Engineers look to help artic ships assess ice buildup

The R/V Melville, a now-retired Navy research vessel, was used by the Cornell research team to test their model for inferring a ship’s role gyradius.

As global temperatures rise and arctic ice melts, more ships are taking advantage of expedient, yet dangerous ocean routes that are opening in the polar region.

One of the main hazards of sailing in freezing temperatures is topside icing, in which water blown from the ocean freezes once it contacts a ship, potentially accumulating enough ice to put the vessel at risk of capsizing.

No tools have existed for ships to accurately monitor topside icing, but now Cornell engineers have demonstrated a novel method to do so using a combination of applied mathematics and computational mechanics. The results are published in the February edition of the journal Applied Ocean Research.

‘If you know something about the excitations occurring in a seaway that load a ship, and we can measure some response of the ship to those excitations, we may then be able to infer the current condition of the vessel,’ said Christopher Earls, professor of civil and environmental engineering and co-lead author of the paper. Engineers refer to this as inverse problem solving - using data from an effect to infer something about the cause.

In topside icing, an effect is that the motion of the ship is changed due to the weight of the ice. ‘So we solve an inverse problem by using the inertial motion unit of the ship and a computer vision sensor that looks at the near wave field around the ship, and then use a model that turns that into an excitation,’ Earls explained. ‘So we have an excitation and a response to infer how much ice must be on the ship.’

To demonstrate the inversion framework in the real world, Earls and his team applied it to the R/V Melville - a 279-foot Navy research vessel operated by the Scripps Institute of Oceanography prior to its retirement in 2015. The goal was to accurately determine the ‘roll gyradius’ of the ship and its smaller 1:23-scale model, essentially predicting each ship’s weight distribution about its axis of rolling. And while certain mass properties of a ship may be estimated based on design assumptions, those estimated properties are uncertain once a ship is seafaring due to factors such as varying fuel and hydraulic fluid levels, and how heavy equipment is stowed. Because of this, the inversion framework could be put to the test without the use of ice.

Prior to the full-scale demonstration, the research team began exercising the inversion framework using the small model of the R/V Melville, and the data began to roll in. ‘That was exciting, but not a guarantee of meaningful results yet. Then, incrementally, there were results that showed promise and also showed new things to consider,’ said Yolanda Lin, a doctoral student in structural mechanics and the study’s co-lead author. After some revisions, the team was able to accurately predict the roll gyradius of the ship within the standard deviation of error.

After validating the inversion framework at the model-scale, the team ran the framework on the full-scale ship, using the R/V Melville’s onboard inertial motion unit. ‘Using telemetry, we were collecting data as the ship made some pretty severe maneuvers. It made the crew a little sick and the captain mused that onlookers must wonder if he was drinking and driving,’ said Earls of the testing conducted near the coast of San Diego. The full-scale results were more difficult to validate since the massive ship couldn’t be lifted from the ocean and placed on a pendulum to measure the vessel’s precise roll gyradius, ‘but it was within the expectations that we would have in our minds,’ Earls reported.

Earls and Lin are now taking the successful proof of concept a step further by conducting new experiments using ships with more sophisticated equipment. ‘It’s essentially a plug-and-play framework so you can put any seakeeping modeling tools into it,’ said Earls, who added that the new tests include applying topside icing to a scale-model ship.

‘The ultimate goal is to develop a full framework that can help detect when the surface of a ship has accumulated so much ice that the ship is in danger,’ Lin said. The data produced could also help captains determine the capabilities of a ship, such as what maneuvers it can safely make at any given time.

Earls and his research group are also using inverse problem-solving to attack other maritime challenges, such as using the magnetic signature of the ocean to infer the internal wave structure deep under the surface, or to infer the condition of a ship’s propeller as it interacts with the wake field it generates.

Syl Kacapyr is public relations and content manager for the College of Engineering.

Share

  • Facebook
  • LinkedIn
  • Reddit
  • StumbleUpon
  • Twitter
  • Printer-friendly version
  • Send by email
  • Bookmark

 
 
Jobs listed on