2020
Prediction-uncertainty quantification methods are reviewed and implemented using two chemical process examples: (I) a simple Antoine equation model with uncertain temperature input and (II) a more-complex steady-state n-hexane hydroisomerization model. After the parameters have been estimated, parameter and input uncertainties can then be propagated to obtain information about prediction uncertainties. Linearization-base and MC-based techniques are reviewed for uncertainty quantification in Case I. For Case II, Error-in-Variables-Model (EVM) parameter-estimation methods are reviewed and tested on the examples.
