Overview
SeaChem is an interdisciplinary research initiative developing predictive models for maritime corrosion combined with environmental monitoring and mitigation strategies. It brings together oceanographers, materials scientists, corrosion engineers, and data scientists to forecast corrosion risks for marine infrastructure, vessels, and submerged structures — monitoring salinity, temperature, pH, dissolved oxygen, and pollutant concentration alongside corrosion metrics from test coupons and instrumented structures.
Predictive Modeling
Regression, time-series forecasting (ARIMA, state-space models), and machine learning methods (random forest, gradient boosting, neural networks) model corrosion rate as a function of environmental predictors, with hybrid physics-informed variants where expert corrosion theory constrains the learned models. Prediction uncertainty is quantified via prediction intervals, Bayesian methods, and ensemble variance.
Key Deliverables
A curated multi-source marine-corrosion dataset; trained predictive models with a reproducible forecasting pipeline and open-source code; and a web dashboard for stakeholders to visualize predicted corrosion risk and schedule inspections.
