Our research integrates electrochemistry, graph neural networks, and uncertainty quantification for interpretable, physics-consistent corrosion models.
Embedding Butler-Volmer kinetics and Pourbaix diagrams directly into neural network architectures for physically plausible predictions.
Bayesian Monte Carlo methods generating calibrated prediction intervals for risk-informed asset integrity decisions.
Graph attention networks capturing galvanic coupling, flow-accelerated corrosion, and MIC colony propagation across asset topologies.
Integrating IoT sensor streams with physics-informed models for continuous, real-time corrosion digital twins.
Papers currently in preparation or under review across leading peer-reviewed journals.
Demonstrates how Butler-Volmer and Pourbaix constraints in GNN message-passing layers produce physically consistent predictions across unseen CCUS conditions.
Introduces physics-constrained MC dropout for calibrated prediction intervals, validated on pipeline inspection gauge datasets from offshore assets.
Formalizes multi-material joints as heterogeneous graphs with Pourbaix-derived node features for corrosion state classification with 94% accuracy.
Shows that physics-informed latent representations transfer from Oil & Gas to data-scarce CCUS environments, reducing data requirements by 40%.
Describes the sensor integration pipeline combining ER probes, LPR sensors, and UT measurements into graph-structured PI-GNN input for sub-hourly updates.
Addresses XAI for corrosion ML models, interpreting GNN predictions through electrochemical attribution to answer 'Can I trust AI corrosion predictions?'
Key venues for presenting CorrosionAI research and engaging with the corrosion science community.
The largest corrosion conference worldwide. Primary venue for presenting PI-GNN methodology and field validation results.
Premier offshore engineering event in Houston. Focus on maritime and offshore corrosion prediction applications.
European Corrosion Congress. Annual event rotating across European cities for EU research collaboration.
Greenhouse Gas Technology conferences. Focus on corrosion challenges in carbon capture, transport, and storage infrastructure.
AI for Science workshops at top ML venues. Presenting PI-GNN methodology to the machine learning research community.
International Water Association congress. Focus on water distribution network corrosion prediction and digital twins.
In-depth technical documents for engineers and decision-makers evaluating AI-driven corrosion management.
Comprehensive overview of PI-GNN architecture for technical decision-makers evaluating AI-driven corrosion management solutions.
Practical guide bridging Bayesian Monte Carlo methods with engineering risk management workflows for asset integrity.
Why physics-informed AI outperforms empirical models in novel CCUS operating environments with limited historical data.
Practical integration guide for operations teams using Meridium APM, Synergi Plant, Visions Enterprise, or similar platforms.
BEAI Energy S.L. welcomes collaboration with academic researchers, industry consortia, and standards organizations at the intersection of corrosion science and machine learning.
Joint research programs in physics-informed ML for materials degradation, with focus on CCUS, hydrogen infrastructure, and advanced nuclear.
Field validation studies providing real-world inspection data to benchmark PI-GNN predictions against operational records.
Sensor technology co-development for next-generation corrosion monitoring hardware feeding AI-driven digital twin platforms.
PhD and postdoctoral sponsorship in corrosion electrochemistry, graph neural networks, uncertainty quantification, or sensor fusion.
Standards development contributions to AMPP, ISO, and ASME committees on AI-assisted asset integrity management.