Let's summarize the symptoms. First, the flow from the bullets is difficult below 45% level. The pressure drops in the bullets from about 140 psig to 105 psig at the lowest level. A surge is affecting the pump seal life. It's very difficult to start the pump once it has stopped. The pressure spikes in the bullets when they are filled on a hot day. The pumps have little trouble delivering propane to an underground line, except when the bullets are below 45% level. And, the suction line for each bullet is restricted by a 4-in. valve. Clearly, the problem is pumping from a bullet where neither the level nor pressure is maintained, causing the propane to flash.
A quick calculation with a simulator shows that at 140 psig propane remains a liquid up to about 82°F. At 105 psig the maximum temperature is only 63°F. Add to that the radiation effects and the true temperature of the pipe surface won't be ambient. The heat transfer is more complicated. Flowing propane auto-refrigerates. If it's moving, flow through several hundred feet of 8-in. pipe shows little increase in temperature. So, if it's liquid in the bullets, it will flow. If it's a gas, it won't. This describes why there's a surge in flow and also why weather affects the pump. Perhaps operating with a higher pressure drop would increase the auto-refrigeration?
Another problem is what happens to the pump when vapor forms. When it goes through a pump a 5–10% mixture of gas with liquid will start at 70°F and be 110°F when it exits the pump. Gas typically has one-tenth the heat capacity of liquid on a per-pound basis. Although it's now a liquid, when it goes into the control valve it will become gas again because of the higher temperature.
Leaving 45% in the tank bothers me. Noncondensables may be building up in the bullets. This will only get worse.
The biggest problem with the pumps is the pressure control of the bullets. Some of this could be relieved by moving the vent line from the pump suction to the discharge. Since the mid-1930s, manufacturers have indicated that as the correct placement of the vent. The main problem with this action is that the only good vent is the manual vents on top of the bullets; bottom delivery would be more suitable, at the other end of the bullets, away from the suction. A new pump may be required.
Another option is to increase flow to the bullets to maintain pressure during pump-out. If quality control is a problem the refinery could go to continuous monitoring or a set of isolation tanks to prevent contaminated product from being shipped.
Replacing the valves, either the 4-in. suction or the flow control valve, may reduce the surge but it won't make it easier to load a tanker truck. The flow in the 4-in. suction valve is fine; using the 25N2 rule [where N is nominal pipe size] for maximum design flow, 25×42 = 400 gpm; this is greater than the 150 gpm flowing if we assume equal splits between the bullets — but splits may not be equal. The velocity in a schedule-80 4-in. pipe is 4.2 ft/s assuming equal splits; the velocity in 8-in. is 4.2 ft/s; API RP 14E recommends 3ft/s for suction lines and 9 ft/s for discharge piping (for light hydrocarbons). This means that the suction velocity may be high if API is to be believed.
Avoid changing the rack control valve to a smaller one. Studies have shown that a larger body allows gas to expand more and reduces valve damage. Going with a linear valve will likely require a bigger opening compared to an equal-% valve thus reducing the velocity through the valve and lessening the damage from gas bubbles. There also may be two-phase problems in the flow measurement. Perhaps a vortex shedding or an annubar-type flow meter would be a better choice than the turbine. For more on this sort of problem, check: www.corken.com/media/training/cp226.pdf and www.eng-tips.com/viewthread.cfm?qid=259967.
Dirk Willard, consultant
We're suffering wild yeast problems in our ethanol fermenters. Our clean-in-place system had run flawlessly for 15 years with little attention. The cycle is a rinse with steam/water at 160°F, followed by a weak caustic wash at 210°F, a hot water rinse, a rinse with an organic iodine solution with phosphoric acid, and finally another rinse with hot water. A few months back during the winter we increased the number of times we reused the caustic from three to six before sending the stream to waste treatment. Now, it's spring and we're seeing wild yeast in our cream tanks. We went back to using the caustic three times and then to just once — but that only modestly affected the wild yeast we're seeing in the product. Is the change in the number of times the caustic is used to blame? Or is it something else? How can we solve this problem?
Send us your comments, suggestions or solutions for this question by April 12, 2013. We'll include as many of them as possible in the May 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.