Chemometric Hydrocarbon Stream Classification for Refinery Operations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting attributes of hydrocarbon streams in commercial refineries are slow and costly, taking days or weeks, which hampers rapid identification of fouling and corrosivity, leading to increased operating costs and maintenance needs.
Innovation Solution
A process using analytical methods like mid-infrared spectrometry and nuclear magnetic resonance spectroscopy to transform spectral data into wavelet coefficients, then employing a genetic algorithm to classify hydrocarbon streams based on attributes such as fouling propensity and corrosivity, enabling rapid prediction and optimization of refinery operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional experimental assays are used to predict hydrocarbon stream attributes, then measurement precision is improved, but loss of time increases significantly (days or weeks)
Solution Approach 1:
The patent replaces conventional mechanical/chemical experimental assays with a spectroscopic analysis system using near-infrared (NIR) or mid-infrared (MIR) spectrometry combined with chemometric modeling. This substitution enables rapid prediction of hydrocarbon stream attributes (fouling propensity, corrosivity, sulfur content) within hours rather than days, while maintaining measurement precision through validated chemometric models correlated with reference laboratory data
Solution Approach 2:
The patent introduces chemometric models as an intermediary between spectral data and attribute prediction. The models include preprocessing steps (scatter correction, derivative processing, smoothing), wavelength selection algorithms (genetic algorithms, interval partial least squares), and prediction algorithms (partial least squares regression, principal component regression) that mediate between raw spectral measurements and final attribute predictions, enabling rapid yet accurate assessment
2Measurement precision
If conventional experimental assays are used for hydrocarbon stream analysis, then measurement precision is improved, but productivity decreases due to slow processing
Solution Approach 1:
The patent replaces time-consuming conventional experimental assays with infrared spectrometry and chemometric modeling, reducing analysis time from days/weeks to hours while maintaining measurement precision through validated models. This enables real-time or near-real-time monitoring and decision-making in refinery operations
Solution Approach 2:
The patent performs preliminary action by pre-developing and validating chemometric models using reference laboratory data before actual production use. Once models are trained and validated against conventional assay data, they can rapidly predict attributes without requiring repeated conventional assays, significantly improving ongoing productivity
3Productivity
If rapid assessment methods are implemented, then productivity is improved, but measurement precision may be compromised
Solution Approach 1:
The patent uses chemometric models as intermediaries that maintain measurement precision in rapid assessment. These models are trained and validated against reference laboratory data, creating a robust mapping between spectral data and actual attribute values. The models include error estimation and quality control measures to ensure prediction accuracy meets operational requirements
Solution Approach 2:
The patent implements feedback mechanisms where model predictions are continuously validated and refined. Spectral data is collected, processed through chemometric models, and predictions are compared against actual performance data or periodic reference assays. This feedback loop enables continuous model improvement and ensures measurement precision is maintained over time
Data Source
AI summary
A process for converting a first hydrocarbon feed stream to one or more liquid transportation fuels in a petroleum refinery where the feed stream is analyzed by at least one analytical method to produce data that is transformed to wavelet coefficients data. A pattern recognition algorithm is trained to recognize subtle features in the wavelet coefficients data that are associated with an attribute of the feed stream. The trained pattern recognition algorithm then rapidly classifies potential hydrocarbon feed streams as a member of either a first group or a second group where the second group comprises hydrocarbon feed streams where the attribute or chemical characteristic at or above a predetermined threshold value. This classification allows rapid decisions to be made regarding utilization of the feedstock in the refinery that may include altering at least one variable in the operation of the refinery.