Via Kaizen project
Accelerating the technological development, realization, and adoption of AI-powered ship operation support technology.
The Via Kaizen project
The Via Kaizen project was funded by the Swedish Transport Administration (Trafikverket) to explore how ship operators can improve energy-efficient voyage planning based on artificial intelligence and machine learning. Working in cooperation with industry partners like Molfow and DNV and academic institutions such as Chalmers University, Halmstad University and the University of Gothenburg, an AI-based, semi-autonomous voyage planning system was developed and trialed on two vessels: a car carrier operated by UECC and a Stenersen product tanker.
The project used pre-existing tools, such as YMT’s propulsion optimization system FuelOptTM and performance management and vessel data reporting tool Fleet Analytics, as well as Molflow’s vessel modelling system Slipstream, to enable a higher degree of digitalization and automation in vessel operations. In order to ensure that the technology was streamlined to best support processes and decisions that have the greatest impact on energy efficiency, existing work practices and user needs were analyzed and considered during the design parameters.
The Via Kaizen project has yielded significant information regarding data handling, model development, crew training and corporate processes that can either facilitate or hinder the effective use of AI tools to improve efficiency.
Key takeaways include:
- Incorporating machine-learning algorithms for improved predictive modelling of ship propulsion power can lead to more accurate performance forecasting and optimization.
- Tools must support the existing operation and responsibilities of the crew. Integration of the onboard optimization system with crew workflows is essential to avoid obstacles.
- Constructive collaboration between technology developers and users, as well as between ship operators and their customers, is needed to build a successful use case.
The project has unearthed new knowledge about the unique needs of shipping stakeholders and offers concrete areas of development for digitization and AI. Additional funding has been secured from the Swedish innovation agency Vinnova to further advance our understanding of these topics.
How can AI support ship operators hoping to achieve their most energy efficient voyages?
Making big vessel data available for performance management does not automatically lead to the optimization of energy/fuel efficiency; the data must also be understood, which requires a great deal of effort both on board and ashore. This is where we come in.
As a group of Scandinavian technology entrepreneurs and academics, we are aware of the potential for artificial intelligence (AI) and machine learning to unlock great possibilities when it comes to transforming decision-making and planning in shipping. Therefore, we joined forces to accelerate the research and development of an AI-based, semi-autonomous system for planning and executing more energy-efficient sea voyages.
Yara Marine has been leading this ground-breaking research and development project which commenced its work in August 2020. Titled Via Kaizen, the project derives its name from a combination of the Latin word ‘via’ – meaning ‘way’ – and the Japanese word ‘kaizen’, which is used in situations striving for continuous improvement.
We have been collaborating with Molflow as well as Swedish academics from Chalmers University of Technology, the University of Gothenburg, and Linnaeus University on this groundbreaking project funded by the Swedish Transport Administration. The Swedish Shipowners’ Association is also participating in the project, providing vital insights and input from the Swedish shipping industry, and assisting in the dissemination of research findings and development information to the Swedish maritime industry.
In addition to the main project partners, ship-owners and operators are also involved in the Via Kaizen project, including chemical/product tanker owner and operator Rederiet Stenersen and pure car and truck carrier (PCTC) owner and operator UECC. They enable on board testing by offering their vessels for technology and product trials, the results of which will be directly evaluated within the scope of the project.
AI-powered ship operation support system
FuelOpt, a propulsion automation system that sits ‘on top’ of the existing control system and other systems on the bridge, optimizes the propulsion line dynamically in real-time. This is based on orders given by Route Pilot AI, the ship operation support system developed within the scope of this project. In addition, the FuelOpt system gathers data from Route Pilot AI and other signals onboard, which is then fed into our cloud-based performance management platform Fleet Analytics, where it is shared with Molflow’s vessel modelling system Slipstream.
Slipstream will be trained on the vessel data made available by Fleet Analytics and use deep learning technologies to predict the vessel’s performance in different conditions. As a result, Slipstream will be able to determine the most energy-efficient voyage possible given the constraints of the route and vessel, and calculate the commands that need to be set to reach the destination with the minimal amount of fuel consumed.
Once the perfect simulated journey is determined, FuelOpt will step in and create an interface between the captain and RoutePilotAI, empowering them to cooperate and execute the voyage accordingly.
The entire project is overseen from an academic perspective, with naval architect researchers at the Chalmers University of Technology working in close collaboration with Yara Marine and Molflow on the development of new methods, models, and algorithms.
Management and human-centered artificial intelligence
Adoption of AI technology by shipping organizations
An increasing number of shipping companies are installing advanced systems to help them achieve more energy-efficient vessel operations. However, previous research on shipping has shown that the availability of large information sets related to energy does not automatically lead to optimization of energy use, as interpretation and understanding of the data involved often poses an insurmountable challenge to both the vessel crew and the shore-based organization (Viktorelius and Lundh, 2019).
Humans have a limited ability to see patterns in large amounts of data and evaluate several factors at the same time, which is why organizations and individuals tend to choose what is considered good enough rather than optimal. Therefore, to facilitate the work on decision-making through big data, an AI-based solution is the best way forward.
Understanding existing working methods, requirements, and conditions – and adapting the AI-based support system to these – contributes to new optimized working methods and routines.
There is currently limited research exploring the possibilities of AI being used in a meaningful way within existing working methods and routines to support the decision-making process for a shipping company. Another key point left to be identified is the manner in which an AI-based system should be designed such that decision-makers on board and ashore can feel confident about the increased degree of automation. These are concerns that this project will tackle.
In addition to the technological development, implementation, and evaluation of an AI-based, semi-autonomous support system, the Via Kaizen project integrates social science into its analysis. Researchers from Gothenburg Research Institute (GRI) and the Maritime Academy at Linnaeus University are conducting comprehensive research on the evolution of practices on board and ashore in connection to the implementation of new technologies.
With all these elements in place, the implementation of AI-based optimization technologies can play a groundbreaking role in realizing the most energy-efficient voyages in practice.