Jim Elsey on the mysteries of the twilight zone
by Jim Elsey

People of a certain age, and those that enjoy TV reruns, may remember the show called “The Twilight Zone.” In each episode, there was always a surprise plot twist and a subsequent moral or lesson at the end.

In my career, I have witnessed many pump issues that I have assigned to this zone of unexplained reasons for pump failure.

At first I was “mentally filing” them simply as pump mysteries, but later I just started calling it the twilight zone because no one seemed to know what caused the problem. I learned—after some time, persistence and acknowledgment to the twilight of unexplained things—there was always a good lesson at the end.

Before going further, I want to add a disclaimer: I am not blaming anyone for the issues that occurred. There are guilty parties all around and no names will be used. It is the intent of this article that we all learn from the lessons they teach. All of these examples are true; I present the issues first and the answers follow.

What Is this Twilight Zone?

I am talking about that warped period of time between when the pump is shipped from the factory and subsequently started up. I am talking about the time and place between the pump being removed for repair and when it is reinstalled. I am talking about the unexplained changes that occur when pump “A” is pulled from the system and pump “B” is installed in its place.

This is the interim dimension I previously called the magic and mystery place. Lately, this is the dimension I refer to as the twilight zone.

1. The case of the missing impellers. Do not blink twice, or you will miss it.

In a steel mill in the central Midwest, a six-stage boiler feed pump is pulled from service for routine maintenance to reestablish the original operating clearances. Note that the pump is driven by a multistage steam turbine. The pump operated in excellent fashion for more than eight years of continuous service, but it was time for an overhaul.

The pump was rebuilt and returned to the customer by the OEM repair center. Upon initial startup, the pump operated smoothly, but the hydraulic performance was markedly off. The flow and head were both diminished when compared to the manufacturer’s published performance parameters. The upset customer proclaimed that the repair center had left out a few stages (impellers) during the repair. This pump was originally designed as an eight-stage pump, but was built for this customer as a six-stage pump. This is a common practice where blanks are used for impellers when a lower discharge pressure is required. The customer was suggesting the pump was now a four-stage in lieu of a six-stage pump.

The manufacturer’s engineer was called in to troubleshoot the issue. He arrived at the site and went through all the normal performance checks including vibration, valve lineup, rotor settings and the validation of differential pressure and flow. He considered that it was possible for two stages to be left out of the rotor, but not likely, and besides the OEM repair center vehemently insisted the pump had six stages.

Initially, all the indicators appeared to be an issue with low speed. The customer advised several key points during the troubleshooting process, including that they had not worked on the steam turbine during the outage and that the operating speed was the first thing they checked when the issue appeared. The turbine/pump speed had been validated using a calibrated strobe tachometer (stroboscope) at exactly 3,600 revolutions per minute (rpm). Luckily, the young pump engineer had plenty of experience with steam turbines and stroboscopes.

Solution: No missing impellers
In the first case, the pump factory engineer had extensive experience with turbines, steam governors and stroboscopes. He knew how easy it was to get an incorrect reading from the stroboscope if you were not properly trained. The steel mill stroboscope operator had focused on a coupling bolt to “freeze” the rotor at some speed, which he thought was 3,600 rpm because he had no reason to suspect otherwise. The turbine was really operating at 1,800 rpm, and the stroboscope operator was seeing a multiple image (two coupling bolts appearing as one).

For stroboscope work, it is important to always focus on an object that is unique from a shaft circumference perspective, like a key or key seat. If you must use a coupling bolt, then mark one in some way so it differs from other bolts (glow paint works well in these cases).

The customer did not previously mention that the turbine governor valve (W-TG-10) had been rebuilt while the pump was out for overhaul. The factory engineer who discovered the root cause was given the honor of ceremoniously adjusting the steam governor setting to bring the unit up to the correct speed of 3,600 rpm where it performed like new.

2. The case of the hot transformer, and a cool solution.

Scene: an electric power generation plant on the eastern seaboard. At a 20-year-old power plant, a generator step-up transformer (GSU) has severely overheated at full load since the day it was installed. Consequently, the unit was derated and operated at 85 percent of full load. The derated transformer thereafter limited the power output of the entire power generating station.

Fast forward 20 years and the thermally handicapped transformer is scheduled for new dielectric oil and all new gaskets. Additionally, all six transformer oil circulation pumps are to be rebuilt. (Sidebar: Power transformers use circulating transformer oil systems to serve both as an insulating dielectric and as a heat transfer cooling medium.)

The pumps were sent to the factory repair shop on an expedited basis for remanufacture and returned to the site. The customer performed a pre-installation check for proper rotation and noticed the pumps were rotating in the wrong direction. The customer demanded warranty compensation from the factory to fix the issue. The factory engineer stated that it was impossible for these pumps to be operating in the wrong direction unless the customer had reversed two of the three phases on their power leads at the site. The customer vowed that no lead changes had occurred. The factory engineer was on the next plane to the site.

Solution: The cool solution
The factory engineer was confident the transformer pump motors were correctly wired because of their in-house quality assurance/quality control (QA/QC) program. On arrival at the job site, and to his surprise, the engineer confirmed that the pumps were actually rotating in the wrong direction. In fact, the pumps had been rotating in the wrong direction for 20 years.

Note: a centrifugal pump operating in the wrong direction still produces flow in the correct direction and the customer’s rudimentary sight glass flow indicators could not differentiate between 600 gallons per minute (gpm) and 1,200 gpm. The correct rotation was established for all pumps, the power plant was started up and the transformer was energized. For the first time in 20 years, the transformer did not overheat at full load.

3. The reoccurring seized rotor—the definition of insanity.

At a steel mill in the Midwest, an eight-stage segmented casing pump is sent to a repair facility for a complete overhaul. The pump is returned and installed. During the setup, the millwrights discover that the pump rotor will not rotate by hand. The vendor is informed that the pump rotor is apparently seized.

The pump is sent back to the repair shop to be disassembled and inspected. The pump shop finds nothing wrong—they reassemble the pump and send it back to the customer. But again, the millwrights are not able to rotate the rotor due to an apparent rotor seizure. The pump is sent back to the shop, where again it is disassembled and inspected to find there is nothing wrong. The pump is returned to the customer a third time, but on this occasion the sales engineer decides to accompany the pump.

Solution: What is going on over at the plant?
The engineer accompanied the rebuilt pump to the mill and witnessed the work crew lift the pump off the truck using a small crane with two lifting straps—each attached to one end of the pump shaft. By improperly lifting the full pump weight on the shaft at both ends, the segmented casing rings were shifted out of alignment, consequently locking the pump rotor. Never lift a pump of this type by the shaft.

4. Seized rotor redux: watch out for dead ends.

Similar to the preceding example, this multistage boiler feed for a boiler feed pump at a factory somewhere in the middle of the country was rebuilt in a nearby repair shop and reinstalled. The pump operated for about 10 minutes and seized up. The pump was returned to the repair shop to be rebuilt and sent back, where the same scenario reoccurred. On the third attempt, a pump engineer and consultant were brought in by the customer for oversight purposes.

Solution: Seized rotor
The consultant examined the system and operation and maintenance procedures. It did not take long to determine the reason for the pump failures.

In any piping system when there is a section of pipe that has no flow (also known as a dead leg or trap), it becomes a place for the rust, scale, dirt and debris to accumulate and deposit. The suction line for the boiler feed pump was blocked when the pump was removed, but the system was still operating using the other parallel pumps. The dead leg was collecting the foreign debris in the system. When the pump was reinstalled and the suction line opened, the pump was flushed with the collected debris. The debris lodged in the wear ring fits, and the pump seized.

5. Going the wrong way on a one-way street; you can’t have it both ways.

Scene: the “oil patch” in the upper western U.S. where we have a single-stage group 3 ANSI pump on oil service in a middle stream process application. The local pump distributor contacted the factory with an issue: the pump would run for a day or two and then seize and fail. They pulled the pump, rebuilt and returned it to service. The pump would again operate well for a few days, and the issue would repeat.

The OEM engineer asked for photos of the damaged parts and more information on the operation. Based on those, it appeared as a clear and simple case of running the pump in the wrong direction. But, the distributor had their most experienced serviceman on the job who had personally checked for proper rotation three separate times. Time for some fresh eyes to visit the job site.

Solution: One-way pump
The mechanic was 100 percent correct. The pump was operating in the correct direction. The essential information left out of conversations and drawings was that the system had multiple pumps in parallel with the subject pump. The discharge check valve for this pump would randomly and intermittently stick in the open position. When this pump was not operating under power, it was forced in the reverse direction by flow or pressure from the other pumps. The impeller and casing were now acting as a hydraulic turbine. Because most ANSI pumps use an impeller that is screwed on, the impeller would back off and seize to the casing during reverse operation.


I have a long list of pump issues that fit into this twilight zone, but these are some of my favorites. These show that inexperienced personnel are frequently assigned to startup. Assumptions are made, no instructions are ever read, nor is any OEM consulted with prestart questions. Almost no pump installation is “plug and play.” There are alignments to conduct, driver rotational checks to make, mechanical seals to set, oil to add and piping to vent. Educate yourself and avoid that dimension in time and space known as the twilight zone.

Read more Common Pumping Mistakes articles here.