The Adventures of Duke Waters: Chapter 2

It was about 30 years ago when I first encountered the Duke. Once again, I was in over my head on a jobsite; in front of an important client and trying to understand what was causing replacement check valves to slam.

The 16-inch pumps were located in a remote county of Georgia and used to pump finished water from a below-ground wet well to a large water distribution system. The installation of the spiffy new, tilted disc check valves was complete; but lo and behold, the owner reported horrendous slams every time he turned off a pump. I was the lucky guy selected to examine this installation and make the valves perform to the owner’s satisfaction. I yanked the associated drawings and hopped on a plane to Atlanta, readying myself for this formidable challenge.

Upon arrival, I squinted through the hot Georgia sun and saw that the check valve installation appeared to be in accordance with the plans. It’s always good to see tilted disc check valve applications because, with a port area equal to 140% of the pipe, they conserve considerable pumping energy. These valves were a far cry from the aged, energy-consuming control valves they replaced.

I observed six vertical turbine pumps piped in parallel over a wet well containing finished drinking water. Each pump had its own check valve, which was a good thing since only one pump running would mean it would just pump water back though the adjacent pump in a circle. The check valve “FLOW” arrows were pointing in the right direction: away from the pump, which is also a good thing because, believe it or not, check valves are sometimes installed by mechanical contractors backwards, rotated or put in a flow-down pipe, all of which are bad ideas.

I was taking note of the pump nameplate data of 6,900 gallons per minute (gpm) at a total dynamic head of 160 feet of water column when the plant superintendent snuck up behind me and said: “It’s about time you got here. What kind of crappy check valves did you specify for this project?”

“These are tilted disc check valves,” I said with confidence. “We will get ’em running smoothly, sir.”

I did some quick conversions in my head to calculate that the 16-inch check valve would see about 11 feet per second (ft/sec) of fluid velocity at a pressure of about 70 psig. “Piece of cake,” I thought to myself, because these check valves are designed to operate in a range of 4 to 24 ft/sec fluid velocity.

I installed a high-speed pressure transducer on a nearby gauge tap located on the discharge header so I could monitor any significant pressure spikes in the system. Besides, hooking up a transducer and a laptop to a pipeline always gives a client the warm and fuzzy feeling that you know what the heck you’re doing. To be honest, though, I thought I did know what I was doing, given my experience in the lab testing numerous check valves.

Finally, I told the superintendent: “Start the pump and let ’er rip”

The superintendent barked the appropriate command into his radio to someone in the control room, and the monster pump roared to life. I saw on the pressure gauge that the pump generated about 50 psig of pressure in a matter of seconds and then built the pressure up to the required 70 psig over the next 30 seconds. So far, so good. I figured they must have variable frequency drives on the pumps, which start the motors at maybe 50% and then ramp them up to full speed.

The superintendent subsequently gave the command to shut the pump down and after a couple seconds, we heard the power cut from the pump motor. A fraction of a second later, a water hammer shook the pipes and generated a slam that could be heard for several city blocks.

Being the brave engineer I was, I didn’t jump or even flinch. I just slowly looked down at the laptop screen and reported, “It looks like we saw a pressure spike of about 140 psig on that trial.” I asked if he could ramp the pumps down to 50% before tripping them off. He agreed, and we repeated the test with the same resulting slam.

Little did this superintendent know that I was in deep doo-doo here because the pump shutdown and the resultant check valve closure should not have produced a slam or water hammer. I covered my rear by saying, “Let me do some quick calculations, and we will run another trial after lunch.”

That’s when I happened to meet Duke Waters for the first time. I noticed a distinguished-looking, darkly tanned worker in jeans and a flannel shirt leaning over next to one of the nearby pump motors, reading the nameplate. At the time, he just looked like a pump technician recording data on the pumps.

I was intrigued, though, at the twinkle in his eye and the grin he gave me as he noticed that I just about crapped my pants when the check valve slam occurred. Ignoring his glances, I pretended to analyze the data on the laptop in a futile attempt to explain what I observed. Given the system parameters and layout, there should not have been a slam. I saw that we went from the full velocity of 11 ft/sec to slam in about one third of a second, which translates into a deceleration of about 33 ft/sec2. That was extremely high for running a single pump into a long distribution system.

I slapped my laptop closed and wandered over to the mysterious figure watching me. I noticed he had the telltale golfer’s untanned left hand and a scar over his right eye. “This guy just doesn’t quite fit this picture,” I thought. I politely asked: “Are you working on the pumps?”

“Nope.”

I followed up with: “Then, do you work for the District?”

“Nope.”

I began to turn away and cut my losses with this guy when he finally said, “You know, check valve slam is caused by the sudden stoppage of reverse water velocity through the check valve. For every 1 ft/sec of flow velocity stopped suddenly, a water hammer of about 50 psig can result.”

“OK, thanks,” I said, “But I am somewhat familiar with Joukowsky’s transient pressure equation, which says that the magnitude of the pressure surge is directly proportional to the change in fluid velocity.” I continued: “It appears there was too much reverse flow through the check valve when it closed but pumping into a distribution system like this one through a fast-closing tilted disc check valve should not cause a slam.”

The stranger countered with: “Well, then I think we need to think of reasons why this particular system might have a high-velocity deceleration.” I was surprised that he knew what flow deceleration was. His comment made me think for a minute before replying: “We had only one pump running into a very long, relatively flat-water distribution system with high friction, so there should not be high fluid deceleration. They also ramped down the pump to 50% before tripping it and the check valve still slammed.”

He asked, “What about the pressure?”

“The static pressure is about 50 psig,” I replied, “But with the length of the piping system creating most of the pressure due to friction, that pressure should not generate a slam-causing deceleration.”

“Then you need to look further at the pumping system,” he said.

I looked around and saw a typical pump station; nothing out of the ordinary.

“I’m just not seeing it,” I commented.

“Did you notice that large, cigar-shaped tank over there behind you on the other side of the parking lot about 100 yards down the pipeline?” the stranger said.

I glanced over my shoulder. “You mean that 10-foot diameter tank keeping that sweet red vintage Jaguar XKE convertible out of the sun?”

“Yep, that’s the tank I am talking about. My leather seats don’t take kindly to the Georgia sun.”

This took me back, and I asked, “Who the heck are you?”

He quietly introduced himself as, “Duke Waters, an independent consultant who occasionally works for the District to troubleshoot various problems.”

I introduced myself and asked: “What brought you to the District today?”

“Pleasure to meet you Johnny,” said Duke “The superintendent is my friend, and he asked me to come over and troubleshoot the pumps because he observed vibrations he didn’t like. It turned out to be a simple concentricity problem with the pump shaft coupling. He and I go way back and I help him out when I can.”

I let the “Johnny” slide out of respect for Duke’s age, then confessed, “I am the engineer responsible for troubleshooting these new check valves, but have no clue what the big tank on the other side of the parking lot shading your car has to do with my problem.”

Duke began to explain. “That is a surge tank, and it is directly connected to the water line and charged with water and a layer of air pressure so that whenever there is a pump trip, it sends water back to the pipeline rapidly to prevent a vacuum pocket or column separation from occurring. In doing so, the tank produces an extremely high, reverse velocity that is getting past your check valve before it closes. By the sound of that last check valve slam, I’d say it’s about a 150 psig water hammer.”

This guy was right on the money. “I actually recorded a pressure surge of 142 psig over the static pressure,” I said.

Duke wasn’t done explaining. “Because of that surge tank, these valves are seeing extremely fast flow reversal and are not able to close fast enough to prevent the reverse flow from rising to an unacceptable level.” He continued to explain that the control valves I replaced were electrically wired to the pumps so that the pump ran while the valves slowly closed and pinched off the flow; then the pump was tripped off. Hence, no slam.

I grimaced and said, “There were no surge tanks on the set of plans I was given.”

“Surge tanks are often specified and simply installed by a contractor downstream of the pump station,” he replied.

I paused, then said, “I guess I need to find a way to make these check valves close faster, before the reverse flow gets too high.”

Duke explained further: “In theory that is true, but if you talk to the manufacturer you will see that the valve is already closing very fast. It is simply a matter of installing either a top-mounted or bottom-mounted oil dashpot on the check valve. Oil dashpots control the closure of the disc to slowly dissipate the reverse flow over several seconds.”

“Well, given the magnitude of the slam,” I suggested, “we probably should install both a bottom- and top-mounted oil dashpot.”

Duke set me straight once again. “No, you need to pick one or the other. The top-mounted dashpot independently controls the full opening and closing strokes in about 5-30 seconds. But this would allow the surge tank water to blow back through the pumps on electrical power failure. It would be better to use a bottom-mounted oil dashpot, which allows the valve to close fast, but slows down the last 10% of travel in a couple of seconds, thereby slowly cutting off the reverse flow and preventing the slam. I call it the hurry-up-and-wait solution,” he said.

After lunch, I took Duke’s advice and told the superintendent about the need for a bottom-mounted oil dashpot, which is easily field-installed.

Duke gave me his card and said, “Well, Johnny B., it sure was a pleasure to work with you on this case, and feel free to give me a shout whenever you are in the need of assistance.” I knew then that I would be making use of this new resource, and that I could live with the “Johnny B” nickname. It beat the usual “engineering kid.”

The superintendent called me later in the month to explain he’d removed the bottom inspection cover on the valve and installed the bottom oil dashpot to the valve, a simple matter. “The valve rapidly closed 90% against the bottom dashpot snubber rod, and then closed the last 10% over three seconds as the reverse flow was dissipated. The valve closed quietly and the problem was solved.”

That was good to hear; I closed that field report and chalked one up in the “Win” column. I rifled through my briefcase and found Duke’s card, intent on thanking him.

On the other end of that call, I heard, “Duke is out. Leave a message.”

I left a long-winded message about the success of the field repairs and figured the dude was probably on a golf course somewhere working on his tan.

 

Up next in the series: Duke helps Johnny B. look at problems with a new water pump generating only about half of what it’s designed to produce.