We circulate 50% caustic through the tubes of a vertical 1-1 shell-and-tube condenser using 50% steam reduced from 400-psig steam. The exchanger contains 60 tubes, each ½-in. diameter with 0.083-in. walls, but 20 have been blocked off. The caustic enters the bottom of the exchanger and, after exiting, dumps into the top of a storage tank. Regular deliveries ensure that tank is seldom below 75% full. A 3-in. globe valve controls the steam flow. The worst-case ambient temperature is -15°F.
Recently, after almost a decade of service, a single 316L stainless steel tube failed, causing caustic to flow into the condensate return pump. An inspection revealed no fouling in the tubes. However, trend data show temperature spikes that go above the 200°F top value of the chart.
The carbon steel caustic tank exhibits severe corrosion above the normal liquid level and only minor weld pitting below. The roof is strongly corroded.
Inspection of the failed tube revealed erosion a few inches around the hole. The tube wall thickness was about 0.015 in. at the edge of the ¾-in.-long lens-shaped hole. The hole is about 1¼ in. from tube sheet on the outside of the sheet. We were surprised the failure occurred at that spot and there wasn’t any sign of stress corrosion cracking.
What do you think caused the failure? What can we do to avoid this problem in the future? Since we got nine years out of the tubes, is this really a significant problem?
Answer Some Key Questions
The problem should be broken into two parts: the heat exchanger and the storage tank.
As you stated that one tube failed after almost ten years and the trend data show there were a number of temperature spikes; I see a couple of possibilities: some of the tube plugs may be loosening; some tubes may be developing micro-cracks; or the tube-to-tubesheet joint may be a leak source. I believe you may have checked these potential leak sources during your last shutdown to plug the last tube that failed.
Very minute cracks sometimes are hard to catch by pressure check with nitrogen or water. Some practitioners have had success using helium to detect micro-cracks. However, use of helium isn’t that straightforward for many plants.
Check the velocity of the caustic solution. For 50% solution, good practice is to keep it at 7–10 ft/sec. On the shell side, an impingement plate at the steam entrance is a common feature of many heat exchangers.
To decide whether this is a significant problem, you must answer several questions: How often do you experience episodes of temperature spikes? What’s the remaining life of the exchanger, considering it already has been in service for nearly ten years? (You probably should discuss this with its manufacturer.) What are the possibilities of tube erosion getting worse and causing larger and more frequent leaks? Keep in mind that leaks contaminate your condensate return system and this, in turn, could affect the reliability of the boiler system (and hence your entire unit or plant).
As for the storage tank, it’s not clear whether it’s coated. Coating failure could have accelerated corrosion. Typically, uncoated carbon steel tanks are considered a relatively poor choice for storing 50% caustic solution. With corrosion of the carbon steel, caustic solution picks up iron that could affect your downstream processing. If storage temperature isn’t excessive, think about switching to a polypropylene tank — if one is available in the size you need.
If you plan to stick with uncoated carbon steel, consider eliminating or minimizing air entry into the tank (nitrogen blanket and split-range pressure control) and not exceeding a storage temperature of 120–125°F. Carbon steel suffers from caustic embrittlement above these temperatures. Stress relieving of welds also will help minimize caustic embrittlement. Finally, provide secondary containment.
GC Shaw, senior HSE adviser, Wood
See The Big Picture
Most engineering problems are rooted in a poor design. This is one of those compound problems. The information looks a little thin, so let’s start with some assumptions: 1) tank dimensions of 30-ft ID, 24-ft straight-side with 2-in. of fiberglass insulation on the shell and roof; 2) a steam supply of 300 psig, reduced to 30 psig; 3) a pressure drop of 3 psig across the shell of the condenser; 4) 8 psig at the discharge of the shell to the condensate pump; 5) 5-psig drop in the line delivering the steam and a standard globe valve for the control valve; and 6) a maintenance temperature of 150°F in the tank — below this temperature, the caustic gets too thick for a centrifugal pump. Based on this information, I come up with a maximum demand of 80,000 BTU/hr. Assuming 60°F average temperature, I get only 45,000 BTU/hr needed to maintain the tank temperature. Note that the superheat temperature for reduced superheat steam is 53°F (instead of 327°F for saturated 30 psig steam.) The thermal conductivity of ≈ 2–5 BTU/hr-ft-°F is about 1/200th of the typical 1,000 BTU/hr-sq.-ft-°F (per “Kern’s Process Heat Transfer”); that’s a serious loss in heat transfer area if the steam isn’t desuperheated. A condenser’s overall heat transfer coefficient always is larger than that for heat transfer with a gas phase because the resistance comes from the condensing liquid not the gas film; the Prandtl number (µCp/k) is 0.7 for air and 7.5 for water.
Now, let’s look at questionable design choices. Why is a 3-in. steam valve necessary? A small 3-in. equal-percentage globe valve operating at a maximum Cv of 71 will deliver 4,600 lb/hr of steam. So, for 80,000 BTU/hr, the valve will have to be <10% open. Of course, it will pop open and closed! Replace the 3-in. valve with a 1-in. one. It’s still over-sized and will operate at 10–20% open for a 16-psig drop — add a restriction upstream to put it in the 80% range. Even if this tank is much larger or without insulation, a 3-in. valve is way too large; I calculated 700,000 BTU/hr if the tank had no insulation at -15°F.
As for the piping design, change it. Inject the caustic from the heat exchanger a few feet below the liquid level. With 75% at 18 ft, put the caustic at 12 ft after running a hydraulic analysis to verify the flow is adequate through the exchanger tubes. This change should decrease the corrosion at the top of the tank and reduce stratification of the caustic inside the tank.
Getting nine years from thin-walled tubing is pretty good. A temperature of 150°F is well below the 264°F temperature for the onset of stress corrosion cracking. Instead, it seems more likely that a poor weld caused the tube failure.
Speaking of injection, why not get rid of the heat exchanger altogether? For 80,000 BTU/hr, you only need 86 lb/hr of steam. You could inject steam directly into the caustic tank via the circulation pump. That would eliminate the desuperheated steam problem but introduce the dilution of caustic. Still, it’s worth thinking about. You could set the refurbished exchanger aside for when the process isn’t using caustic.
Dirk Willard, consultant
We use an axially split centrifugal pump to handle a catalyst slurry in naphtha. We were running with one pump on, one on standby, until the one in service tripped a high temperature alarm on the bearings. Flow fell drastically. We switched to the standby pump; it also developed low flow but the thermal sensors didn’t trip. Both incidents resulted in a serious leak around the shaft seal of each pump.
Doing a post-mortem on the incident, we’ve run into an impasse. The maintenance manager suspects the seals were the culprit. He flushes the seals with a clean light oil. He never has been a fan of using an API Plan 62 with an oil quench. He would prefer a Plan 32 and has been working with the seal salesman for years to change the seal arrangement. However, production likes things just the way they are. Projects says they don’t have the money. And the corporate environmental engineer wants to reduce emissions by installing a double seal plan plantwide.
Do you think the maintenance engineer is right? Did the seals cause the failure? Is this a problem we should be worrying about, given the money crunch? What do you think?
Send us your comments, suggestions or solutions for this question by February 8, 2019. We’ll include as many of them as possible in the March 2019 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, 1501 E. Woodfield Rd., Suite 400N, Schaumburg, IL 60173. 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.