MSC Internship assignment: "Prediction of wave spectra around crew transfer vessels using machine learning methods”
Publication date20 Feb 2023
LocationWageningen, Gelderland
Employment40 hours
ContactBulent Duz, B.Duz@marin.nl
For MARIN Academy we are looking for a student for the following MSc internship/assignment:
Prediction of wave spectra around crew transfer vessels using machine learning methods
Introduction
Motion sickness on board a vessel limits people’s ability to perform their work properly or enjoy the voyage. Ranging from navy vessels and cruise vessels to yachts and smaller Crew Transfer Vessels (CTV), sea sickness reduces their operability window, effectiveness or joy. In particular for CTVs taking technicians to wind turbines, real-time knowledge of local wave field is valuable as this is one of the primary factors causing motion sickness. Therefore, obtaining information on the local wave field from various data sources such as ship motions has gained a lot of interest in the last couple of years.
![]()
Researchers at MARIN have been working on estimating the local wave field in real time from ship motions using machine learning methods [1-5]. The concept, called ship-as-a-wave-buoy (SAWB), allows predicting the directional wave spectra from time series of ship motions using convolutional/recurrent neural networks (see Figure 1 for an impression of the SAWB approach). In this project, we will apply this approach to estimate the directional wave spectra around CTVs.
The measured ship motions from the project ‘improving Safety and Productivity of Offshore Wind Technician Transit’ (SPOWTT) [6] will be used in this study. SPOWTT was an extensive project aimed to widen the workable weather window for CTVs and to improve the productivity of technicians performing service activities on offshore wind turbines. For the directional wave spectra, ERA5 [7] data will be used.
Goal
The goal of this project is to estimate the sea state around crew transfer vessels (CTVs) by adopting the ship as a wave buoy (SAWB) approach.
Motivation
When CTVs are transferring crew to the windfarms, they sometimes encounter rough seas. This might cause issues such as motion sickness, which affects the safety and performance of the technicians. In order to prevent this, a decision is made in transit whether to continue the voyage to the wind farm or return to the port. This decision is based on the characteristics of the local wave field. Hence, a method that predicts the local wave field in real time from easily-available data sources is highly desirable. This is the main motivation for this work.
In addition to the main motivation highlighted above, there are some important distinctions between this study and the previous SAWB projects carried out at MARIN. These distinctions can be summarized as:
Figure 2: United States Coast Guard cutter Bertholf (left), and research vessel Akademik Tryoshnikov (right).
Prediction of wave spectra around crew transfer vessels using machine learning methods
Introduction
Motion sickness on board a vessel limits people’s ability to perform their work properly or enjoy the voyage. Ranging from navy vessels and cruise vessels to yachts and smaller Crew Transfer Vessels (CTV), sea sickness reduces their operability window, effectiveness or joy. In particular for CTVs taking technicians to wind turbines, real-time knowledge of local wave field is valuable as this is one of the primary factors causing motion sickness. Therefore, obtaining information on the local wave field from various data sources such as ship motions has gained a lot of interest in the last couple of years.
|
Figure 1: Artist's impression of SAWB approach.
|
Researchers at MARIN have been working on estimating the local wave field in real time from ship motions using machine learning methods [1-5]. The concept, called ship-as-a-wave-buoy (SAWB), allows predicting the directional wave spectra from time series of ship motions using convolutional/recurrent neural networks (see Figure 1 for an impression of the SAWB approach). In this project, we will apply this approach to estimate the directional wave spectra around CTVs.
The measured ship motions from the project ‘improving Safety and Productivity of Offshore Wind Technician Transit’ (SPOWTT) [6] will be used in this study. SPOWTT was an extensive project aimed to widen the workable weather window for CTVs and to improve the productivity of technicians performing service activities on offshore wind turbines. For the directional wave spectra, ERA5 [7] data will be used.
Goal
The goal of this project is to estimate the sea state around crew transfer vessels (CTVs) by adopting the ship as a wave buoy (SAWB) approach.
Motivation
When CTVs are transferring crew to the windfarms, they sometimes encounter rough seas. This might cause issues such as motion sickness, which affects the safety and performance of the technicians. In order to prevent this, a decision is made in transit whether to continue the voyage to the wind farm or return to the port. This decision is based on the characteristics of the local wave field. Hence, a method that predicts the local wave field in real time from easily-available data sources is highly desirable. This is the main motivation for this work.
In addition to the main motivation highlighted above, there are some important distinctions between this study and the previous SAWB projects carried out at MARIN. These distinctions can be summarized as:
- Compared to the ship geometries considered in the previous studies, CTVs are smaller and have different types (Mono-hull, Catamaran or Swath). In that sense, their seakeeping behaviour is different, and how that affects the feasibility of the SAWB approach is of importance.
- In-service ship motion measurement data was collected from 14 ships at 5 wind farm locations (see Figure 3 for the locations of the wind farms). In the previous studies, a United States Coast Guard cutter and a research vessel were used (see Figure 2).
- The ground truth for wave spectra will be the ERA5 data [7]. In the previous studies, wave spectra were usually obtained from a wave scanning radar mounted on the ship.
Figure 2: United States Coast Guard cutter Bertholf (left), and research vessel Akademik Tryoshnikov (right).