Oct-2024

Diagnosing a premature flood in an atmospheric crude tower

Diagnosing and rectifying the flooding in a crude tower emphasises the importance of conducting plant tests, as well as gamma scanning investigations and analyses.

Daniel Hussman, Gord Bruce, Abdullah Abufara, Komi Chandi and Joshua Donohue Parkland Refining (BC)
Henry Z Kister Fluor

Article Summary

This investigation of a premature flood near the diesel draw of an atmospheric crude tower demonstrates the importance of conducting plant tests, judicial gamma scanning investigations and analyses, and closely checking theories against plant data and targeted tests to diagnose fractionator problems.

In a 2020 turnaround, the planned mechanical work included replacing the wash section two-pass sieve trays 8-10 with in-kind, but the tray material was changed from 410 SS to carbon steel. Afterwards, the tower operated at high crude rates for 18 months without issues. In October 2021, there was a planned turnaround of an adjacent unit, during which the tower was drained, purged, and safe-parked.

Upon return to service, the tower struggled to maintain high crude rates due to increased pressure drop across the bottom two-thirds of the tower. The pressure drop progressively rose over the next seven months, forcing crude rate cuts of 5,000-6,000 BPD. At higher rates, the tower flooded, causing sudden loss of overflash and diesel draw (level in the side stripper). Gamma scans showed flood initiating on tray 10, right above the diesel draw tray, propagating up the tower when the rates were raised.

Several theories requiring different remedial actions were proposed. The theories were evaluated by thoroughly checking against plant tests and data, together with quantitative analysis of the gamma scans. Based on this evaluation, the team concluded that entering the tower to clear foulants or repair damage or fouling was the best path forward.

Upon entry, the active areas on trays 10 and 9 (diesel draw) were found heavily fouled, to the point where the sieve holes were invisible. The foulant thickness varied across the diameter of the column and could be easily removed with a metal scraper. The deposits were clay-like and sandy. The downcomers were relatively clean. Trays above and below this zone were found to be quite clean.

Process description and problem definition
Figure 1 is a schematic of the crude tower with typical rates and temperatures for the 18 months before the October 2021 outage. The tower differential pressure, measured between the flash zone and the vapour space above tray 24, was steady, typically 2.3-2.5 psi, or about 0.13 psi/tray, a little on the high side, but within the range of normal operation. The crude rate was high (30,000 BPD), the diesel pumparound rate was high (5,700 BPD), the diesel side stripper level was steady, and the overflash was steady at about 500 BPD.

After the October 2021, turnaround the tower differential pressure (dP) became erratic and crude rates had to be limited to 29,000 to prevent dP excursions. The diesel pumparound rate was reduced to about 4,500 BPD with a colder outlet temperature of about 225ºF.

The vapour space between trays 8 and 9 has two temperature indicators that historically read the same. Following the October 2021 outage, the one on the northeast started reading 15ºF higher than the one on the southwest. Two other temperature indicators were in the vapour space between trays 6 and 7, they kept reading the same temperatures.

Things took a turn for the worse following another outage in December 2021. The dP became more erratic, often rising to about 3.5-4 psi. Liquid appeared to be accumulating above the diesel draw with a loss of diesel product, wash to the lower trays, and overflash. The high dP was often accompanied by some diesel ending in the jet draw from tray 16. The accumulation appeared sudden, erratic, and unpredictable. Drawing more diesel brought the dP down, but the overflash would not re-establish until the coil outlet temperature (normally about 650-655ºF) was lowered by about 5°F. The dependence of the flood on the coil outlet temperature suggests strong sensitivity to the vapour loading.

The frequency and severity of the flooding above the diesel draw tray increased at higher crude charge rates and high diesel pumparound flow rates. This forced the refinery to gradually cut charge rates, from 28,000 BPD after the December outage to 25,000 BPD just before the tower was shut down in July 2022. Diesel pumparound flow rates were further reduced to about 2,200 BPD.

Troubleshooting and testing
The active areas of the trays were gamma-scanned on January 25 (flooded) and January 26 (‘normal operation’). During both scans, the crude feed rate was 28,000 BPD, and the diesel pumparound flow rate was 3,800 BPD. The key parameter changed was the coil outlet temperature. In the flooded scan, the flash zone temperature was 653ºF, lowered to 646ºF to give the ‘normal operation’ scan. The pressure drop at the flooded scan ranged from 2.8 to 3.7 psi, compared to 2.3 psi and stable in the ‘normal operation’ scan.

The flooded scans (see Figure 2, blue and red pens) showed flooding propagating from tray 10 upwards to trays 17-19 (Figure 2 was cut off at tray 15, but the scans continued all the way up the tower). In the flooded region (tray 10 and up), the scans show little difference between the northeast (NE) and southwest (SW) chords. On the unflooded trays below (6 through 9), the SW side (red pen) showed taller and denser froths than the NE side (blue pen). The NE sides of trays 6, 7, and possibly 9 looked dry, while the SW sides looked normal and not heavily loaded.

The ‘normal operation’ scan showed flooding on tray 10 only. Trays 11 and higher were not heavily loaded, with little difference between the NE and SW chords. This verifies that the flooding was due to a restriction near tray 10. The taller and denser froths observed on the SW of trays 6-9 in the flooded scan persisted on tray 9 in the ‘normal operation’ scan but became much less pronounced on trays 6-8. Only tray 6 in the ‘normal operation’ scan NE side approached drying.

In search for the condition at which the downcomers began to fill up, the centre downcomers of trays 6-15 were scanned (see Figure 3) two weeks following the active area scans. The crude feed rate was set at 28,400 BPD and the flash zone temperature at 651ºF, similar to those in the flooded active area scans. To unload the downcomers and keep them out of the flood, the diesel pumparound flow rate was lowered to 2,700 BPD compared to 3,800 BPD in the active area scans. For the ‘more loaded’ scan (blue pen), the pumparound return temperature was set the same as during the active area scans, 211ºF. For the ‘less loaded’ scan (red pen), this temperature was raised to 244ºF. The pressure drop at both tests was about 2.4 psi and stable.

Figure 3 shows no flooding in the tower, as confirmed by the low dP. The active area above the tray 10 centre downcomer approached flood, but the active areas above the centre downcomers of trays 12 and 14 were not heavily loaded. This again verifies that the first active area to flood is that above tray 10.

Figure 3 shows that the froth height in the downcomer from tray 10 in the more loaded scan was about double that in the less loaded scan and approached the top of the downcomer. The froth in the more loaded scan was also much denser than in the less loaded scan. In contrast, the loading difference had little effect on the froth heights and froth density in the tray 12 downcomer (which also serves the pumparound), and its froth did not approach the tray above. This strongly supports a restriction in the tray 10 downcomer, plugging on tray 10, or both.