(125b) Molecular Based Assays for the Practical Correlation and Prediction of Crude Oil and Petroleum Fraction Properties | AIChE

(125b) Molecular Based Assays for the Practical Correlation and Prediction of Crude Oil and Petroleum Fraction Properties


Chen, C. C. - Presenter, Aspen Technology, Inc.
Dziuk, S., Aspen Technology, Inc.

Molecular Based Assays for the Practical Correlation and
Prediction of Crude Oil and Petroleum Fraction Properties

Keywords: crude oil assay, molecular characterization,
planning and scheduling, process simulation


For the first time, a comprehensive molecular based
methodology has been developed for practical correlation and prediction of assays
and properties for crude oils and petroleum fractions. Based on limited and
commonly available crude oil assay characterization data such as distillation
yield curve, API gravity, and viscosity, the method identifies chemical
compositions of pre-determined surrogate model hydrocarbon molecules designed
to mimic major chemical constituent components for the purpose of
characterizing physical and chemical properties of crude oils. The resulting molecular
representation of the crude oil then provides a self-consistent molecular basis
to interpolate, extrapolate, and predict all physical and chemical properties
of crude oils, petroleum fractions, and blends.

The determination of crude oil assay is a lengthy,
tedious and costly process. The conventional approach to perform an assay
consists of a limited set of measurements on the crude oil and its fractions.
Most often only a few boiling points, densities, and other property
measurements are available for selected distilled fractions or the whole crude.
Therefore it is necessary for crude oil assay experts to predict or estimate
missing properties to meet various business needs, such as refinery planning and
scheduling and refinery process simulation. Typically lower order polynomial
expressions are used for interpolation and arithmetic probability scale is used
for extrapolation of boiling point curves. Statistically derived predictive
methods also have been extensively used in the industry for the prediction or
estimation of crude oil assays. Analytical methods may predict crude oil
properties by correlating the data obtained with rapid surrogate measurements
(usually spectroscopic) to those of reference crude assays.

The key elements of this molecular based assay
characterization approach include:

identifying optimal classes of surrogate model
hydrocarbon molecules, practical rules to construct these model molecules from
molecular ?repeat units? or "segments," and distribution functions of
the model molecules required to account for properties of crude oils around the

applying a thermodynamically-consistent
framework to accurately estimate key physical properties of the individual surrogate
model molecules and their mixtures, and

developing an
engineering methodology to robustly identify compositions of the surrogate model
molecules from commonly available assay data.  

Implemented in commercial software, molecular based assays
have been successfully validated against hundreds of crude assays worldwide including
many heavy crude oils. The methodology not only facilitates extraction of
molecular insights for the crude oils, it also allows self-consistent
interpolation, extrapolation, and prediction of a comprehensive set of physical
and chemical properties such as distillation yield curve, API gravity,
viscosity, Reid vapor pressure, Watson K, PNA contents, sulfur content, nitrogen
content, acid content, asphaltene content, C/H ratio,
etc. The methodology offers a superior alternative to conventional and
empirical assay characterization technologies. Furthermore, it suggests a
highly prized new frontier over the traditional ?pseudocomponent,?
?hypothetical component,? or ?micro-cut? modeling approach for planning,
scheduling, and process simulation of oil production and petroleum refining