Discover how, by integrating machine learning (ML) models into process simulation models, you can extend the potentiall applications for processes where first-principles models may either struggle to capture the complexity accurately, for example having accurate kinetic data for reactions, may be slow and incur excessive computational effort, or where the process contains a manufacturer-specific IP, which needs to be obscured. Machine learning models can rapidly generate accurate predictions from data sets, which you can integrated into process simulations for use throughout design and operation. AVEVA™ Process Simulation provides a state-of-the-art simulation platform in which you can leverage hybrid models. Models are built and integrated into the AVEVA Process Simulation flowsheet using ONNX interchange format adapter. This allows for the easy integration of models built from a variety of different ML toolchains. ISU Chemicals has been key lighthouse customer for this capability. It created ML models to transform sample assay data to feed component structure. Further ML models then predict reactor yield. The results were extremely successful. Reactor yield could be predicted with 99.7% accuracy together with catalyst performance. The model is accessible through state-of-the-art user experience which enables easy use for engineers and operators.