2so 6PM_ STfllpPtft


ho I

INSTALLATION A 10-foot ID caustic wash tower (Figure 12.16.1) which removes a small quantity of C0_ and traces of H-S from light j hydrocarbon gases at about 200-300 psig. The tower contains 40 single pass valve trays at 24 inch tray spacing. Each tray is equipped with a 6 inch wide downcomer. Overheads from the column flows via a knockout drum to a compressor. The plant was at its initial startup.

PROBLEM At low plant rates the column operated well. As rates were increased to 60-70 percent of design, massive entrainment into the knockout drum was observed. This could not be tolerated because of the risk of damaging compressor blades.

INVESTIGATION The symptoms suggested premature flooding. Column j differential pressure was not being measured. Calculations indicated i the column was adequately sized. However, it was suspected that some 1 degradation of the caustic solution could have occurred near the bottom of the column, giving rise to the formation of a viscous emulsion. This, together with the extremely, small downcomers (2 percent of tower area), was believed to have caused premature flooding.

SOLUTION A larger downcomer area was sought to overcome the premature flooding problem. Retraying with larger-downcomer trays appeared the logical solution, but would have caused a 6-weeks delay ! and a week long shutdown. Downtime was extremely expensive and undesirable. As an alternative solution, the plant and startup team then decided to construct home-made downpipes (Figure 12.6.2) in the bottom few trays, where liquid load was greatest and potential for emulsion formation was highest. The pipes were constructed by "burning" a few holes in the tray near the outlet weir, and fitting 6-inch vertical pipes in them. The pipes pretruded a small distance above the tray, simulating a weir, and extended to a small distance above the tray below, simulating a downcomer apron area. The decision was made to go ahead because there was little to lose, the worst that could happen was mechanical damage to a few trays, but new trays would have been needed anyway to enlarge the downcomers.

IMPLEMENTATION The bottom few trays were modified (Figure 12.16.2). This was done in a 3-day shutdown. When the column came back on line, the problem completely disappeared, and the column attained full rates without any problem.

Was this article helpful?

0 0

Post a comment