Patrick Ryan is Senior Vice President of Engineering and Technology for ABS, one of the largest classification societies in our industry, whom he is leading through their digitalization.
He has authored seven U.S. and international patents on industrial augmented reality technology, and in 2017, Smart Industry Magazine named him one of the “Top 50” industrial innovators in Smart Technology.
We talk about digital transformation and machine learning in the classification industry. Specifically the work Patrick and ABS are doing on modelling and simulation to improve safety and sustainability in the maritime industry. We dig into the multi-physics simulation approach and the combination of machine learning with physical models using simulated data.
Class is about safety - of the mariner, environment, and property. Class societies like ABS ensure that vessels comply with rules and regulations, through surveying and issuing class certificates.
Modeling and simulation can unlock a safer maritime industry. Software has become a critical component in many vessel systems, but it’s challenging to verify security of software because it’s intangible, compared to physical assets.
If we can see the world through a digital lens, we can see and do things off board and do it more robustly. Simulation is a way of achieving that.
Decarbonization has been the big uptake for this technology at ABS. With existing regulations like MRV and the upcoming CII (2023), class is a natural third party to help shipowners comply. Multiphysics simulation enables class and owners to understand the physics of components like design, wind assist and hull cleaning and how each contributes to any given decarbonization score. It also allows ABS to advise customers on future action and to apply findings across whole fleets.
Patrick uses the term “closing the loop to machine learning” to describe how simulation allows machine learning to see and train on enough varied data so that it can perform robustly. Simulating sensor failure and the follow-on reaction at, say, a piece of rotating hardware, generates data that a machine learning model can use to gain understanding of critical patterns and effects in the real world. The connection of (multi-physics) simulation and machine learning is what creates “true” digital twins. Combining real and simulated data is challenging but potentially gives you the best of both worlds.
Rather than reinvent the wheel, ABS combines and integrates off-the-shelf tools, such as Amesim and Windchill, so that maximal energy can be invested in value creation within ABS business models. ABS runs many joint development projects looking at streaming data and simulation, which are proving that the approach is viable.
By integrating OTS tools with in-house expertise building proprietary models and components within and around those tools, as well as the rules of classification, ABS is able to disrupt e.g. the old-fashioned physical use of blue prints, and is able to create effective digital twins and create new solutions. The resulting system is called “ABS Freedom”.
ABS is looking into non-traditional class services such as dry dock planning optimization, vessel maintenance optimization and so on.
The maritime industry will see a blurring of lines between segments as a result of data and AI enabling new value creation. Class is uniquely positioned to take advantage of this because it spans all of the traditional sectors - owning, building, operating, marine equipment, etc.
Looking across to other industries, such as aerospace and automotive, we can leverage their learnings and apply it in our industry; an example is autonomous navigation.
The IMO 2030 and 2050 goals are tough and noone is going to reach them on their own.
The engine of converting data from potential to what Patrick terms “kinetic” energy, i.e. generating value, lies in obtaining a license to use that data. How to manage this is a work in process in the maritime and other industries.
The main challenge around AI for the C-level and top management is mainly figuring out how to apply it and in which order. When it comes to building AI capability and talent base, a good first step can be to invest in existing technology, such as software tools. In terms of in-house training, it can be easier to teach a maritime SME about AI than to teach an AI SME about the intricacies of a maritime business.
“Modeling and simulation can unlock a safer maritime industry.” - Patrick Ryan.
“We can leverage a lot of the work that the aerospace industry has done on decarbonization.” - Patrick Ryan.
“Machine learning and data analytics is [...] a key [...] piece of the overall landscape today.” - Patrick Ryan.
If you wondered what the inside joke was about the “can modeling and simulation can unlock a safer maritime industry” question, this was the title of an article Patrick wrote for Digital Ship (page 14 in that issue).