Prediction of fuel properties
Inventors
Morris, Robert E. • Hammond, Mark H. • Johnson, Kevin J. • Cramer, Jeffrey A.
Assignees
Publication Number
US-10288588-B2
Publication Date
2019-05-14
Expiration Date
2035-09-29
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
A system is described that includes a known fuels database of data from gas chromatography-mass spectrometry analyses of a library of fuels with known fuel properties for a multiple known fuel samples. Gas chromatography-mass spectrometry equipment can acquire gas chromatography-mass spectrometry data for an unknown fuel sample. A metaspectrum module can accept and transform the gas chromatography-mass spectrometry data collected by the gas chromatography-mass spectrometry equipment for the unknown fuel sample into a single metaspectrum for the unknown fuel sample, wherein the metaspectrum is a quantitative representation of every compound detected in the unknown fuel sample. A correlation module can correlate the metaspectrum for the unknown fuel sample to a plurality of fuel properties of known fuel samples using a regression model to predict fuel properties for the unknown fuel sample. A reporting module can report the fuel properties for the unknown fuel sample to a user.
Core Innovation
The invention describes a system and method for predicting fuel properties using data obtained from gas chromatography-mass spectrometry (GC-MS) analyses of fuel samples. The system includes a database of known fuels with GC-MS data and associated fuel properties, acquisition of GC-MS data for unknown fuel samples, transformation of this data into a single metaspectrum representing every detected compound, and correlation of the metaspectrum to known fuel properties using regression models to predict properties of the unknown fuels.
The problem addressed arises from the need to reduce time, manpower, and sample amounts required to measure ASTM standard fuel properties. Previous methods commonly relied on near-infrared (NIR) spectroscopic data, which do not directly provide chemical composition and require extensive modeling adaptation to accommodate different fuel types, especially with increasingly diverse and unpredictable non-petrochemical fuels.
The invention solves this by providing a granular chemical-by-chemical assessment of fuel composition using GC-MS data and chemometric modeling. It produces a quantitative metaspectrum of compounds detected in fuels, enabling multivariate regression modeling to robustly predict multiple fuel properties. This approach provides improved adaptability to future fuels and more accurate property prediction compared to indirect spectroscopic methods.
Claims Coverage
The patent contains multiple independent claims covering computer-implemented methods and systems for predicting fuel properties using GC-MS data transformed into metaspectra and correlated via regression models, including blending functionality.
Use of metaspectrum for quantitative representation of compound abundances
The method and system transform GC-MS data from an unknown fuel sample into a single metaspectrum that quantitatively represents the abundance of every identified compound, using a metaspectrum module implemented in a computer system with executable instructions.
Correlation of metaspectrum to known fuel properties using regression models
A correlation module correlates the metaspectrum to a plurality of fuel properties from known fuel samples using regression models to generate fuel property predictions, implemented in a computer system with executable instructions.
Identification and quantification of chromatographic peaks against a mass spectral database
The metaspectrum module locates chromatographic peaks in the GC-MS data, identifies each peak with a mass spectral database, and calculates peak areas to determine amounts of each compound, optionally disregarding peaks below a threshold area for match quality.
Mathematical blending of GC-MS data from two fuels to predict composite blend properties
A blending module enables mathematical blending of GC-MS data from two fuels to predict fuel properties and report the composition of the composite blend.
The independent claims collectively cover a computer-implemented approach that transforms GC-MS data into compositional metaspectra, correlates them using regression models to predict fuel properties, and includes identification, quantification, and blending capabilities to enhance robust fuel property prediction for unknown samples.
Stated Advantages
Enables rapid, accurate prediction of multiple fuel properties from a single small fuel sample using a single analytical technique (GC-MS).
Improves adaptability to future fuels, including non-petroleum and blended fuels, by relying on direct compositional analysis rather than indirect spectral data.
Reduces the need for extensive model adaptation or redevelopment for different fuel types.
Allows for simulation of blended fuels to predict composite blend properties, aiding fuel certification and compatibility assessment.
Provides a user-friendly standalone software tool to perform automated data processing and property prediction without requiring operator intervention.
Documented Applications
Point-of-use performance characterization of fuels.
Point-of-production performance characterization of fuels.
Prediction of overall suitability for Navy fuel certification procedures.
Short-term and long-term fuel stability prediction and monitoring.
Accelerated assessments through multiple simultaneous property predictions.
Lowering per-fuel analysis costs.
Investigation of blended fuel components and possible adulterants.
Prediction of blending stocks necessary to attain desired performance characteristics.
Interested in licensing this patent?