Session: 05-02-01 Aquaculture and Related Technology I

Submission Number: 156607

Multi-Spatial Measurement Points Based Structural Health Monitoring in the Tendons of an Offshore Aquaculture Farm’s Fish Cage 

The increased demand for nutritious food and high-quality aquaculture products while preserving of the ocean environment’s safety has led to the development aquaculture facilities including offshore fish farms. Very few such farms exist worldwide, and they are composed of one or multiple semi-submersible fish-breeding cages. These cages are placed offshore due to the requirement of space which results to a greater separation between the fishes and to the reduction of the negative impact caused by illnesses such as lice. Due to severe environmental conditions, structural endurance in the fish cages is more challenging with important cage components such as the net and tendons (steel ropes) getting damaged. Such damages lead to the escape of many fishes with dire economic and biologic effects. Thus, Structural Health Monitoring (SHM) in the offshore fish farms is crucial for detecting damages early and reacting before severe accidents happen.

Currently, due to the severe environmental conditions and increased distance from the shore, SHM in fish cages is sporadic, time-consuming and costly as it is achieved via visual inspection with the use of divers and via the analysis of lengthy videos and a larger number of high pixel images obtained by a Remote Operating Vehicles (ROV). Additionally, the changes in the values of tension signals acquired via loads cells are used for achieving SHM in tendons but the effectiveness of this approach is limited to damages corresponding to tendons breaking close to load cells.  Due to the disadvantages of divers, ROVs and load cells, efficient and automated methods based on vibration data from measurement points in a fish cage are required for continuous and economical SHM and predictive maintenance in fish cages.

The current paper presents a case study investigating the detection of a damaged vertical tendon in an offshore fish farm’s cage via an automated machine learning method equipped with multiple statistical AutoRegressive models (MM-AR) and simulated acceleration data corresponding to measurement points on the fish cage. The considered farm is based on the Norwegian salmon farm Ocean Farm 1 whereas the cage’s finite element model is used for generating acceleration signals for healthy and damaged cases and under varying current speed and wave height. The examined damage cases correspond to uniform degradation (fatigue damage) along the whole tendon and to degradation at specific points in the tendon. The degradation is simulated by the reduction of the tendon’s area thus leading to the tendon’s stiffness, with stiffness reduction of [5,15,20,25,30,40,50]% along the whole tendon and stiffness reduction of [0,5,15,30,50]% at specific points in the tendon. 27 healthy and 184 damaged cases are examined with the MM-AR based method detecting all the healthy cases, all the damage cases corresponding to stiffness reduction along the whole tendon and the damage cases corresponding to stiffness reduction [0,30,50]% at specific points in the tendon.

Presenting Author: Rune Schlanbusch NORCE - Norwegian Research Centre AS

Presenting Author Biography: Dr. Rune Schlanbusch works as a chief scientist in NORCE Norwegian Research Center and as deputy research director of the DARWIN group. Rune received his MSc within Space Technology at University of Tromsø (UiT), Norway and PhD within Engineering Cybernetics at the Norwegian University of Science and Technology (NTNU), Norway. His research interests include control design and stability analysis of unmanned aerial systems including aircraft and spacecraft, and condition monitoring of machinery. Rune is currently leading the group of research and innovation within UAS Norway, the Norwegian national RPAS organization. He has (co)authored more than 50 research papers on topics related to control, condition monitoring and multi-physics modeling.