Current Technique for Feedstream Analysis:
Analysis Performed – Refractive Index, Distillation, Specific Gravity
Calculation Obtained – Watson K-Factor
Outcome: aromatic carbon number, aromatic hydrogen number, total hydrogen content
Proposition: Detailed hydrocarbon analysis for kinetic model development.

Our experience is that an improved and useful analysis can be obtained from NMR
analysis. There are several ways to approach the NMR analysis and the
chemometric approach to correlating NMR data to physico-chemical parameters of
use to process control.

RCC Feedstream Analysis by 1H and 13C NMR: Multivariate Prediction of Chemical and Physical Properties

Presented at the 236th ACS National Meeting, Philadelphia PA, August 17-21, 2008

John C. Edwards Ph.D.
Process NMR Associates LLC
87A Sand Pit Rd, Danbury, CT 06810

Jincheol Kim,
SK Energy Co., Ltd, SK Energy Technology Center,
140-1, Wonchon-dong, Yuseong-gu, Daejeon 305-712, Korea

Summary of Slides

1) 60 MHz process NMR data available from online NMR unit.

2) 300 MHz 1H NMR data available from standard NMR experiments on laboratory NMR system at Process NMR Associates

3) Expansions of 1H NMR data on RCC Feeds

4) Calculated 1H NMR Parameters Represented as an alternative “spectrum” for use in chemometric modeling and linear regression.

5) 13C NMR Data obtained on RCC Feeds

6) Expansions of 13C NMR Data

7) 13C NMR and Average Molecule Parameters Calculated from 13C Spectrum

Calculated 13C NMR Parameters Represented as an alternative “spectrum” for use in chemometric modeling and linear regression.

9) Correlations of Spectra and Calculated Parameters to Physico-Chemical properties of RCC Feeds

10) Correlation of Spectra and Calculated Parameters to Density

11) Variable Selection for Linear Correlation of Calculated NMR Parameters to Physico-Chemical Properties of RCC Feeds.

12) Correlation of Calculated 13C NMR Parameters (C-Type and Average Molecule) with both high resolution 1H and 13C NMR spectra. Models created can be utilized by personnel with no NMR experience to calculate NMR parameters directly from the spectrum without any prior knowledge of integrations or calculations to be performed.

Summary
Chemical and Physical Properties of RCC Feedstreams can be determined
by 1H NMR (at 60 and 300 MHz) and by 13C NMR
H-Type and C-Type Parameters do not provide as good a correlation as is observed
by full spectrum regression. This is due to loss of resolved chemical shift information
when the spectrum is reduced to larger integral areas.
1H NMR can be combined with PLS regression modeling to provide detailed carbon
type analysis for RCC Feeds
Regression analysis of 13C NMR data can be utilized to fully automate the prediction
of 13C NMR type analysis : reducing the necessity for considerable knowledge and
analysis time on the part of the analyst.

For further Detail Contact John Edwards