Second only to electric motors, centrifugal pumps are usually the simplest machines in a typical process plant. Pumps have fewer parts and are not nearly as complex as modern aircraft jet engines, for example. However, even a senior reliability professional may not understand why process pumps often experience more repeat failures and are often less reliable than a jet engine with more than 8,000 parts.
This four-part series will explain repeat failures. It uses a real-world field example involving boiler feed water pumps. It highlights many deviations from best practices that often become an accepted routine at many facilities. Deviations or oversights can range from seemingly insignificant to stunningly elusive. These can combine and often cause costly failures. Therefore, remedial steps should extend beyond pump restoration, and upgrading should be considered. Upgrade steps must be targeted to future failure risk reductions—a reliability engineer’s job.
A Defined Operating Range
Just because pumps can operate at low flow does not mean that doing so is beneficial. Compare this practice to running an automobile capable of going 20 miles per hour (mph) in sixth gear and 60 mph in first gear. If operated this way for too long, a high price will be paid. The same is true with pump hydraulics and with internal fluid recirculation issues.
Fluid machinery has limits, which can be defined analytically or by factory testing. A simple sketch, originally hand drawn by Irving Taylor in 1977 (see Figure 1), will assist in visualizing pumps’ limits.1 A practical engineer, Taylor deserves credit because his approximations explain in one image what others have often attempted to convey with complex mathematical formulas.
NPSH Is Not the Whole Story
Although more precise calculations are available, Taylor’s trend curves of probable net positive suction head required (NPSHr) for minimum recirculation and zero cavitation-erosion (see Figure 1) are sufficiently accurate for most fluids. The curves merit the attention of reliability professionals who want pumps to run within safe margins.2 However, the NPSHr actually needed to prevent damage to impellers and other pump components can be many times the amount published in the manufacturer’s literature. A factor of 2.5 or more is often recommended by field-wise engineers who reverse costly selection errors in services such as carbamate.2, 3
For reliable, long-term operation, the net positive suction head available (NPSHa) at the impeller inlet vane tips must exceed that of the true NPSHr at all times. Keeping the NPSHa-over-true-NPSHr ratio at more than 1 is very important for fluids that readily release a large volume of dissolved gas. Maintaining the ratio at more than 1 is less important for fluids that have little or no entrained gas.
A pump manufacturer’s NPSHr plots are similar to Figure 1. They reflect the anticipated performance of a particular pump impeller and pertain only to that impeller’s geometry. The pump manufacturer reports or anticipates the NPSHr based on observing a 3 percent drop in discharge head. Two or three intersecting head (H)/flow (Q) points are plotted from tests, and the lines connecting these points are drawn in later. Issuing the resulting performance curve with the head achieved at zero flow is customary, although operating a pump at zero flow would destroy it. Head can be converted to pressure. Dividing the head by the specific gravity of the fluid being pumped allows an end user to calculate the differential pressure from the published and plotted head-versus-flow capability of an impeller.