(64h) Atom Explicit Compositional Models of Heavy Oils: Making and Using State-of-the Art Measurements

The considerable interest in molecule-based models of hydrocarbon structure and reaction is motivated by the need to predict the properties of a given composition. This is because the molecular composition is an optimal starting point for the prediction of mixture properties. The potential advantages of molecule-based model­ing are thus clear. Less readily apparent, how­ever, is that the development and operation of molec­ular models comes with a large requirement for state-of-the-art analytical chemistry measurements and model construction and solution time. This presentation considers current approaches to both the quantitative measurement of molecular compositions and the use of this information in reaction models.

The challenge of building molecule-based models is due to the staggering complexity of the complex reaction mixtures. There will often be thousands of potential molecular and intermediate (e.g., ions or radicals) species. The sheer size of the thus-im­plied modeling problem creates a con­flict between the need for molecular detail and the formulation and solution of the model. Clearly, the use of the computer to not only solve but also formulate the model would be helpful in that it would allow the modeler to focus on the ba­sic chemistry, physics and approximations of the model.

While the fine details vary from one research group to another, all approaches address four general issues. The first is the measurement of the feed molecules. The analytical chemistry will generally be assisted by software tools that transform bulk analytical chemistry into a molecular description of the feedstock. This casts the modeling problem in molecular terms. Reactivity information is then often organized in terms of quantitative structure-reactivity relationships. Network building tools allow for the model equations to be built and coded on the computer. A solution environment allows for reactor simulation, which provides a prediction of the molecular composition that can then be organized into any desired commercially relevant outputs.

This approach is illustrated through the development of molecule-based models for thermal cracking, reforming and hydrotreating.