as emerging ones. Hopefully, in the chapters of this book the readers will find useful insights and valuable knowledge about the true potential of digital twins for ship decarbonisation, and subsequently use it to further research and practice. Bill Karakostas Inlecom Group, Belgium Takis Katsoulakos Inlecom Group, Belgium REFERENCES Antonopoulos, A., Karakostas, B., Katsoulakos, T., Mavrakos, A., Tsaousis, T., & Zavvos, S. (2023). A digital twin enabled decision support framework for ship operational optimisation towards decarbonisation. Eight International Congress of Information and Communication Technology (ICICT), London. 10.1007/978-981-99-3091-3_38 Armstrong, V. (2013). Vessel optimisation for low carbon shipping. Science Direct. https://www.sciencedirect.com/science/article/pii/S0029801813002643. Daniel, L., Pfeiffer, J., Tinsel, E., Strljic, M., Sint, S., Vierhauser, M., Wortmann, A., & Wimmer, M. (2022). Digital Twin Platforms: Requirements, Capabilities, and Future Prospects. IEEE Software. IEEE. Energy Transitions Commission. (2018). Mission Possible - Reaching Net-Zero Carbon Emissions From Harder-To-Abate Sectors By Mid-Century - Sectoral Focus Shipping. Energy Transmission Commission. https://www.energy-transitions.org/publications/mission-possible-sectoral-focus-shipping/#downloadform Fonseca, A. Í., & Gaspar, H. M. (2021). Challenges when creating a cohesive digital twin ship: A data modelling perspective. Ship Technology Research, 68(2), 70–83. doi:10.1080/09377255.2020.1815140 Global Industry Alliance. (n.d.). GIA to Support Low Carbon Shipping. GIA. https://greenvoyage2050. imo.org/about-the-gia/ International Transport Forum. (2018). Decarbonising Maritime Transport. Pathways to Zero-Carbon Shipping by 2035. ITF, Paris. xix Preface ISO. (2018). Ships and marine technology — Shipboard data servers to share field data at sea. ISO. Mauro, F., & Kana, A. A. (2023). Digital twin for ship life-cycle: A critical systematic review. Ocean Engineering, 269, [113479]. https://doi.org/. doi:10.1016/j.oceaneng.2022.113479 Mofor, L., Nuttall, P., & Newell, A. (2015). Renewable Energy Options for Shipping. IRENA. doi:10. 1080/00207543.2018.1443229 Paul, G., Walsh, C., Traut, M., Kesieme, U., Pazouki, K., & Murphy, A. (2018). Assessment of full life-cycle air emissions of alternative shipping fuels. Journal of Cleaner Production, 172(20), 855–866. Taylor, N., Human, C., Kruger, K., Bekker, A., & Basson, A. (2019). Comparison of Digital Twin Development in Manufacturing and Maritime Domains. In: International Workshop on Service Orientation in Holonic and Multi-Agent Manufacturing, (pp. 158–170). Research Gate. xx 1 Copyright © 2024, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. Chapter 1 DOI: 10.4018/978-1-6684-9848-4.ch001 ABSTRACT The evolution of ship computerization towards digital twinning (DT) has been gradual, having its roots in the 1970s, when the first automated navigation and control systems were developed. Over the course of the past decades, the increased automation of ship functions, coupled in ICT advances, paved for the development and analysis of highly realistic ship models. These models are now enhanced and supported by sensor technologies that provide real-time data from ships. This chapter explores the transformative potential of digital twin technology to create virtual replicas of ships, their systems and broader shipping processes. These digital twins empower decision-makers in various stages, including ship design and operational management, both onboard and ashore, regarding ship and fleet management as well as optimised integration in multimodal transport networks. Additionally, they facilitate optimized integration within multimodal transport networks. The chapter provides insights into current state-of-the-art (SOTA) solutions, recent advancements, and emerging approaches in the maritime industry. Furthermore, the chapter delves into the regulatory aspects associated with the adoption of digital twins in the shipping sector, shedding light on potential risks and limitations. To assist in understanding and implementing digital twins effectively, the chapter introduces a comprehensive shipping digital twining architecture and a capabilities model. These frameworks can accommodate diverse technologies, enabling different levels of ambition and customization in the realm of shipping digital twins. Shipping Digital Twin Landscape Takis Katsoulakos Inlecom, Belgium Georgia Tsiochantari Inlecom, Belgium Fearghal O’Donncha https://orcid.org/0000-0002-0275-1591 IBM Research, Ireland Eleftherios Kaklamanis EURONAV, Belgium Allesandro Maccari RINA, Italy Marcin Mucharski RMDC, Poland This chapter published as an Open Access Chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/) which permits unrestricted use, distribution, and production in any medium, provided the author of the original work and original publication source are properly credited. 2 Shipping Digital Twin Landscape DIGITAL TWINS In the most general term, a digital twin (DT) represents a digital counterpart of a physical entity, encompassing its entire lifecycle, continually updated with real-time data, and serving as a decision support tool (IBM-A, nd). The digital twin concept has been around since the 1960s; however, it has started to gain traction only relatively recently, due to technological advances that we discuss later in the chapter. Digital Twin models are gaining more and more interest for their potentials and strong impact in application fields, such as manufacturing (Lu et al., 2020), aerospace (Li et al., 2022), healthcare (Copley, 2018), and medicine (Hempel et al., 2019). The potential advantages that organizations can gain through the implementation of digital twins are substantial. In the medium to long term, having a digital twin is anticipated to become a prerequisite for competitiveness, particularly in manufacturing, transport logistics, and shipping. Indeed, digital twin technology earned recognition as one of the top ten strategic technologies by IT market research firm Gartner (Gartner, 2017). Digital twins now stand as a business imperative that covers the entire lifecycle of assets and processes, forming the foundation for connected products and services. SAP underscores that companies failing to embrace the digital twin mandate risk falling behind (Forbes, 2017). This growing imperative is driven by the increasing significance of data and digital technologies in contemporary business operations. Furthermore, digital twins empower companies to bridge physical assets with digital data, enabling the development of new products and services that are more responsive to customer needs. The maritime industry has already witnessed the applications of digital transformation in automated procedures, operational measurement and control, and the creation of digital products and product information models (Erikstad, 2018; 2019). The primary drivers for the adoption of digital twins in shipping encompass: 1. Technology: The development of new technologies, such as the Internet of Things (IoT), Edge Computing and advanced data analytics, has made it possible to create and use digital twins for maritime applications. This has led to a growing number of companies offering digital twin solutions for ships, which can provide valuable insights into ship performance and operations. 2. Business drivers: Ship owners and operators are increasingly recognizing the benefits of digital twin technology for improving efficiency and reducing operational costs. The use of digital twins can help to identify areas for improvement, reduce maintenance costs, and minimize the risk of operational downtime. These benefits are driving the adoption of digital twin technology in the maritime industry. 3. Environmental Regulation: The maritime industry is highly regulated, and there is a growing focus on improving safety and reducing the environmental impact of shipping. The use of digital twins can help to support these regulatory efforts by providing real-time monitoring and analysis of ship operations, allowing operators to identify and address potential safety or environmental issues before they become problematic. 4. Shipping value chains and ecosystem: The shipping/maritime industry is a complex ecosystem made up of a diverse range of stakeholders, including ship owners and operators, shipbuilders, classification societies, and port authorities. The use of digital twin technology is being embraced by these stakeholders as a way to improve efficiency, reduce costs, and enhance safety and environmental performance. 3 Shipping Digital Twin Landscape In summary, digital twinning creates substantial business value in shipping by effectively managing the complexities arising from the interdependencies between technical, operational, chartering, safety, and regulatory perspectives, all of which impact ship operators’ performance metrics. Projected benefits, as indicated by RINA (nd), include the potential for a significant reduction in ship operating expenditures (up to 40%) and a decrease in port time (up to 30%). Equally important, DT-enhanced shipping companies are expected to handle increased volumes, resulting in higher revenue and profitability, while shipbuilding costs may decrease by 15-20%. This chapter is dedicated to exploring the current state and future potential of digital twins in the shipping and waterborne sector. Subsequent sections will delve into:: The historical background to digital twin for shipping (Section 2). General concepts and principles related to shipping digital twins (Section 3). Approaches for the development of digital twins (Section 4). Digital Twin development drivers in shipping industry (Section 5). The application landscape for digital twins in a shipping context (Section 6). An architecture and capabilities model for shipping DTs (Section 7). Current trends, conclusions and directions and future outlook of digital twins in shipping/maritime (Section 8). HISTORICAL BACKGROUND OF SHIPPING DTS AND OF THEIR PRECURSORS While shipping digital twinning is currently at its infancy, ship digitisation has been widely researched and is increasingly used today (DNV, 2021). The use of computer technology on ships has its roots in the 1960s, when the first automated navigation and control systems were developed. These early systems had limited capabilities, but they started the integration of computer technology in the maritime industry. In the 1970s and 1980s, more advanced computer systems were developed for use on ships, including automatic radar plotting aids (ARPA), shipboard management information systems (SMIS), Advanced Alarm Handling (AAH), and ship condition monitoring and planned maintenance systems. These systems were designed to improve the efficiency and safety of shipping operations by automating various tasks and processes. In the 1990s and early 2000s, Integrated Bridge Systems (IBS), combined multiple ship systems and functions into a single ship management platform. This allowed for greater automation and increased situational awareness for the officers and crew, improving safety and efficiency. Notably, e-Maritime introduced by the European Commission working closely with industry stakeholders, member states, is a concept that predates digital twins in shipping, and it aims to harness the potential of digital technologies by ameliorating complexities that hinder networking of different stakeholders, helping to increase automation of operational processes particularly compliance management and facilitating the management of information from disparate sources to assist decision making. Closely associated with the Ship DT concepts are also developments in e-navigation, initiated in the late 90’s and becoming part of the International Maritime Organisation’s (IMO) strategy. E-navigation is defined as “the harmonized collection, integration, exchange, presentation and analysis of marine information on board and ashore by electronic means to enhance berth to berth navigation and related 4 Shipping Digital Twin Landscape services for safety and security at sea and protection of the marine environment.” (IMO, nd). E-navigation is “intended to meet present and future user needs through harmonization of marine navigation systems and supporting shore services. The development and implementation of the concept is coordinated by the International Maritime Organization in accordance with the “E-Navigation Strategy Implementation Plan – Update 1” adopted in 2018 by the Maritime Safety Committee (MSC). Potential synergies are self-evident particularly on ship information standardisation. More recent developments include the creation of ‘smart’ ship’ or cyber-enabled ships, referring to ships equipped to monitor the state of the vessel through onboard sensors enabling increasing ship automated operation. A smart ship is equipped with automation that provides multiple systems control and data driven decisions (Reilly and Jorgensen, 2016). In smart or cyber-enabled ships (Lloyds Register, 2018), onboard sensors monitor the state of the vessel through and are essentially utilising the maturing developments for autonomous ships (IBM-B. nd). Digital technology relies on the ship-to-shore connectivity using hardware equipment on board vessels and sharing data and information to support decision-making either on board or at shore. Integration of data from navigation, propulsion, cargo management and energy systems, provides the means to manage effectively ship and fleet performance and operations. A typical ship monitoring system is depicted in Figure 1. A layered model of shipping processes produced by the eMAR EU project (cordis.europa.eu/project/ id/265851) is shown in Figure 2, providing an overview of the diversity of processes and technologies interacting in a ship operational management context. Figure 1. Significant ICT developments facilitating shipping digitalisation 11 Shipping Digital Twin Landscape Addressing problems like berth allocation models, improving punctuality, or optimising the use of cargo space, can be achieved with digital twin models. A real-time data-fed DT can help coordinate port operations with synchronized operational planning (coordinated arrivals, slow steaming etc.). Owners of cargo, good dealers, and customers waiting for the deliveries constantly need to track their shipments with transparency over their movement across ports. To meet such requirements DTs, allow simulation and analysis of operations, allowing real time adjustments to plans. Connected DTs, can report to key stakeholders on infrastructure investments and serve as an investigation tool across ports linked like the web. This enriches the awareness of each situation in the long-term run and creates a portal for collaborative decision-making over goals such as emission control measures. The dynamically updated DT makes it easier to understand operational parameters of remote assets such as ships, oil rigs etc. This makes possible the estimation of risk levels, structural reliability, conduct tests to make it more productive by just analysing data. DT can help optimize inspection regimes, making the cost of expenditure transparent and providing visibility and traceability. DT ARCHITECTURES AND DEVELOPMENT APPROACHES DT Data Management Building on the previously discussed technologies, a digital twin platform requires the underpinning of technologies such as Industrial Internet of Things (IIoT) as well as ship blueprint models, both for the ship as a whole, as well as for its individual subsystems. The vast amounts of data generated by ships and their systems must be collected, processed, and stored in a centralized database to allow for real-time monitoring and analysis. This requires the implementation of robust data management systems, including data security protocols, to ensure the integrity and confidentiality of the data, Increasingly, ship data used to build the digital twin will be collected automatically, from sensors. Internet of Things (IoT) is a digital technology that allows the control of machinery remotely by using machine to machine communication using digital signals. When IoT is used on ships, it allows remote and unmanned operations of machines thereby making the operation safer and efficient, reducing maintenance, downtime, and fuel consumption. This effectively allows the reduction of carbon emissions as a result of efficient operations (Plaza-Hernández, et al., 2021). As the volume and heterogeneity of data continues to grow, data gravity is an increasing challenge facing organisations. To eliminate connectivity limitations and ensure continued operations across 100s of ships each containing 1000s of sensors and devices, edge computing is developing as part of a multicloud paradigm that ensures that ship stakeholders can leverage digital twin capabilities in any locations, with safety mechanisms around software security and data sovereignty. An important area for development is specifying a ‘standard’ data acquisition system identifying the key sensors to be used for each ship subsystem. To keep track of metadata and to structure the storage and processing of sensor data, there is a need for a unique identification of sensors as well as the components and systems subject to monitoring by sensors. In that direction, a standardised sensor naming scheme, developed by DNV (Geneva, 2017) provides a reference in this area. Data used to build the digital twin will need to be based on some reference schema(s). Standardisation is expected to speed up the adoption of digital twinning. Standard industry wide schemas will therefore 12 Shipping Digital Twin Landscape need to be introduced, while for data types that are shared within different domains (e.g., geospatial data) existing models can be utilised. Ship data models are used by classification societies and provide an important baseline in this area. A notable example is the DNV Functionally Oriented Vessel Data Model (DNV, nd) utilising the principles outlined in ISO 15926A. Ship functions include structural integrity, anchoring, propulsion, navigation, fire prevention, etc. Remote monitoring with the use of IoT devices and other technologies, enables monitoring of a ship’s performance and operations from a remote location enabling optimization of ship operations and reduce costs. Edge computing (subsection 6.4??), supports decision making on the ship and provides autonomy in case of connectivity limitations or outages. Approaches for Developing Digital Twins Various types of modelling techniques can be employed for the construction of digital twins. ‘White Box’ approaches rely on creating analytical models of the ship where relationships between the model variables are represented by mathematical models (e.g., differential equations) that are derived theoretically and are validated from experimental data, e.g., from towing tank or sea trials, or from actual operational data collection (e.g. ‘noon reports’). Computational Fluid Dynamics (CFD) are used to model the flow of fluids, such as air or water, around a ship. to predict the performance of a ship’s propulsion system, including its speed and fuel consumption, as well as its manoeuvrability and stability (Wang and Dan, 2020). In contrast to white box methods, ‘Black Box’ or data-driven models utilise mainly statistical techniques to derive the relationships between the digital twin data, without reliance on a prior model of each component or process that generates them. Black box approaches establish relations (functions) between variables that can be used for the prediction of future states. For instance, consumed energy or fuel can be modelled as a function of the speed and other ship related input variables (e.g., trim, displacement, hull condition), along with the environment related input variables (wind, current, wave…). The function need not be known from the start, but can be derived, e.g., using regression, based on data collected from the vessel in question. Amongst the black box model prediction techniques. We can distinguish three types of data-driven depending on intended usage: Descriptive models provide a snapshot of the current state or behavior such as ship health condition, navigation status, port status, cargo status etc., based on real-time monitoring of the ships, subsystems and voyage information (sensor data integration, data aggregation, and visualization techniques) Predictive models that use historical data to forecast future behaviors or conditions of the system. They enable proactive decision-making by anticipating issues before they occur using Time-series forecasting, machine learning models, and statistical algorithms Prescriptive models that not only predict future outcomes but also recommend actions to achieve desired outcomes using prescriptive analytics and optimization algorithms, Machine learning algorithms, such as reinforcement learning, can be applied to digital twins to enable autonomous learning from new data Federated learning provides a privacy-preserving technique that allows multiple ship digital twins to collaboratively train a shared model while keeping their data decentralized Machine Learning (ML) based algorithms have found significant impact in many fields of engineering including naval engineering, to estimate the ship’s power and fuel consumption. Accurate voyage 13 Shipping Digital Twin Landscape fuel consumption prediction under variable sea environment has been reported using Artificial Neural Networks (ANNs) and Multi-Regression (MR) techniques (Panda, 2023). The use of black box approaches in machine learning and artificial intelligence come with several disadvantages, with the need for comprehensive quality datasets being one of them. Also, predictions are made without clear explanations and many black box models, especially deep learning models, require significant computational resources and time for both training and inference. Finally, hybrid (or ‘grey box’) models, combine the benefits of the white box and black box approaches discussed above, meaning that the physical laws governing the relationships between the ship variables (white box) can be used where feasible, together with statistical approaches (black box), in order to decrease the error margin and give better prediction accuracy with reasonable computational times (Leifsson et al., 2008). In practice, when the grey model relies more on the physical ship model than the volume of the historical data, it requires deep initial information about the ship’s physical characteristics to obtain good results. On the other hand, when the model employs fewer physical assumptions, and relies more on historical data and/or empirical rules, it requires a broader set of data describing as many as possible of the operational conditions. Real-time simulation models of a ship’s behaviour allow for the continuous monitoring of key parameters, such as speed, fuel consumption, and propulsion performance, and the prediction of how these parameters will change over time (Erikstad, 2017). System Dynamics: simulation method that focuses on the behaviour of complex systems over time. It is used to model the interconnections between the various components of a ship, including the propulsion system, cargo handling, and power generation, to predict how the system will behave under different conditions. This allows designers to identify potential safety issues and make changes to the design to improve the ship’s stability, manoeuvrability, and other safety-critical parameters. DT DRIVERS The shift from traditional Information Technology (IT) to digital transformation is a major trend in the maritime industry, driven by advances in AI, IoT, network connectivity, and edge computing. A digital culture can help to drive innovation and improve the efficiency of maritime operations by providing real-time insights, optimizing operations, and improving safety and sustainability. The World Economic Forum estimates that the digital transformation in logistics will be valued at $4 trillion: $1.5 trillion for logistics stakeholders and $2.4 trillion worth of societal benefits. The datadriven information services will bring $810 billion to the industry as these services are used to optimise routes, reduce costs and improve utilisation (World Economic Forum, nd) Essentially, Digital twining can create business value by managing complexity when dealing with the interdependencies between technical, operations, chartering, safety, regulations, and other aspects that affect all performance metrics for ship operators. This can generate 40% reductions in ship operating expenditure, and 30% reductions in port time (RINA, nd). Equally important, the enhanced shipping company capabilities are expected to increase cargo volumes, revenue and profitability, while shipbuilding costs are expected to decrease by 15-20% (RINA, nd). The trend towards the adoption of digital twin technology in the maritime industry is being driven by the need for increased efficiency, safety, environmental performance and competitiveness. The environment and competitiveness are two key battlefields that shipping companies try to navigate. The industry has identified several economic advantages in driving the energy transition – by implementing 14 Shipping Digital Twin Landscape more energy efficient technologies. These drivers include increased demand for eco-efficient vessels from shippers and charterers, as well as hesitancy and concern from insurers and investors regarding stranded carbon assets (International Transport Forum, 2018). In this section we identify key stakeholders, considerations, opportunities and pitfalls of DT introduction in the shipping sector. Digital Transformation of Waterborne Assets Digital Transformation of waterborne assets is driven by production needs, access and use of data, continuous evolution of digital technology and the development of new business-driven applications answering a variety of goals (environmental, societal, economical). Digital Transformation can support the optimisation of assets and operations, but the commercial return on investment may be a long-term, multi-stakeholder and complex journey, sometimes requiring the introduction of new business models. Therefore, Digital Twins need to be considered as part of an integrated and evolving process, i.e., iterative, interactive and cyclical, in which the developments in design and operation are interacting and converting to reach the expected business expectations. Digital twins and associated data are ship and application specific, and this increases their complexity. The general risk is the proliferation of isolated, one-off developments leading to a lack of coherence or even incompatible solutions. A proper digital governance would enable coordinated development and integration of specific digital applications and technologies, with due regards to industry goals and strategies and business requirements. Legislative Imperatives Such as Decarbonisation Related Regulation Digital twining can play a pivotal role in supporting the transition to zero emissions shipping, which is addressed in Chapter 2 of this book. New vessels are gaining in complexity particularly as they introduce novel decarbonisation technologies. Optimal performance is only achieved when all subsystems are working optimally; together as one system. The assessment of the functioning of all subsystems and overall system behaviour is getting increasingly difficult to MANAGE requiring new simulation models to ensure success and paving the way for the use of DT solutions (Giering and Dyck, 2021.) Green shipping corridors are attracting attention as a way to accelerate decarbonisation and adoption of green ships, because these can help mitigate the risks associated with the introduction of zero-emission ships/fuels, as the risks are shared by all the stakeholders operating in the green corridor (Joerss et al, 2021). Digital Governance: A Regulatory Perspective A modern, future-oriented, harmonised regulatory framework is needed to provide continuous, seamless and transparent governance and regulatory support to stakeholders, towards faster and easier implementation of digital models and applications. Digitalisation governance in shipping should ideally involve not only authorities and regulators (IMO, EC, flag state administrations, port authorities…) but also, all key stakeholders: ship owners, ship operators, shipyards, ship designers, ship system suppliers, cargo owners, port operators, classification 15 Shipping Digital Twin Landscape societies. It is noted that, so far, THE International Association of Classification Societies (IACS) has not approved the establishment of a dedicated working group on Digital Twins and related issues. The digital transformation of the waterborne sector should follow a transversal process, based upon harmonised and standardised regulations to enable a consistent and future proof development, validation and implementation of tools, models, systems and technologies. The involvement of regulatory bodies is essential to streamline these applications on a level playing field, enabling continuous compliance but also allowing a continuous evolution to match the technology progress (e.g., Digital Twin models and software simulations). Classification Societies: A Potential Catalyst for Adoption of Ship DTs Classification societies are responsible for developing and enforcing Class Rules and provide guidelines, Interpretations, and Additional Class notations supplementing the safety and environmental regulations issued by the International Maritime Organization (IMO), the European Commission or the flag Administrations. Classification societies may also act on behalf of the national flag Administrations as Recognized Organization for the implementation of the statutory requirements. It should be noted that industry standards are usually developed and issued by international standard organizations (ISO, IEC...) and that Unified interpretations and Recommendations are harmonized at IACS (International Association of Class Societies) level on some key IMO regulations. In this context Classification societies may promote the standardisation of ship models such as the Functionally Oriented Vessel Data Model based on the principles outlined in ISO 15926A. Stakeholders for Shipping Digital Twins The shipping DT stakeholder user groups can be described as shown in Figure 3 below. Figure 3. Digital twin shipping stakeholders 16 Shipping Digital Twin Landscape • Ship Owners/operators: They benefit from the more complete knowledge and insights about their ships, that digital twins bring about in order to improve their decision making related to ship financial and operational aspects and for planning and implementing decarbonisation and competitiveness strategies. The HQ departments, e.g., operations, technical, chartering and sustainability, will be DT users, often utilising collaborative decision-making support. The ship master and officers are also potential users, even though their interaction with the system is likely to differ from company to company and will evolve with time. • Shipping Customers They benefit from associating themselves with greener ships and thus elevating their public profile. Potentially, thy can link up with ship DTs to be better informed and create better decision support for using more environmentally friendly and seaworthy/safe ships, they improve the performance of their logistics and supply chains. • Ship Designers & Builders: They develop and use ship design digital twins, but arguably they also learn from deployed digital twins and use such knowledge to optimise and improve the parameters of the actual ship design or improve the simulation models for development of future ships. • Ship Services and equipment providers (equipment providers, engineering services, retrofitters, maintainers): This category of organisations benefits from the detailed and accurate knowledge contained in a digital twin that improves all engineering (e.g., maintaining, retrofitting, decommissioning) and other ship related activities. • Decarbonisation solution providers including green fuels, emerging carbon capture solutions and fuel/ hydrogen cells. Expected to use DTs for the design and deployment/operation of their solutions. It will be important to provide standard ‘connectors’ enabling integration of their products in new ships or retrofits and to enable robust assessment of the performance of a product for specific types of ships. • Ports are developing their own DTs to support their own operation and decarbonisation operations. Integration of port and ships DT is an essential route to increased ship voyage efficiency and improved integration of ships in supply chains. Further ports are directly involved with green fuels supply. • Classification societies: main role is to assess DT applications on safety and regulatory implication and develop/provide e-compliance services and support services including guidelines and standardisation as outlined earlier. Also, likely to be users of DT technologies, to support their own approval and survey/ inspection services. • EU and national government policy makers could use DTs to monitor decarbonisation progress and assess potential policy instruments; data-based evidence of the efficiency of different technologies and policies with respect to specific objectives such as decarbonisation. • Authorities including Safet and Emergency support: These include organisations such as authorities directly interacting with the ship’s operations such as port authorities or legislation setting organisations (national or supranational). The former, such as IMO, provide the regulations that constitute the main driver for decarbonisation. As users, they could link-up with ship DTs to benefit from a more complete (and hopefully accurate and representative) understanding of the ship with a compliance view. Safety and emergency support services including traffic management can be efficiently integrated in DTs. 17 Shipping Digital Twin Landscape Related Standards for DTs in Shipping Relevant initiatives that may be considered as parts of the ship digitalisation strategy are listed below. GAIA-X (https://gaia-x.eu/) GAIA-X is an EU-supported initiative proposing an open Forum. It developed a Digital Transformation framework of control and governance and implements a common set of policies and rules that can be applied to any existing cloud/edge technology stack to obtain transparency, controllability, portability and interoperability across data and services. The framework is meant to be deployed on top of any existing cloud platform that decides to adhere to the GAIA-X standard. Synergy with requirements for a shipping Digitalisation governance framework outlined in the previous section worth exploring. IMO Compendium (https://imo.org/en/OurWork/Facilitation/Pages/IMOCompendium. aspx): the International Maritime Organisation has developed a reference data model that is used to harmonise information exchanges between the authorities sector (World Customs Organisation), the trade sector (UNECE), and the ship operation sector (ISO). Common Maritime Data Structure (CMDS) (http://s100.iho.int/home/s100-introduction): International Hydrographic Office (IHO) maintains the S-100 framework that is also used by other organisations such as IALA and IEC to develop information models and protocols in the maritime sector. S-100 is generally acknowledged as the realisation of the CMDS from IMO’s e-navigation definitions. Maritime Information Technology Standards (MITS) http://www.mits-forum.org/architecture. html: Provides background information on ship architectures. OXC Consortium (https://3docx.org/) - The Open Class 3D Exchange Format. Shipyards and classification societies must modify the traditional design documentation and review process and enable a direct 3D digital classification process to improve the exchange of information between the different stakeholders and ultimately accelerate the classification process. The Open Class 3D Exchange (OCX) standard represents a step-change in this context. The OCX is a vessel-specific standard addressing the information needs by the classification society and is a key enabler to replace traditional 2D class drawings with a 3D model. Uniquely, OCX addresses the needs of the classification society and shipbuilders for fully digital information exchange. Effectively, OCX acts as a conduit between the design tools and class confirmation tools, highlighting the structural information the class society requires and idealizing and formatting it in an efficient way that can be easily processed. Major Challenges Inhibitors and Constraints to DT Developments These can be summarised as: • Lack of trust among stakeholders on holistic DT solutions applied to the shipbuilding industry (Giering and Dyck, 2021). • Because vessels have long lifespans with an average of 20 to 25 years of operating life (Joerss, et al 2021), transition towards zero-emission ship designs can be slow. It is necessary to act quickly to prevent locked-in emissions for a long time (Istrate, et al., 2022) • The lack of standardised quality assurance and validation models for DT systems may limit adoption of digital twins particularly for shipbuilding (Giering & Dyck, 2021) • Increased risk of cyber-attacks (Markets & Markets, 2022). 18 Shipping Digital Twin Landscape Close collaboration between stakeholders will be key to minimise the risks of deploying new zeroemission vessels: collaboration between cargo owners, vessel operators and fuel producers will be crucial to achieve zero-emission shipping (M. Joerss, et al., 2021). This means that federation between different DTs is needed which may take longer to achieve. Under the same theme, typical information silos can limit the extent of uptake of DTs in the shipping industry. Most of the data obtained during shipbuilding is lost before the vessel is transferred to the shipowner, who would need to develop their own DT. Additionally, classification societies use their own system, creating further redundancies and complexities (Giering and Dyck, 2021). Cost of establishing DTs must be proportionate to ensuing benefits with rapid return on investment. In Europe, most shipyards (≈98%) are small or medium-sized (Markets & Markets, 2022.) limiting their investment capacity in new technologies/solutions. The same applies to the large number of small ship operators. Cost must include any additional training. DT APPLICATIONS Business Decision Support With DT Analytics As explained so far, shipping digital twins are dynamic, data driven, multi-dimensional digital replicas of a physical entity such as a ship, a subsystem such as ship propulsion, or digital representations of complex processes such as the ship voyage management or fleet management, potentially covering the continuum of ship design, construction and lifecycle management. DTs can support predictive or reactive decision-making at different operational, tactical and strategic levels in order to achieve the performance goal of a company or broader community or industry goals. A ship DT can support reactive and predictive decision-making tasks at different levels in order to achieve multiple performance goals, including enhanced economic and safety KPIs and minimisation of environmental impact. Shipping companies can use federations of DTs for fleet optimization on the grounds of its cargo-carrying capacity. DTs can support market intelligence and sensitivity analysis of market transactions in terms of past, present, and future scenarios. DTs can be used to detect trade patterns and enhance both operational and strategic decision-making. Analytics applications for anomaly detection can detect atypical situations such as weather reports and coordinate with ship control systems. At the operational level, predictive decision-making involves using real-time data from sensors and systems to anticipate and prevent undesirable incidents. For example, a predictive maintenance system can use data from sensors and historical trends to predict more accurately the remaining useful life of a component allowing proactively maintenance and preventing system failure. Reactive decision-making at the operational level involves responding to unexpected equipment failures. At the tactical level, predictive decision-making involves using historical and real-time data to optimize ship operations, taking into account weather patterns and trend of fuel prices to determine optimal energy management strategies. Reactive decision-making at the tactical level involves responding to unexpected changes in the environment or operational conditions, such as a sudden change in weather conditions. At the strategic level, predictive decision-making involves using historical and market data to make long-term planning decisions, to optimize fleet deployment and fuel procurement strategies and plan for new facilities. Reactive decision-making at the strategic level involves responding to unexpected changes 19 Shipping Digital Twin Landscape in the market or regulatory environment, such as changes in fuel emissions regulations or geopolitical changes that affect supply chains. Supply Chain Visibility With the establishing development in standards for IoT connectivity, smart containers can be realised. Digital twins containing smart container data can be used to optimize the fleet, create awareness in different situations and address various port and terminal operations. Throughout the supply chain, the containers transit through various shipping hubs, while flowing data stream generated by connecting containers globally can provide important data that can be utilized by digital twins. Digital twins can therefore further optimise the supply chain by providing opportunities for the stakeholders to select their ideal route for serving customers and coordinating transport buyers. In turn, this could facilitate the establishment of strategic relationships between the transport producers and the hubs for trans-shipment. Digital twins can also monitor the efficient movement of the empty containers, further reducing their environmental footprint. Cyber-Security As cyber-physical systems and connections expand, digital twins can assist tackle the growing worry concerning cybersecurity issues. Outside networks are increasingly compromising corporate information technology and operational technology systems as more assets are remotely overseen, managed, and preserved through the Industrial IoT. Digital twins can supervise the security of digital assets and aid the security upgrades and control of remotely accessible system data. Additionally, digital twins not only manage the exterior cyber-possessed security threats such as cyber-attacks done intentionally with some third-party interest, but also the cyber safety threats which develop due to internal complexities and properties, emerging and residing within the integrated systems (NIST, 2023). The priority is to restrict hazards at their early stage when their risks are minimal, which stops the operations from being manipulated. So, a digital twin which has capabilities such as testing based on simulation and verification yields the best solution possible. Advanced Future DT Enabled Applications Trends in networking, edge computing and AI offer many opportunities for enhanced DT capabilities for shipping to enhance sustainability, efficiencies, safety, and customer satisfaction. The development of 5G capabilities and future evolution of 6G, 7G, and beyond enable high fidelity interaction between ships and HQ to disseminate highly realistic simulations or digital twins of every aspect of the ship operation. Remote operations and augmented reality guided maintenance can streamline complex processes on the ship and greatly enhance the productivity of key engineers and personnel. Mobile computing has revolutionised our world since the first iPhone was released in 2007. Together with the proliferation of IoT devices, and the growth of highly elastic cloud computing, it enabled computing and disparate software applications to penetrate people’s daily lives. Today, edge computing is bringing computing to the enterprise edge and allows organisations to process data, deploy AI models, and augment decision support tools to every portion of their operations. 20 Shipping Digital Twin Landscape AI has been described as having its “Netscape moment” in 2023 (NY Times: https://www.nytimes.com/interactive/2023/05/19/business/what-is-a-netscape-moment-artificial-intelligence.html#:~:text=There%20 are%20parallels%20between%20today%E2%80%99s%20fervor%20for%20A.I.-powered,around%20 an%20existing%20technology%2C%20leading%20to%20new%20innovation.) with the release of highly powerful chat assistants such as OpenAI ChatGPT or Google Bard. Foundation models, also known as pre-trained language models, are core to this and have demonstrated impressive performance across tasks. While they have gained recognition for their performance in language-related tasks, they can also be customized for various domains beyond NLP, including computer vision, IT operations, and industry applications. The advantage of foundation models is that since they are pre-trained on vast volumes of data, they can be fine-tuned to downstream applications at a fraction of the cost. Foundation models can be trained to ship-specific data to create generalisable representation of ship processes and their interfaces with other factors such as environment, land-borne transport, and customers. As these models become more energy efficient, and more explainable, they can be deployed on the edge of even the smallest vessel and have sufficient robustness to be trusted by the largest of organisations. A SHIPPING DIGITAL TWINING ARCHITECTURE AND CAPABILITIES MODEL Overall, Shipping Digital Twinning systems consist of several key components that work together to create a virtual replica of a ship and its operations. Creating a shipping Digital Twin (DT) involves a sophisticated infrastructure and various services to ensure efficient data management, orchestration, scheduling, version control, and continuous integration/continuous deployment. Figure 4, provides a Shipping Digital Twining architecture which highlights four areas: • (Open) Shipping Digital Twining Infrastructure Support Tools Figure 4. Shipping digital twining architecture (SOURCE: EU project DT4GS https://dt4gs.eu) 27 Digital Twin for Selection, Deployment of Decarbonization Solutions INTRODUCTION International shipping provides 80–90% of global trade, but strict environmental regulations around NOX, SOX and greenhouse gas (GHG) emissions create new imperatives for short-, mediumand long-term emission reduction targets. As shown in an EU survey (EU, n.d.), in 2018 the global shipping emissions represented 1076 million tonnes of CO2, which represents the 2.9% of global emissions caused by human activities. In that review it was also projected that, if no actions will be taken, the emissions from shipping can be increased by up to 130% of the 2008 baseline by 2050, which are far from the EU targets. The shipping industry emissions generation continues to rise due to increased global trade and the growing demand for maritime transport. The greatest source of GHG emissions are the container ships, bulk carriers, and oil tankers, however due to their larger engines, their emissions intensity (emissions per unit of cargo transported) is often more favourable compared to smaller vessels (Olmer et al, 2017). The pathway to achieving the international target of 50% GHG reduction by 2050, which has been recently upped to 100%, is not certain, but numerous promising options exist. Efficiency measures, for GHG reductions can be classified along three axes: • Operational efficiency, e.g., slow-steaming. • New, more efficient ship designs and, • Utilisation of renewable resources, such as wind, and carbon free or low carbon emitting fuels. There is clearly no single route, and a multifaceted response is required for managing decarbonisation pathways for each ship. The scale of this challenge is explored by estimating the combined decarbonisation potential of multiple options. A recent study by Transport and Environment on Decarbonisation pathways for EU-related shipping concludes that a mix of different technologies is required to achieve the International Maritime Organization’s (IMO) and EU targets leading to 2050 zero emission shipping. For instance, 50% decarbonisation with LNG or electric propulsion would likely require four or more complementary efficiency measures to be applied simultaneously. Broadly, as GHG reductions need to be achieved at increasing rates over the next 30 years we can differentiate between shortand long-term approaches. Short-term approaches can include operational changes and fuel switching, while long-term efforts involve developing and adopting new technologies and infrastructure. The complexity of the challenge in achieving the IMO required emission reduction rates leading to full decarbonization underscores the need for a holistic and multifaceted approach. It also highlights the importance of ongoing research and development, investment in clean technologies, and global cooperation to transition the maritime industry toward zero-emission shipping by 2050 and beyond. The Chapter is organised as follows: The next section discusses the most important energy efficiency measures set by organisations such as IMO which current and future ships must adhere to. These provide the yardsticks against which decarbonisation technologies will be evaluated. ships and new buildings are considered. Furthermore, the chapter illustrates the application of DTs to specific use cases, namely energy production, distribution, and recovery onboard process management with the help of a simulator, and hull performance prediction utilising simulation. 28 Digital Twin for Selection, Deployment of Decarbonization Solutions Section 3 outlines a methodology that employs digital twins to select and evaluate decarbonization measures for the physical ship. Then it describes methods to assess the effectiveness of specific decarbonization technologies and how these results can have added value employing a knowledge hub within a dataspace. Two applications of DTs for the selection and deployment of decarbonisation solutions are described in sections 4 and 5, pertinent to waste heat recovery and hull performance, respectively. The final section of the chapter reviews the potential of the presented methodology for the optimal selection of current and future decarbonisation technologies and resources. ENERGY EFFICIENCY ASSESSMENT AND COMPLIANCE Global shipping decarbonisation measures driven by the IMO aim to improve the energy efficiency of ships. Since 2015, all newly delivered ships must meet the Energy Efficiency Design Index (EEDI), a minimum design energy efficiency standard, which becomes more stringent every five years. The EEDI is an index formulated for new ships at the design stage to cut down the amount of emissions from these ships. In that direction, DNV (2018) has developed a tool to compute the EEDI for the whole fleet. The EEDI is expressed in grams of CO2 per tonne-mile, or in other words, the ratio of environmental cost to the benefit for society. Each ship has to obey to different limits accordingly to the type of ship. However, the values of these limits will be reduced by 30% by 2025 compared to the zero phases in 2013 (Puisa, 2015); therefore, different technologies are required for ships to comply to significantly reduced EEDI limits (El-Gohary, 2013; Palomares, 2011; Papanikolaou et al., 2011). Figure 1. Number of ships and their carbon emissions by category in 2017 (Source: Balcombe et al. (2019) 29 Digital Twin for Selection, Deployment of Decarbonization Solutions Moreover, for existing ships, since the 1st of January 2023, the Energy Efficiency Existing Ship index (EEXI) has entered into force and applies to all vessels above 400 gross tonnage (GT) falling under MARPOL Annex VI. The decision to adopt this short-term measure has been taken at the IMO MEPC 76 and proposed in the MEPC 75 (IMO, 2020). The EEXI is a one-time certification targeting design parameters. The calculation guidelines refer to the corresponding EEDI guideline for new buildings. Most of the guidelines have been finalized at MEPC 76, while some are still not, according to (DNV, 2019). Also, since 2015, the IMO has introduced voluntary guidance on the Ship Energy Efficiency Management Plan (SEEMP) that applies to both new and existing ships and aims to improve energy efficiency via operational measures such as optimising routes and speeds. The Ship Energy Efficiency Management Plan (SEEMP) establishes a cost-effective mechanism to improve the ship’s energy efficiency (IMO, 2016). SEEMP is based on four steps: planning, implementation, monitoring, and evaluation and improvement. Firs, it is important to define the current status of energy consumption and how it can be reduced. Secondly, implementing the SEEMP is the responsibility of the involved stakeholders, and then monitoring the effectiveness of the implemented SEEMP. Finally, it is necessary to evaluate the previous stage’s results to check the effectiveness of the applied SEEMP to improve the plan. This procedure is applied to new and existing ships to manage ship and fleet efficiency performance over time, based on the Energy Efficiency Operational Indicator (EEOI) as a monitoring tool. This operational measure can optimize and improve voyage planning, introduce the timing of ship hull cleaning, and suggest installing new types of equipment onboard (Tikka, 2011). The EEOI is considered a monitoring tool to support the decisions in SEEMP (IMO, 2009). It is used to monitor and identify the operation and personnel performance as well as the quality control procedures. It aims to provide a transparent and recognized approach to assessing the level of GHG emissions from the ships in their real operating condition along the route. Tran (2017) developed an open tool to support the computation of the EEOI for different types of ships based on the fuel type, the amount of cargo carried, the distance of the voyage and the sailing speed. For an accurate estimate of EEOI, real data is important to be provided by the fleet to take the right action (Perera et al., 2015). Such data relate to factors such as biofouling, which can reduce the propeller performance by 30% (Owen et al., 2018). Another metric that has been introduced is the Environmental Ship Index, ESI, developed by the World Ports Climate Initiative (WPCI), complements existing indicators, like EEXI (WPSP, 2019). This index aims to identify seagoing ships with better performance by monitoring real-time emissions data from ships participating in port-based initiatives. Since 1st January 2023 it has been mandatory for all ships not only to calculate their attained Energy Efficiency Existing Ship Index (EEXI), in order to measure their energy efficiency, but also to collect data for reporting their annual operational Carbon Intensity Indicator (CII) and their CII rating (fig. 2) (Czermański et al., 2022). CII index will be used to rate ships on a scale from A to E,. This is shaped to drive improvements in vessel operations, e.g., by technology upgrades. The decision to adopt this shortterm measure has been taken at the IMO MEPC 76 (IMO, 2021) and proposed in MEPC 75. The CII requirements have taken effect for all cargo, RoPax and cruise vessels above 5,000 GT and trading internationally. The metric addresses the actual emissions in operation by measuring how efficiently a ship transports goods or passengers and is given in grams of CO2 emitted per cargo-carrying capacity and nautical mile [ g t nm CO2 ⋅]. Then, the ship is given an annual rating ranging from A to E, whereby the rating thresholds will become increasingly stringent towards 2030. If the ship fails to com- 30 Digital Twin for Selection, Deployment of Decarbonization Solutions ply with the CII limitations, she will be asked to revise before returning in service (Qi et al., 2021). As argued in Wang et al. (2021), ongoing research is essential to refine and improve the CII. This includes developing more sophisticated models and using real data to create versions of the CII that are effective in driving emissions reductions. This obviates the key role digital twining and industry dataspaces can play. The first reporting year for the CII is 2023, this means that the first annual reporting will be at the end of 2023, and the first rating given will be in 2024. Reviewing the effectiveness of EEXI and CII will be required to develop further amendments. As has already been done for ships that their operational profile justifies the usage of correction factors and exceptions in the calculation of the attained CII value, as shown in MEPC 78 in the resolutions 352 to 355. As shown in Figure 2, IMO expects a 20% reduction in emissions by 2030, a 70% reduction by 2040 (compared to 2008 levels), and the ultimate goal of achieving net-zero emissions by 2050. METHODOLOGY Rationale The methodology for Digital Twin-aided assessment of decarbonization technologies and ship-specific solutions is based on a data-driven and proactive approach which selects and deploys decarbonisation solutions that takes into account technology options that match ship profiles and company strategies. By leveraging real-time data and simulations, ship operators and their consultants can make informed decisions, optimize performance, and contribute to a more sustainable and decarbonized shipping industry. The approach is based on a classification of decarbonisation technologies, Table 1 shows a two-phase classification of decarbonisation technologies for retrofitting of the existing ships, and a single phase for the design of new zero-emission ships. Figure 3 illustrates the potential contributions of different decarbonisation strategies, emphasising the early importance of energy efficiency and operational improvement, which are included into the emission reduction by reduced energy demand section, contrasted with the longer-term prominence of green fuels combined with new power and propulsion technologies, that can be summarised in the emission reduction by use of carbon neutral fuel section. For existing ships, phase 1 uses mainly the first two categories with minimal investment and importantly establishes a critical timing for the second phase intervention that ideally will enable the ship to operate to the end of her life in compliance with emissions regulations. Figure 2. CII calculation 31 Digital Twin for Selection, Deployment of Decarbonization Solutions The triggering factor for phase 2 interventions, is maintaining level C in CII or alternative target. The CII computation is an important element of the approach, and its calculation is also presented in this section. Actual CII rating can be directly estimated by the DT, since the required parameters are inherently available within the DT. But most importantly, the digital twin can forecast the CII based on its decisions and adapt or optimize its decisions, accounting not only for other KPIs but also for the CII. In this way the combination of a DT and CII calculation method for real time and forecasted situations, can effectively support the transition to zeroemission shipping. The classification of some of the possible decarbonisation technologies is detailed in Table 1, as mentioned above. CII Calculation In this section, the CII calculation is described, utilising a Digital Twin platform for real time, automated input from ship sensors’ data streams, in contrast to the typical approach where usually manually measured and registered noon report data are used. The information required for the calculation include the distance travelled, the type, quantity, and specifications of the fuel used. The latter holds especially in the case of a non-standard fuel (MEPC.308(73)) (IMO, 2018). Furthermore, for specific ship operation profiles more data is needed, as shown in MEPC.355(78) (IMO, 2022), in order to calculate the attained CII, taking into account any corrections or exemptions that are applied to the reporting vessel. The CII calculation, can be utilised in various cases, such as: • To assess the current CII rating, that includes data the beginning of the current year. • To predict the CII rating for the next three years, assuming the same ship use, as required by regulations. Figure 3. Contributions of different decarbonisation technologies to zero emissions transition 32 Digital Twin for Selection, Deployment of Decarbonization Solutions • To estimate the future CII rating at the end of a certain period using data-driven models that calculate future fuel consumption and other factors for the CII calculation. The methodology to calculate the CII illustrates the required data intensity, which is compatible with ship performance optimisation data. The aforementioned methodology can be broken down to the following steps: STEP 1. The total consumption of each fuel is calculated and then it is converted to grams of CO2 according to MEPC.308(73). The basic characteristics of the fuels, as described in MEPC.308(73), including representative values from ships participating in DT4GS, are shown in the Table 2 below: STEP 2. Calculate the attained CII value using the MEPC.352(78), as shown in the equation bellow: Table 1. Classification of decarbonisation technologies 33 Digital Twin for Selection, Deployment of Decarbonization Solutions CII M CF C D g t nm ship jj j t CO            2 where: CIIship: The CII attained from the specific voyage, according to MEPC.336(76), in g t nm CO2 ⋅ ◦Mj: The mass of the j type of fuel that has been consumed, in grams. ◦CFj: The carbon coefficient Cf of the j fuel according to MEPC.308(73) ◦C: The capacity of the vessel as described in MEPC.353(78) ◦Dt: The distance traveled by the vessel in the reporting period, in nautical miles (i.e., the trip that we want to calculate the CII) ◦j: The different fuels that are consumed in the reporting period (i.e., the trip that we want to calculate the CII) STEP 3. In case that the corrections and exemption of MEPC.355(78) are applicable for the reporting vessel, the attain CII (CIIship) value can be calculated using the equation bellow: CII C FC FC TF y FC ship jFj j voyage j j i electrica         ,. .0 75 0 03 ll j boiler j others j i m c iVSE t FC FC f f f f Capacity D , , ,                 Dx where: • j: The fuel type. • CFj: represents the fuel mass to CO2 mass conversion factor for fuel type 𝑗, in line with those specified in the 2018 Guidelines on the method of calculation of the attained EEDI for new ships Table 2. Fuel characteristic according to MEPC.308(73) Type of fuel Reference Lower calorific value [kJ / kg] Carbon content CF [tCO2/tFuel] Diesel ISO 8217 Grades DMX through DMB 42,700 0.8744 3.206 LFO ISO 8217 Grades RMA through RMD 41,200 0.8594 3.151 HFO ISO 8217 Grades RME through RMK 40,200 0.8493 3.114 LPG Propane 46,200 0.8182 3 LPG Butane 45,700 0.8264 3.03 LNG 48,000 0.75 2.75 Methanol 19,900 0.375 1.375 Ethanol 26,800 0.5217 1.913 34 Digital Twin for Selection, Deployment of Decarbonization Solutions (resolution MEPC.308(73) as amended by resolutions MEPC.322(74) and MEPC.332(76)), as may be further amended); • FCvoyage,j, Dx, TFj, yi: the factors for any potential exemption as described in the MEPC.355(78). • fi, fm, fc, fiVSE: The corrective factors, as described in the MEPC.355(78). • FCelectrical,j: The correction for FCelectrical,j refers to 3 main categories: ◦Refrigerated Containers ◦Cargo cooling systems on gas carriers and LNG carriers ◦Electric cargo discharge pumps for tankers • FCBoiler: For cargo heating and discharge pumps on tankers ◦In the case of boilers used for cargo heating, the amount of fuel used by the boiler (FCBoiler) should be measured by accepted means, e.g., tank soundings, flow meters. ◦For tankers which use steam driven cargo pumps, the amount of fuel used by the boiler (FCBoiler) should be measured by accepted means, e.g., tank soundings, flow meters. Note that boiler consumption should not include consumption during voyage adjustment periods. • FCothers: For discharge pumps on tankers powered by their own generator, the amount of fuel used for the period that the discharge pumps are in operation (FCothera) should be measured byaccepted means, e.g., tank soundings, flow meters. STEP 4. Calculate the baseline CII (or reference) as dictated in the MEPC.353(78), for the year of 2022: CII a capacity g t nm reference cCO         2 Where: CIIreference: Is the baseline (or reference) CII rating, as dictated in the MEPC.353(78), in g t nm CO2 ⋅ ◦capacity: The capacity of the vessel as described in MEPC.353(78) ◦a,c: Coefficients depended to the ship type, as described in MEPC.353(78) Then we calculate the future (y year) baseline CII (reduction targets) as dictated in the MEPC.338(76): CII ZCII g t nm reference y reference CO ,            1100 2 With Z the percentage required for reduction each year. 35 Digital Twin for Selection, Deployment of Decarbonization Solutions STEP 5. Calculate the limits for the different CII ratings, according to MEPC.354(78), where we multiply the required CII reference with the adequate set of factors (exp(d1), exp(d2), exp(d3), exp(d4)) according to the ship type. In this way, the upper limits for each CII rating are produced. The same procedure is applied to calculate the corresponding limits for the y year in the future, but this time using the CIIreference,y instead of CIIreference. Furthermore, regarding the third use case, due to the continuous connection of the ship with the Digital Twin platform the user can have access to the running CII rating from the start of the reporting year till the present moment. The real added value that the Digital Twin platform can provide to the user, regarding the CII calculation, can be seen in the precise calculation of the attained CII value by the end of a trip and the ability to optimize future voyages of the ship considering the CII value. STEP 6. Identify, in the long term (future 3-year plan), if the CII rating is projected to be below the C rate which will trigger selection, evaluation and deployment of one or combination of decarbonization technologies. A Methodology for Developing and Deploying an Integrated Ship Performance Model for Reduced Emissions According to Naito [in Tsujimoto & Orihara, 2019], ship performance is categorized as propulsive performance, safety performance, seakeeping performance, and manoeuvring performance. Among these, propulsive performance in actual seas is more prominent, since it affects fuel consumption and hence CGH (Tsujimoto & Orihara, 2019). In our approach we define an Integrated Ship Performance Model as a model that includes the above types of performance parameters and their relation to decarbonisation technologies (DEs). A DE according to this approach impacts positively or negatively one or more performance parameters. The exact impact can be described using simulation and/or analytical models that are informed by the ship’s DT data. Therefore, the integrated ship performance model provides a holistic view of how discrete decarbonisation technologies and/or their combination affect critical ship performance parameters. Key steps and components of this approach, depicted in Figure 4, are: 1. Connect the Ship to the Headquarters Digital Twin: a. Identify the already existing sensors on the ship and examine if they are adequate for the implementation. In that direction, additional sensors may be identified as required and deployed on the ship to enhance monitoring capabilities. b. Establishment of a digital connection between the ship and the shore-based office enables real-time data exchange, remote monitoring, and decision-making support. c. Data management services are set up to handle the vast amount of data generated by various ship sensors and systems. d. In case of retrofitting with installation of new decarbonisation solutions additional sensors will be introduced to monitor the performance of devices or applications. 2. Establish Ship Profile and Performance Model: a. Historical data is used to create a ship profile that includes routes, ports of call, cargo capacity, fuel consumption and CO2 patterns as well as typical weather conditions. In addition, the expected utilisation of the ship for its remaining lifecycle (locations and numbers of voyages, frequencies of voyages and planned maintenance activities) are also included. 36 Digital Twin for Selection, Deployment of Decarbonization Solutions b. To establish a benchmark against which decarbonisation solutions and their combinations can be assessed, actual data is used to develop the ship’s base performance model. c. Performance models from shared industry data spaces can be leveraged to increase development efficiency and possibly enhance accuracy. 3. Calculate Carbon Intensity Indicator (CII) and rating: a. A CII calculator is employed to calculate the ship’s CII for the current reporting period and project the CII value and its respective rating after future voyages. b. This data is critical for guiding the decarbonization transition strategy. 4. Specify Decarbonization Transition Strategy: a. With reference to SEEMP Decarbonization Transition Strategy, based on the ship type, status, i.e., years in operation, expected remaining life, an annual minimum performance threshold table is created to set performance targets. b. Technologies of interest are identified for two phases, as mentioned in the section 3: Phase 1 involves minimal investment, while Phase 2 focuses on compliance with emission reduction regulations over the ship’s lifespan. c. Multiple objectives are defined to achieve optimal ship performance, including efficiency, load factors, operating cost reduction, and emission targets. 5. Evaluation of Decarbonisation Technologies – produce Decarbonisation Strategy: a. Simulation is used to evaluate the impact of various decarbonization technologies, and their combinations, based on ship profiles and base performance models. b. Evaluation criteria may include the ability to meet performance thresholds, cost-effectiveness, technological maturity, and crew acceptance. 6. Create an Integrated Performance Model: a. A Knowledge Graph (see Chapter 3 of this book) is utilized to represent the interdependencies of all ship subsystems affecting performance variables. b. This integrated model provides a holistic view of how control variables interact and affect ship performance. The Integrated Performance Model depicted in Figure 4 is designed to optimise the ship’s propulsive performance (i.e., in terms of GHG emissions) given as control inputs the different decarbonisation technologies adopted, (described as a Decarbonisation Strategy) in the diagram. The control objective therefore is to select an optimal mix of decarbonisation technologies that when applied to the Base Performance Model, as shown in 2b above, yield a performance improvement above a given threshold, while satisfying the constraints imposed by the other performance parameters. The controller is therefore model based and more specifically incorporates the Integrated Performance Model of the ship that includes the main interactions between the ship model parameters (including environment parameters such as financial: CAPEX, OPEX. The Integrated Performance Model therefore, integrates the ship’s characteristics, operational data, and the performance parameters, energy consumption, propulsion performance, and emissions, of selected decarbonisation options and is supplied with data obtained from the Ship’s Digital Twin. The multi-objective optimisation process is applied both for the design of the Decarbonisation Strategy and in support of continuous operational ship optimisation. The formulated mathematical optimisation program is set in a rolling time-horizon. This means that the optimization process considers real-time data and adapts to changing externalities over time. This enables the system to make informed operational and maintenance decisions based on up-to-date information. 43 Digital Twin for Selection, Deployment of Decarbonization Solutions request from the Heat Users, may benefit the most from WHR strategies. The section places special emphasis on Waste Heat Recovery (WHR) systems, a technology that, although not new, continues to hold promise and has the potential to deliver significant efficiency enhancements. Benefits and Requirements of DTs in Energy System Optimization The Digital Twin provides enhanced capabilities for optimization purposes through the lifecycle of the object: design, operation, retrofitting, retire. Among other reasons, this is achieved by considering in detail a multitude of parameters beyond those associated solely with the specific system under examination. Another advantage of employing energy system optimization within a DT framework is the utilization of its structured information about regarding subsystem and components availability and overall configuration. The primary scope of the Digital Twin based simulator is the benchmarking of the energy efficiency of different WHR configurations, modelled through a limited set of design parameters, over a large set of service conditions representative of the operational profile of the ship. The WHR simulation aims to evaluate the overall electric / heat energy balance of the ship and as such it will be able to manage production, distribution and consumption of Fresh Water, Steam and Electric power. In general, there are two types of simulators that can be used: • Disregarding transient states, a logical state-machine solving a system of algebraic equations with given known terms to determine the resulting balance state. • Modelling in real-time the internal processes occurring in each involved system and the regulation feed-back loop of the relevant mass flows and associated temperatures implemented to achieve the dynamic equilibrium; time-domain simulation solving a system of differential equations driven by initial conditions to analyse the dynamic evolution of the energy processes. Focusing on the steady state option, each simulation considers a static set of operational conditions denoted as a “voyage condition”, which could either represent a stay in port or a navigation leg. A chain of sequential voyage conditions will therefore represent a typical “voyage” of the ship whereas a chain of voyages will represent the “operational profile” of the ship that is the set of operational / environmental conditions which the ship will likely encounter during a year of standard service. In this context, the information concerning loads in terms of power requested for propulsion and the electric load must be seen as input for the energy simulation, indeed, the requested values must be fulfilled in the different operative condition. Hence, knowing the requested of powers the user can modify the layout of onboard devices to assess the change a ship energy performance. In detail, the operation profile, consists of: • a navigation leg to be completed within an expected time. • a set of environmental conditions (i.e., sea / wind / current) likely to be encountered. • the service load expected in the leg. The service load will have in general the following breakdown: 44 Digital Twin for Selection, Deployment of Decarbonization Solutions • propulsion power (in case of Diesel-Engine propulsion); electric power requirements; propulsion load (in case of Diesel-Electric propulsion)]; hull & engine services load; hotel load • Fresh Water requirements • Cooling Water requirements • Steam requirements In particular, the voyage simulator provides Power requested to the Main Propulsion Plant and the Electric Power requested to the Power Generation Plant to the WHR simulator. Based on this, the WHR simulator estimates the Waste Heat produced by the Power Generation Plant and by the Main Propulsion Plant (in case of Diesel-Engine propulsion), which will be used as an input to the WHR Plant model. Based on this input (and the prevalent internal / external environmental conditions), the WHR Plant model provides an estimate of the electric power and thermal power recovery. However, it must be considered that, when electric energy recovery from the waste heat of Diesel Generators is implemented, the recovery process should be analyzed iteratively as the resulting reduction in the electric load will in turn reduce the waste heat and thus the recovery. Modular Model High Level Description As there is a diversity of possible WHR Plant configurations, depending on ship types and operational profile, it is of primary importance for the simulation tool to be easily configurable, even in automated way controlled by the DT. For this purpose, the components registered in the DT as power producers or consumers are modelled in terms of thermodynamic interaction (mass and energy, input and output). Indicatively, such components are the following: • Generation units ◦Diesel Engines ◦Alternators • Recovery units ◦Heat Recovery Units ◦Electric Recovery Units • Users ◦FW Users ◦CW Users ◦Steam Users • Similarly, in the context of DT as decarbonization methods, thermal and electrical recovery units are considered, all falling under the category of waste heat recovery units. These units can be modelled in the energy simulator and include: Steam Economizers • HW Economizers • Evaporators • Heat Exchangers for Heating / Re-Heating / Pre-Heating of FW • Absorption / Adsorption Chillers Units • Organic Rankine • Steam Power Turbines 45 Digital Twin for Selection, Deployment of Decarbonization Solutions • Gas Power Turbines CASE STUDY: HULL CONDITION PREDICTION SIMULATOR In this section it is demonstrated how typical procedures that are currently used, e.g., for hull condition assessment can be enhanced by incorporating them into DTs and, on the other side, how the DT can facilitate their application. The basis of the study is the ISO 19030 (ISO 19030-1, n.d.) (ISO 19030-2, n.d.) that accounts for the hull and propeller performance assessment. It pertains to the comparison of performance of a specific ship to herself, over a certain period of time. This can be conceived through the relation between the ship’s underwater condition and the power needed to move the ship through water at a specific speed. For a given speed, variation in the needed power may be observed, due to changes in the underwater hull and propeller condition, leading to increased hull resistance and alterations in the propeller efficiency. It is important to highlight that data-driven methods are well-suited for hull assessment within a DT. Nevertheless, model-driven (‘white-box’) approaches continue to hold value, especially when used in conjunction with data-driven methods, giving rise to grey-box methods. Flows Data Requirements In a ship’s Digital Twin (DT), the top priority is to gather all available data from navigation instruments (GPS, echo sounder, speed log, anemometer, gyro and ruder indicator), draft sensors, and shaft power meters or flow meters. These sources provide the fundamental information needed to construct an accurate representation of the ship’s operational status. In addition, the DT is typically integrated into a dataspace with weather and oceanic data providers. Lastly, in the DT, various constructive, geometrical, and hydrodynamic particulars and references are documented, as these are essential for creating a comprehensive digital representation of the vessel. This includes data such as the height of the anemometer, the relationships between draft, trim, list, and displacement and sea trials. On the other side, the ISO 19030 procedure primarily considers the ship’s speed through water and the delivered power as its key parameters. Additionally, it takes into account several secondary parameters, including the ship’s speed over ground, relative wind speed and direction at the height of the anemometer, significant wave height, direction, and spectrum (for wave resistance corrections), swell height, direction, and spectrum (for wave resistance corrections), water depth, water temperature and density, loading conditions, dynamic floating conditions, and rudder angle/frequency of rudder movements. Furthermore, the digital twin typically relies on advanced data collection and processing techniques, leading to filtered, normalized, and synchronized data. According to the ISO 19030 the measurement procedures consist of data acquisition, data storage and data preparation, in terms of ensuring the same sample rate. The next crucial part of this first, preparatory stage, is the data filtering process, during which outliers and invalid data is excluded from the calculations thereafter. To this end, and exploiting Statistic’s theory, the document of ISO 19030 dictates the use of Chauvenet’s criterion, which is based on the calculation of the probability for the occurrence of any value within the data. The process prescribes the calculation of the mean values and standard deviation of consecutive, non-overlapping subsets of data. 46 Digital Twin for Selection, Deployment of Decarbonization Solutions It is evident that the typical data availability and data processing tools within the typical Digital Twin exceeds the requirements of ISO 19030, making its application significantly more straightforward and accessible. Brief Model Description After the stage of filtering and validating the raw data obtained from the ship and ensuring uniform sample rate, the next step of the described process is the calculation of true wind speed and direction, at the height of the anemometer, which in turn allows the calculation of the relative wind speed and direction at reference height. The latter are exploited directly for the calculation of the wind resistance coefficients, which are utilized to derive the components of the wind resistance components and therefore for the corrected delivered power The described process is graphically shown with the aid of Figure 6, which is literally the backbone of ISO 19030. Wind Resistance Calculation The relative wind speed and direction at the current loading condition of the vessel, serve as input for the corrections of the measured power (∆Pw). The wind resistance corrections eventually lead to the subtraction of that component of the delivered power that is required to overcome the wind resistance. The wind resistance correction is calculated by the equation: Figure 6. Process for the assessment of hull-propeller performance, as described in ISO 19030 47 Digital Twin for Selection, Deployment of Decarbonization Solutions PR R v P w rw w g D D DM D           ( ) 0 0 0 1    where Rrw is the wind resistance due to relative wind, R0w is the air resistance in no-wind condition, vg is the ship speed over ground, 𝜂D0 is the propulsive efficiency coefficient in calm condition and 𝜂DM is the propulsive efficiency coefficient in actual voyage condition. Rrw and R0w are given by the following expressions: R v A C rw a wr rw wr ref        1 2 2   , R v A C w a g rw0 2 1 20      ( ) where 𝜌a is the air density, A is the transverse projected area in current loading condition, Crw(Ψwr,ref) is the wind resistance coefficient, dependent on wind direction of relative wind and finally Crw(0) is the wind resistance coefficient for head wind (0°). We can directly notice that wind resistance expressions follow the general form of any resistance-component expression. Speed Loss Calculation The next step of the process is the calculation of the percentage speed loss (Vd). It is calculated as the relative difference between the measured ship speed through water Vm and the expected speed through water (Ve). The latter is obtained by a speed power reference curve. In detail, these non-dimensional parameters are given by the equation: VV V V d m e e 100 where, Vm is the measured vessel speed through water and Ve is the expected speed through water. For the calculation of Ve, the well-known Admiralty Coefficient can be also exploited. Besides, the speed-power reference curve of the ship, as well as data from the trim and stability booklet of the ship are indispensable. Definition of the Reference Periods The evaluation of the hull condition is carried out by comparing its performance with reference conditions. These reference conditions are extracted from intervals where the following criteria are simultaneously met: • water temperature greater than 2°C, given that the vessel does not navigate in ice 48 Digital Twin for Selection, Deployment of Decarbonization Solutions • the true wind speed must be between 0m/s and 7.9m/s (or equivalently 0 and 4 BF) • the water depth must hold: h max BT V g M s        3 2 75 2 , . , where h is the water depth, B is the ship breadth, TM is the draft at midship, or mean draft, Vs is the ship speed and g is the gravitational acceleration Calculation of the Performance Indicators There are four performance indicators (PIs): the first one assesses the effectiveness of a dry docking, while the second determines the in-service performance, through the assessment of the effectiveness of the underwater hull and propeller solution (e.g., hull coatings), including any maintenance activities. The third and fourth PIs refer to the crucial issue of the maintenance of the ship structure, particularly whether it is necessary or not to trigger a maintenance activity and to evaluate the effectiveness of a possible maintenance activity. Specifically, for the dry-docking performance, the period following directly after the latest dry-docking is the evaluation period. The period following directly after the previous dry-docking is the reference period. All periods are to be of the same length of 1 year. For the in-service performance, the period following directly after the latest dry-docking is the reference period. The period following the reference period until the end of the same dry-docking period is the evaluation period. The reference period and the evaluation period shall both be of minimum 1 year. For the maintenance trigger, the period following directly after the latest dry-docking is the reference period. A period after the reference period in the same dry-docking interval is the evaluation period. The reference period and the evaluation period shall both be a minimum of 3 months. Finally, for the maintenance effect, the period following directly the maintenance event is the evaluation period. The period preceding the event is the reference period. The reference period and the evaluation period shall both be a minimum of 3 months. From the mathematical point of view, a PI is defined as the difference between the average percentage speed loss of the reference period and the evaluation period. The average percentage speed loss over the reference period, is calculated from: V k n V d ref j k i n d j i   1 1 , where k is the number of reference periods, j is the reference period counter, n is the number of data points in the processed data set under reference conditions in the reference period j, i is the counter of data points in reference period j, Vd j i, is the percentage speed loss for data point i in reference period j and Vd ref is the average percentage speed loss over the reference period. Regarding the average percentage speed loss over the evaluation period: V n V d eval i n d eval i  1, 49 Digital Twin for Selection, Deployment of Decarbonization Solutions where, n is the number of data points in the processed data set under reference conditions in the evaluation period, Vd eval i, is the percentage speed loss for data point i in a data set of the evaluation period and Vd eval is the average percentage speed loss in data set of the evaluation period. Finally, the corresponding PI is obtained by the equation: k V V HP d eval d ref   The results are always anticipated to describe both qualitatively and quantitively the hull and propeller degradation. Specifically, plotting the extracted PI versus time allows a visual representation of the continuously decreasing ship performance. Besides, it enables the involved entities to identify what the exact decrease of the ship’s performance is (in %) and to evaluate whether a dry-docking is needed or, for example, what the effectiveness of any possible underwater maintenance activities was. CONCLUSION AND FUTURE OUTLOOK The diverse array of decarbonization technologies at various stages of maturity, necessitates the use of decision support tools to aid in the selection of the most suitable combination of technologies for a specific ship profile, time period and decarbonisation requirements. The use of Digital Twin-based decision support tools streamlines the process of evaluating and selecting decarbonization technologies, making it possibly more efficient and accurate. By leveraging real-time data, simulations, and optimization algorithms, Digital Twins enable the maritime industry to navigate the complexity of decarbonization and transition toward a sustainable, low-emission future. It is important to note that the maturity, potential availability, and cost of zero-emission fuels are subject to ongoing research, development, and market dynamics. Government policies, regulations, and incentives will also play a significant role in shaping the future landscape of zero-emission shipping technologies and fuels. The proposed methodology for assessing and deploying decarbonisation technologies emphasises the following: • Gather data on the ship’s operational profile, including routes, ports of call, cargo capacity, fuel consumption and CO2 patterns and produce the ship Base Performance Model to be used in simulations that can provide insights into areas for improvement. • Produce decarbonisation models, to meet decarbonisation goals / strategies utilising industry knowledge hubs that maintain a comprehensive database of available decarbonization technologies, including alternative fuels, energy-efficient systems, waste heat recovery, and emission reduction solutions. • Use of an Integrated Performance Model to assess the ship’s performance and emissions under various operational scenarios, considering different technologies and fuel options. • Use of the DT to continuously monitor the effectiveness of decarbonisation solutions to continuously optimise ship operations, while sharing experiences and data within the Shipping Industry. 50 Digital Twin for Selection, Deployment of Decarbonization Solutions REFERENCES Abdelkareem, M.A. (2021). Lootah, ET Sayed Fuel cells for carbon capture applications. Science of The Total Environment. Elsevier. Al Baroudi, H. (2021). A review of large-scale CO2 shipping and marine emissions management for carbon capture, utilisation and storage. Applied Energy. Elsevier Balcombe, P., Brierley, J., Lewis, C., Skatvedt, L., Speirs, J., Hawkes, A., & Staffell, I. (2019). How to decarbonise international shipping: options for fuels, technologies and policies. Energy Convers. Manag., 182, 72–88. https://doi.org/. enconman.2018.12.080. doi:10.1016/j Czermański E. (2022). A Oniszczuk-Jastrząbek Implementation of the Energy Efficiency Existing Ship Index: An important but costly step towards ocean protection. Marine Policy. Elsevier DNV. (2018). Energy efficiency design index calculator. Accessed 11 September 2023. https://www. dnv.com/se rvices/energy-efficiency-design-index-calculator-140598 DNV. (2019). EEXI – energy efficiency existing ship index. Accessed 11 September 2023. https://www. dnvgl.com/ma ritime/insights/topics/eexi/calculation.html El-Gohary, M. M. (2013). Overview of past, present and future marine power plants. J. Mar. Sci. Appl., 12(2), 219–227. doi:10.1007/s11804-013-1188-8 European Union. (n.d.). Reducing emissions from the shipping sector. European Commission. https:// climate.ec.europa.eu/eu-action/transport/reducing-emissions-shipping-sector_en#:~:text=In%20 2018%2C%20global%20shipping%20emissions,%E2%80%A2%E2%80%A2%E2%80%A2%20by%20 2050 Hoang, A. T. (2020). Analyzing and selecting the typical propulsion systems for ocean supply vessels. 6th International Conference on Advanced Computing and Communication Systems (ICACCS). IEEE. IMO. (2009). Guidelines for Voluntary Use of the Ship Energy Efficiency Operational Indicator. EEOI. IMO. (2016). Energy efficiency measures. IMO. http://www.imo.org/en/ourwork/environ ment/pollutionprevention/airpollution/pages/technical-and-operational-measures IMO. (2018). Resolution MEPC.308(73) (adopted on 26 October 2018). Guidelines On The Method Of Calculation Of The Attained Energyefficiency Design Index (Eedi) For New Ships. IMO. Accessed 11 September 2023. https://wwwcdn.imo.org/localresources/en/KnowledgeCentre/IndexofIMOResolutions/ MEPCDocuments/MEPC.308(73).pdf IMO. (2020). Amendments to the annex of the protocol of 1997 to amend the international convention for the prevention of pollution from ships, 1973, as modified by the protocol of 1978 relating thereto. IMO. https://wwwcdn.imo.org/localresources/en/OurWork/Environment/Documents/Air%20pollution/ MEPC.324(75).pdf IMO. (2021). Marine Environment Protection Committee (MEPC 76), 10 to 17 June 2021 (remote session). IMO. Accessed 11 September 2023. https://www.imo.org/en/MediaCentre/MeetingSummaries/ Pages/MEPC76meetingsummary.aspx 51 Digital Twin for Selection, Deployment of Decarbonization Solutions IMO. (2022). Marine Environment Protection Committee (MEPC) - 78th session. IMO. https://www. imo.org/en/MediaCentre/MeetingSummaries/Pages/MEPC-78th-session.aspx ISO. (2023). Ships and marine technology - Measurement of changes in hull and propeller performance – Part 1: General principles. ISO. https://www.iso.org/standard/63774.html ISO. (2023). Ships and marine technology - Measurement of changes in hull and propeller performance – Part 2: Default method. ISO. https://www.iso.org/standard/63774.html Lindstad, E. (2022) Ship of the future - A slender dry-bulker with wind assisted propulsion. Maritime Transport Research, 3. doi:10.1016/j.martra.2022.100055 Masaru, T., & Orihara, H. (2019). Performance prediction of full-scale ship and analysis by means of onboard monitoring (Part 1 ship performance prediction in actual seas). Journal of Marine Science and Technology, 24(1), 16–33. doi:10.1007/s00773-017-0523-1 Olmer, N., Comer, B., Roy, B., & Mao, X. (2017). Greenhouse gas emissions from global shipping, 2013–2015 Detailed Methodology. Int Conference on Clean Transportation. IEEE. Owen, D., Demirel, Y. K., Oguz, E., Tezdogan, T., & Incecik, A. (2018). Investigating the effect of biofouling on propeller characteristics using CFD. Ocean Engineering, 159, 505–516. doi:10.1016/j. oceaneng.2018.01.087 Palomares, M. (2011). Challenges & opportunities: What EEDI and SEEMP mean for the reduction of GHG emissions from ships. Materials Technology, 48, 54–58. Papanikolaou, A., Harries, S., Wilken, M., & Zaraphonitis, G. (2011). Integrated design and multiobjective optimization approach to ship design. In: International Conference on Computer Applications in Shipbuilding, (pp. 31–42). IEEE. Maja, P., Vladimir, N., Jovanović, I., & Koričan, M. (2022). Application of fuel cells with zero-carbon fuels in short-sea shipping Applied Energy, Volume 309. Elsevier. Perera, L.P., Mo, B., Jønvik, P.C., Kristjansson, L.A., Svardal, J.Ø. (2015). Evaluations on ship performance under varying operational conditions. In Proceedings of the ASME 2015 34th International Conference on Ocean, Offshore and Arctic Engineering (OMAE2015), pp. 1–10. St. John’s, Newfoundland, Canada Puisa, R. (2015). Description of Uncertainty in Design and Operational Parameters. Report number: FAROS D6.3. Application of Real Options for valuation of ship design alternatives. June 2015. doi:10.13140/RG.2.1.3436.1442 Qi, J., Wang, S., & Psaraftis, H. (2021). Bi-level optimization model applications in managing air emissions from ships: A review. Commun. Transp. Res, 1, 100020. doi:10.1016/j.commtr.2021.100020 Roussanaly S. (2021). Towards improved cost evaluation of Carbon Capture and Storage from industry. International Journal of International Journal of Greenhouse Gas Control, 106. Elsevier. Singh, D. V., & Pedersen, E. (2022). A MCFCs MCFCs review of waste heat recovery technologies for maritime applications. Energy Convers. Manag. 52 Digital Twin for Selection, Deployment of Decarbonization Solutions Tadros, M. M., Ventura, M., & Soares, C. G. (2023). Ventura, C. Guedes Soares. Review of current regulations, available technologies, and future trends in the green shipping industry. Ocean Engineering, 280, 114670. doi:10.1016/j.oceaneng.2023.114670 Tikka, K. (2011). Green evolution: Lessons learned from development of environmentally friendly tankers. Materials Technology, 48, 82–84. Tran, T. A. (2017). A research on the energy efficiency operational indicator EEOI calculation tool on M/V NSU JUSTICE of VINIC transportation company. Vietnam. J. Ocean Eng. Sci, 2(1), 55–60. doi:10.1016/j.joes.2017.01.001 Wang, S., Psaraftis, H. N., & Qi, J. (2021). Paradox of international maritime organization’s carbon intensity indicator. Commun. Transp. Res., 1, 100005. doi:10.1016/j.commtr.2021.100005 WPSP. (2019). Home. Environmental ship index (ESI). https://www.environmentalshipindex. org/Public/ Home/AboutESI. 59 Shipping Digital Twin Data Management With Use of Knowledge Graphs example, using a green fuel may result to reduced CO2 footprint but increased operational cost. Building a vessel with a dual fuel engine, may result to lower CO2 emissions and increased initial capital but the operational costs are to be estimated. The parameters currently considered are the CO2 footprint in CO2e g/ton-mile, VOYEX (costs directly related to a voyage), OPEX (operational costs that do not depend on voyage), and CAPEX (capital expenditures). With these arrangements both real time operation and lifecycle analysis can be implemented. Vessel performance assessment is differentiated for each user type. By applying the appropriate weights to the above parameters, indicators applicable to different type of stakeholders can be produced. Operational Requirements “Operational requirements” is the basic input. An interpretation of this term is “how the vessel should behave”. Operational requirements include: • regulatory obligations, such as ballast treatments or emissions constraints • contractual or business obligations, such as arrival time, FOC or vessel speed and • other operational constrains, such as required stoppage for bunkering, crewing, supplies, maintenance, etc. Environmental Conditions Apart from the operational requirements another input is the environmental conditions which are the external factors such as weather conditions and bunker prices. Depending the context, these parameters can be acquired for the present time from an external data source or can be based on a forecast (e.g., for the weather), can be variables of a scenario, or can be assumed. Setpoints (Operational Conditions) Operational requirements, along with environmental conditions determine the setpoints, which are parameters that describe machinery operation. Setpoint are the settings to achieve the operational requirements, given the environmental conditions. Engine speed is a fundamental operational condition. Optimization Measures Optimization measures is the set of available measures to optimize vessel performance and reduce CO2e. Such measures include: • JIT arrival • Trim optimization • Weather routing • Under Water Cleaning and/or Propeller Polish • Modifying maintenance frequency • Determining fuel type • Optimizing bunkering location/quantity 60 Shipping Digital Twin Data Management With Use of Knowledge Graphs Application of Decarbonization Technologies While optimization measures consider existing vessel assets, the application of decarbonization technologies refer to the application of an addition or retrofit. Such measures may include: • M/E retrofit to utilize different fuel or fuel blend • Wind assist • Carbon Capture / Fuel cells • Application of coating • Installation of energy saving devices (duct, fin, etc.) KNOWLEDGE GRAPH FOR SHIPPING DIGITAL TWINS Linking Physical Assets, Digital Assets, and Domain Concepts In the following sections it is provided a detailed description of the KG. However, in higher level, the underlaying schema of the knowledge graph and it’s linkage to the operational metamodel, is shown in Figure 2 Main classes of the knowledge graph are the predictive models and the corresponding vessel’s systems, in two “parallel” structures. In addition, variables, that are model’s inputs and outputs are correlated to actual vessel sensor reading, composing the digital twin and the digital shadow, respectively. Despite that the functional metamodel, described in section 3, is a conceptual model, it is applicable to lower, functional levels being part of the knowledge graph. Each node, such as operational requireFigure 2. The underlaying schema of the knowledge graph 61 Shipping Digital Twin Data Management With Use of Knowledge Graphs ments, optimization measures, etc are classes of variables which are interconnected to Model’s inputs/ outputs and vessel sensor reading. For example, vessel speed is an operational requirement, connected to the corresponding sensor’s reading and input/output of one or more prediction models. The knowledge graph is also used to detect which models should be utilised and their execution sequence in order to conduct a calculation given the provided inputs and requested outputs. For this functionality model library includes all required model properties to enable automated model selection. This functionality provides a solution to address the diversity of model that may exist in a library especially within the context of co-simulation. The Vessel Domain Model Data used to build the digital twin, needs to be based on some reference schemas. Standardization speeds up the adoption of digital twinning. Standard industry-wide schemas need to be introduced, while for data types that are shared within different domains (e.g., geospatial data) existing models can be utilised. Ship data models are used by classification societies and provide an important baseline in this area. A notable example is the DNV Functionally Oriented Vessel Data Model (DNV n.d) based on a hierarchical ship’s functions structure, a library of components and interrelations between functions and components, adhering to ISO 15926A (ISO, 21003) principles. Ontologies have become a standard for knowledge representation across several domains (RodríguezRevello et al, 2023). In maritime in particular, ontologies have been utilized for solving collision situations at sea (Hatlas-Sowinska, 2023) and for knowledge extraction for shipyard fabrication workshop reports (Hiekata et al, 2010). In our approach, we define an ontology layer to represent vessel functions and their corresponding variables, and secondarily, other domain-related concepts such as fuels, weather, voyages, and more. This ontology layer is also encoded in the Knowledge Graph, aiming to support interoperability between DT models. Figure 3. Vessel components/functions and variables ontology layer 62 Shipping Digital Twin Data Management With Use of Knowledge Graphs Vessel Components and Functions Metamodel The term “Functions” refers primarily to ‘vessel functions’, i.e., the ability to perform or prevent certain actions. Examples are structural integrity, anchoring, propulsion, navigation and fire prevention. However, the term Function is also used in a broad sense covering elements such as administrative items and compartments. While building a DT, we consider for a collection of every function that we are interested in, such as pumping, the main engine, the auxiliary engines, but also green functions such as heat recovery, wind assistance, etc. Some functions may be applicable to decarbonization solutions. For example, hull coating (silicon hull paints) or energy saving devices (Gaggero &Martinelli, 2021), such as propeller boss cap fins can be characterized as decarbonization technology. This information is included in the knowledge graph either as a property or preferably by linking the relevant nodes. Variables Associated to each function, there are variables that describe the condition, performance, or status of the functionality. For example, the exhaust gas heat recovery function is described by gas mass rate and temperature, exchanger inlet and outlet temperature etc. There exist variables that instead of being related to a “function/component” node, are related to elements of the graphs described below, for example “significant wave height” of the weather-related graph, or “required speed” of the “Charter Party Agreement” graph. Vessel Particulars Vessel geometrical properties, tanks sounding tables, deadweight scale, and other information that is required input for simulation models. Operational References This aspect of the knowledge graph encodes various reference conditions related to vessel operation, categorized by component, i.e., M/E, DGs, boilers, hull and by source: • Official Tests ◦Shop test ◦Sea trials ◦Maker reference • User reference ◦Based on trials ◦Based on data analysis • System ◦Model weights ◦Constants library 63 Shipping Digital Twin Data Management With Use of Knowledge Graphs Fuel Information Graph including available fuel types, properties, prices and ports availability. Shipping Sector Information • Commercial practices ◦Charter Party Agreement (CPA) terms ◦Company common practices ◦Ports – vessel interaction (required paperwork, info exchange). • Voyage particulars ◦events/operations • Regulations ◦CO2 SOx NOx emissions metrics ◦Discharged water/wastes ◦ECAs ◦Safety ◦Ballast treatment • Cargo related information ◦Cargo categorization, locations, and freights • Infrastructure ◦There exists relationship with other classes for example, bunkering locations are also referenced in the fuel graph. ◦Terminals (container, refinery, passengers, vehicles) Weather Weather services providers and linkage to the weather related variables. Encoding Use Cases Into the Knowledge Graph As per the operational metamodel, a use case comprises requirements, setpoints, and metrics. Figure 4 illustrates the encoding of use cases in the knowledge graph. Relevant variables are associated with the use case using corresponding relationships such as “operational requirement”, “setpoint”, etc. The Models Layer of the Knowledge Graph In a digital twin framework, the ability to simulate various operational and environmental scenarios plays a crucial role. This dynamic digital representation requires the integration of advanced simulation models to mirror the real-world behavior of vessels, their interactions with the environment, and the myriad of operational conditions they encounter. By registering simulation models into the framework, the system can predict outcomes, optimize operations, and enhance decision-making processes with accuracy and efficiency. 64 Shipping Digital Twin Data Management With Use of Knowledge Graphs The knowledge graph is used to store the available models metadata and more importantly, the relation of model inputs and outputs with the variables of the vessel domain model, presented previously. shows the graph related to the operational phase where the model is used to predict component operation. Figure 5 shows a model that has as an input a variable and its output is fed as input to another model estimating the final variable. It is also shown an alternative model that estimates the final variable, using only the first variable, ignoring variables related to the component in the middle. In Figure 5 (b) and (c) is shown model instantiation for a vessel, which is discussed in the next section. The Layer of Vessel Instantiation: The Incorporation of the DT The vessel domain model may include all vessel functions and variables that can be considered. However, these may not be applicable to every application of vessel. On the other side one vessel may have multiple instances of a single item. In this position it is introduced the instantiation of a vessel. A new node, shown in dark red in Figure 5, representing the actual vessel is created in the knowledge graph. New entities, shown in dark green represent the actual measurements, instances of the variables of the generic model. The focus so far has been on the theoretical components, e.g. variables not related to a specific vessel or measurements. Now, we shift our attention to an instantiation, which involves a specific vessel equipped with various components. Measurements that describe the operation of these components are explicitly linked to relevant variables in the knowledge graph. This establishes a reliable reference for understanding the nature of these measurements. Figure 4. The relationships of a use case with the variables of the ontology encoded in the KG 65 Shipping Digital Twin Data Management With Use of Knowledge Graphs The nodes within the scope of the DT serve a dual purpose: they house both the actual measured values and the simulated values, that stem from the execution of the corresponding model. Knowledge Graph for Digital Twin Linking The utilization of the knowledge graph, which connects different aspects and instances through the ontology layer, offers a notable advantage in terms of transferring knowledge across DT. Figure 5. Models and their relationships within the KG (a) Connections with the ontology layer, (b) instantiation of a model for a specific vessel, (c) automated modification of models’ instantiation after change of vessel assets 66 Shipping Digital Twin Data Management With Use of Knowledge Graphs In the example of Figure 6, there are two vessels, one of which is equipped with a new green technology. A simulation model is tuned to accurately predict the effect of this technology on vessel performance, based on the specific use case. The knowledge graph provides the opportunity to identify interdependencies and enables the conditional utilization of the model weights in the second vessel, which is not equipped with the green technology. This ‘knowledge transfer’ allows for the evaluation of the green technology on the second vessel. Model Pipeline Creation Using the Knowledge Graph Part of our research objective is the development of a framework capable to accommodate a wide context in terms of decarbonisation technologies and applications, intended to provide a unified representation of vessel operational aspects. To achieve these two objectives, it is required to constantly produce and consume the available knowledge. However, this is complicated, especially when considering the models that are available. A solution suggested to address these difficulties is using knowledge graphs at the metalevel of the models. Including model blueprints into the knowledge graphs and connecting model inputs/outputs with the variables of the ontology layer it is achieved to provide a semantic layer to the user and tune system operation by detection of the shorter model execution path. The knowledge graph stores a register of variables and a library of models used in the DT. Model inputs/outputs are related to the variables of the register. Vessel measurements are also related to variables of the register and models are instantiated for specific vessels. When the DT is required to simulate vessel operation, it is required to determine a model pipeline, in other words model sequence, which will be executed for the estimation of the required variables (outputs) using the available variables (inputs). The input and output variables are determined by the use case described in previous sections. For the detection of the model pipeline, the knowledge graph is employed. Since the models are connected to the ontology variables with “input of” and “output of” relationships, the interconnections between models can be inferred. In the knowledge graph is stored which variables have a known value Figure 6. Knowledge transfer in terms of simulation model weights in different vessel via the dataspace 67 Shipping Digital Twin Data Management With Use of Knowledge Graphs and which are the outputs. Therefore, a special query at the knowledge graph can return the model sequence required to calculate the outputs based on the available inputs. In addition, it is possible to assign a computational cost metric and accuracy metric. Technically, the metrics are attributed to the relationships rather that nodes. Then, a query can be applied to provide the “shortest path” considering the weights defined by the application (accuracy vs execution time). A SAMPLE APPLICATION OF THE APPROACH The Application of Variable Frequency Drives for Cooling System Optimization Historically, the design and construction of marine vessels did not prioritize the energy efficiency of auxiliary systems, leading to the implementation of systems on existing ships that lack optimization for reduced fuel consumption. This trend persists, with many ships still being constructed without a significant focus on energy-efficient solutions. Furthermore, shipyards often overlook the long-term ownership costs associated with vessels. In the absence of specific owner demands for incorporating certain technologies into the design specifications, the energy efficiency potential of these ships remains underutilized, despite the fact that investing in additional equipment could result in substantial savings within a year. Currently, many marine installations adjust for environmental fluctuations using inefficient methods like ‘throttling’ and ‘by-pass loops’. Shipboard systems that offer significant opportunities for enhancing energy efficiency include those with large pumps and fans, which do not need to operate continuously at maximum capacity. Equipping electric motors with Variable Frequency Drives (VFDs) can lead to more efficient operation of pumps and fans under partial load conditions, such as during reduced-speed sailing or when there is less need for ventilation. The electrical power usage of a pump correlates with its volumetric flow rate as per the affinity laws. Decreasing the pump’s speed results in a squared reduction in system pressure (Head) and a cubed decrease in electrical power consumption. For instance, reducing the pump’s speed by 10% can lead to a 27% reduction in power usage (Inal et al,2023). The cooling water system aboard a vessel, primarily composed of pumps, stands as a significant energy consumer on ships. Seawater serves as the primary coolant for the machinery within the engine room. A typical oceangoing merchant vessel’s central cooling water system is divided into three subsystems: the seawater (SW) cooling system, the low-temperature fresh water (LTFW) cooling system, and the high-temperature fresh water (HTFW) cooling system. The SW system uses seawater to cool the LTFW system’s fresh water, while the LTFW system, in turn, cools the HTFW system. Each of the three systems—SW, LTFW, and HTFW—has two dedicated pumps, all of which are electrically powered. The pump capacity is dictated by the ship’s size and the main engine’s power output. Ship systems must operate within a highly dynamic environment, with the central cooling water system particularly affected by this variability due to fluctuations in engine load, seawater temperatures, and LTFW side cooler temperatures. Engine load variations are often a result of maneuvering conditions, such as when navigating at full speed or slowing down in narrow passages, channels, straits, and during berthing. Additionally, since ships traverse international waters, they encounter a wide range of climates and seawater temperatures, directly impacting the central cooling water system. The cooling system of a ship is engineered to function under extreme conditions for safety reasons. Under these design conditions, the pumps operate at full capacity, constant speed, and constant mass flow 68 Shipping Digital Twin Data Management With Use of Knowledge Graphs rates regardless of the ambient conditions. From an energy efficiency standpoint, this results in suboptimal performance when conditions are not tropical. Typically, temperature regulation of the fresh water in varying environmental conditions is managed by bypassing the water. However, adjusting the mass flow rate of the seawater pumps in response to changing seawater temperatures can lead to substantial energy savings. Implementing variable speed seawater pumps could achieve this efficiency. Knowledge Graph Entries for Cooling System Optimization The foundational layer of the knowledge graph is dedicated to the vessel domain model, encompassing the components and functions being analyzed, as well as the variables necessary to characterize their operations. Concentrating on the High Temperature Fresh Water Cooling System, to enable Digital Twin adaptation it is essential to incorporate into the knowledge graph the Main Engine (M/E) and variables such Figure 7. Knowledge graph nodes related to the cooling system optimization use case 75 Intelligent Ship-Edge Computing Enabling Automated Configuration Cloud vs. Edge Computing The primary difference between cloud computing and edge computing is the location where data processing occurs. In cloud computing, data is processed on a central cloud server, which is usually located far away from the source of information. Traditionally compute took place on centralized cloud services such as Amazon EC2 instances. Hybrid cloud is a mixed computing environment where applications are run using a combination of computing, storage, and services in different environments—public clouds and private clouds, including on-premises data centres or edge locations. Hybrid cloud computing approaches are widespread because almost no one today relies entirely on a single public cloud. Hybrid cloud solutions offer the flexibility to seamlessly migrate and manage workloads across diverse cloud environments, empowering organizations to tailor their infrastructure to meet specific business requirements. By adopting hybrid cloud platforms, organizations gain the ability to lower costs, mitigate risks, avoid vendor lock-in, and leverage existing cloud-native developer skills and CI/CD pipelines to drive successful digital transformation initiatives. In today’s landscape, the hybrid cloud approach has become a prevalent infrastructure setup. As organizations undergo cloud migrations, hybrid cloud implementations naturally emerge, enabling a gradual and systematic transition of applications and data. With hybrid cloud environments, enterprises can continue utilizing on-premises services while harnessing the advantages of public cloud providers like AWS, Azure, and GCP, which offer flexible options for data storage and application access (Google, n.d.). Edge computing extends the hybrid cloud paradigm to address the unique requirements of enterprise and consumer applications. Although they possess individual traits, edge computing and hybrid cloud can collaborate to establish a comprehensive and adaptable computing infrastructure. A notable aspect of contemporary edge computing solutions is their adoption of cloud native development practices specifically designed for the edge. By leveraging cloud native development practices, applications can be built and deployed using the same skills and tools that have been honed for developing cloud native applications in hyper-scale cloud environments or private data centres. This allows for the seamless extension of these practices to the edge, enabling organizations to leverage their existing expertise and resources for edge computing deployments. Hence, aspects such as distributed computing, data processing, management, and integration, as well as workload placement and optimisation are enabled in an accelerated and scalable manner. Concept and Technology Edge architecture encompasses all the active components of edge computing, including devices, sensors, servers, and clouds, spread throughout the network wherever data is processed or utilized. By bringing processing closer to the data source, edge architecture significantly reduces latency and cost: applications and programs running at the edge can swiftly and efficiently respond to user interaction and data without the need for transferring data across a wide area network, resulting in an enhanced user experience and optimal overall performance. The definition of the “edge” is flexible and context-dependent. For example, in the case of a shipping company, the edge may encompass activities at docks where shipments are loaded and unloaded, or it may include on-board processing of ship systems and operations. In both scenarios, processing and analysis occur in near real-time, leading to data-driven decision-making. While the company’s 76 Intelligent Ship-Edge Computing Enabling Automated Configuration headquarters with the main data centre may be located miles away, the edge represents the crucial point where data is collected, processed, and managed to derive insights, irrespective of latency challenges. Why Edge Computing in the Maritime Industry Ship transport accounts for over 80% of global trade in goods and raw materials, making the efficiency and speed of large container ships critical to the value chains and production processes of countries worldwide. However, the environmental impact of shipping, including emissions, fuel efficiency, and noise pollution, is becoming more apparent. To address these concerns, emerging technologies like artificial intelligence (AI) and digital twin are being leveraged to improve the efficiency, management, and environmental sustainability of shipping routes. Whilst digitalisation of the transport sector is key to unlocking efficiencies, this is currently hindered by data silos, technological limitations, and barriers of complexity. DT4GS promises to enable stakeholders in shipping to actively embrace the full spectrum of Digital Twins innovations to support smart green shipping in both the upgrade of existing ships, as well as the building of new vessels. These solutions will be deployed across the cloud continuum from centralized public cloud services to decentralized edge computing devices. At one end, ship owners and stakeholders can leverage public cloud services provided by major cloud providers that offer centralized computing power and storage resources that can be accessed from anywhere. On the other end, edge computing will be deployed closer to where the data is generated to provide compute at every step of the ship system. This provides unique benefits and advantages in terms of data sovereignty and security, resilience to outages and connectivity constraints, innovation opportunities and flexibility to different providers, as well as inherent autonomy that are critically important to shipping. There are multiple complexities inherent to ship operations and decision making. Whilst the central complexity may be the difficulty and uncertainty inherent in decision making in the ocean, there are multiple technological challenges to overcome: • Scale: There are thousands of ships with limited or no technical expertise on board. The solutions implemented must be self-healing and turn-key, requiring minimal intervention and maintenance. • Heterogeneity: Each ship represents a dynamic and unique environment, making it challenging to find a one-size-fits-all model. Solutions need to be adaptable and flexible to cater to the diverse needs and characteristics of each vessel. • Data Gravity: Ships and their instruments generate vast amounts of logs, events, and metric data, often in different formats. Additionally, the sensitivity of some of this data adds another layer of complexity. Handling and processing this data in a secure and efficient manner is crucial. • Resource Constraints: Ships serve as the ultimate edge nodes, characterized by limited communication, computational, and energy resources. These constraints must be considered when designing solutions, ensuring they can operate effectively within the ship’s resource limitations. Fundamentally, edge and cloud technologies promise a solution that seamlessly infuse AI and simulation across the entire spectrum of shipping operations. Agnostic of connectivity or compute capabilities, it provides a continuously available digital assistant to guide all aspects of decision and automation. AI can assist Captains and crew to make better decisions by seeing things humans overlook or don’t fully recognize; by assessing alternatives they may not have considered; by evaluating and optimizing 77 Intelligent Ship-Edge Computing Enabling Automated Configuration choices with more mathematical precision than can be accomplished by the human mind alone; by reacting to events in real-time and at a rate that is much higher than is humanly possible. Perhaps the AI can pre-empt these crises from emerging. And the holy grail of these potential advantages is to inspire decision-makers to ideas they might not have produced on their own. Doing so promises to improve decision-making, which in turn will improve operational efficiency, safety, and decrease burden. With AI we can reduce the current carbon footprint of transporting goods across the oceans. We can incur fewer accidents, limit cargo damage and loss, and most importantly make maritime travel less risky. To do this, we need to harness the power of AI to recognize dangers, evaluate alternatives, and present recommendations in the decision-making process. We need to incorporate AI into the human decision processes. In essence, the AI needs to act as another member of the crew, offering interpretation, advice, and warnings as appropriate to provide meaningful contribution to how crew leaders and the Captain make their decisions. Easier said than done. A prominent example in this regard is the Mayflower Autonomous Ship (MAS400) (Kıcıman et al., 2023; MAS, n.d.). The MAS400 vessel was built without any provision for humans on board and must make all navigational decisions on its own. It uses an array of visual and other signal input to sense its environment. AI is used to perceive and interpret what it “sees”. It classifies a wide range of objects and determines whether they are an obstacle to its current course. It establishes potential strategies for navigating around them, and evaluates each for their safety, conformance to regulations, and optimization. And then, of course, it puts those decisions into action – controlling rudder and speed to achieve the outcome it has determined. However, a potentially larger and more immediate use of this technology is not for vessels that drive themselves all over the world, but rather to assist sea Captains in their own navigational and ship operations responsibilities. One of the lessons from the Mayflower project is that sometimes you do need a human on-board – to fix a broken manifold coupler to the auxiliary engine, or to repair the electrical connection that has frayed and causing a short. And with human lives at risk, you need humans that have the authority and responsibility for protecting those lives – for making the critical decisions necessary to fulfil their mission safely. STATE OF THE ART IN EDGE COMPUTING A forecast by International Data Corporation (IDC) estimates that there will be 41.6 billion IoT devices in 2025, capable of generating 79.4 zettabytes (ZB) of data (Hojlo, 2021). The latency advantages of edge computing provides obvious appeal to extract insight from these vast volumes of data. However there are also multiple other considerations driving edge adoption including security, operational costs, resilience, and flexibility (Rathore et al., 2022). The global edge computing industry is experiencing rapid growth, driven by advancements in technologies such as 5G, IoT, and the Industrial Internet (Rathore et al., 2022). Edge computing entails extending a consistent computing environment from the core datacentre to physical locations in close proximity to users and data. Similar to a hybrid cloud strategy that enables organizations to run workloads across their own datacentres and public cloud infrastructure, an edge strategy expands the reach of a cloud environment to numerous additional locations. This approach allows for a more widespread and 78 Intelligent Ship-Edge Computing Enabling Automated Configuration distributed computing infrastructure, bringing computing capabilities closer to the edge for enhanced performance and responsiveness. There are three categories of edge use cases (Schabell, 2022): • The first is called enterprise edge, and it allows customers to extend application services to remote locations. It has a core enterprise data store located in a datacentre or as a cloud resource. • The second is operations edge, which focuses on analysing inputs in real time (from Internet of Things sensors, for example) to provide immediate decisions that result in actions. For performance reasons, this generally happens onsite. This kind of edge is a place to gather, process, and act on data. • The third category is provider edge, which manages a network for others, as in the case of telecommunications service providers. This type of edge focuses on creating a reliable, low-latency network with computing environments close to mobile and fixed users. • For industry applications, edge computing capabilities can be decomposed into two categories: • Edge Server: This typically refers to IT equipment specifically designed for computing tasks at the edge and operating as an extension to the computation tasks performed in the cloud. It can take the form of a half rack comprising 4 or 8 blades, or it can be an industrial PC. Essentially, it is a piece of IT equipment dedicated to edge computing operations. • Edge Device: An edge device is a piece of equipment built for a specific purpose, such as an assembly machine, a turbine, or a car. While these devices were initially designed for their primary functions, they also possess computing capacity. In fact, many devices that were traditionally considered IoT devices now incorporate increased computational power. For instance, an average car today is equipped with approximately 50 CPUs. These devices often run on the Linux operating system, allowing for the deployment of software containers, and enabling edge computing applications to be executed directly on the devices themselves. Deploying workloads across both cloud and edge resources presents several fundamental challenges for technology solutions: • Consider the scale of the challenge at hand. With an estimated 15 billion devices in the world today, enterprises are faced with the task of managing thousands of devices within their operations. • Heterogeneity at the edge poses another significant factor to address. These devices come in various forms, each serving a different purpose and running on different operating systems. Managing the diverse range of devices and the specific tasks they perform can be a complex endeavour. • Security is a critical concern when dealing with edge devices. Unlike traditional IT data centres, these devices operate outside the confines of controlled environments. They lack the physical barriers, uniformity, and consistency typically found in hybrid cloud environments that aid in ensuring security. Protecting edge devices from tampering and unauthorized access becomes a priority. • Building a robust ecosystem is essential. The role of edge computing is rapidly expanding and will have a profound impact on enterprise computing, much like how mobile technology has influenced consumer behaviour. Establishing a comprehensive ecosystem that integrates and supports edge devices and their functionalities will be crucial for organizations to maximize the benefits and potential of edge computing. 79 Intelligent Ship-Edge Computing Enabling Automated Configuration • Maintenance and ongoing support presents some unique challenges. Software and AI components are complex and require period patches and updates. Software performance can also be impacted by environmental conditions. Having a rigorous approach for software life-cycle management and drift detection is essential to achieve optimal performance and avoid catastrophic failure where nearby IT support personnel is not available. • Ensuring data privacy throughout the deployment process is essential to safeguard sensitive information against data breaches, regulatory compliance, and to enhance public trust within organisations. As workloads are distributed across diverse and interconnected environments, ranging from public clouds to edge devices, the potential attack surface widens significantly. Adhering to stringent data privacy measures becomes imperative to mitigate risks associated with unauthorized access, data leakage, and malicious activities. • The strategic placement and optimization of workloads play a pivotal role in maximizing performance, minimizing latency, and optimizing resource utilization. The deployment of workloads across diverse environments, encompassing cloud data centres and edge devices, presents unique challenges and opportunities. By harnessing the power of intelligent workload orchestration, organizations can unlock the full potential of their computing infrastructure, ensuring optimal performance and resource allocation while seamlessly catering to the unique requirements of individual workloads. FULL STACK EDGE SOLUTION Edge solutions require multifaceted capabilities, including the orchestration of heterogeneous workloads, creation of a scalable, extensible, and robust data management pipeline, and seamless deployment of data inference and AI on the edge. The latter consideration is especially critical as many organisations become AI-driven enterprise and infuse generative AI capabilities across their operations (Yusuf, 2023). The requirements of workload orchestration, data operations, and AI infusion, include: • Edge orchestration: A fundamental requirement is to enable seamless deployment of containerised applications to multicloud and edge environments. Deploying on 1000s of nodes in a scalable, replicable manner requires deep automation for advanced management of 1000s of different software applications across 1000s of compute nodes distributed across the entire footprint of an organisation’s operating remit. Solutions such as Open Horizon (Open Horizon, 2023) enables the autonomous management of more than 10,000 edge devices simultaneously. Core architecture considerations include: ◦The provisioning of Open Horizon includes the management hub that runs in an instance of OpenShift Container Platform installed in the data centre. The management hub is where the management of all remote edge nodes (edge devices and edge clusters) occurs. ◦Managed edge nodes can then be installed in remote on-premises locations to make application workloads local to where critical business operations physically occur, such as docks, ships, factories, warehouses, retail outlets, distribution centres, and more. • Data Operations (DataOps): Edge consist of disparate devices such as engines, propellors, hulls, berths, shipping containers, etc., that are all generating data. Resilient edge solutions require robust data management and organisational strategies for collecting and handling data, ensuring 80 Intelligent Ship-Edge Computing Enabling Automated Configuration compliance with data sovereignty regulations, and providing flexible data quality solutions facilitating the training and deployment of AI models. Further, cognisant of data latency restrictions, data processing on the edge must translate from vast volumes of raw data to digestible subsets of high-quality, high-value features that can be used for AI model finetuning and prediction. As an example, computer-vision based hull monitoring solutions generate large volumes of data, while only small subsets of this data may be relevant for guiding hull cleaning and maintenance. • AI Operations (AIOps): AI-backed decision making will play a critical role in improving the sustainability, safety, and efficiency of shipping. Ship captains and stakeholders can use AI to process large data volumes and make decisions related to route selection, navigation, port arrival and logistics, weather-enforced disruption and mitigation. Further, many aspects can be fully automated such as power management, HVAC system optimisation, cargo and inventory management, predictive maintenance, crew management, and risk analytics. As AI begins to play a central role in shipping, it is critical that 1) ships possess the compute infrastructure to fully exploit these advantages and 2) the AI layer provides reliable inference monitoring mechanisms to assess model performance, detect potential errors or uncertainties, and instil confidence in the decision-making process. • Network Operations (NetOps) plays a crucial role in the seamless deployment of workloads across the cloud to edge continuum. NetOps makes sure the network infrastructure is properly configured, optimized, and secured. It involves activities like network surveillance, traffic control, performance enhancement, security setup, and troubleshooting. Ship owners may ensure effective data transmission, low latency, and dependable connectivity across cloud and edge settings by utilizing NetOps principles, enabling the smooth deployment and execution of workloads. Additionally, NetOps aids in resolving issues with network resilience, scalability, and resource allocation, enabling businesses to fully utilize the cloud to edge continuum. Ultimately, NetOps serves as a crucial link that makes it possible for workloads to be seamlessly integrated and run across a range of computer platforms, improving flexibility, performance. Edge Infrastructure and Compute Recent years has seen the evolution and maturation of IoT technology from primarily a data collection and transmission approach, to devices with significant compute capabilities and the ability to deploy workloads to the data. In traditional IoT setups, data generated by connected devices were often transmitted to the cloud for processing and analysis. This approach worked well for certain applications but introduced challenges like latency, network congestion, and increased reliance on cloud connectivity. Edge computing emerged as a solution by bringing data processing closer to the source, enabling realtime analysis and quicker response times. As edge matures, a critical concern for stakeholders is infrastructure considerations such as server selection for different workloads, security of distributed devices, and loading necessary software onto hardware devices to support various edge applications such as visual inspection, voice interaction, and inference. Compute capabilities within the shipping and logistics space is an evolving state. While container ships may have some computing equipment for basic tasks such as monitoring and controlling ship systems, these capabilities are generally limited in scope and processing power. The computing infra- 81 Intelligent Ship-Edge Computing Enabling Automated Configuration structure on a container ship is designed to support essential functions like engine control, navigation, communication, and safety systems. However, it’s worth noting that with the increasing adoption of digital technologies and automation in the shipping industry, there is a growing trend towards incorporating more advanced computing capabilities on certain types of vessels. For example, larger and more advanced container ships may have additional computing systems for tasks like cargo management, route optimization, and fuel efficiency monitoring. Overall, while there may be some level of compute capability on a typical container ship, it is typically limited and focused on specific operational requirements rather than extensive computing tasks. The bulk of compute-intensive tasks, data processing, and analytics are more commonly performed in onshore data centres or edge computing systems that support the shipping operations. Several companies offer edge solutions that provide integrated hardware and software for edge computing deployments. Some examples include: • Dell EMC Edge Gateway provides a compact and ruggedized device designed for edge computing. It comes with off-the-shelf, pre-configured, pre-certified, ready-to-use products, for different industries or applications (Dell, 2023). • Intel NUC Edge is a compact edge device that promises an out-of-the-box solution ideal for running any critical applications on-premise with immediate high availability (Scale Computing, n.d.). • Microsoft Azure Stack Edge is a solution that combines hardware and software to bring AI, analytics, and IoT capabilities to the edge. It includes an on-premises appliance that can be deployed in disconnected or low-connectivity environments and integrates with Azure services for seamless cloud-edge integration (Microsoft Azure, n.d.). When one considers compute infrastructure for shipping, it is important to consider the types of compute capabilities that may be deployed: edge servers and edge devices (described in more detail in Section 1). Edge Platform and Orchestration Edge, by its very nature, generally involves 1000s of devices. Hence, concepts from the data centre do not translate directly to the edge. Instead, it requires software capabilities to monitor and update a limitless number of edge devices from across the world and new security technology and protocols to keep everything safe (Hines, 2023). An edge computing platform contains many components. These include: • Edge Management and Orchestration: This component handles the management, configuration, and physical deployment of edge devices and placement of applications that will run on them. It ensures efficient resource allocation, software updates, and monitoring of edge infrastructure. • Security and Authentication: Edge computing software incorporates security measures to protect edge devices, data, and communications. This includes authentication, encryption, access control, and threat detection capabilities. 82 Intelligent Ship-Edge Computing Enabling Automated Configuration • Edge Gateway: The edge gateway serves as a bridge between the edge devices and the central cloud or data centre. It provides connectivity, protocol translation, and data aggregation capabilities. • Containerization and Virtualization: To enable portability and ease of deployment, containerisation of software workloads is critical. Software containers or virtual machines encapsulate applications and their dependencies, making them easier to manage and deploy at the edge. Edge management and orchestration are critical components of edge computing. Edge management refers to the management of edge devices, including device provisioning, configuration, and monitoring. Edge orchestration refers to the management of network resources across an edge network by controlling how network resources flow between devices and applications in an edge environment to produce a more responsive and smartly optimized network. As an example, Open Horizon deploys an orchestrator called “management hub” in an instance of OpenShift Container Platform installed in the data centre. The management hub is responsible for controlling the placement of containers to all remote edge nodes, both edge servers and edge devices. Figure 1 provides a typical edge computing architecture. The two most critical responsibilities of the orchestrator are the deployment and monitoring of workloads. Fundamentally, the edge deployment is based on containerisation of workloads. A Cloud Operations team installs the management hub components. Specific applications are developed by data scientists and domain experts and containerised, which are then published to the management hub edge library. Administrators define the deployment policies that control where edge services are deployed. Edge orchestrators use various monitoring techniques to ensure that edge devices are functioning correctly and that network resources are being used efficiently. Some examples of monitoring techniques used by edge orchestrators include: Figure 1. High-level topology for a typical edge computing setup Taken from https://open-horizon.github.io/ 83 Intelligent Ship-Edge Computing Enabling Automated Configuration • Device monitoring: Monitor the status of edge devices to ensure that they are functioning correctly. This includes monitoring device health, connectivity, and performance. • Application monitoring: Monitor the performance of applications running at the edge of the network to ensure that they are meeting performance requirements. • Network monitoring: Monitor network traffic to ensure that network resources are being used efficiently and that there are no bottlenecks in the network. This can also include monitoring of network traffic for security threats and vulnerabilities. • Data monitoring: Edge orchestrators monitor data flows between devices and applications to ensure that data is being processed correctly and that there are no data integrity issues. These monitoring techniques help edge orchestrators to identify issues before they become critical and to optimize network resources for better performance. In the context of shipping, it is critical to consider “Day 0/Day 1/Day 2” of the software lifecycle: • Day 0 activities encompass the initial definition, design, procurement and provisioning of the hardware and software solution. For factory pre-install, this stage includes the initial installation and testing of the hardware and software in the factory prior to shipping it to its final destination. Additional testing are also performed when the device is wired into the network at its final destination and powered up. • Day 1 marks the actual deployment or launch of the software solution. It represents the first day of production use or operation. On Day 1, the software is made available to users or customers, and the system goes live. This phase involves activities such as testing, data migration, user onboarding, and initial training. The goal of Day 1 is to successfully deploy software created during Day 0 and transition from the development phase to the operational phase and ensure that the software meets the required functionality and performance standards. • Day 2 refers to the ongoing operational phase of the software after it has been deployed and is in active use. It represents the period of maintenance, support, and continuous improvement. During Day 2, organisations focus on tasks such as monitoring, troubleshooting, bug fixes, performance optimization, regular updates, and feature enhancements. The emphasis is on managing and maintaining the software to ensure its stability, reliability, security, and efficiency throughout its lifecycle. In the case of shipping, the vast number of deployed edges and the geographical location of each deployment generally makes it prohibitively expensive to provide local IT support at each location to monitor, maintain and update these deployments. Instead, edge technologies must provide a comprehensive solution for administration, managing, monitoring, and securing an almost limitless number of edge servers and devices. The deployment of thousands of Edge devices means that each of those devices are potential entry points for hackers and security breaches. There are some critical considerations to security on the edge. Some of these include (Iyengar, 2023): • The enterprise should have the ability to check whether the edge nodes are operating properly by comparing the current configuration of various resources against the desired state. • The communication between an edge agent and management hub should be signed and encrypted. 84 Intelligent Ship-Edge Computing Enabling Automated Configuration • Each container run on an edge endpoint should be verified against the official container to check for tampering. • Each container should be running in its own “docker” network with self-regulated (application dependencies) connectivity between the components. • Each edge agent should check that it is running the latest version of the container when the endpoint is connected to a network. • The enterprise should create a cryptographic signing key pair and have a plan to rotate them. In the shipping industry, security plays a vital role as any unauthorized access or control over a vessel’s ecosystem can have catastrophic consequences. Ensuring robust security measures is of utmost importance to prevent potential threats and protect the safety and integrity of maritime operations. In contrast to typical centralized Internet of Things (IoT) platforms and cloud-based control systems, the edge control plane is mostly decentralized. Each role within the system has a limited scope of authority so that each role has only the minimum level of authority that is needed to complete associated tasks. No single authority can assert control over the entire system. A user or role cannot gain access to all nodes in the system by compromising any single host or software component. For Open Horizon, the control plane is implemented by three different software entities (IBM, 2021): • Horizon agents: Outnumber all of the other actors in Open Horizon. An agent runs on each of the managed edge nodes. Each agent has authority to manage only that one, single, edge node. The agent advertises its public key in Horizon exchange, and negotiates with remote agbot processes to manage the local node’s software. The agent only expects to receive communications from the agbots that are responsible for deployment patterns within the agent’s organization. • Horizon agbots (agreement robots): Are processes that can run anywhere. By default, the processes run automatically. Agbot instances are the second most common actors in Horizon. Each agbot is responsible for only the deployment patterns that are assigned to that agbot. Deployment patterns consist primarily of policies, and a software service manifest. A single agbot instance can manage multiple deployment patterns for an organization. Deployment patterns are published by developers in the context of an Open Horizon managed user organization. The deployment patterns are served by agbots to Horizon agents. When an edge node is registered with Horizon exchange, a deployment pattern for the organization is assigned to the edge node. The agent on that edge node accepts offers only from agbots that present that specific deployment pattern from that specific organization. The agbot is a delivery vehicle for deployment patterns, but the deployment pattern itself must be acceptable to the policies that are set on the edge node by the edge node owner. The deployment pattern must pass signature validation, or the pattern is not accepted by the agent. • Horizon exchange: Refers to a centralized, but geographically replicated and load balanced, service that enables the distributed agents and agbots to join and negotiate agreements. Horizon exchange also functions as a shared database of metadata for users, organizations, edge nodes, and all published services, policies, and deployment patterns. • The switchboard in Open Horizon serves as a central communication hub, facilitating coordination and orchestration between edge devices and the management infrastructure. It ensures efficient workload placement, dynamic updates, and seamless communication, optimizing resource utilization and enabling effective deployment of workloads. 91 Intelligent Ship-Edge Computing Enabling Automated Configuration CHALLENGES AND RESEARCH DIRECTIONS Rathore et al. (2022) review some of the challenges and future opportunities for edge computing, including: • Security: The distributed architecture, mobility support, and data processing in edge computing raise security concerns. Thorough testing, access control systems, trust management, privacy measures, and intrusion detection systems are crucial. Security protocols, artificial intelligence, and encryption techniques can enhance protection, but fatal defects need to be addressed. • Cost: Configuring, deploying, and maintaining edge computing devices can be expensive. Backup servers for fault tolerance and machine learning-based systems may incur additional costs. Developing cost-effective devices with low communication and computation expenses is essential. • Power: Providing cloud-like remote services in remote areas requires high-power processors, voltage, and three-phase electricity, posing challenges. • Data: Optimizing data usage is vital as edge computing devices process subsets of data, potentially overlooking additional data or wasting raw data. Effective data segregation and sharing across devices are crucial for efficiency. Computation-aware networking can enhance distributed systems. • Heterogeneity: Resolving the heterogeneity of data and the ecosystem within edge computing networks is a challenge. Collaboration between different vendor systems, load balancing, synchronization, resource sharing, data privacy, and interoperability contribute to the complexity. • Trust: User trust in edge computing systems is tied to security and privacy. Developing consumer trust models that foster confidence in adopting edge computing devices is an ongoing challenge. Shipping stands out as a prominent example of edge deployment due to its unique characteristics, including restricted latency, inherent geographic distribution, autonomy demands, substantial data generation capacity, and the necessity for resilience, fault tolerance, and regulatory compliance. Further the shipping industry has ambitious decarbonisation ambitions which impacts all aspects of decision making (IMO, n.d.). CONCLUSION There have been many evolutions in computing from transistor computing to microprocessors, to networking and the internet. The past two decades have seen the rise of cloud computing and the many instances of these from public cloud to private cloud to hybrid and multi-cloud solutions. Edge computing is considered an extension of the cloud continuum because it complements and expands the capabilities of cloud computing. While cloud computing excels at handling massive data storage, complex data analysis, and resource-intensive tasks, edge computing focuses on processing data locally, enabling quick responses, and reducing the need for data transmission to the cloud. These two computing paradigms work together to provide a holistic and scalable solution. By bringing computational power closer to the data sources, edge allows for faster processing and real-time decision-making, enhancing the overall efficiency of the system. The importance of edge enabled capabilities for shipping is obvious. Realizing the potential of edge computing is contingent on two factors: 92 Intelligent Ship-Edge Computing Enabling Automated Configuration • Extending cloud-centric development frameworks such as DevOps, DataOps, and AIOps to the Edge so that developers can seamlessly develop, deploy and improve new solutions within a single user framework. • Leveraging new AI technologies such as foundation models rapidly on the edge. Deploying these powerful capabilities allows the industry to unlock sophisticated natural language understanding, image generation, data analysis, and decision capabilities at a fraction of the effort. REFERENCES Arnold, M. (2020). Towards automating the AI operations lifecycle. Conference on Machine Learning and Systems. Bommasani, R. (2021). On the opportunities and risks of foundation models. ArXiv Prepr. ArXiv210807258. Dell. (2023). New Dell Edge Gateway 5200. Dell. https://www.dell.com/en-ie/shop/gateways-embeddedcomputing/dell-emc-edge-gateway-5200/spd/dell-edge-gateway-5200/emea_edgegateway5200 Google. (n.d.). What is a Hybrid Cloud? Google Cloud. https://cloud.google.com/learn/what-is-hybridcloud Hines, M. (2023). What is the Future of Edge Computing? Built In. https://builtin.com/cloud-computing/ future-edge-computing Hojlo, J. (2021). Future of Industry Ecosystems: Shared Insights & Data. IDC Blog. https://blogs.idc. com/2021/01/06/future-of-industry-ecosystems-shared-data-and-insights/. IBM. (2021). IBM Edge Application Manager. IBM. https://www.ibm.com/docs/en/eam/4.1. IBM. (n.d.). What is AIOps? IBM. https://www.ibm.com/topics/aiops IMO. (n.d.). UN body adopts climate change strategy for shipping. IMO. https://imopublicsite.azurewebsites.net/en/MediaCentre/PressBriefings/Pages/06GHGinitialstrategy.aspx. Iyengar, A. (2023). Security at the Edge. IBM. https://www.ibm.com/cloud/blog/security-at-the-edge Kahn, J. (2021). What an autonomous ship named Mayflower can teach us about building better A.I. Fortune. https://fortune.com/2021/11/30/a-i-lessons-from-mayflower-autonomous-ship-eye-on-ai/ Kahn, J. (2023). The inside story of ChatGPT: How OpenAI founder Sam Altman built the world’s hottest technology with billions from Microsoft. Fortune. https://fortune.com/longform/chatgpt-openaisam-altman-microsoft/ Kıcıman, E., Ness, R., Sharma, A., & Tan, C. (2023). Causal Reasoning and Large Language Models: Opening a New Frontier for Causality. ArXiv Prepr. ArXiv230500050. Kirillov, A. (2023). Segment Anything. arXiv, /arXiv.2304.02643. doi:10.1109/ICCV51070.2023.00371 MAS. (n.d.). Mayflower Autonomous Ship. https://mas400.com/. 93 Intelligent Ship-Edge Computing Enabling Automated Configuration Microsoft Azure. (n.d.). Azure Stack Edge. https://azure.microsoft.com/en-us/products/azure-stack/edge Open Horizon. (2023). Home. https://open-horizon.github.io/ Pearl, J. (2009). Causality. Cambridge University Press. doi:10.1017/CBO9780511803161 Pearl, J., & Mackenzie, D. (2018). The Book of Why: The New Science of Cause and Effect (1st ed.). Basic Books. Raghavan, S., & Shim, C. (2021). A new AI model could help track and adapt to climate change. IBM Research Blog. https://research.ibm.com/blog/geospatial-models-nasa-ai#fn-1 Rathore, V. S., Kumawat, V., Umamaheswari, B., & Mitra, P. (2022). Edge Computing: State of Art with Current Challenges and Future Opportunities. V. S. Rathore, S. C. Sharma, J. M. R. S. Tavares, C. Moreira, and B. Surendiran, (eds.). Rising Threats in Expert Applications and Solutions. Lecture Notes in Networks and Systems. Singapore: Springer Nature. doi:10.1007/978-981-19-1122-4_15 Scale Computing. (n.d.). Intel NUC Enterprise Edge Compute Edition Built with Scale Computing. https://www.scalecomputing.com/landing-pages/intel-nuc-enterprise-edge-compute-edition Schabell, E. D. (2022). 5 reference architecture designs for edge computing. Enable Architect. Redhat. https://www.redhat.com/architect/edge-portfolio-architecture Yusuf, K. (2023). Introducing watsonx: The future of AI for business. IBM Blog. https://www.ibm.com/ blog/introducing-watsonx-the-future-of-ai-for-business/ 94 Copyright © 2024, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. Chapter 5 DOI: 10.4018/978-1-6684-9848-4.ch005 ABSTRACT The chapter is concerned with the potential of alternative fuels, i.e., any other fuel than conventional fossil fuels, for powering ships. The alternative fuels surveyed in this chapter include liquefied natural gas (LNG), methanol, hydrogen, ammonia, as well as synthetic fuels (e-fuels). The chapter explains how digital twin’s simulation capabilities, can be used to model complex energy systems and alternative fuels and compute emissions, power consumption/output, etc., virtually. The chapter provides a comparison of alternative marine fuels, in terms of storage requirements and energy converters (e.g., combustion engines, fuel cells) suitability. Finally, the chapter discusses the role of digital twins in supporting further research and development towards the evolution of alternative fuels. INTRODUCTION Alternative fuels include fuels that can be used as drop-in fuels, like biodiesel, as well as hydrogen, ammonia, methanol, and Liquified Natural Gas (LNG) to replace currently used fuels such as heavy fuel oil (HFO), marine diesel oil (MDO), and marine gas oil (MGO). Alternative fuels do not impact global CO2 emission levels (Sustainable Ships-a), i.e. they are ‘carbon neutral’, or have CHG emissions that are significantly lower than fossil-based fuels. Biomass (the source of biodiesel) for instance, absorbs CO2 directly from the air. This biomass is then transformed into fuel, using a variety of processes, that Shipping Green Fuel Strategies and Benchmarking Supported by Digital Twins Anargyros Spyridon Mavrakos https://orcid.org/0000-0002-8290-9861 Inlecom, Belgium Maxime Woznicki CEA, France This chapter published as an Open Access Chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/) which permits unrestricted use, distribution, and production in any medium, provided the author of the original work and original publication source are properly credited. 95 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins is burned releasing the carbon again in the form of CO2. This is then again, absorbed by the biomass, creating a repeated cycle without adding to the total amount of CO2. Both alternative and fossil-based fuels are used for the powering of ship engines They are consumed within fuel cells, as liquids or gases in internal combustion engines and in external combustion engines. Ship fuels differ from each other in a variety of ways, as their chemical and physical characteristics may be completely different. Furthermore, their environmental impact, renewability and sources can be significantly different. A fundamental aspect to be considered in the use of alternative fuels for marine use is their flashpoint. According to the IMO International Convention for Safety of Life at Sea (“SOLAS”), “low-flashpoint fuel” means gaseous or liquid fuel having a flashpoint lower than permitted under SOLAS regulation II-2/4.2. Typically, bunkers with a flashpoint below 60˚C are deemed unsafe. The usage of marine fuels below 60˚C is already prohibited under SOLAS. Furthermore, the IMO International Code of Safety for Ships Using Gases or Other Low-flashpoint Fuels (IGF Code) provides industry standards for ships that use fuels with a flashpoint of less than 60°C. The IGF Code seeks to regulate the safety changes from the carriage and use of gas fuel, in particular liquefied natural gas (LNG) and other low-flashpoint fuels. The IGF Code sets out mandatory provisions for the arrangement, installation, control and monitoring of machinery, equipment and systems that use low-flashpoint fuels. Currently the IGF Code addresses in detail the requirements for the safety of ships using LNG as fuel. Draft Guidelines have been issued (for fuel cells, LPG and methyl/ethyl alcohol) or are in preparation (for hydrogen, ammonia, low-flashpoint fuel oils) and it is expected that – after a period of non-mandatory application to gain experience – the relevant provisions will be included as new mandatory sections of the IGF Code. Although the usage of alternative fuels is an essential part of the strategies to comply with the stringent IMO and EU requirements targeting the progressive decarbonization of maritime transport, the real focus of the new IMO regulations and draft interim guidelines is on the safety of ships using such alternative fuels, to address and manage the additional risks introduced by their flammability, toxicity and corrosivity. In parallel, IMO is also revising the regulations addressing the transport of such alternative fuels as cargo for their transport in global trading routes, e.g., the IMO Code for the Construction and Equipment of Ships Carrying Liquefied Gases in Bulk (IGC Code). These international regulations have a significant impact on the use and the costs (CAPEX and OPEX) of the alternative fuels on all kind of vessels and – more in general – waterborne transport, including inland navigation. The main characteristic of alternative fuels is that they are, in most cases, renewable, meaning that they don’t run out and can be replenished naturally over time, which contrasts with the nature of fossil fuels that have been formed from decaying plants and animals, over millions of years and thus they exist in finite resources. Alternative fuels, generally, emit fewer greenhouse gases (GHG) than their fossil counterparts, even having net zero footprint for the environment, while, in contrast, the burning of fossil-based fuels produces significant amount of greenhouse gases (GHG). That difference can be even more significant, especially if the alternative fuels have been produced from renewable green sources (‘green fuels’) and utilizing Carbon Capture and Utilization (CCU). In this aspect we must underline the fact that the usage of alternative fuels contributes also to the reduction of the air pollution in general, as the alternative fuels generally emits less, or even zero, air pollutants such as like sulphur dioxide, nitrogen oxides, and particulates. 96 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins The impact of the replacement of fusil-based fuels with alternative ones has a geopolitical angle, as most of the proposed alternative fuels can be domestically produced by most countries, resulting in reduction to imports from foreign countries and contributing to energy security. Moreover, as the alternative fuels can be created from a wide range of different sources, such as biomass, water, sunlight and even waste products, contrasting with the fossil-based ones, that derives from a single source of origin, as most of the fossil-based fuels are produced from a limited range of sources (coal, petroleum, and natural gas). This difference also contributes to the diversification of the energy sources, which also in turn contributes to the energy security. However, there are issues that must be investigated, regarding the advantages of the fossil-based fuel, as the technologies for their production and usage are in a well-developed stage, being widely deployed, while their alternative counterparts require further research and development regarding their production distribution and storage. Although the technology related to the production, transport and use of some alternative fuels is already known and applied in land-based applications and industrial fields (e.g. using ammonia, hydrogen, methanol), marine standards and “marinization” requirements are still at an infancy stage, and therefore the cost of production and infrastructure scale-up, is higher. As the technologies used for fossil-based fuels are mature and the infrastructure extensive and well developed, the cost of using fossil fuels is comparatively lower than that of alternative ones. Moreover, fossil-based fuels are, in general, characterized by high energy density, making them very efficient for transport and energy storage, while the alternative fuels have variable energy densities, and large deviations in their energy content and energy density both energy density by volume and energy density by mass (e.g., hydrogen). Compatibility with the existing engines might also be a problem for the alternative fuels, as there are fuels that can be directly used in existing engines as drop in fuels, or as blends, with minimal or even no modifications required in the engine, while others might need extensive modifications, specialized engines, or delivery systems. In contrast, as the existing technology has been developed and designed around fossil-based fuels, they are completely compatible with the existing engines and infrastructure. Considering the public perception, it is evident that the public favours alternative fuels, in general, due to their environmental benefits over the traditional fossil-based fuels. Moreover, there is strong governmental support in the development of alternative fuels and renewable energy, especially in EU, which for years has been importing fossil fuels from other countries, as the geopolitical and energy security factor becomes more critical. Furthermore, as the global climatological change becomes more and more evident, more actions will be taken in the direction of alternative fuels. As argued in (Mofor et al., 2015). the development of renewable energy solutions for shipping has been hampered by over-supply of fossil fuel-powered shipping and weak investments. The main problems that renewable technology faces in order to increase its usage in the maritime sector is essentially the lack of commercial viability and higher costs well-to-wake. At the same time, the motivation for a fuel shift is very high. There is a compelling need for all stakeholders to decarbonize maritime transport to comply with the progressively stricter emission limits from now to 2050. Therefore, in addition to slow steaming and energy saving/energy recovering on board, alternative fuels are the third–and probably the most effective way to seek regulatory compliance. This Chapter aims at a thorough survey of alternative fuels for shipping, their evaluation and comparison, and a critical analysis of their current status of use and future potential. The Chapter is organized as follows. The next section provides a taxonomy of alternative fuels and focuses on green methods for their production as well as technologies for their storage. Section 3 discusses the key alternative fuels 97 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins with a high potential for the shipping industry as replacements to fossil-based fuels. Section 4 contains a comparison of alternative fuels in terms of energy densities, emissions and requirements for storage and additional technologies for their utilization such as converters. Section 5 discusses the role of digital twins in producing and evaluating alternative fuels, as well as for improving the processes for their storage, transportation and utilization. Finally, Section 6 discusses the current state of maturity of alternative fuels and their possible future developments taking into account financial and technical considerations. PRODUCTION AND STORAGE OF ALTERNATIVE FUELS Green production methods for alternative fuels are particularly important, as the whole lifecycle of the alternative fuels must be considered, rather than simply the phase where the fuel is used (burned). Main green production techniques of alternative fuels include: • Electro fuels (E-fuels) mainly hydrogen derived using water electrolysis powered by renewable electricity sources. • Biofuels from sustainable biomass (forestry residue, agricultural waste, etc.) producing different carbon-neutral fuels such as bio-MGO (Marine Gasoil), bio-LNG and bio-methanol. • Blue fuels from fossil energy by capturing the CO2 during the production process and using carbon capture and storage (CCS) technology, producing mainly ‘blue’ ammonia and ‘blue’ hydrogen. Another green method of production includes renewable energy technologies such as photovoltaics, wind turbines and wave energy devices in offshore areas such as natural and man-made platforms, that produce green electricity, which in turn can be used to produce green fuels using different methods, for example electrolysis to produce green hydrogen. However, renewable energy can also be used directly in waterborne assets. Several potential energy sources have been proposed, including wind, solar, biofuels and wave. The adaptation of the renewable energy sources can be implemented for the existing ships or fleets through refitting and retrofitting, and for the newbuilds through optimised ship designs. Renewable energy can be used to produce green fuels, like green hydrogen via electrolysis of sea water with electricity provided by offshore or onshore wind turbines. In that direction there are numerous investments like that of Siemens Gamesa and Siemens Energy (Mallouppas & Yfantis, 2021). Furthermore, there are various examples of implementation of wind assisted propulsion (WASP), for various WASP solutions, from which, one very interesting example being that of the SEA-CARGO’s SC CONNECTOR which utilizes Flettner rotors in order to achieve 25 to 70% propulsive thrust. Fuel Cells Fuel cells (FCs) are efficient energy converting devices which use pure hydrogen or hydrogen derived from a reforming process of hydrogen-rich fuels, which can be integrated into the fuel cell power installations. Therefore, fuel cells are emerging as a promising technology for ship applications, offering a clean and efficient alternative for onboard power generation. The electricity is produced in the fuel cells, electricity through an electrochemical reaction. Among the different types of fuel cells, proton-exchange membrane fuel cells (PEM), solid oxide fuel cells (SOFCs), and molten carbonate fuel cells (MCFCs) for higher power output, are the most promising options for maritime applications based on metrics for 98 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins energy efficiency, power capacity and sensitivity to fuel impurities. Overall, the main performance metrics for applying fuel cells to maritime applications are the power capacity, the cost and the lifetime of the fuel cell stack. With MCFC, waste heat recovery systems are applied to improve the overall efficiency, possibly coupled with organic Rankine cycles due to the low-grade temperatures. Large capacity hydrogen fuel cells (200kW) are currently deployed to power industrial and transport machinery and have the capacity to provide energy for ship propulsion in suitable configurations (FCWave,2022). Currently, some cruise ships have also installed PEM fuel cells, using reformed methanol, to produce energy in the range of 300-350 kW for non-essential ship services of accommodation areas. TYPES OF ALTERNATIVE FUELS Biofuels Biofuels are produced from organic waste such as plant and animal waste (International Transport Forum, 2018). As of now, the main sources of biofuels are from plant-based sugars and oils, such as palm, soybean and rapeseed (Hsieh and C. Felby, 2017). There are 3 generations of biofuels, with the second and third generations regarded as “advanced biofuels”. The categorization of the biofuels is based on the source of carbon used. The European Biofuels Technology Platform defines first, second and third generation biofuels as follows (Mofor et al., 2015), (Mallouppas & Yfantis, 2021): 1. First Generation: “The source of carbon for the biofuel is sugar, lipid or starch directly extracted from a plant. The crop is actually or potentially considered to be in competition with food.” 2. Second Generation: “The biofuel carbon is derived from cellulose, hemicellulose, lignin, or pectin. For example, this may include agricultural, forestry wastes or residues, or purpose-grown nonfood feedstocks (e.g., Short Rotation Coppice, Energy Grasses).” 3. Third Generation: “The biofuel carbon is derived from aquatic autotrophic organisms (e.g., algae). Light, carbon dioxide and nutrients are used to produce the feedstock, “extending” the carbon resource available for biofuel production.” Advanced biofuels are a very promising and viable solution as a main energy source (Mofor et al., 2015) and they meet the requirements for Very Low Sulphur Fuel Oil (VLSFO) and for Ultra Low Sulphur Fuel Oil (ULSFO) (Hsieh & Felby, 2017). The problem with biofuels in shipping sector refers to the little experience and knowledge on handling and applying biofuels as part of the fuel supply chain, the required volumes of biofuel to fulfil the needs of the shipping sector and the concerns about the storage and oxidation stability of biofuels, as well as the blending different sources and variety of biofuels, leading to the need of further research in that sector (Hsieh and C. Felby, 2017). Another worth mentioning aspect is the fact that sustainable biofuel production is limited considering food price, natural resources (such as availability of land) and social conditions (International Transport Forum, 2018). In a SWOT analysis provided by (Hsieh & Felby, 2017), biofuels have higher prices than fossil fuels, and the situation is expected to remain for at least a medium-term horizon, but with the right policies, regulations, initiatives and technology and infrastructure improvements, there will be a healthy market for biofuels (Hsieh & Felby, 2017). In that direction, market-based measures can be used to accelerate 99 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins the adoption of biofuels in the maritime sector, as part of the decarbonization strategy of the shipping industry (Lagouvardou et al., 2020) Biogas Biogas can be either methane or hydrogen. Biogas is produced from anaerobic digestion within an enclosed environment which consists of microbes that break-down organic material (Vanek et al., 2012), after that, biogas can be further processed to remove impurities like hydrogen sulphide and moisture (Mofor et al., 2015). The produced methane can be liquified in the form of Liquified Bio-Methane (LBM) (Mofor et al., 2015), which can be used for deep decarbonization in the long-term of the shipping industry, as it is produced from renewable sources, in contrast with LNG which is favoured right now but is often considered as a transitional fuel rather than a terminal solution (Mallouppas & Yfantis, 2021). Methanol Methanol, due to the lower investment cost and the relative simplicity of handling - being liquid in normal ambient conditions may be an attractive solution as an alternative fuel compared to hydrogen and ammonia, moreover, its availability and competitive price (FCEnergy, 2018) (Doedee, 2023) makes it a promising alternative to conventional fuels. Furthermore, methanol provides a 25% drop in CO2 emissions compared to HFO coupled with reduction in SOx, NOx and PM by 99%, 60% and 95% respectively (Mallouppas & Yfantis, 2021). Regarding its drawbacks, the fact that is toxic may introduce some additional risks, but on the other side due to the fact that is plentiful, available globally, readily miscible in water, biodegradable and it can be 100% renewable, makes it a very interesting option. Another worth mentioning aspect is that life-cycle environmental footprint of bio-methanol is “greener” compared to LNG (Mallouppas & Yfantis, 2021). The environmental benefits, technology readiness and economic feasibility of methanol as marine fuel are studied in (IMO, 2021). As mentioned, and above and in a study by (FCBI Energy, 2018) methanol is plentiful, available globally and potentially 100% renewable, compliant with short/mid-term emissions reduction regulations, although not carbon-free, while the current bunkering infrastructure needs only minor modifications to handle it with relatively modest costs compared to potential alternative solutions. Liquified Natural Gas (LNG) Liquified Natural Gas (LNG) is already in use in maritime industry and is the cleanest available fuel for shipping available right now in meaningful volumes (Shell, 2020). LNG is stored in -162° Celsius and provides a 20-30% reduction of CO2 emissions compared to HFO, while having lower NOx and PM emissions and 90% drop in SOx emissions. The problem with LNG is that due to the methane slip, the environmental benefits are moderate, hence (excluding green LNG from Carbon Capture and renewable energy sources), LNG is considered by many like a transitional fuel rather than a terminal zero-carbon solution. In this point it must be mentioned that right now the only comparable in cost with HFO fuel is LNG, while other fuels such as bio-methanol and biofuel are compatible in price with HFO only when a tariff of 300 USD/ton CO2 is applied on HFO consumption (Mallouppas & Yfantis, 2021). 100 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins Hydrogen Hydrogen, on the other hand, can be used in internal combustion engines and in fuel cells, as described above. A recent study by CE Delft (Delft, 2020), has investigated the availability and costs of LBM and Liquified Synthetic Methane (LSM), also known as e-methane. In this study the LSM is assumed to be produced from CO2 and H2, where hydrogen is produced from renewable energy sources and CO2 is recycled from carbon capture (Delft, 2020). The study has shown that the future projected supply of LBM and LSM can exceed the future demand of the maritime sector, if biomass will be used to produce methane and sufficient investments are made in renewable electricity production (Delft, 2020). Furthermore, the cost of the production of LBM and LSM may not be significantly higher or even be comparable to the production costs of other lowand zero-carbon fuels (Delft, 2020). Finally, if the costs of bunkering infrastructure and ships are comparable as well, then LSM and LBM fuels may be viable candidates to achieve decarbonization in the shipping sector (Delft, 2020) (Mallouppas & Yfantis, 2021). As of 2020, the majority of hydrogen (∼95%) is produced from fossil fuels by steam reforming of natural gas, partial oxidation of methane, and coal gasification. These methods produced, so-called ‘blue hydrogen’. Other methods of hydrogen production include biomass gasification and electrolysis of water according to Wikipedia (https://en.wikipedia.org/wiki/Hydrogen_production). Hydrogen, even though in many cases it can be proved advantageous, as it produces no CO2, particulate matter (PM) or SOx, when burned, presents many challenges to be addressed mainly about the refuelling, safety, storage (embrittlement of materials) as well as the production and usage (especially in Internal Combustion Engines - ICEs). Most of the hydrogen produced today comes from fossil fuels, making it impactful for the environment. The solution might be the so-called “green” hydrogen, which, in contrast with all the other hydrogen’s types of production (Brown, Grey, Yellow, Blue, Pink), is produced solely from renewable energy sources, like solar, wind, etc., but at a higher cost than any other method of production. Another issue is that although hydrogen includes more energy per kilogram than any other proposed solutions, excluding nuclear power, it has very low volumetric energy density (Energy Transitions Commission, 2019). The answer to this problem may be liquid hydrogen (LH2), which is stored in high pressure and very low temperatures (-253 Celsius), or compressed hydrogen (CH2) at high pressure (300-700 bar), but this obviously creates some new challenges, which can technically and financially be overcome, as it was shown by a project by Kawasaki Heavy Industries, using liquified hydrogen as cargo (Kawasaki, 2010) Hydrogen is already in use to power offshore crane vessels (Sustainable-ships.org-c, 2021) and inland vessels (Sustainable-ships.org-e, 2021). Infrastructure for hydrogen production has begun development in various locations such as the green hydrogen refinery in Rotterdam (Sustainable-ships.org-d, 2021) and in BP’s Lingen Refinery in northwest Germany (Sustainable-ships.org-f. 2021), as well as in planned offshore platforms (Sustainable-ships.org-I, 2021). The European Commission estimates that €13-15 billion will be invested in electrolysers to increase hydrogen production capacity to 40 GW by 2030 (Charbonneau, 2021). The European Maritime safety Agency (EMSA) commissioned and published in 2023 a very comprehensive study on the “Potential of Hydrogen as Fuel for Shipping”, which details the technical issues, regulatory frameworks, and state of play for application of hydrogen as a fuel. 107 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins GREEN FUELS AND DIGITAL TWINS This section discusses the potential for utilizing digital twins (DTs) for performance analysis of alternative green fuels, as well as for improvements in their manufacturing and transportation (i.e. in terms of required infrastructure), which are pre-requisites for their wider adoption by the maritime industry. In the context of fuels, digital twins can provide detailed, high-fidelity models of the fuel (e.g. its chemical composition and properties), as well as of all processes related to its production, transportation, storage, and utilization. Such models are connected to actual data captured from the fuel through chemical analyses, as well from its direct environment, via sensors. Digital twins allow high fidelity simulations and exploration of different ‘what-if; scenarios regarding the fuel, its behaviour and properties, allowing more efficient and environmentally safe scenarios and methods for its utilization. According to (Lamanga et al., 2021), regarding carbon-neutral fuels, the aim is the development of digital twins of the power generation plant, containment system and fuel supply system. The deployment of these DTs will enable simulations of the vessel responses in actual voyages with regards to engine response, consumptions (daily rate and total), boil-off rates and power demands for the fuel supply. Based on this a systematic variation/optimization of variables including but not limited to containment system volume, pressure, boil off rate, consumptions and application/selection of handling machinery such as compressors, sub-coolers, shaft generators etc., and auxiliaries such as ventilation systems, automation systems, safety/monitoring/control systems etc. can be conducted. After the generation of an adequate number of design variants each of them can be assessed by simulation and a multi-objective decision making, and design selection can be conducted with the use of utility functions. In addition, DTs can be used to validate the control and safety schemes, and regulatory compliance of green fuel production techniques such as electrolysis and their integration with industrial plants. (Charbonneau, 2021). Finally, DTs can be used in research and development of advanced green fuel power conversion technologies such as fuel cells and the storage of them. Currently DTs are used in surrogate modelling methods that combine a state-of-the-art threedimensional physical model and a data-driven model by different researchers. The result obtained in (Lamanga et al., 2021), concerning the DT of a proton exchange membrane fuel cell, can predict its outputs with a root-mean-square error from 3.88% to 24.80%. Similarly, in (Lamanga et al., 2021), a DT model was made for a solid oxide fuel cell (SOFC), starting from a 1 kW SOFC data were used to regress the parameters to scale-up the model to 25 kW. The final obtained DT was validated by steady-state data and applied to on-site operation prediction with very high accuracy. Table 5. Methanol blends compliance Fuel name % renewable in blend Compliance year limit (as of Feb. 2023 thresholds) ng_meoh 0 Never Compliant meoh_25e 0.25 2035 meoh_50e 0.5 2042 meoh_75e 0.75 2047 emeoh 1 >2060 108 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins CONCLUSION AND FUTURE OUTLOOK State of Maturity of Green Shipping Fuels Fossil-based fuels have been essential for industrial transportation, but in order to have a sustainable transport sector and to counteract their detrimental effects, we must move to the greener options that the alternative fuels provide. In order to offset the initial high cost of investment and the high cost of the alternative fuels right now, international organizations and states are developing legislations and rules that benefits the usage of them, but there will be comparative small progress if no market-based measures won’t be taken. One very important question cited in (Lindstad, 2022) regards the impact, to the total global GHG emissions, of decarbonizing the shipping sector, using renewable energy sources, compared to decarbonizing other sectors, namely the electricity production. As it was argued in (Lindstad, 2022 the essential for the decarbonization of shipping resources, if used for the production of electricity can offer 7 to 10 times larger GHG reductions compared to the GHG reduction achieved from decarbonizing shipping. Under this prism, we can understand there are more urgent needs of sector decarbonization than shipping. Moreover, it must be highlighted that the GHG footprint of biofuels (or e-fuels) depends largely on the raw materials that have been used in their production. For example, biogas made from waste has close to zero GHG emissions, while biodiesel made from palm oil can potentially have more than double the emissions compared to the fossil fuels (Lindstad, 2022). An aspect worth considering is that due to the dependency of the production of energy from renewable sources on the climate, there is an inherited volatility of the production of energy, which is heavily affected by the climate change, which in our days, is more evident than ever (Lindstad, 2022). It is evident that there is a long way to go, both in the technological/research point of view and from the economic/ logistics point of view to have a vial and efficient usage of the renewable energy sources in order to decarbonize not only the shipping industry in particular, but the greater issue of energy production and dependency in general, in a holistic approach. Another consideration is with regards to the substantial investment in additional equipment and storage space that must be made to accommodate green fuels such as hydrogen and ammonia, especially in the case of hydrogen (Energy Transitions Commission, 2019) Solutions have been proposed such as using large floating solar fields in key areas around the world close to existing shipping hubs and ports. These could provide independent energy for seasonal storage and can provide synthetic, circular fuel to existing infrastructure (shippinh.org-j) However, the feasibility of such proposed solutions needs to be investigated further. Digital twin technology can support this, by analysing “what if” design scenarios, validate optimized control strategies, enable cost-effective regulatory compliance, validate staff operating procedures, optimize preventive maintenance practices, and upskill operations staff. (Charbonneau, 2021). Recommendations for Fuel Selection Based on the previously discussed factors: energy densities, safety and toxicity aspects, availability, logistic requirements, volumes footprint, GHG footprints and conversion systems, as well as taxation policies vs. incentives, it becomes clear that there might not a single solution meeting the maritime decarbonisation targets, even if some of them, namely methane, ammonia and methanol seem to be interesting options. 109 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins Indeed, different strategies might be used for different ship types and related operational requirements, taking inro account: • Fuel characteristics (volume, production pathways, blending, ….) • Conversion system types • Optional technologies and systems (Wind Assistance, Cold Ironing, …) And also, the ship’s operators preferences and constraints, i.e.: • Loss of payload space and ship operational characteristics (e.g. due to changed • bunkering requirements, operational range, geographic location, etc.) • GHG Regulatory limitations and related penalties, • CAPEX, OPEX and financial performances. Time to commercial availability of greener fuels will therefore depend largely on level of R&D efforts, and even more importantly on the level and speed of which environmental regulations/policies are implemented and incentive schemes are developed. (DNV, 2022) REFERENCES CE Delft. (2020). Availability and Costs of Liquefied Bio-And Synthetic Methane: The Maritime Shipping Perspective. CE Delft, Delft, Charbonneau Loic. Digital twin technology transforms hydrogen production. Emerson. https://www.emersonautomationexperts.com/2021/sustainability/digital-twintechnology-greenhydrogen-production/ Vincent, D. (2023). The State of Methanol as Marine Fuel. Methanol Decarbonizer Emissions, March 23. DNV. (2022). Energy Transition Outlook – A global and regional forecast to 2050. DNV. https://www. dnv.com/energy-transition-outlook/index.htm Energy Transitions Commission. (2019). Mission Possible - Reaching Net-Zero Carbon Emissions From HarderTo-Abate Sectors By Mid-Century - Sectoral Focus Shipping. ETC. https://www.energytransitions.org/publications/mission-possible-sectoral-focus-shipping/#downloadform. (Accessed 6 September 2022). European Commission. European Commission, “Regulation of the European Parliament and of the Council on the use of renewable and low-carbon fuels in maritime transport and amending Directive 2009/16/EC (COM(2021) 562 final 2021/0210 (COD) Proposal),” 2021. Falko, U. (2021). Potential and risks of hydrogen-based e-fuels in climate change mitigation. Nature Climate Change, 11(5), pages384–393. doi:10.1038/s41558-021-01032-7 FCBI Energy. (2018). Methanol as a marine fuel report for the Methanol Institute. FCBI Energy.https:// www.methanol.org/wp-content/uploads/2018/03/FCBI-Methanol-Marine-Fuel-ReportFinal-English.pdf 110 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins FCWave (2022). 200 kW Hydrogen Fuel Cell. [Online] https://www.sustainableships.org/stories/2022/ ballard-fcwave-200kw Hsieh, C.-C., & Felby, C. (2017). Biofuels for the marine shipping sector. IEA Bioenergy. IMO - Methanol as marine fuel. (2021). Green Voyage. https://greenvoyage2050.imo.org/wpcontent/ uploads/2021/01/METHANOL-AS-MARINE-FUEL-ENVIRONMENTAL-BENEFITS-TECHNOLOGYREADINESS-AND-ECONOMIC-FEASIBLITY.pdf International Transport Forum. (2018). Decarbonising Maritime Transport. Pathways to Zero-Carbon Shipping by 2035. ITF. Kawasaki Heavy Industries, Ltd. (2018). Hydrogen Road. Kawasaki Heavy Industries, Ltd. Kim, K., Roh, G., Kim, W., & Chun, K. (2020). A preliminary study on an alternative ship propulsion system fueled by ammonia: Environmental and economic assessments. Journal of Marine Science and Engineering, 8(3), 183. doi:10.3390/jmse8030183 Lagouvardou, S., Psaraftis, H. N., & Zis, T. (2020). A Literature Survey on Market-Based Measures for the Decarbonization of Shipping. Sustainability (Basel), 12(10), 3953. doi:10.3390/su12103953 Lindstad E. (2022). Zero carbon E-Fuels. Marine technology. Liu, X., Elgowainy, A., & Wang, M. (2020). Life cycle energy use and greenhouse gas emissions of ammonia production from renewable resources and industrial by-products. Green Chemistry, 22(17), 5751–5761. doi:10.1039/D0GC02301A Mallouppas, G., & Yfantis, E. A. (2021). Decarbonization in Shipping Industry: A Review of Research, Technology Development, and Innovation Proposals. Journal of Marine Science and Engineering, 9(4), 415. doi:10.3390/jmse9040415 Lamagna, M., Groppi, D., Nezhad, M. M., & Piras, G. (2021, November). A comprehensive review on digital twins for smart energy management system. International Journal of Energy Production and Management. Mofor, L., Nuttall, P., & Newell, A. (2015). Renewable Energy Options for Shipping. IRENA. Perčić, M., Vladimir, N., Jovanović, I., & Koričan, M. (2021). Application of fuel cells with zero-carbon fuels in short-sea shipping. Applied Energy. Shell. (2020). Decarbonising Shipping: Setting Shell’s Course. Shell International. Siemens Energy. (n.d.) Green Hydrogen Production. https://www.siemens-energy.com/global/en/priorities/future-technologies/hydrogen.html Stolz, B., Held, M., Georges, G., & Boulouchos, K. (2022, January). Techno-economic analysis of renewable fuels for ships carrying bulk cargo in Europe. Nature Energy, 7(2), 203–212. doi:10.1038/ s41560-021-00957-9 Sustainable-ships.org-a. (n.d.). Frequently asked questions. Sustainable Ships. https://www.sustainableships.org/key-insights/faqs-biofuel 111 Shipping Green Fuel Strategies, Benchmarking Supported Digital Twins Sustainable-ships.org-b. (2023). The State of Methanol as Marine Fuel. Sustainable Ships. https://www. sustainable-ships.org/stories/2023/methanol-marine-fuel Sustainable-ships.org-c. (2021). Hydrogen Powered Propulsion for an Offshore Crane Vessel. Sustainable Ships. https://www.sustainable-ships.org/stories/2021/hydrogen-sleipnir Sustainable-ships.org-d. (2021). How to build a green hydrogen refinery for the maritime industry in Rotterdam. Sustainable Ships. https://www.sustainable-ships.org/stories/2021/green-hydrogenrefineryrotterdam Sustainable-ships.org-e. (2021) Which will be the First Hydrogen-Powered Inland Vessel in Rotterdam? Sustainable Ships. https://www.sustainable-ships.org/stories/2021/inland-vessel-hydroge Sustainable-ships.org-f. (2021). BP and Ørsted launch green hydrogen project at German oil refinery. Sustainable Ships. https://www.sustainable-ships.org/stories/2021/bp-orsted-hydrogen-partnership Sustainable-ships.org-g. (2021). What Is Green Hydrogen And Will It Power The Future? Sustainable Ships. https://www.sustainable-ships.org/stories/2021/cnbc-green-hydrogen Sustainable-ships.org-h. (2021). BP Lingen on green hydrogen. Sustainable Ships. https://www.sustainableships.org/stories/2021/bp-lingen-green-hydrogen Sustainable-ships.org-I. (2020). Poseidon Pilot - Offshore Hydrogen. Sustainable Ships. https://www. sustainable-ships.org/stories/2020/poseidon-pilot-offshore-hydrogen Sustainable-ships.org-j. (2019). The Case for Floating Solar. Sustainable Ships. https://www.sustainableships.org/stories/2019/8/20/the-case-for-floating-solar Transport & Environment. (n.d.). Ships. Transport & Environment. https://www.transportenvironment. org/challenges/ships/ van Biert, L., Godjevac, M., Visser, K., & Aravind, P. V. (2016). A review of fuel cell systems for maritime applications. Journal of Power Sources, 327, 345–364. doi:10.1016/j.jpowsour.2016.07.007 Vanek, F. M., Albright, L. D., & Angenent, L. T. (2012). Energy Systems Engineering: Evaluation and Implementation (2nd ed.). McGraw Hill. 112 Copyright © 2024, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. Chapter 6 DOI: 10.4018/978-1-6684-9848-4.ch006 ABSTRACT The chapter explains techniques and approaches to optimize a ship’s voyage in terms of environmental and business parameters, utilizing the digital twin (DT) concept. It demonstrates how voyage planning and navigation management, in general, is enhanced by taking into account vessel state in real time as reflected and analyzed by the digital twin ecosystem. The theoretical backbone of voyage planning entails a multitude of state-of-the-art processes from trajectory mining and path finding algorithms to multi constraining optimization by including a variety of parameters to the initial problem, such as weather avoidance, bunkering, Just in Time (JIT) arrival, predictive maintenance, as well as inventory management and charter party compliance. In this chapter, the authors showcase pertinent literature regarding navigation management as well as how the envisaged DT platform can redesign voyage planning incorporating all the aforementioned parameters in a holistic digital replica of the en-route vessel, eventually proposing mitigation solutions to improve operational efficiency in real-time, through simulation, reasoning, and analysis. TRADITIONAL METHODS AND CHALLENGES IN VOYAGE PLANNING Voyage planning involves the holistic enhancement and optimization of a vessel voyage by considering various factors such as weather conditions, fuel efficiency, vessel performance, cargo considerations, and safety protocols. It encompasses a broad perspective and aims to create an overall efficient and effective voyage experience. A subset of voyage planning is route optimization that aims to determine the path that minimizes travel time, reduces fuel consumption, and enhances overall operational efficiency, Enhanced and Holistic Voyage Planning Using Digital Twins Dimitris Kaklis DANAOS, Greece Antonis Antonopoulos Konnecta, Greece This chapter published as an Open Access Chapter distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/ licenses/by/4.0/) which permits unrestricted use, distribution, and production in any medium, provided the author of the original work and original publication source are properly credited. 113 Enhanced and Holistic Voyage Planning Using Digital Twins taking into account factors like weather patterns, currents, wind conditions, traffic congestion, and other navigational challenges. In recent years, there have been several state-of-the-art solutions for route optimization and weather routing in the maritime sector. One popular approach to routing optimization is to use dynamic programming that takes into account various parameters, such as the ship’s speed, the sea conditions, and the distance to the destination. These models can be used to generate optimized routes that minimize fuel consumption while ensuring that the ship arrives at its destination on time. One example of such a model is the BunkerOpt system developed by DNV GL, which uses a mathematical model to calculate the optimal speed and route for a ship based on real-time data. Another approach to routing optimization is to use machine learning algorithms to predict weather patterns and optimize routing accordingly. For example, the Weather Intelligence for Shipping (WIS) system developed by StormGeo uses machine learning to predict weather conditions up to ten days in advance, allowing ships to adjust their routes to avoid adverse weather conditions and reduce fuel consumption. Pertinent literature regarding smart and efficient transportation in the industry (supply chain optimization, autonomous vehicle, logistics management, voyage planning) as well as regarding societal frameworks (traffic management, public transportation networks, suburban mobility), concerns a variety of multi-constraint optimization methods varying from Genetic Algorithms (Xing Wei and and Huang, 2021), Simulated annealing (Fermani et al.,2021) and Particle Swarm Optimization (Chondrodima et al., 2020) to AI integrated decision making using Reinforcement Learning (Kaklis et al., 2022, Kaklis et al., 2023, Bai et al., 2019) or Natural Language Processing (NLP) based approaches (Garg et al., 2021). The aforementioned practices and methodologies attempt to exploit the proliferation of IoT devices and state-of-the-art communication networks (5G) to build self-contained Information Hubs and provide a sustainable, safer and cost-effective transportation. In the maritime sector, the optimization criteria adopted in the context of the ship routing problem deal with the minimization of voyage time, fuel consumption (or Fuel Oil Consumption, FOC) and voyage risk. The approaches, which have appeared so far in the literature, can be classified into three broader categories: • Vessel-based optimization, which aims in optimizing a given route with respect to vessel characteristics, e.g., vessel speed, main-engine rotational speed, draft, trim and sea-keeping behavior: roll, heave and pitch motions, (Roh and Lee, 2018); • Environmental-based optimization, which aims in optimizing a given route by taking into account environmental conditions, e.g., wind (speed, direction), wave (height, frequency, direction), currents. (Kim and Kim, 2017); • Holistic optimization that combine the two previous approaches in a common context. (Vettor and Soares, 2015; Varelas et al., 2013). • Analytical approaches trying to tackle the problem with the use of exact (NP-complete) and/or heuristic algorithms like label-setting algorithms, non-linear integer programming, or simulated annealing (Shin et al., 2020). In order to incorporate more constraints, several methods split a vessel’s voyage into areas of critical interest, involving for example zones of extreme weather conditions, emission control areas (ECAs, SECAs), high-risk zones (piracy), etc. Then, they seek for Pareto optimal solutions from a set of routes 114 Enhanced and Holistic Voyage Planning Using Digital Twins that are optimal in terms of Expected Time of Arrival (ETA), FOC, and safety, or they use Genetic Algorithms (Kim et al., 2017) in order to find the best route, as a composition of optimal route segments. Methods like PSO (Particle Swarm Optimization) (Zhao et al., 2020) are also employed in order to solve the multi-constraint, non-linear optimization problem of optimal route planning. The techniques employed in the literature for estimating FOC based on vessel characteristics and/or environmental conditions can be grouped into the following categories: • Data-oriented approaches that combine vessel-trajectory data, gathered from sensors, satellites (AIS data), or Noon Reports, with Machine and Deep-Learning algorithms. These techniques range from simple Regression analysis like Support Vector Regression, Lasso Regression, and Polynomial Regression to ensemble non-parametric schemes like Random Forest (RF) regression, Decision Trees, or AdaBoost. Some studies have also experimented with baseline sequential Artificial Neural Networks (ANN) by tuning a number of hyperparameters (learning rate, number of neurons, number of layers, activation function). (Jeon et al., 2018, Gkerekos et al., 2019). • Approaches where machine learning (ML) methods (also known as black-box models - BBM), are combined with theoretical models (also known as white-box models - WBM), such as the equations of motion of a freely floating body moving with constant forward speed, in order to increase the prediction accuracy. The proposed models are known as gray-box models (GBM) (Coraddu et al., 2017, Kaklis et al., 2019). ANNs have been at the center of attention lately in many research areas. As far as vessel FOC is concerned, not many studies utilize the computational power of ANNs to approximate FOC mainly due to the problem of missing historical data. The studies found in pertinent literature dealing with FOC estimation from a deep learning perspective are presented briefly below. Some studies experiment with baseline sequential ANNs by applying a dropout in the weights in order to achieve better generalization error (Gkerekos et al., 2020) or by tuning a number of hyperparameters (learning rate, number of neurons, number of layers, activation function) utilizing brute force methods like randomized grid search (Papandreou et al., 2020, Jeon et al., 2018). In (Yongjie et al., 2020) a Recurrent NN is employed in order to estimate FOC but without further research as far as the architecture, or the generalization capabilities of the neural proposed. The majority of the approaches found in literature regarding Operational Optimization and Navigation Management in the maritime sector, concern the implementation of standalone services in the sense that they are employing isolated information silos that lack the support mechanisms and enhancement of a centralized Information Hub that exploits the upsurge of IoT and Industry 4.0 advancements, to train, validate and update these services in real-time. Furthermore, they are usually tested on a single vessel and therefore lack the generalization capabilities of models evaluated in a variety of ships that are able to adjust and adapt to the underlying function that describes the relationship between FOC and each specific vessel, continuously, by exploiting the vast amount of data collected by IoT installations. Frameworks and technological advancements regarding the continuous monitoring of the vessel are inextricably linked with the emerging concept of the so-called Digital Twin in the shipping industry, as they employ a digital replica of the en-route vessel that is able to simulate-project and validate in real time the majority of the operational procedures. In recent years, there has been a growing interest in using Digital Twins to optimize routing in the maritime sector. Digital twins serve as virtual replicas of physical systems, offering a dynamic platform 115 Enhanced and Holistic Voyage Planning Using Digital Twins to simulate, analyze, and control real-world conditions. Unlike traditional methods that depend on historical data or pre-set conditions, digital twins make use of real-time analytics. This live data feed can include everything from a vessel’s hull condition and load status to the available fuels and engine performance. The implication is a quantum leap in the level of granularity and customization that voyage planning algorithms can achieve. By embracing this real-time, vessel-specific approach, maritime operators can leverage sophisticated algorithms designed to interpret and act upon a cascade of live data points. As a result, voyage planning becomes a dynamic process, continually updated to reflect the vessel’s current actual condition. This increased level of detail not only allows for more precise route optimization but also leads to safer and more efficient voyages. Whether accounting for the wear and tear on a hull over a journey or optimizing fuel consumption based on real-time metrics, digital twins provide a holistic and up-to-the-minute view that is revolutionizing voyage planning. In the table below we summarize the approaches found in pertinent literature regarding Navigation Management and Operational Optimization and provide the reader with a comprehensive-consolidated breakdown analysis based on the specific methodology adopted as well as input data utilized on each category. Towards this direction, a novel approach that aims to extend a Digital Twin framework with a Voyage Planning module in the context of a multimodal-adaptive Digital Twin ecosystem will be proposed. This extension enables stakeholders to simulate diverse scenarios and optimize routes using real-time data, considering factors like weather, traffic, fuel consumption, hull conditions, and commercial considerations. Through the integration of various data sources and the application of machine learning algorithms to forecast future conditions, stakeholders can dynamically optimize routes, leading to cost reduction, enhanced safety, and improved efficiency. Section 2 outlines the proposed perspectives to enhance voyage planning beyond route optimization and weather routing, leveraging the opportunities provided by the digital twin platform. Section 3 provides a brief overview of the digital twin framework proposed for accommodating the voyage planning application. Table 1. Navigation management and operational optimization approaches in the literature Category Sub-Categories Approach/Methodology Input Data Example Models or Systems Smart & Efficient Transportation (Beyond Maritime) Multi-constraint optimization methods: Genetic Algorithms, Simulated Annealing, Particle Swarm Optimization, AI integrated decision making, NLP. Real-time traffic data through various sensor installments on EDGE Xing Wei and Huang, 2021 - Fermani et al., 2021 - Chondrodima et al., 2020 - Kaklis et al., 2022, 2023 - Bai et al., 2019 - Garg et al., 2021 Ship Routing Optimization Vessel-based optimization Multi-constraint optimization methods Weather data/ Operational Data/ Charter Party contracts Roh and Lee, 2018 - Kim and Kim, 2017 - Vettor and Soares, 2015 - Varelas et al., 2013 Environmentalbased optimization Dynamic Programming Holistic optimization RL, Dynamic Programming 116 Enhanced and Holistic Voyage Planning Using Digital Twins Section 4, we delve into the methodologies for estimating fuel oil consumption and weather routing, both fundamental components of any voyage planning tool. We explore how these methodologies can be extended to leverage the benefits of the digital twin. Section 5 offers considerations for potential advancements that could enhance the value of the proposed solution, as well as acknowledging certain limitations. The chapter concludes Section 6 with a summary of voyage planning using digital twins. INTEGRATION OF VESSEL OPERATIONAL AND COMMERCIAL ASPECTS IN VOYAGE PLANNING In the traditional paradigm of voyage planning, strategies often rely on static data models and broad generalizations. These generalizations, while useful for general navigational purposes, fall short of accounting for the dynamic and ever-changing conditions of individual vessels. In this section, we propose specific vessel operational aspects that hold value for integration into the digital twin, aimed at enhancing voyage planning. It’s notable that some aspects discussed are innovative for voyage planning, and we’ll demonstrate how this innovation is further enhanced by the application of digital twins. Trim Optimization Trim is defined as the draft at the stern (or aft of the ship), minus the draft at the bow (or forward). Trim optimization is one of the approaches considered by the industry to improve the energy efficiency of ships, having a potential in both reducing operational costs and to decrease the emissions of the ship. Trim optimization is the selection of trim with the goal of fuel consumption reduction, by ballast water management and load distribution, which can be done without significant changes to the ship structure. The application has been investigated by Gao (2019) and Islam (2019). The MEPC (2008) has estimated that optimizing a vessel’s trim and draft can result in fuel consumption reductions ranging from 0.5% to 3% for most vessel categories. In the case of ships operating with partial loads, these savings can soar to as high as 5%. Coraddu (2017), in their research, have even demonstrated the potential for surpassing a 2% improvement in fuel consumption for handymax chemical tankers. A case study conducted by DNV-GL in 2013, which assessed the effectiveness of the commercial optimization tool known as the ECO Assistant, revealed impressive fuel savings ranging from 2% to an astonishing 14% across various draft and speed combinations for handymax bulk carriers. Furthermore, research by Yuan (2018) indicates that trim optimization for Very Large Crude Carriers (VLCCs) can lead to a notable 1.8% reduction in fuel consumption. The work of Du (2019), revealed that trim optimization has the potential to save between 5% and 6% of bunker fuel for 9000 TEU container ships. Similarly, Gao (2019) reported significant bunker fuel savings of 3% to 7% for Pure Car and Truck Carriers (PCTCs). In conclusion, it is evident that the impact of trim optimization on reducing fuel consumption can be substantial, with the extent of savings contingent upon factors such as the vessel’s type, operational profile, and maneuverability in achieving the desired trim. Typically, the optimum trim is often ascertained through reference trim tables. These tables are derived either from model-scale towing experiments or, in certain instances, from computational fluid dynamics (CFD) simulations. In the past years, alternatives to trim tables in the field of trim optimization have emerged. A range of commercial trim optimization solutions have been introduced to the market, 123 Enhanced and Holistic Voyage Planning Using Digital Twins Simulation and Modelling Indicatively, simulation is applied for the following purposes: • Forecast future conditions: Simulation is used to predict and model the future behavior of the system or process based on current conditions and known parameters. • Estimate parameters required for generating the digital twin but not directly available: Sometimes, certain parameters essential for building the digital twin may not be directly obtainable from data. Simulation helps estimate these parameters through modeling and analysis. • Estimate the effect of a change in performance: Simulation allows for testing different scenarios and changes in the system to understand their potential impact on performance and outcomes. Assist in the acquired data validation or event detection (failure) by detecting patterns that deviate from the simulation results: The digital twin can aid in data validation by identifying discrepancies between real-world data and simulated results. These deviations can indicate data quality issues or reveal anomalies that require investigation. On the other side the same approach can be utilized for the detection of deterioration events assisting in their early detection. The modeling framework for the digital twin system must have the following main characteristics: • Support for single moment simulation, time steps, and events based: The framework should be capable of conducting simulations at specific time points, discrete time intervals, or in response to specific events or triggers. • Support for dockerized code, Python, Java, FMI, R: The framework should be versatile and support different programming languages and technologies, allowing flexibility in implementing and integrating simulation models. • Integration with data-driven model instantiation and versioning systems (such as MLflow): Seamless integration with data-driven models and version control systems enhances reproducibility, transparency, and management of different model versions. • Description of model source (e.g. binary) and versioning: The framework should provide mechanisms for storing and managing simulation models, including version control to track model changes over time. An important aspect of modeling is its utilization for optimization purposes. Optimization is a crucial requirement for DT systems, and the following approaches are considered: • Repetitive procedure for determining the best solution based on predefined scenarios and variable ranges: This approach involves iteratively testing different combinations of variables within specified ranges to identify the optimal solution. • Similar approach based on Monte Carlo principles instead of using predefined scenarios: In this approach, random sampling is used to explore a wide range of possible scenarios, allowing for a more comprehensive optimization analysis. • Utilization of genetic algorithms, such as NSGA II: Genetic algorithms mimic the process of natural selection to iteratively evolve and refine potential solutions, making them well-suited for multi-objective optimization problems. 124 Enhanced and Holistic Voyage Planning Using Digital Twins By employing these optimization approaches, the digital twin system can identify optimal configurations, settings, or decisions that lead to improved performance, efficiency, or other desired outcomes for the modeled system or process. System Monitoring and Diagnostics A Digital Twin (DT) system requires robust functionality for self-monitoring and diagnostics to ensure its own health and performance. The system should continuously monitor critical metrics, such as resource utilization, response times, and event logs. Real-time performance monitoring, resource tracking, and health checks are essential for identifying anomalies and potential issues within the system. The DT system should generate alarms and alerts for abnormal conditions and include an auto-recovery mechanism to handle failures autonomously. It must offer diagnostic tools and historical performance data for troubleshooting and trend analysis. Scalability and security monitoring are crucial to ensure the system can handle increasing workloads and maintain data integrity. Integration with IT operations tools facilitates centralized management and analysis. By implementing these functionalities, the DT system can maintain its reliability, optimize performance, and provide a stable foundation for effective digital twin operations in diverse scenarios. Compliance and Standards According to the degree a DT interacts with the physical vessel, the corresponding shipping industry standards, elicited by the IMO, the Institute of Electrical and Electronics Engineer (IEEE) and rules of Classification Societies, should be adopted. The lowest level of interaction is observed in the case of simple data acquisition, while at the other end of the spectrum, there are aspects related to autonomy, which significantly increase the requirements related to compliance to regulations. The Digital Twin as a Dataspace Actor As per Nagel (2021), dataspace can be defined as “A data ecosystem, specified by a sector or application, whereby decentralized infrastructure enables trustworthy data sharing with commonly agreed capabilities”. Thus, a dataspace is produced by an ecosystem of actors that interact through the sharing of data. A Shipping Dataspace provides the data infrastructure, specifically data connectors, for creating, managing and interacting with actors, such as weather data providers, fuel availability and pricing data providers, route and port congestion data providers, commercial brokers, and supplies availability data providers. Integration of DT in the dataspace can assist in Information exchange between them, in a uniform and secure way, providing them with collective knowledge. Another aspect cohering to the DT functionality and the dataspace is model sharing and co-simulation on an open or commercial basis. Furthermore development of applications interacting with DTs is enhanced by the standardization that the dataspace offers. Setting up a dataspace with weather data providers, fuel availability and pricing data providers, route and port congestion data providers, commercial brokers, and supplies availability data providers provides seamless integration of these sources of information. Dataspace components provide the appropriate infrastructure to collect, store, and analyze data from various sources in real-time. It uses a distributed architecture that enables the processing of large 125 Enhanced and Holistic Voyage Planning Using Digital Twins volumes of data while ensuring scalability and reliability. Dataspace provides a variety of processing capabilities, such as filtering, aggregation, and data augmentation, that can be used to further optimize the Routing Optimization module. THE APPLICATION OF THE DIGITAL TWIN FRAMEWORK IN VOYAGE PLANNING The following paragraphs focus on the realization of the specific case of FOC estimation and Routing Optimization, by consolidating the aforementioned components of the broader DT4GS frame, towards a holistic Operational Optimization Digital Twin suite that aims to improve voyage efficiency and environmental compliance. Reference Implementation In the present section is described a DT implementation conducted within the scope of the DT4GS research project, which is in progress at the time that this chapter is authored. In Figure 2 is provided the implementation of the framework in terms of building blocks. A messaging system is engaged to distribute data, events, and triggers to the intended peripheral components. Storage, permanent and temporary, for both configuration and data, is achieved by the combination of a time series database, a nosql database, a knowledge graph for metadata storage and a cache. The ingestion is achieved using internal connectors for data originating from the vessel, external connectors for external data sources such as CRM systems or the internet (sources not possible to be integrated into the dataspace). Data is also exchanged with the dataspace via an IDSA compliant connector. Peripheral components exist also for the following purposes: • Web app backend • Data catalog/lineage • Platform monitoring tools • Processing units, assisting ingestion, producing composite variables and simple calculations Task Scheduler • Model Execution Engine Fuel Oil Consumption Estimation Methodology Feature Selection In order to unveil the relationships between the independent variables as well as their importance and role in estimating FOC, we conduct an initial exploratory analysis with Random Forest regression as the feature ranking algorithm. Then calculate the correlations between the most important features and conclude to an ideal feature set that consists of independent variables that will be utilized accordingly in the context of FOC approximation. 126 Enhanced and Holistic Voyage Planning Using Digital Twins Decision Trees (DT) is a popular classification or regression algorithm that takes into account the importance of features. More specifically, the feature importance defines the order in which features are selected for splitting the initial set of samples to subsets, from the tree root to the leaves. It is defined by the decrease in (tree) node impurity, which is weighted by the node probability. This probability is the number of samples that reach the node, divided by the total number of samples. Higher decreases in impurity denote more important features. Assuming only two child nodes (left, right) for each node, the node importance is given by the following equation: nij= wjCj – wleft(j) – wleft(j)Cright(j) where nij is the importance of node j for feature i, wj is the weighted number of samples reaching node j and Cj is the impurity of node j. Impurity is measured using Gini Index or Entropy. The Random Forest (RF) algorithm extends the concept of Decision Trees, for high-dimensional data, by constructing many individual decision trees during training, using each time a different random subset of the initial set of features. It then collectively examines the predictions of trees in order to make the final prediction. Respectively, RF can be used to evaluate the importance of each feature across all the trees and provide a more comprehensive ranking of feature importance. In Table 2 we depict the experimental results from conducting regression analysis utilizing RF regression in order to rank the importance of the aforementioned features in estimating FOC. Besides selecting the most important (i.e. informative) features, we also aim to avoid selecting highly correlated ones. For this purpose, we utilize the Spearman’s Rank Correlation (SRC) coefficient. The Spearman’s rank-order correlation is the non-parametric equivalent of the Pearson product-moment correlation (ρ) and assesses the strength and direction of the monotonic relationship between two ranked variables R(Xi), R(Yi) using covariance and standard deviation σ, and is calculated as follows: 𝜌R(X),R(Y)= cov(R(X), R(Y)) / 𝜎R(X)𝜎R(Y) Figure 2. Overview of an implementation 127 Enhanced and Holistic Voyage Planning Using Digital Twins Assembling the ranking of features depicted in Table 2 and the correlation coefficients calculated, depicted in Figure 3 using Algorithm 1, we conclude with a subset of the initial feature set that combines feature importance and independence. Table 2. Feature ranking using RF Ranking Feature Importance 1 STW 0.94 2 WS 0.13 3 DRAFT 0.011 4 VSLH0.005 5 COMBH 0.0058 6 SWH 0.0054 7 CS 0.004 8 WAVEH0.0039 9 SWP 0.0036 10 COMBD 0.0032 11 SWD 0.0028 128 Enhanced and Holistic Voyage Planning Using Digital Twins Data Cleaning Raw data, collected from the sensors of the vessel, are in time-series (minutely) form and tend to be ``noisy’’ (high variance, high standard deviation from the mean) and in some cases even erroneous. In order to remove noise, we employed a fit\filter technique that effectively ``cleaned’’ the data but at the same time kept the bulk of information needed for training robust predictive models. Data filtering was implemented in two stages. First, assuming that the dataset follows a normal - like distribution, we keep the data points that lie within the 99\% confidence interval around the mean. Then we apply an appropriately designed Decision Tree based algorithm in order to further cancel the noise in FOC target distribution caused by the flowmeter sensor on the vessel. Then, we proceed to transform our dataset into 15-min rolling window averages in order to further smooth out any spikes and outliers that occur in the feature set from sensor installments. Note that the use of rolling window averages is consistent with the use of the FOC prediction model within a WR algorithm, in which decisions are based upon average values of FOC and not momentary consumption. The raw data of the vessel’s speed and corresponding FOC collected from the sensors, versus the mean values per speed range (+/-0.25 V$) and the 15 min rolling window averages are depicted in Figure 4. Red circles are indicative of the number of observations found for a particular range of speed. Model Implementation The dynamic estimation of FOC based on vessel state and environmental conditions can be examined as a multivariate time-series prediction problem that takes into account the actual values as well as their recent history, and captures the information hidden in the values’ evolution over time. Based on the superiority of Long Short-Term Memory Neural Network (LSTM) models over traditional time-series prediction methods (e.g., ARIMA) as suggested by Siami (2014), LTSMs are chosen as the basis of our solution. Figure 3. Spearman correlation heatmap 129 Enhanced and Holistic Voyage Planning Using Digital Twins The initial feature set, collected by sensor installments on-board the vessel, comprises the vessel speed through water, draft and heading and some basic weather features, mapped from external services (i.e. NOOA), such as wind speed and direction. The sampling rate of the sensor based operational data corresponds to minutely measurements. In order to take maximum advantage of this feature set, we employ a LSTM architecture, using a pre-training step that extracts information from the original features, using spline-based regression (Friedman J. 1999). In what follows, we describe how LSTM is used for FOC estimation and detail the proposed LSTM model and its novel aspects. LSTM is a variation of traditional Recurrent Neural Network (RNN) architecture, which has been extensively used for time-series prediction tasks. Unlike standard feed forward neural networks, LSTM also contains feedback connections and can process single data points (e.g., images) as well as entire sequences of data (e.g., speech, video or object trajectories). Compared to RNNs, Hidden Markov Models and other sequence learning methods, LSTMs are not so sensitive to the length of gaps between important events in a time series, which makes them more preferable in numerous applications. To this end, we adopt an LSTM architecture for the prediction of FOC values from the consecutive observation, corresponding to the aforementioned features, in a time window, as described in the following paragraphs. The input of the LSTM network at timestep tu comprises N time-series, one for each feature of interest (speed through water, wind speed, wind angle etc) and in order to use the recent history of values in each feature, we employ a fixed-length time-window (time-lag of length $m$). As a consequence, the window contains the values for each time step for the weather and vessel state features that are used for the estimation of FOC at time tu, resulting in N time-series, of length m+1, of the form [FN(u–m), …, FN(u–1), FN(u)], for each feature FN. Given a sequence of consecutive time-steps, and a multivariate feature set, we get the following correspondence between the input and the output of the LSTM: Figure 4. Raw data values vs. mean values vs. rolling window average values 130 Enhanced and Holistic Voyage Planning Using Digital Twins F F F F F F u m j u m N u m i j i N i 1 1                      ... ... ... ...   FF F F FOC FOC u j u N u u m i 1                                ... ...   FOCu                 Weather Routing Aspects and Integration of the Solution In order to validate the approach in the context of a real-world application, the data driven FOC LSTM model has been coupled with a WR algorithm to support vessel routing decisions towards the reduction of FOC. The WR algorithm that has been utilized is based on the isochrone principle (Hanssen et al., 1960). It builds upon a predetermined basic route; this route can be the original route planned by the vessel’s master or provided by a basic routing algorithm. In the context of this work an initial route was employed on the basis of shortest path principles. The original (initial) route is then broken into segments, with respect to a given time step (indicating the master’s routing decision horizon, e.g., every 6 hours), and a graph is built around it that enables course and speed deviations, while ``following’’ the direction of the vessel’s original course. To this end, for each node of the original route, a set of nodes is added in a ``parallel’’ fashion on both sides of the route (i.e., parallel to the direction of the original route). Edges are added between all nodes of subsequent sets. Note that nodes that are identified to be on land as well as edges that go above land segments are naturally excluded from the graph. Once the graph is created (Figure 5), LSTM NN is used to obtain the FOC of each edge of the graph, i.e., of each corresponding sea route, given the vessel’s STW, draft and corresponding weather conditions along that sea route. After scoring each sea route (i.e., graph edge), a variation of Dijkstra’s algorithm for the shortest path problem is utilized to obtain the route that minimizes the total route FOC (i.e., considering the calculated FOC of each edge as its corresponding ``edge weight’’ or ``distance’’). Note that since the algorithm is isochrone, the produced route also satisfies any constraints concerning the time of arrival (if any). Note also that the decision variables for the WR algorithm are only the STW and the vessel’s direction, since these are the aspects that the vessel’s master can control. Obviously, any change in the vessel’s speed affects FOC directly (since STW is a basic feature of the corresponding model). However, changes in speed and direction also affect FOC indirectly, since they alter the spatio-temporal state of the vessel and hence the corresponding weather conditions. Preliminary Experimental Results We continue by demonstrating the results of the WR optimization algorithm demonstrated briefly above. We compare the total FOC of an initial transatlantic voyage conducted by the vessel’s master, with the suggested optimized route produced from the WR algorithm by utilizing the aforementioned LSTM FOC 131 Enhanced and Holistic Voyage Planning Using Digital Twins Figure 5. Graph construction comprised of alternative waypoints (red circles) for an example route Figure 6. Initial (blue) and Optimised (red) route for one leg TAMPA (FLORIDA U.S) - TANGER MED (MOROCCO) Table 3. Estimation based on weather service (NOAA) Voyage Date Latitude Longitude Departure 2019-09-21 27.7°N 82.5°W Arrival 2019-10-03 35.8°N 6°W Basic Comparison Actual Route Estimation Optimized Route Estimation Distance 4787.4 4369.36 Time (hours) 289.997 264.63 Avg. Speed (kt) 16.52 16.51 Total FOC (MT) 774.53 759.97 CO2 (MT) 2411.88 2366.54 132 Enhanced and Holistic Voyage Planning Using Digital Twins model. Furthermore, we calculate the total distance traveled, the estimated time of arrival, the average speed and the emissions emitted for the two alternative routes and we exhibit the results in Table 3. Consecutively we demonstrate the weather (wind speed (m/s)) of the initial and the optimized route per hour, in Figure 7 We can clearly see that the optimized route attempted to avoid ambient weather conditions on average while at the same time complied with ETA constraints. The accuracy of the estimations we depict in the table below rely heavily on the accuracy of the FOC model we have demonstrated in previous sections. The model showcases promising approximation capabilities, and we can therefore incorporate its prediction to the heuristic function of a path finding algorithm, being confident that the simulation results correspond to the reaction of the physical system with minimal margin of error. FUTURE TRENDS AND IMPLICATIONS As we look ahead, the convergence of digital twins with other cutting-edge technologies promises to revolutionize voyage planning. Further advancements are expected in the near future in data integration. The digital twin ecosystem is extending beyond the vessel itself. Ports, fuels, and spares suppliers will increasingly become integrated into the digital dataspace. This interconnectedness will enable smoother transitions during port calls, efficient refueling, and timely maintenance. Real-time data feeds from satellites, ocean sensors, and onboard IoT devices are increasingly made available and can be exploited by the digital twin improving the on-the-fly adaption on current sea conditions, weather patterns, or other unexpected challenges, ensuring optimal navigation and safety. As standards develop, digital twins will become more interoperable across different platforms and systems. This means that a digital twin created by one shipping company could be easily used by another, leading to a more collaborative and efficient maritime industry. Additionally, the rise of blockchain and decentralized technologies can enhance the security and integrity of the data used in digital twins. This ensures that the information is tamper-proof and comes from verified sources, increasing the reliability of voyage plans. Figure 7. Weather comparison (wind speed) for the initial and optimised route 235 Shipping Applications of Digital Twins collected and disposed of according to waste management regulations. These methods include the use of non-biocidal antifouling paints, silicone elastomer-based coatings, ultrasonics, water dock systems, reactive cleaning in water, and preventive cleaning in water (hull care). • Antifouling paints without biocides: Biocide-free antifouling paints provide hull and propeller protection but their effectiveness in highly polluted waters is unverified. • Coatings based on silicone elastomer: These coatings create a non-stick surface, preventing organism adhesion, but application is complex and may emit volatile organic compounds or risk oil leakage. • Ultrasonics: Ultrasonic waves disrupt and prevent biofouling without chemicals, but require substantial investment, power source, lifting of the ship, and multiple transducers for larger vessels. • Water dock: Isolated from outside water, water docks prevent organism sedimentation, but are limited to motor and pleasure boats, excluding larger vessels. • Reactive cleaning in water: Fast cleaning with brushes, water jets, or robots while anchored, but requires waste collection to prevent harmful organism and microplastic release. • Preventive cleaning in water (hull care): Grooming the hull removes surface biofilms, effective for small biofouling thickness, but lacks organism collection equipment and requires regular implementation. Methods for cleaning up biofouling in water can be categorized into three groups: (1) Manual cleaning is typically used for small vessels; manual cleaning involves using brushes or scraping devices to remove biological fouling organisms. However, complete removal of biofouling is challenging. In a survey conducted by Song & Cui (2020), it was found that approximately 60% of organisms were removed using a hand brush during manual cleaning, (2) Electric cleaning systems with rotating brush: Mechatronic technology has advanced underwater cleaning techniques for larger vessels. Robot cleaning systems, large cleaning devices, and hand cleaners have been developed. Systems with large rotary brushes driven by hydraulic motors are commonly used for fast cleaning of flat or slightly curved areas, while smaller brushes are suitable for propeller cleaning. It is important to choose the appropriate brush based on the characteristics of the biofouling. (3) Non-contact cleaning methods have been proposed to avoid damaging welds, protrusions, and the mechanical integrity of the hull. These methods include high-pressure water jet, cavitation water jet, and ultrasonic cleaning. Compared to rotating brushes, these techniques cause less damage to the hull coating. Under the category of non-contact cleaning technologies, subcategories for cleaning biofouling in water include: • Ultrasound method: Utilizing ultrasonic pulses at different frequencies, this method generates alternating positive and negative pressures that create tiny bubbles, effectively removing biofouling from the hull. • Laser method: This approach involves using high-energy rays to clean the hull by targeting and removing biofouling. • High-pressure water jet method: Using the force of high-pressure water, this method removes biofouling from the hull. Examples include the HullWiper, which collects biopollutants and sprays water at pressures of 50-450 bar (HullWiper, 2022), and the Magnetic Hull Crawler, a remotecontrolled system with high-pressure jets reaching up to 1000 bar (Cybernetix, 2022). 236 Shipping Applications of Digital Twins • Water jetting method with cavitation: This method employs specially designed nozzles that convert high-pressure water into cavitation water. The resulting bubbles burst near the hull, generating high local stresses for enhanced cleaning power compared to conventional high-pressure water jets. Currently, there are no autonomous robotic systems available on the market that incorporate nozzles for cavitation water jets. These subcategories offer diverse approaches to non-contact cleaning technologies, providing options for efficiently removing biofouling from ship hulls while minimizing damage to the hull coating. The Rapid, Automated Inspection System for Biofouling Assessment Description of System for Hull Biofouling Assessment The proposed inspection system for hull biofouling assessment is a technological advancement in underwater monitoring, combining underwater robotic platforms, image analysis, and AI algorithms to provide a highly efficient and accurate assessment of biofouling development on ship hulls. The center of this system is an autonomous underwater robot specifically designed for biofouling inspection, equipped with advanced sensors, cameras, and mapping capabilities. In this way, the robot can navigate along predetermined paths on the hull, meticulously scanning and analyzing every surface. Therefore, the mapping and pattern recognition techniques ensures that all affected areas are identified. Integration of Underwater Robotics, Image Analysis, and AI Algorithms The integration of underwater robotics, image analysis, and AI algorithms is key to the rapid and precise evaluation of sea growth on the ship’s hull. As the robot explores the underwater environment, it captures high-resolution images of the hull surface. These images are then processed using advanced image analysis algorithms that leverage AI techniques to accurately identify and measure biofouling organisms. In more details, by leveraging extensive datasets, the AI algorithms integrated into the system Table 1. Advantages of the proposed system compared to traditional methods Advantages Rapid, Automated Inspection System Traditional Methods Efficiency Reduces time and labour required for inspection Time-consuming and physically demanding processes Accuracy Relies on objective data and sophisticated algorithms, minimizing subjective interpretation and human error Subjective interpretation and higher chances of human error Coverage Covers larger areas in less time Limited coverage due to manual processes Data Quality Provides detailed and accurate data on biofouling extent and types Data may be less comprehensive and may lack detailed insights Maintenance Optimization Enables optimized hull cleaning schedules and improved coating performance Limited data for proactive maintenance planning Operational Efficiency Facilitates proactive maintenance and enhances overall operational efficiency Potential delays and reduced efficiency due to reactive maintenance Health and Safety Minimizes human involvement in potentially hazardous inspection tasks Exposes workers to risks associated with underwater inspections 237 Shipping Applications of Digital Twins have been trained to accurately differentiate between various types and levels of sea growth, including algae, barnacles, and other marine organisms, enabling quick identification of biofouling presence and extent while providing detailed insights into the condition of the hull. Furthermore, with its capability for real-time data processing, the system enables on-site analysis of collected information, leading to immediate assessment, decision-making, and enhancing overall system efficiency and responsiveness to address biofouling issues promptly. The system’s ability to perform on-site analysis and immediate assessment, coupled with its real-time data processing capability, further amplifies its advantages by enabling prompt decision-making, enhancing overall system efficiency, and bolstering its responsiveness in effectively addressing biofouling issues, as presented in Table 1. It should be noticed that it not only offers the advantages mentioned earlier but also prioritizes health and safety by minimizing the need for human involvement in potentially hazardous inspection tasks. Implementation of the Rapid, Automated Inspection System Underwater Robotics Platforms Used in Biofouling Assessment Leveraging robotic technology, outfitted with advanced sensors and equipment, not only eliminates the requirement for human divers to perform lengthy and dangerous underwater inspections but also enables navigation in underwater environments and captures detailed images of a ship’s hull. As a result, it improves efficiency while reducing the risks associated with human involvement in such tasks. In more details, high-resolution cameras capture detailed images of the ship’s hull, providing valuable visual data for analysis and assessment of biofouling. Fluorescence spectrum sensors are employed to detect the fluorescence emitted by marine organisms on the hull’s surface, aiding in the identification and quantification of specific biofouling agents. To ensure precise navigation and orientation, the robots are equipped with accelerometers that measure acceleration and provide information on motion in the underwater environment, and gyroscopes that measure angular velocity and assist in maintaining stability and orientation during underwater operations. To assess the structural integrity of the hull, the robots utilize LVDTs (Linear Variable Differential Transformers), which measure displacement or strain, and pressure sensors to monitor the pressure exerted on the hull, providing insights into its condition and integrity. Additionally, LIDAR (Light Detection and Ranging) technology is employed to generate three-dimensional models of the hull’s surface. By utilizing laser beams, LIDAR facilitates accurate mapping and identification of biofouling areas, enhancing the system’s ability to detect and address biofouling effectively. While the integration of specialized sensors provides the necessary hardware for comprehensive inspections and evaluations of biofouling, it is equally essential to develop the appropriate software that will conduct the evaluation process. The software component plays a crucial role in analyzing the data captured by the sensors, employing advanced image analysis techniques and AI algorithms to accurately assess the levels and types of biofouling. By developing robust and intelligent software, the system can extract meaningful insights from the sensor data, enabling informed decision-making and optimizing maintenance strategies for the ship’s hull. The synergy between hardware and software is crucial for a successful and effective biofouling evaluation system, ensuring a comprehensive and data-driven approach to hull assessment and maintenance. 238 Shipping Applications of Digital Twins Role of Digital Twins, Image Analysis and AI Algorithms in Evaluating Sea Growth The heart of the system lies in its ability to analyze captured images and evaluate the extent of biofouling on the hull, using image analysis techniques, combined with powerful AI algorithms. Fluorescence spectrum analysis is an image analysis technique which is used to assess and analyze biofouling on surfaces, particularly in the context of marine environments. It involves detecting and studying the fluorescence emitted by marine organisms present on the surface, providing valuable insights into the composition and extent of biofouling. In practice, fluorescence spectrum analysis is typically conducted using specialized sensors and equipment. The surface of interest is illuminated with specific wavelengths of electromagnetic radiation, known as excitation light. When the biofouling organisms on the surface are excited by this light, they emit fluorescence due to de-excitation of the atoms’ electrons from the highest energy levels to the lowest, which is then captured and measured by fluorescence spectrum sensors. These sensors are designed to detect and quantify the emitted light across a range of wavelengths. Different types of marine organisms, such as algae, barnacles, or mollusks, emit distinct fluorescence patterns, allowing for the identification and differentiation of various biofouling species. This analysis offers several advantages for biofouling assessment. 1. It is a non-intrusive and non-destructive analysis. There is no need for physical sampling or direct contact, and therefore reduces the risk of damage the surface of the hull/propeller. 2. Allows for rapid assessment over large areas. 3. Facilitates real-time monitoring of biofouling growth and enables efficient mapping and identification of biofouling hotspots on surfaces. 4. Aid in understanding the effectiveness of cleaning and antifouling treatments. By comparing fluorescence measurements before and after treatment, it becomes possible to evaluate the impact and efficacy of the applied interventions. Fluorescence spectrum analysis, combined with the power of AI algorithms, further enhances the capabilities of biofouling assessment. While fluorescence analysis provides valuable data on the composition and extent of biofouling, AI algorithms take this information to the next level by leveraging machine learning techniques. AI algorithms and machine learning are powerful tools that enable systems to learn from data, recognize patterns, and make intelligent decisions or predictions without explicit programming, driving advancements in various fields. The process of training algorithms involves several key steps. Firstly, a diverse dataset of images/samples representing various instances of sea growth, including a wide range of biofouling types, densities, and conditions, is collected to ensure comprehensive training. Then, each sample should carefully label or annotated with the corresponding biofouling type and level, providing ground truth information for reference during the training process. Then a categorization of the labelled samples is carried out according to their relevant characteristics, such as texture, shape, colour, or any other distinguishing properties that differentiate different types and levels of marine growth, to capture the basic characteristics of the biofouling. Depending on the nature of the data and the classification, appropriate machine learning algorithms (Convolutional Neural Networks (CNNs), Support Vector Machines (SVMs), or other classification algorithms are selected. During the next step of the process, the selected algorithm is trained using the labelled dataset and extracted features, enabling it to recognize patterns and correlations between features and the corresponding biofouling types and levels through 239 Shipping Applications of Digital Twins iterative optimization techniques that adjust internal parameters to minimize classification errors and improve accuracy. In order to assess the performance of the trained algorithm, separate validation datasets are used, and if it is necessary iterative refinements are conducted, where adjustments (e.g., adjusting hyperparameters, or augmenting the training dataset with additional samples) are made to further improve the algorithm’s performance. Training the algorithms using previous data to recognize and classify different types and levels of sea growth brings significant advantages to the evaluation process. It ensures accuracy by exposing the algorithms to a diverse dataset encompassing various biofouling types and levels, enabling precise identification and classification of biofouling instances. The algorithms learn from patterns and characteristics present in the training data, facilitating accurate assessments in real-world scenarios. Furthermore, the trained algorithms exhibit generalization capabilities, allowing them to apply their learned knowledge to new instances of sea growth, thus making informed evaluations. The adaptability of the algorithms is crucial, as they continuously refine their evaluation process using newly collected data, leading to improved accuracy and performance over time. Additionally, the scalability of the algorithms enables efficient processing and analysis of large-scale datasets, facilitating comprehensive assessments across different hulls and environments. Lastly, as decision support tools, the trained algorithms provide valuable insights for maintenance operations and cleaning strategies, aiding operators in making informed decisions based on the severity and type of biofouling. Mapping and Pattern Recognition Techniques for Navigation and Identification Navigation and inspection of the hull’s surface in a systematically way requires the utilization of mapping and pattern recognition techniques, which determine the predefined inspection paths and then the underwater robot follows a strategic route, ensuring comprehensive coverage of the hull. Mapping techniques involve the generation of three-dimensional models of a ship’s hull surface. Such models can be provided by the ship’s DT. These models can be augmented through the utilization of advanced technologies such as LIDAR (Light Detection and Ranging), which enables the collection of precise measurements and data points. The acquired information allows for the creation of comprehensive and accurate representations of the topography of the ship’s hull. Consequently, these models greatly aid in identifying areas affected by biofouling and facilitate targeted inspection and maintenance interventions. Pattern recognition plays a crucial role in the system as it enables the identification and tracking of key reference points or objects in the underwater environment, thereby facilitating accurate positioning and navigation. These patterns may encompass landmarks, contours, or distinctive features that assist in precise navigation and localization. Consequently, the robotic platform can effectively adhere to predefined paths, systematically cover the hull’s surface, and navigate around obstacles or challenging areas. This enables the system to make well-informed decisions concerning navigation adjustments, route optimization, collision avoidance, and adapt its navigation strategies to ensure the safe and efficient exploration of the hull’s surface. Furthermore, they significantly contribute to the overall efficiency of the system by reducing the reliance on manual intervention. The automated recognition of patterns for navigation minimizes the necessity for human intervention or constant supervision, enabling continuous and uninterrupted inspection operations. The combination of mapping and pattern recognition techniques simplify the inspection process by ensuring full coverage of the hull and accurate identification of biofouling hotspots. In addition, they 240 Shipping Applications of Digital Twins support decision-making as they provide, together with other assessment parameters, information relevant to the assessment of severity, and therefore support the selection of appropriate response actions such as scheduling clean-up operations and applying targeted treatments. Methods for Collecting and Analyzing Data on Hull Biofouling In the context of the collection and analysis of data on hull biofouling a comprehensive approach is necessary to ensure accurate assessment and monitoring. This approach includes both data collection methods for bio-fouling assessment and data collection for monitoring robot performance during the inspection process. By combining these datasets, a holistic understanding of the hull’s condition and the inspection process can be achieved. For biofouling assessment, high-resolution cameras capture detailed images of the hull’s surface, providing visual data that enables the assessment of biofouling growth and distribution. Fluorescence spectrum sensors detect and measure the fluorescence emitted by marine organisms, aiding in the identification and quantification of specific biofouling agents. These data sets are combined to provide insights into the extent, type, and distribution of biofouling on the hull. In parallel, data collection for monitoring the robot’s performance involves sensors such as accelerometers and gyroscopes, which measure motion and orientation. LVDTs and pressure sensors assess the structural integrity of the hull. Three-dimensional mapping using LIDAR technology generates detailed models of the hull’s surface, aiding in navigation and monitoring the robot’s position. • Correlation of Image Analysis and Motion Data: By correlating the captured images with the motion data from accelerometers and gyroscopes, it is possible to analyze the biofouling distribution in relation to the robot’s movement, and thus providing insights into areas that were effectively covered during the inspection and areas that may require further attention. • Integration of Biofouling and Structural Integrity Data: Combining the data from fluorescence spectrum sensors, LVDTs, and pressure sensors allows for a holistic assessment of the biofouling impact on the hull’s structural integrity. As a result, correlations between biofouling levels and potential structural vulnerabilities can be revealed, assisting in the prioritization of maintenance and cleaning operations. • Visualization of Biofouling Distribution on 3D Models: The three-dimensional models generated by LIDAR can be overlaid with the biofouling data collected through image analysis. This visualization provides a comprehensive view of the biofouling distribution, highlighting areas of significant growth and concentrated infestation. • Data-Driven Reporting and Decision Support: By integrating all the collected data sets, including images, fluorescence spectrum measurements, motion data, and structural integrity data, comprehensive reports can be generated. These reports provide accurate information on the biofouling condition, hull integrity, and inspection performance. Data-driven decision support systems can leverage this information to optimize maintenance schedules, prioritize cleaning efforts, and enhance overall hull performance. This aspect will further be analyzed in section 0. Taking all into consideration, a combination of data from both hull biofouling and robot performance monitoring ensures a holistic approach to understanding the hull’s condition and inspection process, 241 Shipping Applications of Digital Twins enabling enables accurate assessment, proactive maintenance planning, and optimized decision-making for effective biofouling management. Utilization of Image Analysis and AI Algorithms for Precise Assessment Utilization of image analysis and AI algorithms gives a precise assessment of hull biofouling. The collected data (specifically the images captured by high-resolution cameras and the fluorescence measurements obtained by spectrum sensors) can be harnessed, enhancing the accuracy and efficiency of the evaluation process. On the one hand, with image segmentation, feature extraction, and classification, the algorithms can distinguish different types and levels of marine organisms and encrustations, providing a more comprehensive understanding of the biofouling state on the hull. In this line, AI algorithms, powered by machine learning, are employed to analyze the collected data from the sensors, enabling a precise assessment by leveraging the algorithms’ ability to recognize complex patterns and make informed decisions based on the data. Therefore, a synergistic effect is revealed, meaning that the accurately captured images, along with the precise fluorescence measurements, serve as inputs for the algorithms, which then utilize their learned knowledge to analyze and assess the biofouling state with enhanced accuracy and efficiency. An example will now be provided, focusing on the assessment of biofouling caused by the macroalga species Ulva lactuca. Sea lettuce, scientifically known as Ulva lactuca (Ulvaceae, Chlorophyta), is a macroalga that is widely distributed in marine and estuarine environments. It is a prevalent species found in coastal benthic communities across the globe, exhibiting its ubiquity in these ecosystems (Cruz de Carvalho et al., 2022). When assessing hull biofouling, high-resolution cameras will capture detailed images of the ship’s hull surface, revealing the presence of green patches, while image analysis techniques are used to identify and quantify the extent of Ulva lactuca biofouling. Additionally, fluorescence spectrum sensors will detect and measure the specific fluorescence spectrum emitted by Ulva lactuca, which is 681 nm (Cruz de Carvalho et al., 2022). This will then be analyzed by AI algorithms trained on previous data and will accurately quantify the abundance and distribution of Ulva lactuca. In this way, an accurate assessment of the impact of Ulva lactuca biofouling is achieved, allowing informed decisions to be made regarding appropriate mitigation measures. Reporting Capability and Benefits of Accurate Information The reporting capability of the system enables the generation of detailed and accurate reports that provide comprehensive information about the biofouling condition of the ship’s hull. The reports are not only data presentations but analyzed data, ensuring an effective reporting. This can be achieved by employing different kinds of techniques: Written reports present a holistic summary of the biofouling assessment, a detailed analysis of the collected data, results interpretation, and key findings. They can incorporate charts, graphs, and images to enhance the understanding of information. In contrast, visual representations utilize text, images, diagrams etc. to highlight the key findings, trends, and observations from the biofouling assessment. These can be used when a more dynamic and interactive approach is desired. When a more dynamic, user-friendly, and interactive approach is desired, data visualization tools such as infographics or interactive dashboards, can be utilized, allowing relevant stakeholders to gain a deeper understanding of the biofouling assessment. These include interactive charts, maps, and other visual elements, which enable the user to explore data, visualize trends, and extract 242 Shipping Applications of Digital Twins insights. Finally, due to the rapid progress of information technology (IT), significant advancements have been made in the reporting capabilities for hull biofouling assessment. Specifically designed online platforms and software applications facilitate efficient and standardized reporting procedures, ensuring consistency and ease of use, by customizing templates, data input functionalities and automated report capabilities. The choice of the best reporting technique depends on various factors, including the specific needs and preferences of the stakeholders, the nature of the data being reported, and the intended purpose of the report. Each technique has its advantages and considerations. However, in the context of hull biofouling assessment, a combination of techniques may be beneficial to cater to different audiences and enhance the effectiveness of communication. Regardless of the specific type of reports utilized in hull biofouling assessment, there are numerous benefits associated with accurate information compared to traditional methods as presented in Table 2. DISCUSSION AND FUTURE OUTLOOK In this section we discuss the challenges and possible future directions with respect to predictive maintenance, cargo hold cleaning and hull antifouling treatment using digital twins. In all the above discussed applications, digital twins play a pivotal role. As emphasised in several chapters of this book, a digital twin is a faithful digital replica of a ship or a ship subsystem. The digital twin guides and streamlines the above operations by providing accurate and realistic representations of the ships engines (and their operational behavior), 3D models of the cargo holds which lead to more targeted cleaning operations, and also more accurate understanding of the fouling conditions of the hull which leads to more efficient cleaning operations. However, as all the discussed DT applications are in their infancy there are potential strengths as well as pitfalls that are examined in this section. Challenges of Predictive Maintenance With the Help of DTs Predictive maintenance is all about anticipating to failures and taking the necessary preventive actions Overall, the DT supported predictive maintenance approach proposed in this Section shares similar benefits and pitfalls with other data driven maintenance operations that utilise AI/ML techniques. These can be summed up as data availability, coverage and quality. Table 2. Benefits of the accurate information compared to traditional methods Benefits of Accurate Information Modern Techniques Traditional Methods Informed Decision-Making Enables data-driven decision-making Relies on subjective observations Proactive Maintenance Planning Facilitates proactive planning Reactive approach to maintenance Cost Optimization Optimizes resource allocation May result in inefficient resource use Performance Enhancement Improves vessel efficiency Potential performance degradation Compliance with Regulations Demonstrates adherence to standards Potential non-compliance issues 243 Shipping Applications of Digital Twins Data availability is a crucial issue as equipment failures are rather rare phenomena. Thus, it is hard to find datasets that contain failure states as well as normal operating states of components and equipment. Datasets must therefore be collected over large periods of time, potentially over several years. Additionally, datasets can be borrowed from other DTs modelling ships with similar equipment. In addition, not only data quality but also model quality is important. An accurate DT model should precisely reflect the properties of its physical counterpart. For individual subsystems and components such as ship engines, these models can be obtained from the manufacturers and are expected to be of sufficient quality, as manufacturers themselves dedicate substantial resources to develop accurate models (as well as digital twins) of their products. However, the quality of the models must also address subsystem assemblies and their interactions, for instance the faithful modelling of the behaviour of coupled engine, driveshaft and propeller subsystems. A digital twin requires remodelling with any change in equipment’s configuration or element state Any modification affecting equipment performance requires a change to its model and underlying algorithms. Such modifications – at a machine level (replacing original parts with made-to-order ones) or at a factory level (changes to the operational policy) - are not always reflected in factory specifications, thus, cannot be precisely simulated, which escalates the risk of errors. Although deploying a digital twin-based predictive maintenance is time-consuming and labour-intensive, the technology offers the ability to timely recognize disruptions in asset performance, forecast potential problems and simulate various maintenance scenarios. It helps enterprises eliminate machine downtime, reduce equipment maintenance costs, improve equipment reliability and extend its lifespan. Challenges and Future Outlook of Cargo Hold Cleaning With the Use of Robots and DTs The second use case presented in this Chapter pertained to the use of digital twins and other automation (such as robots) towards making the operation of cargo hold cleaning more safe, efficient and effective. This use case argued about the pitfalls of the current cleaning method (mainly manually operated, slow and with safety risks for the operators). Instead, a robotics based operation can speed up the process, improve the quality of cargo hold cleaning and reduce safety risks for the crew. The use of digital twins, automation and robotics in cargo hold cleaning can potentially improve safety by reducing the need for human entry into enclosed spaces and improving the accuracy and efficiency of cleaning processes. There are still significant risks and limitations that need to be addressed. For instance, a totally autonomous robot operation may not be feasible for every type of cargo hold and cleaning operation, and therefore a semi-automated approach with human participation is required. Also, new or existing sensing technologies need to be added to the cleaning robots in order to detect the cleanliness state of the surface in order to apply the correct type and/or quantity of the cleaning substance. Therefore, more research is needed to fully understand the potential benefits and risks of the approach, as well as how to enhance the effectiveness of commercial robot based cleaning systems with DTs. Challenges and Future Outlook of Automated Antifouling Hull Survey and Treatment One of the three use cases presented in this Chapter is monitoring and measuring the biofouling development in underwater body of the ship using a rapid, automated inspection system for hull biofouling 244 Shipping Applications of Digital Twins assessment. This cutting-edge system utilizes underwater robotics equipped with image analysis and AI algorithms designed to evaluate sea growth. The results of these assessments allow for a more optimized hull cleaning schedule and improved coating performance. The robot deployed to inspect the sea growth on the hull, makes use of mapping and pattern recognition techniques to guide its navigation along a predefined path in order to identify all affected areas. The system includes a reporting capability, providing detailed and accurate data on the state of the hull. Biofouling assessment results can be used for developing an optimized cleaning schedule. Indeed, a regular scheduling of ship hull maintenance and cleaning plays an important role in maintaining hull performance and mitigating biofouling-related problems. For this reason, the data provided by the biofouling assessment makes it possible to optimize the cleaning program, determining the optimal frequency and schedule for cleaning operations and ensuring preventive and targeted interventions. Specifically, by monitoring and analyzing the rate of development and severity of biofouling, stakeholders can make more effective decisions regarding ship cleaning, minimizing financial burdens, excessive biofouling risks, drag, and energy consumption. This enhances hydrodynamic efficiency and overall boat performance. Additionally, the evaluation results provide valuable information about the effectiveness of different cleaning techniques under specific conditions. The DT can play several roles in aiding antifouling treatment. One of the most important is the ability to maintain historical data of hydrodynamic performance, correlated with data about the ship voyages (locations, sea temperature, time of the year/season). This allows the identification of relationships between hull fouling parameters and ship voyage profile and allowing for a more effective future antifouling treatment based on the planned and expected voyage schedule. Therefore, the development of an optimized cleaning program with the aid of DTs can support the execution of cleaning operations at appropriate time intervals, optimizing the use of resources, reducing operating costs, minimizing environmental impact, extending the life of hull coatings, and increasing the ship’s performance. These outcomes contribute to sustainable maritime practices and the financial prosperity associated with them. However, further advances are required to sensing and robotic technologies in order to obtain clearly defined benefits over the alternative methods in use. REFERENCES Alghamdi, S., & Quijada, R. (2019). The Impact of Biofouling on Marine Environment : A Qualitative Review of the Current Antifouling Technologies. [Dissertation, World Maritime University]. https:// commons.wmu.se/all_dissertations/1201 Alonso, J. (2011). Evaluación de efectos de biocidas contenidos en recubrimientos “antifouling “(AF coatings) en ecosistemas marinos. Babin, M., Roesler, C. S., & Cullen, J. J. (2008). Real-time coastal observing systems for marine ecosystem dynamics and harmful algal blooms : theory, instrumentation and modelling. In Oceanographic methodology series. Britannica. (2023). Salinity distribution. Britannica. https://www.britannica.com/science/seawater/ Salinity-distribution 251 Digital Twin With Multizone Combustion Model for Pollutant Emission Estimation for 2-stroke and exhaust for 4-stroke engines). The use of these techniques has resulted in the reduction of the bsfc penalty because of retarded injection timing. . The use of sophisticated tuning, especially in the last decade, following the advances in engine control capabilities via electronic systems, results to variations between even engines of the same model that affect both brake specific fuel consumption (bsfc) and emissions trends with engine load. This was verified in multiple studies both by individual researchers and organizations (De Lauretis et al., 2019; Grigoriadis et al., 2021). In these works, following extensive data collection, high deviations between bsfc and NOx emissions were detected between older and newer type marine engines that were attributed to tuning choices for NOx emissions control. However, for NOx control there exists a number of technologies that have been employed in CI engines some of which are currently applied to 2-stroke marine engines. In Table 2-1 a summary of the measures used generally in the most common diesel engines to control most common pollutant emissions is provided along with the methods’ advantages and disadvantages. One of the main measures applied is the adjustment of injection timing. Retarded start of injection will lead to lower NOx emissions, but also possibly increase soot formation (Helmut, 2010) and bsfc. The goal of retarded injection is to limit the peak pressure and thus in-cylinder temperature. This occurs because fuel injection takes place closer to expansion, which lowers the pressure. Injection retard reduces peak temperature, however the fuel and air mixing are negatively affected which leads to higher soot and particulate formation (Helmut, 2010). The decreased oxidation of soot due to the lower temperatures further increases the soot emissions. For advanced injection timing, combustion initiates the pressure rise when the piston is still moving upwards, thus pressure also increases due to compression. The higher pressure results in higher peak temperature values, that drives NOx formation. The adverse effects of retarded injection timing, as mentioned above, can be partially compensated using advanced injection technologies. A common technique is the reduction of injection duration. Increasing injection pressure enhances fuel mass flow rate, shortening injection duration. This allows for better mixing of fuel and air so the number and range of fuel rich mixture regions is decreased, leading to lower soot formation. However, the shorter injection duration leads to faster combustion, placing the combustion around Top Dead Centre (TDC), increasing thus peak temperature. The effect of injection Table 1. Measures for optimising diesel engine combustion and their effects on emissions and consumption (Helmut, 2010) [Document text truncated for crawler view.]