M.K. Cavanaugh

Aqueous Corrosion Testing and Neural Network Modeling to Simulate Corrosion of Supercritical CO2 Pipelines in the Carbon Capture and Storage Cycle

A database was constructed from tests in aqueous electrolytes simulating the damage that may occur to ferrous transport pipelines in the carbon capture and storage (CCS) process. Temperature and concentrations of carbonic acid (H2CO3), sulfuric acid (H2SO4), hydrochloric acid (HCl), nitric acid (HNO3), sodium nitrate (NaNO3), sodium sulfate (Na2SO4), and sodium chloride (NaCl) were varied; the potentiodynamic polarization response, along with physical damage from exposure, was measured. Sensitivity analysis was conducted via generation of fuzzy curves, and a neural network model also was developed. A correlation between corrosion current (icorr) and exposure tests (measured in the form of weight and thickness loss) was observed; however, the key outcome of the work is the presentation of a model that captures corrosion rate as a function of environments relevant to (CCS) pipeline, revealing the extent of the threat and the variables of interest.

A Quantitative Study on the Effects of Environment and Microstructure on Pit Initiation in Al-alloys

Potentiostatic testing was employed to examine pit initiation on AA7075-T651 exposed to chloride solutions. The influence of chloride concentration, potential, and temperature on pit initiation was studied. An automated method for counting current transients is presented, and the applicability of potentiostatic testing in Al alloys is generally discussed.

The Influence of Iron, Manganese, and Zirconium on the Corrosion of Magnesium: An Artificial Neural Network Approach

A total of 71 custom alloys were prepared and tested in order to produce a statistically relevant spread of compositions containing a range of iron (Fe), manganese (Mn), and zirconium (Zr) additions to magnesium (Mg). Alloys were produced using Mg-Fe/Zr/Mn master alloys and were tested using potentio-dynamic polarization and mass loss (immersion) testing to ascertain the relative rates of corrosion. The rationale was to empirically explore the concept of threshold or tolerance limits, namely any variation in tolerance limits depending on the relative Fe, Mn, and Zr content, with direct relevance to aluminum (Al) free Mg-alloys. Data was analyzed using an artificial neural network (ANN) model. It was shown that Mn has a moderating effect on Fe with regard to the acceleration of the corrosion rate, even in the simple Mg-Fe-Mn system and in the absence of Al.