Overview

Tom Angle, Director of Technology at Weir Specialty Pumps, was faced with an unexpected and ongoing problem with a series of high-pressure pump failures in a customer’s system. Tom and his group of engineers used PIPE-FLO to model the system and communicate to the customer that the problem was the exact opposite of what was expected. The problem occurred in the piping system rather than in the pump.

"With PIPE-FLO’s simulated system view we were able to perform a series of ‘what if’ types of analysis to recognize a number of different solutions", said Tom Angle, Weir. "The use of PIPE-FLO allowed us to identify the problem and communicate the solution to the customer quickly and to everyone’s satisfaction." 

The Challenge

A series of catastrophic cavitation and mechanical failures occurred on a high-pressure water pump (HPP) shortly after installation in a mineral processing plant in Europe. The pump never lasted more than two months before failing. The failures always consisted of severe cavitation damage and actual component mechanical breakage. The customer was convinced that the problem persisted in the pump.

The system consisted of a pump, connected by approximately 150 feet of 4-inch pipe to a manifold that lead to 14 high-pressure bladders used to squeeze slurry through a filter screen. The process was more batch than continuous, since each filter went through a series of high and low pressure stages as well as a vacuum stage. The low-pressure stage uses a low-pressure pump which is connected to the bladders by means of a totally separate piping system. The speed of the HPP is typically 5500 rpm and kept within a few hundred rpm in either direction by a VFD.

Initially the low pressure pump was used to increase the NPSHA of the HPP, and to provide the low pressure process fluid through a completely different line. This was changed after the first two failures of the HPP when it was discovered that the pressure on the HPP suction was going to zero during periods of low-pressure fluid demand. When this change was made a separate suction tank was provided for the HPP approximately 27 feet above the pump centerline level. This provided approximately 61 feet of NPSHA. The pump itself would not run out of NSPH until a flow of roughly 425 gpm was reached. However, the ongoing failures suggested that, in fact, this might be the problem. View an image of the initial system below.

PIPE-FLO's Solution (Tom Angle)

Because of the ongoing failures, a visit to the customer’s plant was necessary. Once on site we were able to thoroughly go over the customer’s system and easily produce a detailed model using PIPE-FLO®. As a new user somewhat unaware of PIPE-FLO’s reliability, I was surprised to find out that the PIPE-FLO model indicated the opposite of what everyone had expected. The flow was less than 100 gpm no matter how much the friction data was "tweaked".

Because of PIPE-FLO’s contradiction with the customer’s suggestions and our original inclinations of the problem, I began to distrust PIPE-FLO’s model and calculated results.

Throughout the day we continued the study of this unusual problem and found that indeed the PIPE-FLO model was correct, "we were now PIPE-FLO believers", and there was too little flow in the system.

PIPE-FLO allowed us to recognize another very important unexpected problem. At certain observed conditions flow would actually reverse in the main 4 inch line as the pump was slowed down by the VFD. Because of the check valve just downstream of the pump and upstream from a minimum flow bypass orifice, this would lead to the pump operating against a dead head part of the time. This in turn would lead to significant heating of the fluid in the pump.

At this point we told the customer that the flows were not anywhere near what they had thought. We requested that they obtain a flow meter and verify our position. They followed this recommendation and discovered that the flows in the discharge line ranged from a high of 27 gpm to a low of negative 10 gpm! In addition, the flows were very unstable, varying widely within this range. We also noticed that the pump discharge would get very hot and then cool rapidly over 60 second cycle.

At this time, part of the cause of the failures became obvious. Cavitation was occurring not as a result of the pump running out on its curve, but as a result of the pump operating against a dead head and heating up the liquid inside it, which then vaporized. The mechanical component failure was due in some degree to this problem, however the majority of the problem occurred with the pump operating at roughly 5% of its best efficiency point. An additional problem was at the low flow rate there was basically no friction in the main line.

This lead to the system curve being very flat at the end of each filter cycle. Because the pump curve was also flat, both curves were basically co-linear which resulted in unstable flow conditions.

Finally, it came time to present our findings to the customer’s engineers in the UK. PIPE-FLO allowed us to easily demonstrate and communicate that the problem was the exact opposite of what was expected. In a situation like this, there are always two challenges. The first is the technical problems of determining what exactly is happening. The second, and often more difficult challenge, is convincing the customer that the fault is with the system and not the pump.

With PIPE-FLO’s simulated system view we were able to perform a series of ‘what if’ types of analysis to recognize a number of different solutions. "The use of PIPE-FLO allowed both of these very difficult problems to be resolved quickly and to everyone’s satisfaction." Tom Angle

* The Resolution of a Series of High Pressure Pump Failures gives reference to Tom Angle, Director of Technology for Weir Specialty Pumps in Salt Lake City, UT.