sign in sign up
<<
Home , Cruiseferry, Maritime transport, Road transport, Skogn (municipality)

By Namireddy Praveen Reddy, Mehdi Karbalaye Zadeh, Christoph Alexander Thieme, Roger Skjetne, Asgeir Johan Sørensen, Svein Aanond Aanondsen, Morten Breivik, and Egil Eide

ECAUSE IS AN INTEGRAL PART OF DAY-TO-DAY LIFE FOR humans, civilizations have generally flourished either in coastal areas or near basins. Today, about 40% of the world’s population lives B near a coast and many of the biggest cities are located near the sea or along . In , a high percentage of people live in coastal areas. There are many waterways, such as and fjords, in these coastal areas. Constructing over these water channels is expensive and may hinder marine . Therefore, in the coastal cities, urban are much more than tourist attractions. They are an integral part of the city’s multimodal transportation system and essential for easing traffic and congestion. The 2.1 billion per year. By comparison, com- mercial across the world carry 2.3 billion passengers per year. Ferries in Lon- don carry about 10 million passengers every year. The number of ferry rides annually in , Sidney, Hong Kong, and Istanbul comes to 3 million, 16 million, 47 mil- lion, and 108 million, respectively (Figure 1). In many expanding cities, ferry ridership will increase significantly as the population grows. London’s ridership is predicted to reach 20 million by 2035, double the number today. San Francisco’s ferry ridership has grown 74% in the last five years and is predicted to increase by up to 900% by 2035. Urban ferry transportation systems are undergoing a revival in many coastal cities Zero-Emission Autonomous Ferries for Urban Water Cheaper, cleaner alternative to bridges and manned vessels.

Digital Object Identifier 10.1109/MELE.2019.2943954 Date of current version: 2 December 2019

32 IEEE Electrification Magazine / DECEMBER 2019 2325-5987/19©2019IEEE xxeconomic viability, due to lower life-cycle cost, com- pared with constructing bridges and manned ferries xxcentral supervision of several through a control room located onshore. An autonomous ferry that can see kayakers and and shows up right when you need it could be an ingenious substitute for footbridges. However, there are several challenges in designing fault-proof zero-emission autonomous passenger ferries. Autonomous systems by nature are complex. An autonomous system must accom- modate various kinds of software, hardware, and actua- tors and different groups of people. All these subsystems interact with each other to some degree and tracing all these interactions across the system is virtually impossi- ble. During normal operation of an autonomous system, this may not be a problem. However, a single failure in one computer may lead to cascading failures in the software system, resulting in an erroneous actuator output. This output itself may not be harmful, but with a user not too familiar with the system, it may lead to an accident. Antic- ipating such interactions beforehand and consequentially eliminating them will not be possible without a systemat- ic, traceable, and holistic assessment process. Therefore, the task of assessing the safety, reliability, and security of autonomous systems cannot rely on the tradi- tional methods that focus on single failure events or on iso- lated security threats. For autonomous systems, security considerations, especially those pertaining to cybersecurity, will be different from those for conventional systems. Threat agents may exploit the preprogrammed behavior, and detection mechanisms may be insufficient to differen- tiate between a random encounter and an intentional dis- ruption. The complexity of the system may mask hacking attempts and create vulnerabilities through insufficient and inadequate implementation of a subsystem. More challenges can be found with respect to the moral and ethical behavior of the autonomous systems. In case of unavoidable accidents, what behavior is appropri- ate? Here legislation could give answers. The International Maritime Organization (IMO) is still working on its code for autonomous maritime surface ships. One requirement ©ISTOCKPHOTO.COM/LOWESTOCK soon to be universally accepted is that autonomous ships due to increasing ridership, advancements in technolo- gies, and the replacement of aging fleets. The concept of zero-emission autonomous passenger 3 ferries in urban areas is a more flexible, cost-effective, and London 10 environmentally friendly alternative to bridges or manned Sydney ferries. Autonomous ferries can also be coordinated with, for 16 instance, autonomous in an urban intelligent-trans- Hong Kong 47 portation system, reducing commuting time and improving Istanbul 108 quality of life for people living in cities. Zero-emission autonomous passenger ferries are a promising alternative 020406080 100 120 for the following reasons: xxlow or zero Figure 1. A chart showing annual ferry ridership in millions (2017) in xxless or no hindrance to marine traffic five major cities.

IEEE Electrification Magazine / DECEMBER 2019 33 should be as safe as manually operated Norway as a Projects Related to ships. The challenge will be to demon- Zero-Emission and strate that this is the case since very maritime nation has Autonomous Shipping limited experience and data are avail- In Norway, several ongoing projects able on autonomous systems and their been at the forefront are focused on developing zero- operation. Simulations and verification in developing and emission energy solutions and au­­ may assist in the attempt to demon- tonomous systems especially for strate safety, but these need to be ac­­ adapting pioneering passenger ferries, but also for mari- companied by real trials and tests in the time transportation in general. Nor- operational environment. Furthermore, technologies to way as a maritime nation has been autonomous ships may also change improve maritime at the forefront in developing and traffic patterns, due to different naviga- adapting pioneering technologies tional strategies. transportation. to improve maritime transporta- These challenges are, to some ex­­ tion. Much of the research related to tent, valid for several autonomous seg- these projects is being performed at ments, including autonomous ships, autonomous , NTNU’s Centre for Autonomous Marine Operations and underwater , and aerial systems. Although there Systems (AMOS), which was established in 2013 with the may be differences in the operational environment, the aim of conducting multidisciplinary research to create a acceptable performance, and the types of failures and world-leading research center for autonomous marine accidents that may be encountered, these autonomous operations and control systems. AMOS contributes to systems share commonalities that require new ap­­proaches fundamental and interdisciplinary knowledge in marine to safety, reliability, and security. Interdisciplinary and hydrodynamics, ocean structures, and control theory. The interindustrial approaches are required to address these research results are being used to develop intelligent challenges. Key enablers for successful and acceptable ships and marine structures, autonomous unmanned systems are as follows: vehicles (underwater, on the surface, and in the air), and xxrecognizing and understanding risks associated with robots for high-precision and safety-critical operations in autonomous system operation extreme environments. xximplementing safe solutions from early design phases In 2016, the fjord area was designated as xxmonitoring and reviewing risk levels exceeding the world’s first testbed for autonomous vessels by the acceptable risk levels Norwegian maritime authority in consultation with indus- xxestablishing regulations and procedures to ensure try and research organizations. This will boost research safe operation and development in this area and raise the prospects for xxcommunicating with society to establish trust in autonomous shipping to be competitive with road trans- autonomous systems. for passengers and goods. The following list summa- In 2018, the Norwegian University of Science and Tech- rizes the various related projects involving autonomous nology (NTNU) initiated a pilot project, Autoferry, as a vessels in this area and in the region: platform for research and development to tackle these xxAutonomous Transport at Trondheim (ASTAT) is challenges. The project focuses on autonomous passenger a project initiated by a consortium of partners, includ- ferries in urban waterways. A full-scale ferry with a capaci- ing the Trøndelag government. ASTAT’s aim is to ty of 12 passengers with their is under construc- examine possibilities for operating small zero-emis- tion. This is expected to be operational in the Trondheim sion autonomous ships in the fjord at distances rang- by 2021. Furthermore, NTNU researchers are collabo- ing from 80 to 90 km. The main objective is to design rating with industry to develop autonomous technologies both bulk and break-bulk shipping-based transport and with the Norwegian Maritime Authority to define the that will be competitive with road-based trans- regulations for autonomous shipping. port. The three selected case studies for this project Once the full-scale project is successfully demonstrat- (Figure 2) are 1) transport of containers and break bulk ed, the NTNU Autoferry project should open new possibili- between the Jøsnøya terminal for coastal ships and ties for both urban and coastal transportation. In general, Trondheim or and thus link coastal routes the use of autonomous vessels should reduce the cost of to transport systems in the Trondheim fjord; 2) trans- ferry services and help revitalize the urban water trans- port lumber from Orkanger to Follafoss or and port system and stimulate further development of coastal eventually replace older manned ships and for areas. Autonomous ferries could prove to be transforma- large volumes of ; and 3) carry gravel from Verra- tive by fostering technological development, benefiting botn to Trondheim for tarmac production and thus society, and creating new opportunities. Autono- replace trucks and older ships. mous ferries of tomorrow could also include zero-emis- xxNTNU Autoferry is a pilot project to develop, design, sion vessels, a particular focus of this article. and build a full-scale zero-emission autonomous

34 IEEE Electrification Magazine / DECEMBER 2019 passenger ferry for safe and easy urban transport intended for the fully autonomous operation of the in Trondheim. The main goal of this pilot project is full-scale ferry. to demonstrate autonomous technologies for pas- xxYara Birkeland will be the world’s first zero-emission senger ships through a ferry operating in the Trond- autonomous . It is named after its own- heim canal. Figure 3(a) shows the concept design ers, Yara International, and Yara’s founder, Norwegian of Autoferry. Figure 3(b) shows the route map scientist Kristian Birkeland. It is designed by Marin

Short Sea Transport Route 1

Short Sea Transport Route 2

Short Sea Transport Route 3

Figure 2. A map exhibiting the designated short-distance sea transport routes in Trondheim’s fjord area for testing autonomous ships. (Map data © Google.)

y

Cruiseferr

Autonomous Shuttle Fosenkaia

Autoferr y Ravnkloa To Nidaros Cathedral Vestre Kanalkai

(a) (b)

Figure 3. The Autoferry pilot project. (a) A drawing illustrating the concept design of the autonomous ferry. (Source: Petter Mustvedt; used with permission). (b) A photograph with labels that show the intended route of autonomous operations in the Trondheim canal.

IEEE Electrification Magazine / DECEMBER 2019 35 Teknikk, with equipment from Kongsberg industry. The ferry was remotely piloted through a test Maritime. It is expected to enter service in early 2020 area near , proving that human over- with a small crew on board for trials. Thereafter, it is sight of vessels from anywhere is achievable with expected to operate autonomously by 2020. The today’s technologies. design prototype of the Yara Birkeland is shown in Fig- xxThe Urban Water Shuttle aims to develop a rapid zero- ure 4(a). The intended route for fully autonomous emission vessel that will help reduce urban conges- operation is shown in Figure 4(b). With this new tion, emissions, and city costs. This autonomous battery-driven container vessel, Yara will could be realized in cities close to rivers and other replace 40,000 journeys of diesel-powered truck haul- waterways, making the concept relevant for most cit- age from the road to the sea. This will reduce noise ies in the world. This project has been developed by a and particle emissions and improve road safety. joint industrial partnership involving the business xxDNV GL ReVolt is a project to build a zero-emission cluster NCE MaritimeCleanTech. The partners are unmanned vessel for short routes by sea. The full- Wärtsilä, Fjellstrand, Servogear, Grenland Energy, CFD scale ReVolt is a 60-m container vessel with a capacity Marine, Sapa, and Norsk Hydro. of 100 20-ft equivalent units (TEU) and a sailing range xxTransport: Advanced and Modular is a project funded of 100 nmi on battery power. DNV GL, in collaboration through the ’s Horizon 2020 research with NTNU, built a 3-m-long ReVolt model to test program. The project’s main objective is to develop autonomous technologies, such as sensor fusion and and validate a concept for waterborne modular design collision avoidance, for autonomous surface vehicles. and construction with a focus on zero-emission ves- xxSafer Vessel With Autonomous Navigation is a dem- sels in protected , coastal areas, and inland onstration of the world’s first fully autonomous ferry waterways. The project will result in a new fully elec- by Rolls-Royce Commercial Marine (now Kongsberg tric high-speed vessel with zero emissions that will Maritime in Norway) and Finnish state-owned ferry operate between Stavanger and Hommersåk on the operator Finferries in the city of , . The west coast of Norway. In addition, the project aims to Falco vessel is equipped with a range of advanced develop new methods to reduce pro- sensors for situational awareness to conduct collision duction and costs. The vessel will be built avoidance through sensor fusion during its voyage by the Norwegian shipyard Fjellstrand. between Parainen and Nauvo. The vessel also has xxThe Zero-Emission Fast Ferry project (ZEFF), initiated demonstrated automatic berthing with an autono- by five cluster partners at Norwegian Centre of Exper- mous navigation system. The situational awareness tise (NCE) Maritime CleanTech, was awarded 10.5 mil- picture is relayed to a remote control room on shore, lion Norwegian kroner through the Pilot-E scheme. All where a captain monitors the autonomous operations partners—Norled, Selfa Arctic, LMG Marin, Hyon, and and can take control of the vessel if needed. Servogear—are members of the NCE Maritime Clean- xxThe II is an ice-class passenger ferry in Tech cluster. This project aims to develop fast zero- Helsinki. ABB tested a marine pilot control system on emission ferries: a battery version for shorter routes the Suomenlinna II to cultivate the acceptance of and a hydrogen version for longer routes. Today, fast autonomous operation systems in the maritime ferries represent 1% of the oil consumption in Norway,

(a) (b)

Figure 4. (a) The design prototype of the Yara Birkeland. (Source: SINTEF Ocean; used with permission.) (b) A map of the intended route for the autonomous vessel Yara Birkeland. (Map data © Google.)

36 IEEE Electrification Magazine / DECEMBER 2019 which amounts to 86 million liters of fuel oil per year. ferry routes, where a battery-powered ferry is not viable. The goal of ZEFF is to eliminate these emissions over Table 1 summarizes the selected projects presented in the next 10 years. this section. xxMF Ampere is the world’s first battery-powered ferry with conventional propulsion. The vessel, owned by Zero-Emission Energy Solutions Norled and built by Fjellstrand shipyard, commutes for Autonomous Shipping daily on a route between Lavik and Oppedal in the Zero-emission autonomous ships with all-electric power Sognefjord. and propulsion systems have generated considerable xxMF Folgefonn is the first commercial ferry in the world research and development interest in recent years. All- operating with automatic high-power wireless induc- electric power and propulsion systems in zero-emission tion charging capability for its batteries. Wärtsila in autonomous ships are powered by batteries alone or by Norway has developed the automatic wireless induc- fuel cells as the main energy sources in combination tion charging system. MF Folgefonn commutes daily with batteries to compensate for the slow dynamics of between the islands of Stord, Tysnes, and Huglo in the fuel cells. Fuel cells and batteries make no noise and Hordaland area of Norway. require less maintenance than conventional systems Meanwhile, Stranda municipality in Norway has eval- because they do not use rotating equipment. Fuel cells uated the concept of a hydrogen-powered passenger ferry and batteries are designed modular, which makes them in Geiranger to eliminate emissions in the World Heritage resilient, easy to maintain and replace, and suitable for fjord. Furthermore, Stranda has proposed Hellesylt scalable designs. These are all considered very impor- Hydrogen Hub to produce hydrogen through local surplus tant benefits for autonomous shipping. Fuel cells have hydropower. Similarly, the Sogn og Fjordane district several advantage that make them especially appropri- municipality in Norway is supporting the development of ate for applications: high efficiency a hydrogen-powered speedboat. The Norwegian Public for most of the power range, quiet operation, fuel flexi- Administration awarded a development contract bility, and suitability for air-independent propulsion sys- for a hydrogen-powered ferry aimed at long-distance tems. Moreover, hydrogen has high gravimetric energy

TABLE 1. Selected projects involving technologies for zero-emission and autonomous shipping in and around Norway. Project Goals

ASTAT (2017–2019) Ÿ Designing a zero-emission autonomous ship for both bulk and break-bulk transport Ÿ Making shipping competitive with road-based truck transport NTNU Autoferry Ÿ Creating an on-demand ferry that is as easy to use and safe as an (2018–2022) Ÿ Developing zero-emission all-electric propulsion with automatic charging of batteries Ÿ Crafting an integrated autonomous system that includes automatic docking Ÿ Perfecting a safe and reliable anticollision system Yara Birkeland Ÿ Operating the world’s first zero-emission all-electric and autonomous container ship (2017–present) Ÿ Moving transport from road to sea, thereby reducing noise and emissions while improving the safety of local roads ReVolt (2013–present) Ÿ Facilitating an unmanned zero-emission ship concept for short-distance sea shipping Ÿ Easing road congestion by making shipping competitive with Safer Vessel With Autono- Ÿ Establishing the world’s first fully autonomous ferry mous Navigation (2018) Ÿ Improving the safety and reliability of autonomous vessel operations The Urban Water Shuttle Ÿ Reducing urban congestion and infrastructure costs (2017–present) Ÿ Proving the feasibility of a zero-emission all-electric propulsion system Ÿ Demonstrating a prototype vessel with the length of 25–30 m for 180 passengers and an operat- ing speed of 20 kn Transport: Advanced and Ÿ Developing and validating a concept for modular design and production of zero-emission all- Modular (2018–2022) electric fast-moving vessels Ÿ Focusing on coastal areas and inland waterways ZEFF (2019–2021) Ÿ Developing a ferry that operates without emitting , nitrous oxides, sulfur oxides, and particulate matter; can cruise between 25 and 45 kn; consumes significantly less energy than current vessels per passenger per kilometer; has a modular design accommodating 100–300 passengers; and can be battery-powered for short-range and hydrogen-powered for long-range use

IEEE Electrification Magazine / DECEMBER 2019 37 density and can be produced from Batteries are maintenance requirements and enable micro- to megascales by using re­­ zero-emission operation. newable energy sources for achiev- the most common ing truly zero emissions. Batteries Batteries are preferred over other energy-stor- energy storage Batteries are the most common ener- age devices for transport applica- in all-electric and gy storage in all-electric and hybrid tions due to high energy density and ships. Batteries significantly reduce wide power range. hybrid ships. fuel and maintenance costs and emis- Passenger ships, such as ferries, sions. They also help improve ship cruise ships, and liners, must com- responsiveness, resilience, operational ply with stringent emission regulations as they operate performance, and robustness in critical situations. Battery in either emission-controlled areas designated by the cells are categorized into four types for maritime applica-

IMO or urban areas. Therefore, an all-electric power tions: lithium cobalt oxide (LiCoO2;LCO), lithium manga-

and propulsion system powered by batteries alone or nese oxide (LiMn2O;4 LMO), nickel manganese cobalt

together with fuel cells is the most suitable alternative oxide (LiNi1xyMnxCoyO2; NMC), and lithium phos-

as it offers emission-free operation. phate (LiFePO;4 LFP). LCO cells have many advantages: relatively high gravi- Conventional With metric energy density and volumetric energy density, low Cleaning System or Scrubbers self-discharge, high discharge voltage, and good cycling The diesel engine, the most commonly used power performance. The main limitations for this type are high source in about 95% of ships today, has several advan- price, low thermal stability, and fast capacity reduction at tages: high reliability, wide power range, high efficiency, high current rates or during deep cycling. LCO batteries and low energy consumption. The main disadvantage of are mostly used in consumer electronics where 80% of the using diesel engines in autonomous ships is the need capacity use is sufficient. for frequent maintenance due to rotating parts. Also, LMO batteries cost less, have better power capability, the engines emit a significant amount of greenhouse and feature high thermal stability. However, these batter- gases and particles. Operators of diesel engine-based ies have lower energy density compared with LCO batter- ships have three main alternatives to reduce emissions: ies. Also, LMO batteries have a short cycle life at high they can modify ships to use as temperatures. fuel; they can continue to use high-sulfur fuel oil (HSFO) NMC batteries, compared to LCO batteries, cost less and process air emissions through an exhaust gas and have a higher gravimetric energy density, while offer- cleaning system; or they can switch from HSFO to a ing a similar operating voltage. It has high specific energy lower-sulfur fuel, such as marine gas oil or low-sulfur due to the attribute of its constituent nickel. Manganese is fuel oil. Each option has its costs and benefits, depend- doped to improve thermal stability. The composition of ing on market conditions and regulations, which can nickel, cobalt, and manganese can be changed to obtain change in ways difficult to predict. Moreover, they do different tradeoffs between power and energy density, not result in zero emissions; merely reducing emissions cost, and thermal stability. in urban areas is not enough. Furthermore, with diesel LFP batteries are known for high thermal stability engines, unmanned operation of ships is not possible as and high power capability. The main disadvantage of the crew is required on board for maintenance. There- LFPs is low average potential and low electrical and ionic fore, the shipping industry and new ship buyers are conductivity. These limit the energy density, a major looking toward alternative zero-emission energy sourc- drawback. Table 2 compares different types of lithium- es, such as batteries and fuel cells, that can reduce ion batteries.

Fuel Cells TABLE 2. A comparison of different types of lithium-ion batteries. In the literature, different fuel cell technologies have been Lithium-Ion Battery Types LCO LMO NMC LFP evaluated and compared for Energy density (Wh/kg) 195 150 205 130 maritime applications. The most Energy density (Wh/l) 560 420 580 333 promising fuel cell technologies for maritime applications are Cycle life 500–1,000 300–700 1,000–2,000 1,000–2,000 those involving proton exchange Thermal runaway (°C) 150 250 210 270 membrane fuel cells (PEMFCs), Safety Low Moderate Moderate High high-temperature PEMFCs (HT- PEMFCs), and solid oxide fuel Performance Moderate Low Moderate Moderate cells (SOFCs).

38 IEEE Electrification Magazine / DECEMBER 2019 PEMFCs are promising energy Recently, However, for longer missions, the bat- sources for zero-emission transporta- teries must be larger compared to fuel tion due to their low operating tem- researchers have cell systems as the required energy perature, high power density, and storage increases. This is because the modularity (or scalability). The main shown significant energy density of batteries is less than disadvantage of PEMFCs is price interest in zero- one-tenth that of liquid hydrogen at because of the platinum required to the same power level. Therefore, hy­­ catalyze the electrochemical reac- emission ships, brid power and propulsion systems tion at low operating temperatures. powered by fuel cells as the main Another disadvantage, because of the including battery- energy source combined with batter- low operating temperature, is intoler- powered and hybrid ies are suitable for long missions that ance of fuel impurities. Today, this is require large energy storage. For fuel the most common fuel cell, mainly in vessels. cell systems, only the size of the fuel vehicular applications. HT-PEMFCs, tank increases for longer-range mis- because they can operate at high sions while the fuel cell itself remains temperature, can use catalysts cheaper than platinum and the same for the required power. The available volume are significantly more tolerant of fuel impurities. The onboard ship is the main constraint in choosing the type main problems for this fuel cell are the high price, short of energy solution, while the mass of the powertrain is not lifetime, and low power density. They are mainly suitable that important as a limiting . Considering the extra for high power applications. SOFCs have operating tem- mass of the hydrogen tank, the overall mass of a power peratures of 800–1,000 °C. These fuel cells, in the develop- powered by fuel cells should be significantly less ment stage, are promising, but mechanical vulnerabilities than one powered by a marine diesel engine, since the and high costs remain obstacles to commercialization. specific energy of hydrogen is about three times that of The main application for SOFCs is stationary power gener- the maritime liquid fossil fuels. This lower mass may ation. Table 3 compares the different types of fuel cells. reduce the power requirement, and it can result in Today’s zero-emission energy solutions can meet the increased cargo capacity if the fuel cell system takes up propulsion power and energy requirements for vessels less space. ranging from small passenger ferries and fishing boats to Recently, researchers have shown significant interest in the large cargo ships. The choice and suitability of zero- zero-emission ships, including battery-powered and hybrid emission energy solutions depend on the size of the ship vessels. In the past few years, several zero-emission ships and the energy requirements of the mission. For a high- powered by batteries alone have been launched around the power propulsion system that does not require large ener- world. For example, in Norway, the passenger ferry MF gy storage, fuel cells are much larger than batteries since Ampere, powered by 1,090-kWh batteries, sails 5.5 km the size of the fuel cell increases significantly with the between Oppedal and Lavik. It can carry up to 350 passen- increase in power level. Therefore, for small vessels that gers and 120 cars. Meanwhile, the autonomous container require high power for short durations, batteries alone are ship Yara Birkeland (under construction) will be powered by sufficient to meet the power and energy requirements. 7-MWh batteries and have a cargo capacity of 120 TEU and

TABLE 3. A comparison of different types of fuel cells. Fuel Cell Types PEMFC HT-PEMFC SOFC

Power range 0.12–5 kW 5–250 kW 1–2,000 kW Operating temperature (°C) 60–130 140–200 600–1,000 Efficiency 40% 40% 60% Cost Decreasing High High Tolerance of fuel impurities Low High High Startup time <1 min <1 min 60 min Applications Transportation High-power applications Large-scale power generation Advantages Ÿ Low temperature Ÿ Cheaper catalyst Ÿ High efficiency Ÿ Quick startup Ÿ Quick startup Ÿ Fuel flexibility Challenges Ÿ Expensive catalyst Ÿ High temperature Ÿ High temperature Ÿ Short lifetime Ÿ Long startup

IEEE Electrification Magazine / DECEMBER 2019 39 EEMS

Minimize Minimize Minimize and Minimize Ensure Safety, Extend Lives Life-Cycle Fuel Manage Component Security, and of Components Operating Consumption Emissions Losses Resilience Cost

Figure 5. A chart presenting objectives for an EEMS.

a deadweight tonnage of 3,200 t. A study of the Norwegian minimizing power losses in system components, extend- ferry fleet has concluded that it is economically feasible to ing the useful life of components, and minimizing emis- convert 84 of 180 ferries to battery-powered propulsion and sions (Figure 5). 43 to hybrid propulsion. Most of the existing research has considered one or some of the objectives mentioned. However, all of the Energy and Emission Management objectives must be considered simultaneously for a real System for Autonomous Ships autonomous ship. The energy and emission management An energy and emission management system (EEMS) is strategies, in general, can be classified into three categories defined as a high-level control system that commands (Figure 6): 1) rule based, 2) optimization based, and 3) learn- the operation of a hybrid power plant to minimize energy ing based. The rule-based strategies are classified into two usage and emissions while maintaining safety and resil- types: deterministic and stochastic. The deterministic rule- ience requirements and fulfilling the objectives of the based strategies operate on a set of predefined instruc- vessel’s mission. Hence, the EEMS optimizes the perfor- tions: the power plant is modeled according to physical mance of the hybrid power and propulsion system by dis- laws, and uncertain parameters are assumed to belong to tributing the load power required among multiple energy certain bounded sets. Deterministic control algorithms are sources in such a way that each source is optimally used then designed to ensure stable closed-loop behavior of the and overall emissions are minimized. The EEMS, by con- plant as well as robustness to parameter variations within trolling the dynamic behavior of the hybrid power sys- the bounded sets. Stochastic rule-based methods, on the tem, manages the energy efficiency, wear on system other hand, assume that states and parameters are ran- components, fuel consumption, and emissions. There- dom variables with corresponding probability density fore, the selection and development of a suitable EEMS functions. Stochastic algorithms are then typically imple- based on the objective function and requirements of a mented in the form of stochastic filters and state estima- hybrid propulsion system are very important. The main ob­­ tors (e.g., Kalman filter, particle filter, and so on) and with jectives of an EEMS include minimizing fuel consumption, stochastic control algorithms that provide stability and

Control Strategies

Rule Based Optimization Learning

Deterministic Offline Online Supervised

Unsupervised Stochastic Genetic MPC Algorithm or ECMS Dynamic Reinforcement Programming PMP LP/QP/ NLP

Figure 6. A chart displaying the classification of control strategies. MPC: model predictive control; ECMS: equivalent cost minimization strategy; PMP: Potryagin’s minimum principle; LP/QP: linear programming/quadratic programming.

40 IEEE Electrification Magazine / DECEMBER 2019 robustness (and often optimality) on the given probability expectations. Therefore, researchers have become more density functions. interested in learning-based strategies, which can be clas- Optimization-based strategies are typically implement- sified into three types: supervised learning, unsupervised ed in the form of iterative numerical optimization, i.e., learning, and reinforcement learning. Supervised learning genetic algorithms or dynamic programming. Optimiza- is learning from a training set of labeled examples provid- tion-based strategies are mainly classified into two types: ed by a knowledgeable external supervisor. Unsupervised offline optimization or online optimization. Offline opti- learning is learning by discovering the hidden structure in mization relies on the knowledge of predicted sailing pro- collections of unlabeled data. These are very good learn- files. It is very difficult to predict the real sailing profile due ing methods, but alone they are not adequate for learning to several stochastic parameters in waterways, such as from the environment through interactions in real time. unpredictable weather conditions, factors related to Reinforcement learning is a goal-directed method where marine biology, and random ocean currents. However, an agent learns in real time by interacting with the uncer- offline optimization is important for the optimal design of tain environment. a power plant’s configuration, given a set of likely opera- Figure 7 shows an example of a control system ab­­ tion load profiles combined with performance and safety straction for an autonomous ship and its power system. requirements of the hybrid power plant. For online opti- The control layers are classified into three categories: mization of the hybrid power plant during the voyage, an mission layer, online optimization layer, and real-time instantaneous cost function is used. Several numerical control execution layer. On the top level, there is the ves- optimization methods for EEMSs can then be implement- sel mission management system, which acts as a mission ed, including linear, quadratic, or nonlinear programming; layer that keeps of the vessel mission and objectives an equivalent cost minimization strategy; Potryagin’s min- and commands the lower-level systems accordingly. The imum principle; and model predictive control. EEMS performs as an online optimization layer of the Rule- and optimization-based strategies use either hybrid power and propulsion system. The power manage- predefined rules or predicted sailing conditions. These ment system (PMS) is at the next level. It ensures that the methods are not necessarily adaptive to random system load requirement is met at all times and provides refer- conditions and, thus, may produce results short of ences to the main power sources and energy-storage

Mission Layer Vessel Mission Management Autonomous System External Services Mission Methodologies Preventive Mission Planning and Objectives Risk-Reducing Remote Control Center, Replanning Artificial Intelligence Measures Weather Forecasting, (Minimum Risk Power System Mode and Others Condition) and Configuration Perception, Pattern, Recognition, Learning

Online Optimization and Situational Awareness Guidance Layer Dynamic Risk Models Energy and Emission Management Load and External and Monitoring Preventive and Consequential Power Unit Scheduling Environment Sensors Risk-Reducing Measures and Load Distribution Optimization and (Minimum Risk Condition, Onboard Prediction System Reconfiguration and Optimal Data Processing Segregation, and Others) Reference Signals Supervisory Control

Sensor Fusion and Real-Time State Real-Time Control Execution Layer Estimation

Local and Plant-Wide Control Fault-Tolerant Control Power Plant Sensors Consequence Reducing Measures PMS and BMS Control Hybrid Control Signal Processing (Circuit Isolation, Blackout Protection, Battery Equipment, Consumer, Fault Detection and Adaptive Control Protection, and Others) and Load-Sharing Control Diagnosis

Figure 7. An example of control system abstraction for an autonomous ship and its power system.

IEEE Electrification Magazine / DECEMBER 2019 41 devices. If a fault occurs, the PMS prevents a blackout. A communication. communication is suitable for battery management system (BMS) works in parallel with ships with a long sailing range. Radio communication is the PMS. The inappropriate operation of batteries, i.e., sufficient for ships with a short sailing range, such as pas- overcurrent, overvoltage, and overcharging/discharging, senger ships in urban areas. Furthermore, cloud technolo- accelerate the aging process and can result in an unsafe gies for storage and computing can be used for data environment, including risk of fire and explosion. The management and computational purpose. However, this BMS is necessary to ensure the safe and reliable operation may involve additional cybersecurity issues. of batteries. The real-time control execution is at the low- Figure 9 depicts different operation modes for autono- est level, where the power sources and energy-storage mous ships. Autonomous ships can be operated in several devices are controlled individually according to the refer- modes: semiautomatic operation with crew on board, semi- ences provided by the PMS and BMS. automatic operation from a remote control center located Figure 8 shows the control and communication archi- onshore, semiautonomous operation, and fully autono- tecture for an autonomous ship. The autonomous control mous operation. Here, automatic operation mainly refers to system presented is implemented in onboard computers. having real-time feedback control systems that typically These computers can communicate with a remote control are commanded by human operators, whereas in autono- center located onshore through satellite and radio mous operation, the commands are also produced by

Satellite

Remote Control Center Cloud

Radio Tower

Mission Layer

Vessel Mission Management System

Online Optimization Layer EEMS

Real-Time Control Execution Layer PMS BMS

Diesel Engine Fuel Cell Battery Control Control Control

Monitoring

Diesel Engine Fuel Cell Battery Command Propeller Autonomous Ship

Figure 8. An illustration depicting the control and communication architecture for the autonomous ship.

42 IEEE Electrification Magazine / DECEMBER 2019 Remote Control Remote Control Remote Control Center Center Center

(a) (b) (c)

Figure 9. An illustration of autonomous ship operation modes. (a) Fully autonomous operation. (b) Semiautonomous operation. (c) Remote-con- trolled operation. Dotted lines indicate that a selected set of signals is included in the communication. Solid lines indicate that all the signals are included in the communication. Red signifies communication from the vessel. Green signifies communication to the vessel. intelligent algorithms in computers. Semi- or fully autono- Fully autonomous operation of ships, without a crew mous operation can be conducted with or without onboard on board, could significantly reduce the total weight crewmembers, who typically have other duties (e.g., pas- and the space requirement of the vessel. However, this senger comfort) or more supervisory tasks. presents some challenges, chiefly the issue of designing In most of the shipboard power systems, hierarchical all of the ship’s systems to operate with fault-tolerant control is used since different levels of the control system are capabilities. Resilience and survivability are the most decoupled from each other. However, in hierarchical control, important factors to be considered during the design though the low-level real-time control of components is and operation of the shipboard hybrid power system. A handled by independent local controllers, the system-wide zonal electrical distribution system (ZEDS) architec- real-time control and online optimization rely on the coordi- ture could be used (Figure 11) to maximize resilience nation of many local controllers. Based on the coordination and survivability. Here, the shipboard hybrid power among local controllers, hierarchical control can be classified system is divided into several zones interconnected into three categories (Figure 10), described as follows: through a physical electrical connection to exchange 1) centralized coordination control: This is implemented power and a communication interface to exchange through a central controller and a communication information. Each individual zone is a minigrid with network [Figure 10(a)]. Centralized control enables main energy sources, energy storage, loads, and its own system-level optimization, which means the whole system fails if the centralized control fails. Further- more, during autonomous operation, the ships are vulnerable to malicious cyberattacks with centralized Central coordination. Therefore, this is not very suitable for Controller fully autonomous operation of ships. 2) decentralized control: This is achieved through indepen- Controller 1 Controller 2 Controller n dent local controllers without any coordination [Figure 10(b)]. The main advantage of decentralized control is (a) that it offers independence from the communication link and the central controller, so the system can con- Controller 1 Controller 2 Controller n tinue to operate during a single-point failure. The main drawback of decentralized control is the limita- (b) tion on system-level performance since system-level optimization is not possible due to the lack of system- level integration/information. 3) distributed coordination control: This combines the advan- tages of centralized coordination control and decentral- Controller 1 Controller 2 Controller n ized control. In distributed coordination control, the central controller does not exist, but local controllers (c) communicate to perform system-level optimization as well as to maintain system operation during single- Figure 10. A diagram indicating (a) centralized coordination control, point failures, thereby avoiding system-level failure. (b) decentralized control, (c) and distributed coordination control.

IEEE Electrification Magazine / DECEMBER 2019 43 Zone Zonal Operator Communication Remote Control Interface Center Loads

Energy Battery Storage

Z 1 Z 2 Z... Zn–1 Zn Fuel Cell Sources

Main Energy Diesel Engine Communication Link Electrical Connection SSCB

Figure 11. The layout diagrams of a ZEDS for a shipboard hybrid power system. Z: zone; SSCB: solid-state circuit breaker.

zonal operator (local controller). The main energy sourc- configuration, self-healing can be achieved by isolat- es could be a fuel cell to achieve zero-emission opera- ing the faults through sectionalization. When a fault tion or a small diesel engine. In ZEDS, several small occurs, the shipboard power system is sectionalized to diesel engines can be used instead of a few big diesel achieve isolation of the fault. When the fault is engines typically used in manned ships today. In this cleared, the shipboard power system is reconstructed. case, a fault in one diesel engine will not have a signifi- The sectionalizing aims to minimize the isolated area cant effect on the ship’s power system except for the while, at the same time, maintaining the power sup- reduced redundancy. Hence, blackouts can be avoided, ply to healthy zones. which is essential for fully autonomous operation of xxDemonstrate and maintain improved resilience and ships. Moreover, each small diesel engine can have an survivability. integrated battery pack for operating at maximum effi- xxThe design must be modular and scalable. ciency with load smoothing. This will increase resil- xxIt must allow for easier and quicker maintenance. ience and ensure that all of the small diesel engines can However, there are several challenges that need to be be run at top efficiency. Moreover, a failed diesel engine addressed before the successful implementation of a generator set (or another power unit) may then also be ZEDS configuration. The main challenge is that it quickly ejected and replaced at the harbor before a new requires a communication interface for the coordination voyage is conducted. However, studies of life-cycle cost of the local controllers, where communication time and emissions are necessary to investigate the econom- delays and measurement errors may further hamper the ic viability and environmental friendliness of replacing performance of the system-level optimization software a few big diesel engines with several small diesel during normal operation and of the protection system engines. Each zonal operator or local controller can be during faulty situations. Furthermore, in the case of ship- regarded as an intelligent agent that controls its zone board dc power systems, protection is a challenging task and coordinates with other zones. Combined, these since it requires precise coordination between the solid- agents form a multiagent system. The communication state circuit breakers (SSCBs) and the protective func- interface enables system-level integration through com- tions. Therefore, it is a protection system challenge to munication among all zones and collects the system- achieve fast and efficient reconfiguration, during faulty level information required for optimization during situations, and reconstruction, when the faults are normal operation. Furthermore, during fault situations, cleared. In recent years, a significant research effort has the system will be sectionalized for the isolation of fault been made to develop SSCBs using advanced wideband- through a minimum isolation area, thereby avoiding gap semiconductor materials (such as silicon carbide and system failure. The advantages of using a ZEDS configu- gallium nitride), new power electronic converter topolo- ration include the following: gies, and advanced control techniques. Moreover, mea- xxDistributed coordinated control can be implemented surement errors can be overcome by using outliner data as a multiagent system that combines the advantages detection and reconstruction algorithms. of centralized coordinated control and decentralized control. Conclusions xxSelf-healing is the most important requirement for In urban areas, increased population and economic activ- autonomous ships without a crew on board. In a ZEDS ity are causing an exponential rise in congestion and

44 IEEE Electrification Magazine / DECEMBER 2019 . A large percentage of the world’s population cell/battery hybrid electric using reinforcement learn- lives in urban areas located near the sea or along rivers ing,” in Proc. 2019 IEEE Transportation Electrification Conf. and with many available water channels, such as canals and Expo (ITEC), pp. 1–6. C. A. Thieme, I. B. Utne, and S. Haugen, “Assessing ship fjords. Reviving urban water transport systems and mak- risk model applicability to marine autonomous surface ing them an integral part of a city’s multimodal transpor- ships,” Ocean Eng., vol. 165, pp. 140–145, Oct. 2018. tation system can take away congestion from roads in M. A. Ramos, C. A. Thieme, I. B. Utne, and A. Mosleh, these urban areas. The autonomous operation of passen- “Autonomous systems safety: State of the art and challenges,” ger ferries seems to be a promising solution, especially in Proc. 1st Int. Workshop Autonomous Systems Safety, Trondheim, Norway, Mar. 11–13, 2019, pp. 18–32. with advancements in technologies that enable increased M. K. Zadeh, Stability Analysis Methods and Tools for Power levels of autonomy and emission-free energy systems. Electronics- Based DC Distribution Systems, Applicable to on-Board Thus, zero-emission autonomous passenger ferries could Electric Power Systems and Smart Microgrids. Trondheim, Nor- ease the problem of congestion and pollution in the way: NTNU, 2016. urban areas while reviving the water transportation Z. Jin, G. Sulligoi, R. Cuzner, L. Meng, J. C. Vasquez, and J. M. Guerrero, “Next-generation shipboard DC power system: systems. Besides the environmental advantages, the Introduction smart grid and DC microgrid technologies into autonomous ships offer other benefits to the shipping maritime electrical networks,” IEEE Electrific. Mag., vol. 4, no. 2, industry: reduction of operational costs by up to 30% by pp. 45–57, June 2016. reducing the need for crews; increase in safety and reli- D. Park and M. K. Zadeh, “Dynamic modeling and stability ability especially at night and in bad weather, when analysis of onboard DC power system for hybrid electric ships,” in Proc. 2019 IEEE Transportation Electrification Conf. and visibility is poor; optimal usage of fuel; and expanded Expo (ITEC), pp. 1–6. opportunities to carry cargo by water. However, there are K. Lai and M. S. Illindala, “A distributed energy manage- several challenges that must be addressed to sustainable ment strategy for resilient shipboard power system,” Appl. transportation solutions by autonomous ferries. Some of Energy, vol. 228, pp. 821–832, Oct. 2018. the challenges include ensuring ease of access for pas- M. K. Zadeh, L. Saublet, R. Gavagsaz-Ghoachani, B. Nahid-Mobarakeh, S. Pierfederici, and M. Molinas, “Energy sengers, situational awareness, sensor fusion, interaction management and stabilization of a hybrid DC microgrid with manned ferries for collision avoidance, and optimal for transportation applications,” in Proc. 2016 IEEE Applied operation of emission- and maintenance-free energy Power Electronics Conf. and Exposition (APEC), pp. 3397–3402. systems. The advanced control strategies for energy management can be used to achieve the optimal opera- Biographies tion of shipboard power and propulsion systems for the Namireddy Praveen Reddy ([email protected]) real-time sailing profile. is with the Department of , Norwe- Norway has been at the forefront of developing and gian University of Science and Technology, Trond- adopting technologies for sustainable and intelligent heim, Norway. maritime transportation. This article presented sever- Mehdi Karbalaye Zadeh ([email protected]) is al research projects and community initiatives. Some with the Department of Marine Technology, Norwe- of these projects and activities focus on developing gian University of Science and Technology, Trond- either autonomous or emission-free energy systems. heim, Norway. However, the time has come to combine these in an Christoph Alexander Thieme ([email protected]) autonomous passenger ferry. NTNU Autoferry is estab- is with the Department of Marine Technology, Norwegian lished as a pilot project aimed at expanding the University of Science and Technology, Trondheim, Norway. research and development in the field and Roger Skjetne ([email protected]) is with the Centre a zero-emission and autonomous passenger ferry for Autonomous Marine Operations and Systems, Norwegian designed for urban transport. This has the potential to University of Science and Technology, Trondheim, Norway. create new markets while addressing several pressing Asgeir Johan Sørensen ([email protected]) is challenges of society. The global commercial freight with the Centre for Autonomous Marine Operations and market alone is worth US$208 billion per year. The Systems, Norwegian University of Science and Technology, maritime transportation sector with major crew costs Trondheim, Norway. presents an unexplored market for autonomous ship- Svein Aanond Aanondsen ([email protected]) ping technologies. is with the Department of Marine Technology, Norwegian University of Science and Technology, Trondheim, Norway. For Further Reading Morten Breivik ([email protected]) is with the “Autoferry.” Accessed on: Feb. 7, 2019. [Online]. Available: Department of Engineering Cybernetics, Norwegian Uni- https://www.ntnu.edu/autoferry versity of Science and Technology, Trondheim, Norway. “Norwegian Centres of Expertise Maritime CleanTech.” Accessed on: Feb. 21, 2019. [Online]. Available: https:// Egil Eide ([email protected]) is with the Department of maritimecleantech.no/ Electronic Systems, Norwegian University of Science and N. P. Reddy, D. Pasdeloup, M. K. Zadeh, and R. Skjetne, “An Technology, Trondheim, Norway. intelligent power and energy management system for fuel

IEEE Electrification Magazine / DECEMBER 2019 45