A Bayesian analysis of the compression set and stress-strain behavior in a thermally aged silicone foam

Data obtained from an archived nine-year aging study on S5370 foam were used to develop compression set and stress-strain aging models. Compression set was characterized using a first order kinetic model and the stress-strain relationship was analyzed using a material model previously described by Rusch for flexible foams. The models were fitted to data from the aging study using Bayesian methods, which easily accommodate uncertainties in the test conditions and provide probability distributions of the model parameters. The parameter distributions were sampled using a Markov chain Monte Carlo algorithm and incorporated to effect prediction intervals and compared to data obtained from independent studies for the purpose of validation. Compression set data from the short time study of Patel and Skinner are shown to predict significantly higher compression sets, which are attributed to additional crosslinking reactions and other phenomena that do not dominate the long term aging behavior. Using data from the nine-year study, the time period required to achieve a given compression set at 25 °C is increased by 20 years or more over the predictions of Patel and Skinner. The activation energy applicable near room temperature is similar to that reported by Patel and Skinner, which is consistent with numerous physical and catalyzed chemical mechanisms. Finally, load retention predictions from the stress-strain aging model agree with independent studies at test gaps that are larger than or equal to a zero gradient test gap limit.