a
a
b
Figure 3a Grid gamma scan 3b + gamma scan
b
few gallons of caustic leaked into an idle crude unit (CDU) each day. In normal operation, the caustic was injected to neutralise the acid in the sour crude that the unit processed. The caustic should have been turned off and positively iso- lated when the unit was idling. When it kept going with no crude to neutralise, it caused a major corrosion damage incident and a prolonged outage of the unit. The large car- bon footprint of this incident is impossible to quantify. Seal flush: Another example is shutting off the seal flush to idle pumps. A single pump with a 2in shaft and both in-board and out-board seals can consume 200-300 BPSD of seal flush oil when on standby, which has to be processed in downstream units or risk degrading the prod- uct. For instance, if the flush oil for the residue pumps is gas oil, it represents a degradation from a high-value to a low-value product, with no benefit whatsoever. If the unit produces asphalt, the low-viscosity gas oil can cause the asphalt to go off-spec. Good testing solves problems Still X and Still Y are 5ft ID speciality product finishing columns in the same service. In the early 1990s, a major packing supplier repacked both stills with the same pack- ing heights, two lower packed beds of ‘A’ feet each and an upper bed of ‘B’ feet. The packing was similar to BX wire gauze structured packing, which is commonly applied in fine chemical manufacturing that requires deep vacuum distilla- tions to prevent compound deterioration. At an expected efficiency of five theoretical stages per metre, each column was expected to provide about 50 theoretical stages. In the early 2000s, a test mixture was run in both stills. The results showed that Still X had about 11 effective stages, while Still Y had >40 stages, in the lower beds. The separation efficiency of the lower beds of Still X was 25% of what it should have been. The test was repeated in the spring of 2015 with identical results. With this low effi - ciency, Still X was unsuitable for fine product purification. Still X was gamma scanned in 2001. Typically, packed beds are grid-scanned (see Figure 3a ); however, due to the surrounding structures, only two scan lines (a + scan, Figure 3b ) were shot. The scans found all distributors and collectors at the correct elevations, and all three beds had the proper bed heights. The top bed had a density gradient in the packing and some maldistribution. There was some
Time to ll a pail (about 5 gallons) Location Time (sec) Round 2 1 67 54
2
3
1
2 3 4
150 130 185
131 130
North
4
360+
1 (redo)
52
maldistribution in the top foot (or two) of the middle bed. The liquid distribution in the rest of the middle bed and throughout the bottom bed was good. The packing density was constant throughout the middle and bottom beds. The scanners concluded that there appeared to be no serious mechanical damage, fouling, or major distribution problems that would explain the absence of such a large number of theoretical stages. Neither the gamma scans nor the oper- ation saw any signs of flooding. Still X was taller than Still Y, mostly due to the longer downcomers between the packed beds (about 20ft long) that brought reflux liquid to the feed parting box below. Some felt that this caused the difference in the performance of the stills. It was argued that the problem was the high velocity in the pipes, blowing out the parting box liquid. During an outage, Still X was inspected through the man- holes. It was water-washed with the manholes open and water running through the packed beds. Nothing out of the ordinary was noted. In autumn 2015, a leading expert contractor in mass transfer tested and evaluated Still X. The findings showed a stage count similar to those from the previous two tests. The contractor pointed to potential issues with bed height, drip point density, and other factors. He recommended replac- ing the internals and the three packed beds with the latest low-pressure drop wire gauze packings and contractor- designed internals. Other options considered included add- ing a fourth bed, changing the packings but not the inter- nals, and changing the internals but not the packings. During the autumn 2015 test, discussions were held with Figure 4 In-situ water test of chemical tower (a) visual observation, showing a high liquid flow near the centre (b) results of bucket and stopwatch measurements
101
PTQ Q4 2025
www.digitalrefining.com
Powered by FlippingBook