Computational Tools for Corrosion-Related Ship Maintenance and Design

Corrosion-related maintenance encompasses a fifth of all maintenance costs expended by the U.S. Navy.  In marine environments, such as those experienced by Naval vessels, corrosion can be aggravated via the continuous exposure and high salt content.  Understanding, predicting, and forecasting the level of expected corrosion on a given ship can help inform long-term maintenance scheduling that is based on corrosion severity in contrast to periodic scheduled intervals.  Therefore, the proposed work will address these challenges by creating a computational toolset to aid in both the design and maintenance of Navy ships.  A database of corrosion and mechanical properties and performance will be assembled to train/test the developed models; this database will be delivered to the Navy for unrestricted use.  Machine learning (ML) algorithms will be leveraged to understand key features leading to high corrosion-related maintenance events.  All developed models will be incorporated into user-friendly software tools written in open-source code for integration with existing Navy maintenance systems, such as the Navy Data Environment, Planned Maintenance System, and SKED Planned Maintenance Scheduling Software.  The proposed team includes a wide variety of expertise, including computational modeling and analysis (University of Virginia, CMI Beasy), corrosion-related degradation (Luna Labs, University of Pittsburgh), and ship-related maintenance (Vision Point Systems).  In addition, the proposed work has received letters of support from possible end-users and Phase II collaborators, including Naval Surface Warfare Center (NSWC) Carderock Division, Sandia National Laboratories (SNL), and Lockheed Martin Company.