Fluid Handling / Vacuum systems

Don't Be Stumped by a Sump

Several steps can help handle gasoline-laden vent gas.

Our refinery has a project to fully enclose all API separators and other equipment to meet stringent EPA regulations. This includes a concrete sump, 15-ft long, 10-ft wide, and 7-ft 2-in. deep, built in 1944 and buried in earth. We reckon the slop oil in the sump consists of a mixture of 70 mol-% gasoline and 30 mol-% water. The 8,000-gal sump has two pumps, each eight feet below ground level, operating at 100 gpm each. A structural engineer, with some trepidation, set the pressure limit at 15 inches of water column (IWC) and the vacuum at 1.5 IWC. The normal vent exits the roof, which is 15 ft above the sump, and then must go to an activated carbon bed (ACB) located 60-ft away; the ACBs are on the ground. The bed has a pressure drop of 0.5 IWC at 100 scfm air (60°F). A flame arrestor normally is installed upstream of the ACB and likely would add a pressure drop of 35 IWC at 1,000 scfh air. The emergency vent goes direct to the roof. How can we design these vents?

In my experience the ACB canisters are spent in less than a month. A combustor pilot (burner) would be more appropriate (Figure 1). You can still use the ABC as an alternate protection whenever you have to work on the combustor. An ejector can overcome the pressure of the flame arrestor. You pull the gas with the ejector using fuel gas at a certain pressure as the motive media. This should not affect the sump pressure; on the contrary it would be pulling toward vacuum.

Arbues Maymi, senior process engineer
CITGO Refining and Chemicals Co., Corpus Christi, Texas

This design presents three issues: 1) potentially a large fire-case flow; 2) very low pressure/vacuum design loads; and 3) high pressure drop in the normal vent discharge pipe. All of these issues are possible showstoppers.

Based on the mixture, the fire-case flow with gasoline would be 88,000 scfh air. However, that's with all vertical sides exposed to fire. API-2000 allows for insulation (via F). What would be better insulation than dirt? For eight inches of concrete, typical for this type of sump, I calculated an F of only 0.105.

Then, there is the sump itself. Its wetted area would be 650 ft2. However, only the top is exposed to a fire in the sump room. That area is only 150 ft2.
And the water has an impact. Using Peng-Robinson, I calculated a second molar cut with the following properties: heat of vaporization of 203.9 BTU/lb., saturation temperature of 132.4°F, and molecular weight of 66.6. With 100% gasoline it would be 168 BTU/lb. Interestingly, the latent heat is a flat line between 5% and 50% mole water. The net result is a reduction of the fire-case flow to only 12,800 scfh air. Vacuum always is a limiting factor with pressure relief.

I would size the flow for as little pressure drop as possible. The lowest set point available is 0.9 IWC — start there for vacuum. Try for 15 IWC with pressure. Calculate the vent pressure drop and keep it below the total pressure. The first obstacle is the flame arrestor.

Getting rid of the flame arrester is a priority. At 2,400-scfh air, I estimated about 200 IWC pressure drop through the arrester. API-2000, "Venting Atmospheric and Low-Pressure Storage Tanks," paragraph, gives succinct support: "A flame arrestor is not considered necessary for use in conjunction with a PV [pressure/vacuum] valve venting to the atmosphere because flame speeds are less than the vapor velocities across the seats of PV valves." Paragraph 5.2, API-2210, "Flame Arresters for Vents of Tanks Storing Petroleum Products," provides further support, noting the U.S. Bureau of Mines says the critical velocity is 10 ft/sec. With the flame arrester, no vent is possible.

Sizing the normal vent depends on the in-breathing (vacuum) and out-breathing (pressure) limits established by the process. The flow into the sump sets the in-breathing while out-breathing is based on the pump. Without additional data, use the right end of the (heat flow) curve for two pumps as the flow into the sump. Although assuming both pumps are operating is conservative, this approach usually is acceptable for out-breathing. With the limits set, you can specify the normal and emergency (fire) vents.

Given the conditions, it might be tempting to eliminate the emergency vent and go with the normal vent alone. This usually is not acceptable but it's worth checking into. Crane's method for divided flow won't work but iteration using the flow for the normal vent with the emergency vent set point is effective.

Using a vendor's package, I sized a 3-in. normal vent with 6-in. piping and a 3-in emergency vent. I found that a 6-in. single inlet pipe would feed both vents. I estimated a set point of 11 IWC for the emergency vent, 0.9 IWC for normal vacuum and 4 IWC for pressure. Sizing the emergency vent is tricky because the tank is rectangular and the vendor program calls for a height; use an equivalent diameter based on the top area and ignore the height.

Dirk Willard, senior engineer
Ambitech Engineering, Downers Grove, Ill.

We use a vacuum knockout pot to push vapor produced from a batch polymerization reactor and a neutralization tank to a caustic scrubber (Figure 2). Caustic is recirculated over a chevron demister and through packing. The vacuum is only about 12 psia to avoid losses from the polymer reactor, which contains a solid catalyst. The process periodically suffers from pH upsets in the neutralization tank, corrosion in the vacuum pot, poor temperature control, loss of vacuum from seals, fouling of the vacuum pump rings, and fouling in the demister pad, which only survives a few weeks. We're tearing down this process every week to avoid trouble and the cost is killing us. There's a further complication — we periodically use the batch reactor and tank to make other products. Can you suggest any ways to improve reliability?

Send us your comments, suggestions or solutions for this question by February 13, 2012. We'll include as many of them as possible in the March 2012 issue and all on ChemicalProcessing.com. Send visuals — a sketch is fine. E-mail us at ProcessPuzzler@putman.net or mail to Process Puzzler, Chemical Processing, 555 W. Pierce Road, Suite 301, Itasca, IL 60143. Fax: (630) 467-1120. Please include your name, title, location and company affiliation in the response.

And, of course, if you have a process problem you'd like to pose to our readers, send it along and we'll be pleased to consider it for publication.