Steve, a plant operator in Ohio, called me recently because he had a question that needed a rapid answer. "Lev," he began, "we're going crazy here. I know we're getting the right flow from our system, but the gauges are driving us nuts. What's wrong?"
"Wait a second, Steve," I replied, "and slow down here. Tell me more about this -- and let's see the details."
I knew Steve from our past work together. He is an experienced mechanic, but has little tolerance for anything that wastes time. Knowing that his boss was probably on his case, I couldn't blame him for wanting the answer before I even had a chance to understand what was going on at his plant.
"Well, here's the thing," Steve continued. "We have a big contract to supply a bunch of oil heating systems. We've just completed the first unit and we're ready to ship it." Steve explained that the equipment in question is a mobile skid with a centrifugal pump that transfers oil from a tank to smaller containers. He designed the system for 30 gpm and tested it in his shop. "But when we connected it to the supply tank suction line," he said, "we got funny results."
I reached for my paper pad and started to sketch the system. "Steve," I asked, "tell me about your system. What type of pump do you have?"
"We've got a Goulds 1x1.5-6 with a 5.12-in. impeller to pump oil through the heat exchanger before filling the containers," he replied. "The whole thing is mobile, and we have a short suction pipe, which the customer connects to the large tank."
Steve then explained that he sized the pipes large, especially the suction line, to ensure negligible friction loss. He intends to test each unit at his shop using a flowmeter. There's no flow meter on the skid because they're expensive. The customer will need to estimate flow rate on the basis of pressure gauge readings. "Each unit comes with a certified pump curve," he pointed out, "so estimating the flow from pressure shouldn't be a problem."
That sounded simple enough, and I started a simple sketch while listening.
"I know we're getting 30 gpm," Steve continued, "because I calibrated our flowmeter myself and recalibrated the darn thing to make sure. The discharge gauge reads 44 psig, which converts to about 118 ft. because the oil has a specific gravity of 0.86. If I use this number on the pump's head-capacity curve, I get about 30 gpm."
"So, what's the problem?" I asked. "You get the right flow and your gauges measure the pressure corresponding to the same pump curve flow. Sounds good to me. I assume the suction gauge reads about zero and the 44 psi is the differential across the pump."
"That's the goofy thing, Lev," he answered. "The problem is that the suction gauge shows about 7 psig. If it read zero, I'd be happy. But this way, the differential is 44-7, or 37 psig. According to the pump curve, that implies nearly 80 gpm. If not for the fact we need to start shipping these units, I'd be applying for a patent for a perpetual motion machine."
I jotted the numbers down. Steve was right. The head corresponding to 37 psid was 37 x 2.31/0.86, or 99 ft.. I pulled out my Goulds catalog and verified his reasoning. Was his meter wrong, I wondered? "Steve," I probed, "tell me more about the suction side."
"Well, it's simple," he answered. "The tank level is about 4 ft. above the pump inlet. I know about air entrapment and designed the tank myself to ensure no air is drawn in. I even looked inside. The surface is nice and smooth with no vertices."
"You're right, Steve. It does sound funny," I responded. "Let me think about this for a few minutes and run some numbers. I'll call you back."
"OK, Lev. But please call as soon as possible," he pleaded. "I need to get this resolved."
I hung up and pondered the numbers. There was no smoking gun. But wait, didn't he say the tank level was 4 ft? If so, that's only 4 x 0.86/2.31, or 1.5 psi. With only small losses on the suction side, the gauge should read even less, perhaps 1 psig. Certainly not 7 psig. Where did 7 psig come from?
I had to call him right back. "Steve, I bet it's the suction pipe,” I blurted out. “You must have either a short elbow -- or two -- in front of your pump."
"Well, yeah," he replied. "We don't have much room to maneuver, so we had to get creative and turn the piping around. Why?"
"I think you've got suction swirl," I informed him. "It's produced when sharp radius elbows are too close to the pump inlet. You get the right flow -- but the suction gauge registers higher pressure. In addition to purely static pressure, it picks up a dynamic, rotating component. The result is an artificially high reading at the suction gauge."
"Also," I continued, "your pump's best efficiency point is about 90 gpm and you're operating at one-third of that. When a centrifugal pump operates at low flow, it produces backflow, sort of a swirling flow that actually leaves the impeller tips and moves backward in a rotating pattern. That also adds to your suction gauge reading."
"Darn it. That simple?" he asked. "But what do I do to fix it?"
"Two things," I told him. "First, straighten the suction piping by eliminating elbows, or at least insert flow straighteners. Another thing is ..."
"I know, I know," he cut me off. He was on a roll. "We'll put another straightener between the gauge and pump inlet."
"Yes, that will work," I told him. "Installing a honeycomb or a simple cross-member between the suction gauge and a pump breaks the backflow and eliminates the swirling component in front of the gauge. Then it will read what it needs to read -- about 1 psig, or so."
"Gee, thanks." Steve sounded grateful. "Let me get working on that and I'll let you know how it worked out. I hope you're right."
"Please do," I answered. "By the way, one more thing. Because you understand what's going on, why change anything? Simply accept the fact the suction gauge should be reading approximately zero. Then, develop a calibration curve, plotting flow versus discharge pressure. Do it at your shop, where you can measure the flow. Then, the customer can simply read the single discharge pressure and read the flow off the calibration curve. It's not a ‘clean way,’ but it should work."
"Great idea, Lev," he said. "I guess you're right. Let me talk to the guys here and see what they say. I still want to test your anti-swirl idea, just to make sure."
After he hung up, I knew his mind was on testing the theory. Steve likes to prove things. He called back a few days later and sounded much better. The swirl theory was right. Steve was happy. And, judging from the fact that he already shipped the first units, so was his boss.
Sometimes "seeing is believing" is not always a sure thing.
Dr. Nelik is president of Pumping Machinery Co., Bridgewater, N.J. Contact him at [email protected] and (908) 203-1180.