Inefficiencies associated with corrosion management increase costs, asset non-availability, generation of hazardous air pollutants, volatile organic compounds, toxic wastes, and exposure of workers to hazardous materials. Improved estimates of protective coating service life are needed to improve coating selection and corrosion management of Department of Defense assets and minimize the safety and environmental impact of finishing processes. The objective of this effort was to develop a prognostic model based on statistical survival analysis to predict the functional life of protective coatings. Coating degradation (coating cracking or mechanical damage) followed by loss of inhibitor protection for chromate and non-chrome coating systems was characterized to establish the required input data to predict coating condition. The limits and validity for applying these methods to multiple coating systems was investigated via laboratory testing and outdoor exposures.

A model for the remaining useful life of aerospace coating systems is an enabling technology for materials selection, corrosion prevention, and control planning. The intent was to develop the simplest possible model using available data to provide insight into the protective properties of a coating and the degradation of a specific asset. To do this, in situ measurement techniques were used to verify simplifying assumptions and to more precisely detect coating failure change points, such as the loss of corrosion inhibitor protection and onset of coating blistering. The resulting datasets, including both environment severity measurements (relative humidity, temperature, salt loading) and real-time corrosion rates at a coating scribe were used to train and demonstrate a stochastic model to predict corrosive events that lead to coating failure. These events can then be used as the basis for a “shock model” to predict time to coating failure based on statistical survival analysis.

This work leveraged techniques for real-time measurement of coating performance to verify model simplifying assumptions, detect coating failure more precisely, and develop a preliminary stochastic model for predicting time to coating failure. In this work, event (temporal point process) models were found to be more suitable for coating and corrosion modeling compared to direct modeling of cumulative corrosion.

In the tests performed, non-chrome coatings were not as protective as chromate coatings in every lab and outdoor test in this study; non-chrome coatings did not exhibit persistent protective films. These results indicate that non-chrome coating implementation should consider increased inspection and preventative maintenance relative to structures protected with chromate coatings. Finally, this work confirmed that coating cracking at structural discontinuities and protection of galvanic couples are appropriate assumptions and focus for aerospace coating protection testing and modeling. Further, free corrosion may not be a strong or relevant performance discriminator.

Methods were established to leverage continuous measurements to quantify coating performance and detect time to coating failure. This approach could be used to reduce risk when introducing non-chrome alternatives and to allow for tailored corrosion prevention and control planning based on measured coating properties. Modeling assumptions verified in this work can be used to simplify future coating models and inform coating management. Specifically, the observation of cracking at structural discontinuities that produce localized high coating strains supports a management and modeling approach that assumes coating cracks are present and that service life is dependent on the presence of protective inhibitors. Finally, the datasets collected in this work that include environment severity measurements, substrate corrosion measurements, and weekly coating images can be leveraged in future model development efforts.