(6e) Loss of Float Valves Causing Flood

By Matthew Olsson and Henry Kister, Fluor

An Atmospheric Crude Tower has functioned well since 1989. All the reflux to the top of the tower was supplied by a top pumparound (TPA) loop. In an attempt to lower the naphtha 95% point, the TPA was supplemented by injecting cold reflux from the overhead receiver. As this reflux is believed to contain water, a filter and coalesce were added to remove this water prior to tower entry. However, frequent plugging of the filter necessitated bypassing, causing corrosive water to enter. In the tower it hydrolysed chlorides, forming HCl, that caused extensive corrosion and damage to the TPA trays, resulting in flooding and high naphtha end points.

Pressure drop measurements showed flooding taking place both in the 4-pass TPA trays and also in the 2-pass fractionation section below. A gamma scan confirmed flooding in the upper four-five trays in the fractionation section. Upon shutdown and inspection, extensive corrosion and damage were observed in the TPA trays, as expected. The surprise was that the upper trays in the fractionation section were found clean, but missing most of the valve floats. The missing valve floats were not found anywhere, probably washed away or corroded away. This increased the open area on the trays and should have moved the trays further away from flood, but the flood only occurred near the end of that run, and did not occur in the earlier runs, when the valve floats are believed to have been in place.

This raised the question: could losing the valve floats have promoted flood in the fractionation trays? In an endeavour to gain better understanding to the observation, we applied the Fluor Multipass Maldistribution Model (MMM) to see whether these observations can be modelled. In a previous article¹, we applied the same model to discover multiple steady state vapor/liquid distributions in 2-pass moving valve trays at turndown.

Due to the symmetry of 2-pass trays, a perfect split of both vapor and liquid between the passes is always one possible and well-known steady state distribution. Our analysis shows that uneven or complete float removal on the trays can channel the vapor and liquid in different directions, which may account for the premature flood and above observations.

The trays were repaired and the tower restarted. To prevent recurrence, the refinery stopped supplementing the TPA reflux by injecting cold reflux, even for short periods, and guidelines were drawn for the maximum moisture content in the reflux. The tower has been performing well since.

This experience teaches how damaging the cold reflux can be to column internals in the top sections. On a different topic, it teaches that losing valve floats can lead to channelling and premature flood.

1. Olsson, M., and H. Z. Kister, Multiple Steady States in 2-Pass Hydraulics of Moving Valve Trays” in Kister Distillation Symposium 2017, Proceedings of Topical Conference, AIChE Spring Meeting, San Antonio, Texas, March 26-30, 2017.