The old joke about falling from a tall building is that “it isn’t the fall that kills you, it is the sudden stop at the bottom.” When it comes to cavitation, it is not the formation of the vapor bubbles that kills the pump, it is the subsequent collapse.
There are plenty of articles on net positive suction head (NPSH) and cavitation that talk about the bubble formation and the consequential pump damage in a broad sense, but the details of the damage mechanism are rarely discussed. The focus of this article is looking into why are we so worried about a few bubbles.
Cavitation causes an increase in pump noise and vibration, but more importantly, a drop in performance, efficiency and impeller erosion. Not all of the damage from cavitation is metal loss or metal damage. Sometimes the issue is shortened bearing and mechanical seal life due to the unsteady flows (surging).
Simply defined, “classic cavitation” from the perspective of centrifugal pumps is the formation of bubbles in the pump inlet near the eye of the impeller. The bubbles form because local pressure has dropped below the vapor pressure of the fluid (another way to view this is that the NPSH margin is not sufficient). Less than a fractional second later as the bubbles transit along the low pressure side of the impeller vanes, they enter a region of higher pressure and collapse.
I refer to this as classic cavitation to differentiate it from other causes of cavitation, such as suction or discharge recirculation that manifests on the other side of the impeller vane. Recirculation cavitation is typically due to operating the pump to the left side of the pump operating curve (reduced flows) and away from the best efficiency point (BEP). The approach angle of the incoming flow does not match that of the rotating impeller inlet vane geometry. Consequently, eddy currents and turbulence are generated in between the vanes.
Inside the general area of the eddy current, the velocity increases and the pressure decreases as a result. This action occurs due to the laws of conservation of energy as explained by Bernoulli’s equation and the pressure-velocity relationship. When the local pressure drops below the vapor pressure, the cavitation bubbles are formed. Recirculation cavitation is typically not caused by insufficient NPSH in the classic sense. You could have more than adequate NPSH margin and still experience recirculation cavitation because the pump is being operated away from its BEP.
When this situation occurs, there is a mismatch in the flow angle as compared to the impeller inlet incidence angle. The higher the suction specific speed (NSS) of the pump impeller, the more likely recirculation cavitation is an issue.
For more information on the subject of NSS, see my Pumps & Systems February 2019 article.
Cavitation bubbles that break down in the middle of the impeller passageway collapse symmetrically (equally from all directions), so there is less cause for concern other than potential noise and perhaps some vibration. Similar (but different) to boiling water in an open pan on a stove, the bubble forms at the bottom of the pan, rises to the surface and collapses without issue or harmful effects (technically this is a burst and not a collapse so almost no energy is released).
However, when the vapor bubbles in a pump impeller collapse adjacent to the metal surface of the vane, there is a much higher potential for damage and concern due to metal loss from the substrate. When the bubble collapses near the vane surface, it will collapse asymmetrically. Because of its proximity to the vane surface, the bubble geometry changes and makes the action more lethal. When the bubble collapses, it is not just the surrounding fluid that rushes in to fill that void, it is more importantly that the vapor is changing state from a vapor (back) to a liquid.
I repeat for emphasis that the amount of energy transferred for a change of state is very high. You can calculate the energy using enthalpy equations. Additionally, the collapse of a vapor bubble is exponentially more impactful than if it was an air bubble. With vapor bubbles there is a change of state from liquid to vapor and back, while an air bubble creation or dissipation does not involve a change of state. Further, when the vapor bubble collapses asymmetrically, there is a resulting reentrant microjet burst that on a local, nanoscale level is powerful (the local scale is 1 x 10-9, that is 10 to the negative nine exponent or a billionth).
Local pressure forces involved in the microjet burst can have resultant shockwaves higher than 10,000 pounds per square inch gauge (psig). The bubble collapse phenomena can occur with a high periodicity of 300 times per second and all of this action happens at the speed of sound. The resultant microburst jet almost always directs at the adjacent surface in lieu of the fluid stream. The vane material substrate is subjected to a localized surface fatigue failure. The average lifespan of a vapor bubble from creation to collapse is about 2 to 3 milliseconds. Not everyone agrees if it is the shockwave or the reentrant micro jet burst that creates the damage. Likely, it is the combination.
Hopefully, with this perspective, you begin to understand how cavitation can damage an impeller in short order.
On a scientific level, besides the enthalpy equation mentioned earlier, the energy of the bubble collapse is simply a kinetic energy calculation and is a function of the mass and velocity.
Note that vapor bubbles formed in water at ambient temperature are of a much larger size (mass) than if the water temperature was close to and approaching 200 F. The larger the bubble, the more energy and damage. Therefore, cold water cavitation is much more dangerous than hot water cavitation.
The root cause for vapor bubble evolution is often overlooked. Pumps do not so much generate heat to make the water flash to vapor, but instead it is a result of the drop in pressure near the impeller eye. Remember you can boil water at 33 F if you reduce the pressure low enough.
There is some correlation of cavitation noise (intensity) to impeller damage. I am not presently aware of a conclusive formula or method for accurate determination. I am aware that several people are conducting studies in this subject area. Noise level for cavitation falls in the general range of 10 kilohertz (kHz) to 120 kHz. The general accepted range of hearing for humans is only 20 Hz to 20 kHz. Perhaps I will devote a future article to acoustic detection of cavitation. If you hear cavitation noise, the pump is likely cavitating, but just because you do not hear cavitation noise does not mean it is not cavitating. Some of the most damaging cavitation occurs at noise levels outside the audible range. I also witness many people confusing cavitation noise with turbulent or high velocity flow noises.
Sometimes you just cannot have a cavitation-free system, and you may wish to treat the symptom in lieu of the problem. With all of this energy being dissipated near the surface of the impeller vane, it is important to note that all impeller materials react differently to the exerted force.
For impellers, 300 series stainless is better than cast iron. Higher chrome content steels are better yet, while CD4MCu (duplex alloy) is better than high chrome stainless. There is good information and engineering studies completed in this area. Your empirical results may differ.
Finally, note that even with high NPSH margins, where the NPSH available far exceeds the NPSH required, the pump may still experience some cavitation. It is nearly an impossible task to reduce it to zero.