Active Learning Protocols to Accelerate Galvanic Corrosion Predictions

The current of a galvanic couple is generally treated as a surrogate measure of its resilience to or extent of galvanic corrosion. Experiments to measure cathodic current or obtain cathodic polarization curves incur a very high cost due to the need to explore a wide range of temperatures, salt concentrations, and specimen geometries that are used in engineering applications. To reduce these costs, Finite Element (FE) simulations are used to assess the cathodic current output. However, these simulations use the cathodic polarization curves as boundary conditions, which can only be discerned by performing experiments. Therefore, a protocol to accelerate the assessment of the cathodic current output for different chemistries under in-service environmental conditions is desirable. In this work, we develop an active learning protocol to minimize the total costs associated with performing the experiments and the simulations. We first calibrate a low-cost gaussian process surrogate model for the cathodic current output as a function of the environmental and geometric parameters that characterize the galvanic cell. The surrogate model is calibrated on a dataset of FE simulations, and it is used to calculate an acquisition function that identifies specific additional inputs with the maximum potential to improve the current predictions. The identification of additional inputs for further exploration is accomplished with the help of a staggered two-step workflow – (i) the influence of the geometric parameters are marginalized to identify the best configuration of environmental conditions for discerning the polarization curves, following which (ii) the geometric inputs best capable of refining the current predictions are identified. We demonstrate the efficacy of this protocol by minimizing the number of simulations necessary to obtain accurate predictions of the current output of a AA7075-SS304 galvanic couple. The protocols developed and demonstrated in this work provide a powerful tool for screening various forms of corrosion under in-service conditions. This work was supported by the Laboratory Directed Research and Development program at Sandia National Laboratories. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy National Nuclear Security Administration under contract DE-NA0003525. The views expressed in the article do not necessarily represent the views of the U.S. Department of Energy or the United States Government. SAND2023-11065A