September's Puzzler | August Issue
We are designing a desulfurization process for a client overseas. The client produces a wet coke gas that must be treated for COS, H2S, and HCN before it can be used to make chemicals. A knock-out pot serves double duty — collecting the circulating spent slurry and vapor from the client’s gasification plant. The vapor flow is 50,000 lb/hr and the liquid flow is 519 gal/min. The pot operates at 200°F and 560 psig with a barometric pressure of only 12 psia. A k of 0.27 was used with a velocity allowance of 0.15 to size the tank for vapor separation. The residence time is set at 30 minutes. The calculated diameter is 13 ft with a length/diameter (L/D) ratio of 3.4. The vapor passes through a mesh pad demister that is only partially used because of low vapor flow. Several problems are anticipated with the design: the liquid drains to atmospheric; foaming sometimes occurs, disrupting level measurement; the mesh pad fouls every four months; and the process water used on a timer to keep the pad clean may contain particulates. How would you improve the design?
Address the process issues
For some problems, the best you can do is to figure out how to live with the least bad situation rather than being able to find a solution. Your options depend upon the cost the client is prepared to spend. Some ideas to check the economics of include: vortex tubes for foam reduction; staged clean water washes; vane mist eliminators in series with the mesh pads; multiple vapor exits; extra vessel elevation; overlapping level instruments; nuclear level instruments; and weight cells.
Vortex tube clusters (VTCs) take advantage of direction changes to help collapse foams in a multiple-phase stream. If the source of drum foam is the feed, using VTCs may aid in reducing the amount of foam in the drum.
If the process will tolerate occasionally using a clean wash, then putting in a second wash system to extend the time between plugging on the mist eliminator pads may be justified. Perhaps one wash in four or five might use clean water.
A minimum velocity is needed for wire and vane mist eliminators to effectively remove mist. The higher the pressure drop the smaller the droplet removed. Fouling occurs from the process liquid trapped by the eliminator or particulate in the wash water. The challenge is balancing pressure drop, fouling tendency and efficiency. You could consider using a correctly sized vane eliminator followed by the over-sized wire mist eliminator. The vane mist eliminator has a much greater tolerance for fouling because of its large passages. Of course, this will require a much fancier water-wash system. The purpose of the vane mist eliminator is to reduce the loading of process liquid in the wire mist eliminator. This may extend the mist eliminator run before it plugs.
Multiple vapor exits may also be usable. With the correct valves you could have two or three vapor outlets. If each has a run of four months or so, you have effectively extended the vessel service duration between shut down and cleaning cycles. A single feed in the center of the drum and a vapor outlet at each end would be a good first choice. Of course, if you have less risk exposure to premature shut-down, you may also consider making one end of the vessel correctly sized rather than over-sized for the mist eliminator used.
Extra vessel elevation may be critical in getting liquid out of the drum. If the contents are foam rather than clear liquid, then the static head available for liquid draining is much less. Extra height (and larger exit nozzles) will prevent many gravity fluid-draining problems.
Level indication with foam is notoriously difficult. The first method to deal with liquid level problems is to use something like a displacer or a differential pressure (dp) cell for level measurement and to use more than one of them on multiple, different ranges. If the different level instruments show multiple liquid levels, the lower instruments are probably completely flooded with foam. Instead of measuring level they are measuring an average liquid density across their range. The actual foam level will be in the range of the highest level indicator showing a level. While not perfect, this can be a vast improvement over a single level instrument.
If solids, corrosion or service conditions prevent multiple level instruments, then a nuclear (gamma ray) device may work well. While it is much more expensive, the cost may be justified. Properly set up, gamma ray level instruments can separately show liquid level and foam level.
Finally, in severe cases, weight cells have been used to infer levels in extremely foaming systems. This is a last ditch effort as weight cells on large vessels can be complex to set up. They can also give false readings due to reactions from piping stresses. You are no longer measuring level but system mass inventory directly. From an assumed foam generation rate, you set a maximum system weight acceptable. I prefer nuclear devices to these but in some situations they do work.
Andrew Sloley, principal engineer
CH2M HILL, Bellingham, Wash.
Examine the design carefully
There are a number of problems in this design starting with the k. A k of 0.27 is probably too high for good separation at 560 psig. The Gas Processors Suppliers Association (GPSA) Engineering Data Book suggests de-rating k by 0.01 per 100 psig above 100 psig. This would provide a k of about 0.21 not 0.27. Decreasing k will increase the area needed for separation.
The velocity allowance (de-rating) of 0.15 seems correct for a foam problem; this term is multiplied by the allowable velocity to calculate the area. A barometric pressure of 12 psia corresponds roughly to an altitude of 4,200 ft above sea level. A low barometric pressure increases the required separation area less than 1%.
The residence time seems a little short. Given the flow involved, 519 gal/min, additional capacity might be desired since this drum appears to be the center of the desulfurization process.
Corrosion with H2S, HCN and other compounds may be a problem. A corrosion engineer should review the selection because an attack on the metal might be anticipated at the phase interface. Review NACE International standards for H2S service.
The selection of an L/D of only 3.4 seems wrong. “Coulson and Richardson’s Chemical Engineering” recommends a ratio of three for pressure up to 165 psig; above this pressure the ratio should be five. The reason for the selection is the high cost of a large diameter at high pressure. At 560 psig, a 13-ft diameter head would be more than four-in thick. Welding a thick plate is more of a challenge than welding a thin plate. If length is a problem because of space, a vertical separator, though less efficient than a horizontal separator, can work in most applications. Here’s another reason for a longer vessel: the capture droplet size is smaller. The vapor enters one end of a horizontal drum and exits the other after passing along its length. The longer it is, the more efficient it will be in settling liquid.
Now that the design has been critiqued, let’s consider the three process problems mentioned: de-pressurizing the liquid drained from the drum; reducing the effects of foaming; and avoiding the necessity of cleaning the mesh pad.
De-pressurization can be dangerous, especially at high pressure. Another concern is the wear on the orifice plate, the cheapest device available to reduce the pressure. One solution may be a reducing pot. Certainly more than one plate should be used for safety.
Foaming could be addressed with an additive but perhaps the best approach is to work with a reliable alternative like nuclear or a bubbler. Bubblers are the cheapest option and can work satisfactorily in foams and slurries if you can find a high pressure gas.
The best solution for the mesh pad is probably to be conservative on the tank design, allowing more settling time, and replacing the pad with a chevron. Chevrons can serve for many months without cleaning or plugging.
Dirk Willard, senior process engineer,
Ambitech Engineering, Hammond, Ind.