Nov-2008

Advances in delayed coking heat transfer equipment design

The delayed coking process is used to crack heavy oils, normally vacuum residue, into more valuable light liquid products, with less valuable gas and solid coke as by-products. The first delayed coking plant was built in 1930.

Kenneth A Catala and Mark S Karrs, Lummus Technology Heat Transfer
Gary Sieli and Al Faegh, Lummus Technology

Article Summary

While the delayed coking process has been evolving for 78 years, the last few years have seen changes in feedstock that have had a major impact on the design and operation of delayed coking units. This paper addresses how these changes affect the heat transfer equipment and the coker heater in particular.

The increasing worldwide demand for liquid petroleum products; the resulting increase in the price of crude oil; the differential cost between light and heavy crudes; and the shift toward processing lower quality, lower cost crudes as well as tar-sands bitumen have created a need to convert larger quantities of vacuum residue to higher valued liquid products. This has resulted in an increase in the application of delayed coking.

These lower cost feedstocks typically have higher metals and asphaltenes, and some have higher total acid numbers. The impurities in the crude tend to concentrate in the residue streams, making the resulting coker heater feed “dirtier”, with a higher fouling tendency.

Economics are driving coker operations toward lower coke drum pressures to increase liquid yields and lower recycle rate (throughput ratio). Economics are also dictating larger diameter coke drums (9.15m). Totally enclosed, fully automated top and bottom unheading devices have shortened coke removal time. This has resulted in shorter cycle times, which then require higher severities to produce coke with the same volatile combustible material (VCM). All these factors contributed to the need for larger capacity and higher severity coker heaters.

Delayed coking is an endothermic thermal cracking process that requires large quantities of heat to be supplied to the reacting feedstock at temperatures of 500°C+. All the process heat is supplied by the coker heater. The fluid must move through the heater to the coke drum as quickly as possible to minimise the amount of coke that is deposited in the heater tubes and downstream piping. This coke causes increased tube metal temperature and pressure drop and, moreover, the rate of coke deposition determines the coker heater run length. The heater design must therefore consider feedstock quality, operating conditions, need for future expansion and possible changes in feedstock characteristics.

Changes in heater design
Lummus’ first delayed coker heater designs date back to 1940 and consisted of small, box-shaped heaters with double rows of tubes suspended from the roof and a single row of tubes on each wall (Figure 1). The process fluid was heated only in the radiant section. The double row of tubes was backed with tiles that increased heater efficiency by increasing the convective heat transfer to the tubes. This general design was used for more than 20 years and, even though the shape of the heater later evolved, the double row of tubes suspended from the roof (with tiles backing them) continued to be used until about 1990 (Figure 2.).

Prior to about 1990, delayed coking units were small: few were larger than 1.1 million metric tons per annum (MMTA). Most had a single pair of coke drums, one being filled while the other was de-inventoried. All the heaters were single fired (ie, tubes were heated from one side only). Firing from only one side results in a large difference between the peak heat flux and the average heat flux. The subsequent high peak heat flux results in high film temperature, which leads to an accelerated laydown of coke inside the heater tubes. The average heat flux rates for coker heaters was therefore lower than for general refinery process heaters, in the range of 24 400 kcal/hr m2. The low average heat flux rates required relatively high residence times, which also promote coke deposition. Early designs had a number of small liberation gas burners in the floor and the height of the furnace could be small because the flame lengths were only about half as long per unit heat release as modern, staged-fuel, low-NOx burners.

After 1990, there was a significant new development in the design of the coker heater: the double fired coker heater design. This patented design1 put the coil in the centre of the box and the burners against the wall so the tubes could be heated from both sides. The average heat flux rate could then be increased by 50% with no increase in peak heat flux or film temperature, as illustrated in Figure 3. When the heat flux at points 1 and 7 of the tube heated from both sides equals the heat flux at point 1 of the tube heated from one side, the average is 50% higher. The resulting reduction in coil length reduced pressure drop and residence time and allowed for increased capacity per coil. The reduced residence time in the coker heaters results in a slightly higher outlet temperature required for the same conversion and yields. The patent has been licensed by a select group of furnace designers.

Contemporary heater designs
Contemporary delayed coking units have been designed for capacities larger than 6.0 MMTA and have multiple pairs of large-diameter coke drums. The coker heaters are designed for very high capacity (Figures 4 and 5) in order to limit the number of heaters per pair of coke drums: the resulting heater designs therefore have higher mass velocity and pressure drop. Heater tube diameter has not increased because increasing the tube diameter would raise the film temperature and tube metal temperature, thus accelerating radiant coking and reducing heater run-length. In the resulting tall, narrow firebox designs, the burner operation and flux profile become critical to achieving the required run-length.

Coil materials
Coke deposition inside the heater tubes leads to an increase in the tube metal temperature and pressure drop. The heater run-length will normally be controlled by the End-of-Run (EOR) metal temperature. Older units that processed cleanfeedstocks generally used 21/4 Cr-1Mo or 5 Cr-1/2 Mo tubes. The majority of contemporary units have been engineered with 9% Chrome-1% Molybdenum (T9) tubes due to higher sulfur and higher EOR temperatures.