THIS MONTH'S PUZZLER
We are experiencing water hammer and failure at low-point elbows in our steam-condensate return system. Several old pipe supports have failed. This has become a particular problem in the past few years. The original system ran well with few headaches. The 6-in. schedule-80 pipe carries 75-psig steam condensate back to an old flash tank through a 500-ft pipe. On inspection, I discovered that much of the insulation has been removed and that the header receives condensate from several additional lines, including three 6-in. bare lines running 400 ft, 300 ft and 150 ft. I'm trying to identify the pressure of these condensate lines. It could be 150 psig; the refinery has 400-psig, 150-psig, 75-psig and 25-psig steam. The steam traps, which seem well maintained, are noisy when equipment is running. The condensate return pump at the flash tank is showing signs of wear; it runs nearly continuously and has needed more maintenance than previously. In addition, the backpressure is maintained on the steam return by two pressure transmitters (on the inlet and outlet of the control valve) that wander all over when operating; the operators leave the control valve cocked partially open without bothering with it. What can be done to save this old system? What's wrong with the transmitters?
INSULATE THE LINES
Water hammer is the force generated from a slug of rapidly moving water travelling at the steam velocity that is suddenly stopped. A rapid change in phase, caused by pressure reduction, can create this phenomenon, converting the flow regime from all liquid to slug flow. I am assuming the flash drum runs at 25 psig, the lowest pressure mentioned, because this is the only way all sources you have listed can get into this flash drum. A slower phase change is caused by pipe heat loss.
Insulate the lines. Keep in mind that if the lines are not insulated, the pipe is acting like a heat exchanger. You are creating conditions that can cause potential unstable flow regimes, i.e., slug flow due to excessive cooling, pressure drop due to cooling and the potential of condensing steam back into water under the right circumstances. Insulating the lines creates a uniform pressure and heat transfer profile through the pipes that can help stabilize the flow regime. Having these lines connected can create unstable flow regimes in other sections of the lines as well.
Pipe fittings are extremely susceptible to unstable flow regimes such as slug flow. As a slug of water moves down the pipe, pipe fittings (elbows), control valves and instrumentation can be damaged when this slug is suddenly stopped. The force is tremendous. This is why your pressure transmitters are not functioning properly. In an elbow at the point of highest velocity, you also have the lowest pressure. Because the fluid is saturated, a loss of pressure will flash water into steam and this, combined with the force of stopping the slug of water, is eroding the elbows. When you pull the elbow and inspect it, you should see pits at the bend. Insulating the pipe should help stabilize the flow regime and minimizes the rate of heat transfer that is occurring between the inside fluid and the outside air. Where possible, replace 90° elbows with 45° elbows. This will reduce velocity due to the reduced resistance factor of a 45° elbow compared to a 90° elbow and may improve mechanical reliability. The straighter the pipe run, the better the ability to handle an unstable flow regime because the slugs or waves have nothing to stop them.
Piping has limited space to absorb impacts of the forces generated due to these unstable flow regimes. Connecting steam lines of different pressures together will make water hammer worse because the connection points are where the slugs are stopping and forces are exerted. Instead of tying lines into other lines, try to relocate the lines closer to the flash drum or into the drum, if possible, to help mitigate the forces exerted on the pipe and the system. Make sure you insulate the lines effectively around the drum.
With regards to your pumps, make sure you have adequate net positive suction head available (NPSHA) for your pumps to function without cavitating. Check the liquid level in your flash drum to be sure your suction is flooded, if this is a horizontal pump. The shock wave resulting from the sudden stopping of the water slugs could possibly get far down the lines to impact the pumps. In addition, keep in mind the un-insulated lines could also be delivering cool condensate to the flash drum and cooling the flash steam in the drum rapidly. This could result in forces that could be impacting the pumps and damaging them.
• insulate all lines — I suggest mineral wool;
• relocate lines closer to the flash drum, if possible;
• replace 90° elbows with 45° elbows, if possible; and
• check and validate the NPSHA for pumps to be sure you have enough.
Finally, if you desire to know the amount of flash steam being generated, start at the beginning of the 500-ft run and add in the other sources once you determine what the actual source pressure is at each take-off — the individual runs of 400 ft, 300 ft and 150 ft, respectively. Determine the pressure drop from each run based upon the amount of condensate contribution from each source and its source pressure and temperature. The amount of flash is a function of ƒ (saturation temperature, drum pressure, source pressure and temperature). This can be found using the steam tables, a Mollier diagram or simulation software.
Eric M. Roy, principal engineer
Westlake Chemical, Sulfur, La.
SEPARATE THE CONDENSATE
Install a flash tank upstream of the control valve to trap the condensate. Pad the drum with compressed air and regulate the pressure. Feed the condensate into the bottom of the drum. The condensate can be recovered for the boiler house. The control valve will see only steam.
Donald Phillips, engineer
Phillips Engineering, Melbourne, Fla.
DO THE ANALYSIS
There are numerous problems here: 1) unknown condensate pressure; 2) possible boiler-water corrosion issues; 3) an indication of capacity overload; 4) possible two-phase flow in the 6-in. vent header; 5) steam at multiple pressures added to a steam header; and 6) un-insulated pipe where condensation can occur.
The best solution to this problem is to continue walking down the pipe; get very familiar with the isometrics. Typically, these drawings are a mess. Create accurate drawings. Collect pressure and temperature data. Use an infrared gun to verify which traps aren't working. Identify the sources of the steam, which should give you the pressures. Measure the pressure at the suction side of the condensate pump. Re-calculate the capacity of the flash drum and discuss your findings with one of the steam equipment companies.
Don't forget to get samples of the steam condensate. There may be some problems with the steam-boiler feed water or some contaminant that's entering the return system from a heat exchanger leak. The resulting fouling could exacerbate the water hammer.
After collecting field data, run a heat-and-energy balance and hydraulic simulation of the process, including the pump. If you can't get exact flows, try using information from data sheets and, in desperation, vendor cut sheets. However, both the data sheets and the cut sheets can mislead you for a number of reasons but mostly because equipment seldom is operated at the design capacity. Recently, I compared field flow data averaged over five years with the nameplate capacity of a heat exchanger: 105,000 pounds per hour (pph) compared to 180,000 pph for the nameplate — that's 42% lower and only 58% of capacity. Now, imagine assuming the nameplate pressure drop of 4 psig for the shell. By ratio, the actual drop would be only 1.3 psig!
The simulation likely will indicate a need for knock-out drums in strategic locations, especially at the connection points between the main header and the previously unknown additions to the header. Drums will reduce the two-phase flow that accounts for the water hammer; don't completely re-insulate until you've photographed and inspected the pipe and of course, included it in your new isometrics. Adding more drums will mean additional steam condensate return pumps. Don't expect to get a lot of financial or maintenance support for this kind of project. Steam systems are supposed to run "maintenance-free" — utility maintenance can often be described as run-to-failure.
As for what's wrong with the transmitters, my first suggestion would be to check the grounding. The instruments at least should be grounded to a common ground so the analog signals roughly match. Of course, two transmitters never agree completely, so you'll have to see if you can live with this error or consider a new idea: replace them with a differential transmitter. Reducing the two-phase flow also may improve meter behavior.
Dirk Willard, process engineer
Superior Engineering, Hammond, Ind.
We make an anionic surfactant by reacting oleum (SO3 in H2SO4) with alkyl benzene (C12H25C6H5) in a continuous stirred-tank reactor (CSTR); see Figure 1. The product from our spray dryer is 85% pure, with the remainder being sodium sulfate and water. Ever since we started up this process a year ago, we've had problems: 1) a residue in the product layer in our separator; 2) a heavy black residue in the spent acid at the bottom of the separator — the company that regenerates our H2SO4 has threatened not to take our waste if we don't clean it up; 3) our titration analyzers downstream of the reactor and digester sometimes have fouled tubing — a sample pump failed a few weeks ago because a suction strainer plugged; and 4) startup and shutdown are rough — we can't turn down the capacity of the sulfonator cooler effectively and the smaller jacket control valve can't handle the mid-range cooling requirements. We see a lot of what appears to be burnt product during these periods. Recently, our Coriolis meter, which measures alkyl benzene, corroded out. What can we do to address these problems?
Send us your comments, suggestions or solutions for this question by November 15, 2013. We'll include as many of them as possible in the December 2013 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.