Coke Drum Skirt Insulation Thickness: An Investigation on Cycle Life

Abstract A significant number of variables have been investigated in the design and analysis of coke drum skirt-to-shell connections. These variables include: welds vs forgings; geometry of the forging; hot box sizing; and slots (including distance to the top of the keyhole, number of slots, spacing, width, and keyhole radii). However, consistent throughout the literature has been the assumption that the thickness of the insulation on the exterior of the skirt is equal to the thickness of the insulation on the exterior of the coke drum itself. This assumption has been reasonable, since it simplifies the design of the insulation system by keeping the thickness the same. Furthermore, during the warm-up and feed-in phase of the coking cycle, when the skirt temperature lags that of the vessel as the hot oil level rises, it is beneficial to transfer heat to the skirt (typically via radiation through the hot box) while minimizing heat lost to the ambient through convection. During the steam cooling and quench phases of the coking cycle, the skirt temperature similarly lags that of the vessel shell. However, as the temperatures are significantly lower during this phase, the effectiveness of the radiation heat transfer (being a function of the absolute temperature raised to the fourth power) is reduced, leading to a larger temperature difference between the shell and the skirt. In an attempt to enhance heat loss from the shell during the cool-down and quench phases of the coking cycle, the thickness of the skirt insulation was investigated as a possible variable for consideration. Lower insulation thickness values were investigated to determine the single-variable response of a forged and slotted skirt with keyholes design. The peak stress ranges in both the crotch region of the skirt forging and the top keyhole were evaluated as dependent variables, compared to the independent variable of skirt insulation thickness. Thickness values were evaluated down to zero. Two ambient conditions were also evaluated: natural convection (zero wind) and forced convection. The results show that decreasing the skirt insulation thickness results in a lower temperature of the skirt during the quench phase, leading to a lower temperature difference between the shell and the skirt. This translates to a lower peak stress in both the keyhole and the skirt crotch. A significant reduction in peak stresses is calculated when the skirt thickness is reduced from 6 in (152.4 mm) to 1 in (25.4 mm) and even to no insulation. The results also show that decreasing the skirt insulation thickness from 6 in (152.4 mm) to 2 in (50.8 mm) has minimal changes in the peak stress in both the keyhole and skirt crotch stresses. The results also show a surprising reduction in peak stresses when the insulation thickness is reduced further to 1 in (25.4 mm). However, when the skirt insulation is completely removed, the skirt temperature is marginally lower during the hot oil fill phase, leading to a slightly higher temperature difference, and hence a higher peak stress in both the keyhole and the skirt crotch. There appears, however, to be an optimal insulation thickness, that is lower than the thickness for the coke drum shell itself. These results show that benefits can be obtained by varying the skirt insulation thickness. However, these benefits of reduced peak stresses, and hence increase in fatigue life, are dependent on the actual ambient conditions. Therefore, it is strongly recommended that analyses be performed for each specific site installation, with an appropriate assessment of the range of ambient conditions, to determine optimum insulation thickness on skirts.