Table 1 dispels any doubts that in trays 7 and 9, in both tests, the froth heights were taller, and the froths were denser on the SW. The density counts show that the NE side of Tray 7 was almost dry in the flooded scan. Tray 9 was also low on liquid. In the ‘normal operation’ scan, trays 7 and 9 had more liquid on the NE side than in the flooded scan but significantly less liquid than on the SW. On the even-num- bered trays 6 and 8, the densities again were higher on the SW than on the NE. This analysis established that on trays 6-9, especially the odd-numbered ones, there was a signif- icantly larger quantity of liquid descending on the SW than on the NE. This maldistribution observation concurs with the plant observation that the temperature indicator on the NE read 15ºF hotter than the temperature indicator on the SW. Together, they verified maldistribution between the SW and NE panels, with more liquid descending on the SW. Kister and Olsson⁵ described a post-mortem analysis of an uneven loss of valve floats from an upper fractionation tray in another atmospheric crude tower with two-pass trays. One scenario investigated was of a tray that lost 90% of the valve floats on one panel and none on the other. The mod- elling results showed that a relatively small vapour maldis- tribution (52.8% to the panel that lost the floats, 47.2% to the one that did not) generated a large liquid maldistribution (67.3% to the side that lost the floats, 32.7% to the other) on the two trays immediately below. On the third tray down, the pattern reversed itself, and the liquid maldistribution changed sides and receded in magnitude. Budget and schedule constraints precluded applying a similar analysis here, but the results of the previous anal- ysis 5 can be qualitatively extended to the current case. In both cases, there was a vapour-dominated distribution. There is one significant difference: In the wash section of the current tower (beneath tray 10) the liquid flow rates are much lower, so the resulting liquid maldistribution would be much more severe. New theory: Plugging on one or more trays This theory postulated plugging on one or more trays. Plugging on tray 10 will explain the flooding above tray 10 seen on the scans. However, the theory also needed to explain the maldistribution, the unpredictable nature of the flood, and the vapour and liquid sensitivities of the flood. Referring to Figure 4, complete pressure equalisation between sides A and B can be counted on above tray 10 and above tray 6 (P10A = P10B, and P6A = P6B) where P signifies pressures, psia, the numbers 6 and 10 signify the tray number, and the subscripts A and B signify the side per Figure 4. Normally, one would also expect vapour equalisation on tray 8. However, the sump and outlet weir pickets on that tray block most of the equalisation area. We estimate that the vapour equalisation window above tray 8, including the gap between the picket fences and the sump and the area between the pickets, to be a total of about 3 ft². This is less than half the hole area (machined hole area) on the tray (about 7.7 ft²). So, only a partial equalisation is likely to occur above tray 8, and P8A can be lower or higher than P8B.
A side
B side
11
11
P10
10
10
9
9
3”
6”
8
8
7
7
P6
0˚ N
High froth on #9 South West
Low froth on #9 North East
90˚
SE
N22
Figure 4 Elevation and plan, trays 6-9
Now let us postulate that tray 10 panels on side B are plugged to a greater extent than on side A. The pressure drop on side B will be higher, which will divert more vapour to side A until the pressure drop on trays 6-10 equalises between sides A and B (P6A - P10A = P6B - P10B). Assuming trays 6-9 are not plugged, the pressure drop on side B of trays 6-9 will be less than on side A because of their lower vapour loads. To keep the total pressure drops the same on the A and B sides of trays 6-10, P10B will need to be larger than P10A. The downcomer back-up equals the pressure drop of the tray above, plus the liquid head at the inlet of the tray below, plus the friction at the downcomer apron. 1 This means that if the pressure drop is higher on panel P10B, more liquid
19
Revamps 2024
www.digitalrefining.com
Powered by FlippingBook