A pure process
Fuel production enhanced with nuclear magnetic resonance spectroscopy
By Paul Giammatteo, John Edwards, and Tal Cohen
Clean fuels regulations can impact a refiner’s ability to maximize product draws, yet refiners undercut these draws to ensure minimum error in downstream product blends. Online, high resolution nuclear magnetic resonance (NMR) emerged in refinery processes in 1995 with the installation of two units in Texaco’s Los Angeles area refinery. Since then, integrating proper NMR hardware with properly designed applications has proven economically and operationally beneficial.
Online NMR-based analytical and process control strategies enhance clean fuels production directly at the control display unit (CDU). As ultra-low sulfur requirements force stringent constraints on hydrotreater feeds, NMR can maximize clean diesel recovery by enabling closer-cut point control in the CDU mid-section, thereby reducing dibenzyl thiophene breakthrough.
Integrating proven NMR technology with a focused measurement and control strategy enables crude unit operations to cut chemically closer to the hydrotreater constraint limit. And whether an online NMR sees use as a multi-parameter analyzer or integrates into advanced process control (APC) and optimization, the accuracy, reliability, and performance of online NMR delivers true value throughout the refining process.
As a spectroscopic process analyzer, NMR is similar to other spectroscopic-based analyzers such as near infrared (NIR) and Raman. Typically, by interfacing a spectroscopic analyzer’s output with chemometrics, you can predict detailed chemical and physical information. Obtain chemometric models by correlating a series of spectra with known laboratory analyses. Model performance (accuracy, stability, robustness, etc.) is dependent on these key factors:
- Chemistry/molecular functionality the spectroscopy measures is related to the parameter you are measuring (density, distillation, etc).
- Variability observed in the spectroscopic measurement is related to the sampling variability the primary laboratory analysis measures.
- Primary lab data meets industry standards for accuracy, reproducibility, repeatability, etc.
Predominantly NMR-based chemometrics use partial least squares regression (PLS). These PLS equations generate the parameter measurements (such as density, distillation points, and octane numbers). Typically, each NMR measurement predicts multiple parameters (from 4 to 60) simultaneously.
Clean fuels regulations in the European and American markets have had a substantial impact on a refiner’s ability to maximize product draws at the refinery front end. Extremely low sulfur requirements for gasoline and diesel have resulted in refiners now being more constrained at the hydrotreater. Lack of reliable, focused, measurement and control of critical CDU product draws has forced many refiners to significantly undercut these draws in order to ensure minimum error in the final product blends, especially with respect to total sulfur.
Depending on a refinery’s crude supply and CDU capacity, a conservative estimate of 300-500+ barrels per day of loss in diesel production is typical. With an average of $25 - $35 per barrel margin loss, the economic impact of these clean fuels regulations is substantial. Current and planned online NMR-based analytical and process control strategies for enhanced diesel recovery at the crude distillation unit maximize clean diesel recovery by enabling closer-cut point control in the mid-section of the CDU. This reduces critical sulfur molecular breakthrough downstream to the hydrotreater.
The fundamental properties of crude and petroleum distillation are directly dependent on the distribution of carbon-hydrogen bond types in the numerous molecules that make up these samples. Aromatics, aliphatics, and olefins are easily identifiable. Polynuclear aromatics are easily distinguishable from single-ring aromatics in the NMR spectrum. NMR measurement of this molecular/chemical information uniquely characterizes a wide variety of petroleum streams.
Process NMR for modern refinery
Over the past 14 years, installations of online NMR analyzers have occurred in numerous refinery and petrochemical applications including:
- Crude feed property measurements and crude blending control
- Crude distillation unit product stream measurements and control
- Fluid and resid catalytic cracking unit measurements and control
- Hydrotreating, hydrocracking, and reforming measurements and control
- Product blending: gasoline, jet/kero, diesel
Several of these process NMR applications see use to monitor the crude distillation unit.
Some systems monitor distillates only (kerosene, diesel, automotive gas oil [AGO]), others monitor naphthas (flammable liquid mixtures), and distillates (light/med/heavy naphtha, kerosene), while others monitor feed and product streams (heavy naphtha, kerosene, crude). Several other installations use one NMR on multiple process units. Linked in with advanced process control, 24-hour distillation predictions from one NMR monitoring process streams from two separate CDUs and a hydrotreater provides feedback control to the crude units as well as feed-forward and feedback control to the kerosene hydrotreater.
When a single NMR monitors six process streams from the crude unit to product diesel, it enables constant feedback and feed-forward control for the entire diesel manufacturing process (CDU diesel rundown, coker diesel rundown, hydrocracker feed, hydrocracker product, diesel blend header, and finished diesel). Especially critical is the ability to measure on the feed and product streams of the hydrocracker. The combination of predicting traditional parameters (density, cloud point, and pour point) as well as quantifying the extent of the hydrocracking via the direct measurement of the olefin content (from coker diesel) and aromatic content can offer considerable improvements and returns at the hydrotreater.
The NMR model sets are the same for all six streams; there is one model for each specific parameter for use on all streams and not six separate models of the same parameter for each individual stream.
Benefit in combinations
Combining NMR, APC, and optimization enable operators of a crude unit to operate much closer to constraints with significant economic benefit. Initially, published total benefits of $7-12K/day from online operations were derived primarily from improved measurement of kerosene freeze point ($600K annually) from direct control of kerosene freeze point. With the tighter and more frequent NMR freeze point measurements, plant engineers captured additional off-line benefits by moving circulating reflux flows from their traditional operating levels and thereby avoided unnecessary opening of columns and exchangers during short crude transitions.
Further, NMR-based crude property measurements were used in a feed-forward mode.
Measuring key total boiling point cut points, American Petroleum Institute gravity, and water content on the crude feed (before the preheater and desalter) enabled the operators to further push constraints. No crude and/or crude blend assay data from this refinery were ever incorporated into the NMR models.
In this application, the NMR feed-forward predictions accurately measured the crude blend properties even though no individual crude, crude blends, or on-site lab measurements saw use in the NMR crude property models.
Current strategies are similar to those discussed in the diesel example. Strategic, accurate measurements of conventional parameters (distillation points, cloud points, freeze crude unit mid-section facilitate control of the middle distillates. With accurate measurement of the upper distillation points (T90—final boiling point) from the lower boiling distillates (i.e. kero) with accurate measurement of the lower distillation points (initial boiling point—T10) from the next higher distillate (i.e. diesel), NMR/APC can control the tail regions of each cut and increase lift of the typically higher value lighter fraction (i.e. kero in diesel, diesel in AGO) from the heavier cut. However, most refiners remain constrained at the hydrotreater with respect to sulfur content, and more specifically, with respect to dibenzothiophene content.
The dibenzothiophene class represents some of the most difficult molecules to hydro-treat for effective sulfur removal. Inability to remove these compounds presents downstream difficulties as these species can be present at sulfur concentration levels approaching the final low sulfur specifications for the finished products. Rather than risk exceeding sulfur limits in final product blending, many refiners are significantly undercutting at the front end. Total sulfur measurement, typically by x-ray-based analyzers, usually see applications on various process streams, however, they are not perfect in their application on line.
In an effort to improve understanding of where these difficult sulfur species reside in the various process streams, we have expanded GC simulated distillation (SimDis) laboratory measurements. Fast SimDis methodologies have reduced per-sample diesel analyses times significantly (8-12 minutes vs. 45+ minutes) as well as enabled the simultaneous measurement of carbon number, sulfur, and nitrogen species.
The co-elution of sulfur and nitrogen containing molecules can be a function of carbon number. These techniques have confirmed the dibenzothiophene components retention times corresponding to the C19/C20 region of the SimDis chromatogram.
Spectro-molecular control is our terminology for integrating a unique NMR measurement strategy to enable crude-unit operation to cut and control CDU diesel production as close to the dibenzothiophene distillation limit as possible without the direct measurement of these compounds.
Online NMR offers a viable means for accurate process analysis and control on a variety of upstream and downstream refinery applications from crude feed to finished gasoline property measurements. NMR delivers real-time accuracy with proven control and economic benefits. At the front end of the refinery, NMR measurements are not impacted by sample color, water content, viscosity, and density for crude and lighter streams. Proper sampling technologies enable the analyses of heavier materials such as asphalts, coker feeds, and residuals. With strategic implementation at the crude unit mid-section, online NMR will enable the recovery of additional 300-500 barrels per day of critical distillate products from a typical 100,000-barrel-per-day crude distillation unit.
ABOUT THE AUTHORS
Paul J. Giammatteo, Ph.D. (firstname.lastname@example.org), is principal, and John C. Edwards, Ph.D. (email@example.com) is manager of process and analytical NMR services at NMR Associates, LLC in Danbury, Conn. Tal Cohen (firstname.lastname@example.org) is executive vice president of research and development and BD at Modcon Systems.