Provide troubleshooting and detailed column internals analysis for a poor performing fluid catalytic cracking main fractionator at a major U.S. petroleum refinery.
REFINERY FCC FRACTIONATOR PERFORMANCE IMPROVEMENT STUDY
The client owns and operates a major petroleum refinery in the eastern U.S. and contracted Process Engineering Associates, LLC (PROCESS) to review operating and performance data from a fully-packed refinery fluid catalytic cracking (FCC) main fractionator exhibiting very poor separation between its three liquid products. Operating data from this tower showed large differences in temperatures reported in different quadrants of the tower by sets of thermocouples grouped at several specific elevations. In addition, these temperatures were seen to flip-flop several times per month, at unpredictable intervals, wherein points reading higher would suddenly drop while lower-reading points at the same elevation would simultaneously jump up. Moreover, temperature flip-flops would propagate through multiple beds in the tower within a very short time, making it difficult to ascertain where they were originating.
To begin the study, FCC operating and sample analysis data were retrieved from a day when the FCC unit was running a fairly typical feedstock and a routine weekly material balance was conducted, complete with product sample analyses. PROCESS developed a HYSYS simulation model to match FCC main fractionator performance data, blending products from the material balance together to characterize the reactor effluent stream feeding the tower. Experience from working with previous packed FCC main fractionators was used to override the measured tower pressure drop of 3.2 psi with a more likely value of 1 psi. This greatly improved the temperature match of the model feed zone to the plant measured value.
Next, PROCESS began reviewing the tower drawings, rating the packed beds and rating the liquid distributors in the tower for vapor and liquid conditions from the Hysys model. This actually showed that operating conditions were within the capabilities of the original design. However, there were two highly loaded draw trays which were suspect because of their design. To wit, it appeared that liquid flow across the trays to their draw sumps would be impeded in both cases because of mechanical design choices.
A very complicated Excel spreadsheet was constructed to check these draw trays for hydraulic gradient, an undesirable phenomenon that would cause maldistribution to the liquid distributor parting boxes below these draw trays. Maldistribution would be caused by the placement of three overspill downcomers from each draw tray arranged in a line extending away from the draw sump. Any hydraulic gradient across the tray would cause more liquid to spill into the upstream downcomer relative to the downstream one. A difference of about 20% maximum was deemed acceptable for what the parting boxes could handle without seriously affecting liquid distribution to the packing beds below.
The Excel spreadsheet analyzed these trays as 22 zones which could plausibly have different liquid heads, and balanced the liquid flows for two or three separate flow paths between the zones such that the liquid head loss predicted between each pair of zones was equal for each flow path. The analysis was complicated by the fact that the predicted head loss depends on the flow velocity, which depends on the flow area, which in turn depends on the heads themselves, thus forming a circular definition. Solving this involved using a starting estimate for each zone head and successively updating the head estimates until the calculation converged (which was greatly facilitated by Excel’s “Solver” add-in).
Results from the Excel spreadsheet analysis indicated about a 30% difference in overspill rate into the upstream downcomers versus the downstream ones, deemed unacceptable. PROCESS designed two new chimney trays which would not suffer from hydraulic gradient problems. In addition, PROCESS recommended that a feed zone vapor distributor be installed to mitigate maldistribution of vapor to the bottom-most bed (slurry bed), which was suspected of being an initiator for the observed bed temperature flip-flops.
The final recommendation from PROCESS was to replace the slurry bed liquid distributor. An analysis of the existing distributor indicated it was extremely sensitive to out-of-levelness due to its ‘wineglass’ or ‘arc notch’ shape of the trough distribution elements. The analysis indicated that a 1/8 inch difference in trough liquid head from end-to-end would give a 25% difference in notch flow rate at typical slurry pumparound rates.
The recommendations to replace the draw trays and slurry liquid distributor, and add a feed vapor distributor, are being pursued by the client.
- Petroleum Refining
- Fractionation tower troubleshooting
- Distillation distributor analysis
- FCC fractionator optimization