"More Gasoline from Crude," Feature Article, December 2003

more gasoline
from crude

A watch on the telltale signs of imminent flooding aims to protect a refinery column.

by John DeGaspari,
Associate Editor

When it comes to quenching the public's thirst for gasoline, oil refiners in the United States find themselves between a rock and a hard place. Average gasoline consumption in the country has been climbing steadily for the last decade—at a rate of almost 2 percent a year—while there has been virtually no growth in refinery capacity.

Although 145 petroleum refineries operate in the 50 states, no major refinery has been built in the country in the past 25 years, according to Ron Planting, manager for information and analysis for the American Petroleum Institute in Washington, D.C.

Meanwhile, refineries' average utilization rate—the percentage of total capacity that is being used to make products—runs well above 90 percent. For the first eight months of this year, refineries averaged 92.4 percent of capacity, and it is not unusual for the rate to climb four or five percentage points while the demand for gasoline is most acute, from June through August.

Pushing a refinery distillation column to its limits to meet heightened demand is a tricky undertaking that could result in an unwanted phenomenon called column flooding. As petroleum is distilled, various components of the crude oil remain separated from each other on porous trays inside the column. During a flood, butane, gasoline, asphalt, and other distilled crude oil products begin to commingle with each other inside the column.

A flood results in off-spec products and, because the distillation will have to be redone, wastes the energy that went into boiling the crude in the first place. It can also damage equipment and require the tower to shut down. One way that refiners counteract column flooding is to reduce the amount of crude flowing in, until the system returns to steady state.

Yet refineries have a significant incentive to boost their production during times of high demand. When demand pushes up crude prices, the value of that barrel—and the profits from it—also rise. There's also the question of energy efficiency. Petroleum distillation is an energy-intensive process, consuming about 40 percent of the energy used in oil refining. Improving energy efficiency by a couple of percentage points could save at least half a trillion British thermal units per year.

A start-up company in St. John, Ind., 2ndpoint Inc., says it has developed pattern-recognition software that warns of column flooding. George Dzyacky, who developed the predictor, has spent 25 years in fluid catalytic cracking refining operations.

Crude oil is made of a variety of hydrocarbon products, ranging from light components such as gasoline to heavy ones, including asphalt. In a refinery, crude oil is fed into a distillation column, where it is boiled to separate it into usable products. The heavier the component, the closer it is to the heat source.

Operators know that columns have a severity limit or flooding threshold. Going beyond that point means that the column may go into flood. The traditional way to determine that threshold is to look at delta pressure, or the difference in pressure between the top and bottom of the column. Delta pressure is an inference point of flooding that column operators use as an upper safe limit.

Refineries try to boost their output as gasoline margins rise. A column operator typically tries to nudge production upward, flirting with a flood to take advantage of higher gasoline value. In the worst case, the delta pressure increases exponentially as the column begins to flood. Complicating matters further, the delta pressure point varies, depending on the grade of crude oil. Lighter grades, such as West Texas Intermediate, would tend to flood sooner than a heavier variety of crude oil.

Dzyacky arrived at his idea for a predictor while he was employed as a console operator in the distillation section of a large refinery. "I was operating columns that were flooding constantly," he said. He focused on changes in process variables in the distillation column and looked for patterns that would act as advance signals that the column was moving toward a flood. Dzyacky said that he used the distillation column to verify the accuracy of the flooding predictor, and worked with the control engineer at the refinery to allow the predictor to automatically reduce the column severity at the earliest signs of flooding. The flooding problem disappeared, he said.

Separations Research Program equipment tested the effectiveness of pattern recognition software to predict distillation column instabilities.

The flooding predictor is based on variables of flow, temperature, and pressure measured by standard instrumentation already on the column. The flooding pattern emerges in the rates of change in the column variables. Dzyacky has identified two flooding mechanisms or patterns. One, the more typical, occurs when too much liquid is introduced into the column. The other is triggered by an excess of vapor, which results from too high a temperature at the bottom of the column.

Dzyacky said that, by knowing the actual beginnings of flooding, the operator knows how much he can push the column to distill more products.

The flooding predictor will identify the incipient flood point regardless of the grade of crude oil and will provide the operator with information to reduce throughput when necessary. The key, Dzyacky said, is that the pattern recognition technique provides the operator with the means to identify the changes in the column in real time.

Dzyacky estimates that refineries could push their throughputs from 90 percent to 99 percent of incipient flooding. In other words, it would allow the column to be operated that much closer to the true flood point than possible with the delta pressure method.

The U.S. Department of Energy has awarded a cooperative research and development agreement to 2ndpoint. Under part of the agreement, Dzyacky has subcontracted the University of Texas to test the column flooding predictor at its Separations Research Program facility.

Dzyacky said the Separations Research team has de- monstrated that the flooding predictor works. He is
now working with them to identify a scientific basis for the technology.

The Motiva refinery in Norco, La., has signed on as a partner to run the flooding predictor on a commercial-scale distillation column. Data from those tests will be fed into the ongoing work at the University of Texas.

Bruce Eldridge, director of the Separations Research Program, said that the flooding predictor is a potential improvement over state-of-the-art process control based on a predictive model of the distillation column. The flooding predictor looks at what is actually happening and reacts to changes in the process, he said.

Eldridge noted that the demonstration tests were "idealized," but believes that the predictor has promise. He called the flooding predictor a tool that, if validated, could save a significant amount of energy and increase throughput on some facilities.