Dec-2019

Troubleshooting instability in a debutaniser tower

Successful diagnosis and remediation of premature flood in a debutaniser tower demonstrate the value of good diagnosis for solving tower problems economically.

HENRY Z KISTER and CARLOS TROMPIZ, Fluor
BRIAN CLANCY-JUNDT, Flint Hills Resources

Article Summary

Flint Hills Resources’ (FHR) debutaniser tower in the propane dehydrogenation (PDH) unit in Houston, Texas experienced premature flood at high loadings. A sharp pressure drop rise was encountered when the tower internal vapour rate through the top section exceeded 13000-15000 lb/h. This increase in pressure drop was accompanied by an increase in the bottom C4 content. Instability and fluctuations of key process variables such as tower differential pressure and bottom flow rate occurred at all rates, intensifying when the tower flooded.

To diagnose and solve the problem, and to debottleneck the tower, Fluor, which was not involved in the original design but has extensive distillation expertise, was invited to join the FHR troubleshooting task force.

Plots of tower differential pressure (dP) versus vapour loads were prepared. These confirmed the flooding and suggested that it was most likely occurring above the feed. Hydraulic calculations did not show proximity of any of the major flood limits. The jet flood for all trays was below 50%, downcomer liquid backups were very low, and so were downcomer inlet velocities. So no major flood limit should have been encountered.

Three gamma scans were performed at different rates. The latest of these, on 17 February 2015, was performed at high vapour loads and high dP. The scans looked similar to each other. No flooding was observed in any of the gamma scans.

Closely reviewing the hydraulics revealed extremely low head losses at the downcomer exits of the trays above the feed, about 0.11in of liquid. This raised the suspicion of a possible loss of downcomer seal. Applying the Fluor downcomer unsealing model confirmed that vapour was likely to break the seal on these downcomers and to generate downcomer unsealing flood. Unsealing of downcomers may have also occurred below the feed, but our model showed that it is most likely that there the unsealing only led to tray efficiency reduction, not to flooding.

The downcomer unsealing flood was unable to explain the fluctuations as they also occurred when the tower was not flooded. Further investigation traced the root cause of these fluctuations to the unvented vapour space above the baffled bottoms draw compartment. Trapped unvented gas generated back pressure on the reboiler compartment liquid with intermittent geysering and fluctuations, as detailed in this article.

To solve the problem, and with economics favouring reducing tray spacing to increase the number of trays, the tower was retrayed. Based on the above diagnosis, the new trays had effectively lower downcomer clearances. A gooseneck vent was added to vent the bottom draw compartment, and the downcomer from the bottom tray was shortened, reducing the submergence of the bottom downcomer in the sump liquid.

The tower is now operated with better separation than ever at 10% higher loads than previously, with no flooding, no fluctuations, and at low pressure drops, attesting to the value of good diagnosis for economically solving tower  problems. 

Operating experience
Feed to the PDH debutaniser is about 50% mole C4 hydrocarbons, which end up as an all-liquid overhead product, and about 50% mole heavies, mostly C6 hydrocarbons, which end up in the bottoms. The C4 content of the bottom is below 1% mole. The tower operates at 80 psia at the top and a reflux to distillate ratio of about 3-4:1.

Figure 1 shows key tray and downcomer dimensions. The tower is 32in in diameter and contained large opening, fouling resistant valve trays. Above the feed, the trays were conventional with 14% slot area (as a fraction of the active area), spaced 21in apart, and had straight 4.5in wide downcomers. Weir heights were 1.75in and downcomer clearances were 1.38in. Below the feed, the trays were at 18in spacing, with stepped truncated downcomers. The downcomers in this section terminated 4in above the tray floors, with 32% of the downcomer floor perforated to allow liquid descent. Right under the perforated downcomer bottom there were fixed valves. The fixed valve caps prevented downcomer liquid from weeping through the valves. The slot area of the valves on the trays below the feed was 15% of the tray active areas.

The tower experienced instability and high differential pressures (dP) at high loadings. Figure 2 is a historical record, showing a rapid rise in tower pressure drop at a tower vapour rate of about 13 000-15 000 lb/h. The vapour rates plotted on the x-axis were calculated from the sum of the product and reflux flow meters and are not corrected for reflux subcooling, which is likely to raise the vapour rate shown by around 10%. The blue points that show a rapid rise in pressure drop at about 8000 lb/h and the green points showing another rapid rise at 11000 lb/h are due to fouling episodes and are not addressed in this investigation. The focus here is on the pressure drop rise between 13000 and 15 000 lb/h. This rapid rise in pressure drop conclusively indicates flooding.

There had also been fluctuations of key process variables such as bottom level, tower differential pressure, and bottom flow rate. These fluctuations occurred at all rates, but intensified when the tower flooded.

Troubleshooting
The active areas of the trays in the tower were gamma scanned on three occasions: 28 March 2012, 12 February 2015, and 17 February 2015. The vapour traffic near the top of the tower (unadjusted for subcooling) corresponding to these scan periods was around 9500 lb/h, 12 400 lb/h, and 14 700 lb/h, respectively. The pressure drop ranges were 2.1-2.2 psi, 2.8-2.9 psi, and 3.6-4.2 psi, respectively. The 28 March 2012 scans were performed at the lowest vapour loads, and the 17 February 2015 scans at the highest, about 30-50% higher than the loads on 28 March 2012.

Figure 3 shows the scans for the upper trays and the lower trays on all three dates. The intermediate trays, including the feed tray, are not shown, but their scans showed identical trends and patterns to the trays shown in Figure 3. Figure 3 shows that despite the large differences in vapour loads and pressure drops there was very little difference between the three scans. None of the scans showed any flooding. In both the 2015 scans, the froth heights ranged from 38% to 51% of the tray spacings. If there was any entrainment from the trays, it was light.

During the 17 February scans, the reflux rate was intentionally raised to the point where flooding would be indicated by the dP transmitter, the tower was closely watched, and the following observations were recorded:
•    Reflux set point was increased, causing the dP to climb to 4 psi. The reflux set point was stabilised for approximately 15 minutes before the scan was started.
•    Tower dP varied between 3.6-4.4 psi while the scan was being performed. Under normal conditions, the tower’s dP fluctuation varied between 2.8-3.2 psi. This is nearly a 100% increase in variation.
•    The debutaniser bottoms level fluctuated between 55-85% full while the scan was being performed.
•    Debutaniser bottoms C4 concentration increased from 1% to 2.6% mole nearly 15 minutes after the scan was complete and before the reflux was lowered back to normal.